Best of BJR

AN INTERNATIONAL JOURNAL OF RADIOLOGY, RADIATION ONCOLOGY AND ALL RELATED SCIENCES

Editorial Board

Publications

Available from bir.org.uk

HONORARY EDITORS Dr Jane Phillips-Hughes

Medical

Professor David Bradley

Scientific

Godfrey Hounsfield: Intuitive Genius of CT Written by Stephen Bates, Liz Beckmann, Adrian Thomas and Richard Waltham, and including many recollections from his family, friends and colleagues, this book provides a new insight into the life of one of science’s greatest pioneers.

DEPUTY EDITORS Dr Daniel Birchall Dr Nigel Hoggard Professor Alan Jackson Dr Simon Jackson Professor Paul Sidhu Professor Stuart Taylor Dr David Wilson Professor Alastair Munro Professor Kevin Prise Professor William Vennart Professor Habib Zaidi

Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Radiotherapy & Oncology Radiobiology Physics & Technology Physics & Technology

Radiation Shielding for Diagnostic Radiology This book is an update to the 2000 comprehensive report on the topic produced by a joint BIR/ IPEM Working Party. This report is designed to be a compendium of information for radiation physicists involved in specification of shielding requirements for X-ray facilities.

A scientifically based, realistic and straightforward approach to shielding specification. Sets out a framework for shielding calculations based on patient-dose area product and entrance surface dose information. Includes a review of the dose criteria employed in light of revised ionising radiation legislation and re-evaluates assumptions made in earlier methodologies.

ASSOCIATE EDITORS Abass Alavi USA Zahir Amin UK Richard Amos USA Diane Bergin Ireland David Brettle UK David Butteriss UK Fay Cafferty UK Mark Callaway UK Isabel Castellano UK Gary Cook UK Peter Corr Canada Justin Cross UK John Curtis UK Indra Das USA Marco Durante Germany Tarek El-Diasty Egypt Jonathan Eatough UK Phil Evans UK Sydney Evans USA Mark Farrugia UK Lesley Forsyth UK Nitin Ghonge India Rosalind Given-Wilson UK Stephen Harden UK Christopher Harvey UK John Holemans UK Peter Hoskins UK David Howlett UK Walter Huda USA

Charles Hutchinson Andrzej Kacperek Peter Keston Mayil Krishnam Dow-Mu Koh Iain Lyburn Ranald MacKay Alistair Mackenzie Lesley Malone David Manning Roger Martin Stewart Martin Soren Mattsson Michael Michell Andrew Morgan Shinichiro Mori Hilary Moss Ludvig Muren Kwan Hoong Ng Andrew Nisbet Christopher Njeh Alain Noel Philip O’Connor Allan Odurny Joerg Oestmann Oystein Olsen George Panayiotakis Tufail Patankar Niall Power

UK UK UK USA UK UK UK UK Ireland UK Australia UK Sweden UK UK Japan UK Denmark Malaysia UK USA France UK UK Germany UK Greece UK Ireland

Shonit Punwani Aloysius Raj Madan Rehani John Reynolds Rainer Rienmuller Peter Riley Michael Robbins Kevin Robson Andrea Rockall Carl Roobottom Nicola Russell Anju Sahdev Phillip Shorvon Andrew Slater Malcolm Sperrin David Sutton Katy Szczepura Alberto Tagliafico Nobuyuki Takahashi Paul Taylor Eliseo Vano Anthony Watkinson Elspeth Whitby John Winder Bernd Wittkop Conrad Wittram Wai-Lup Wong Hui-Xiong Xu

UK Malaysia Austria UK Austria UK USA UK UK UK Netherlands UK UK UK UK UK UK Italy Japan UK Spain UK UK UK UK USA UK China

BJR is essential reading for radiologists, medical physicists, radiotherapists, radiographers and radiobiologists. Articles included in BJR cover a wide range of subjects, including diagnostic radiology, radiotherapy, oncology, nuclear medicine, ultrasound, radiation physics, radiation protection and radiobiology.

Authors benefit from rapid decisions, constructive peer-review and global dissemination. For more information or to submit your next article to BJR, go to http://bjr.birjournals.org

Radiobiological Modelling in Radiation Oncology Radiobiological Modelling in Radiation Oncology, edited by Professors Roger Dale and Bleddyn Jones, will help to prepare you for the increasing use of modelling in practical situations, including treatment gap corrections, normal tissue tolerance predictions, optimisation of therapy determined by predictive assays, multi-modality schedule design, the simulation of clinical trials, testing contemporaneous medico-legal problems and teaching general principals of radiotherapy.

MCQ Pocketbook No. 2

MCQ Pocketbook No. 2 The best MCQ study guide from the British Institute of Radiology. Over 300 multiple choice questions. Based on recent journal articles and books, with the majority originating from the following key radiological journals: AJR, BJR, Clin Rad, and Radiology.

Multiple Choice Questions & Answers with References Adrian Mizzi Gavin Low Keh Oon Ong Ananth Shenoy Edited by Simon Ostlere

Published

by

Contents

Best of BJR

Page

Introduction from the Honorary Editors

iii

Review article Watching tumours gasp and die with MRI: the promise of hyperpolarised MR spectroscopic imaging K Brindle

13

C

1

b

Diffusion-weighted magnetic resonance imaging for monitoring prostate cancer progression in patients managed by active surveillance V A Morgan, S F Riches, K Thomas, N Vanas, C Parker, S Giles and N M Desouza

13

b

Evaluation of dose reduction and image quality in chest CT using adaptive statistical iterative reconstruction with the same group of patients L-P Qi, Y Li, L Tang, Y-L Li, X-T Li, Y Cui, Y-S Sun and X-P Zhang

20

b

Bone age assessment by dual-energy X-ray absorptiometry in children: an alternative for X-ray?

D H M Heppe, H R Taal, G D S Ernst, E L T Van Den Akker, M M H Lequin, A C S Hokken-Koelega, J J M Geelhoed and V W V Jaddoe

26

b

Regional grey and white matter volumetric changes in myalgic encephalomyelitis (chronic fatigue syndrome): a voxel-based morphometry 3 T MRI study B K Puri, P M Jakeman, M Agour, K D R Gunatilake, K A C Fernando, A I Gurusinghe, I H Treasaden, A D Waldman and P Gishen

33

b

Prognostic significance of SUV on PET/CT in patients with localised oesophagogastric junction cancer receiving neoadjuvant chemotherapy/ chemoradiation: a systematic review and meta-analysis W Zhu, L Xing, J Yue, X Sun, X Sun, H Zhao and J Yu

37

b

Full papers

2011–2012 The British Institute of Radiology. This publication is copyright under the Berne Convention and the Universal Copyright Convention.

E All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without the prior permission of the copyright owner. Permission is not, however, required to copy abstracts of papers or articles on condition that a full reference to the source is shown. Multiple copying of the publication without permission is illegal—address enquiries regarding photocopying to the Copyright Licensing Agency, Saffron House, 6-10 Kirby Street, London EC1N 8TS, UK (Tel. +44 (0)20 7400 3100; E-mail: licence@cla.co.uk).

All opinions expressed in the British Journal of Radiology are those of the respective authors and not the publisher. The publisher has taken the utmost care to ensure that the information and data contained in this journal are as accurate as possible at the time of going to press. Nevertheless, the publisher cannot accept any responsibility for errors, omissions or misrepresentations howsoever caused. All liability for loss, disappointment, or damage caused by reliance on the information contained in this journal or the negligence of the publisher is hereby excluded.

Subscriptions

Back Issues

For personal subscriptions to the journal, join the BIR at www.bir.org.uk or e-mail membership@bir.org.uk

Advertising

Subscription fulfilment is carried out by Affinity Subscriptions. For institutional subscriptions contact them at: Tel: +44 (0)20 7383 6270; Fax: +44 (0)20 7383 6402; E-mail: subscriptions@bmjgroup.com

All BJR subscribers also have free access to Imaging Online (http://imaging.birjournals.org), the BIR’s review journal.

Founded 1897 Incorporated by Royal Charter Patron: Her Majesty The Queen The British Institute of Radiology has as its aim to bring together all the professions in radiology and allied medical and scientific disciplines to share knowledge, and educate the public, thereby improving the prevention and detection of disease and the management and treatment of patients. Particulars of membership and other information can be obtained from the Chief Executive, BIR, 131–151 Great Titchfield Street, London W1W 5BB, UK, and from the BIR’s website: www.bir.org.uk Tel: +44 (0)20 3586 8325

Back issues can be ordered from Publications Department, The British Institute of Radiology, 131–151 Great Titchfield Street, London W1W 5BB. E-mail: publications@bir.org.uk Price is £70 per issue including postage.

Publications Department, The British Institute of Radiology For more information visit http://bjr.birjournals.org/site/misc/ Advertising.xhtml or e-mail publications@bir.org.uk

President Prof. A Jones Vice Presidents Dr D Wilson (Education) Mr A Rogers (External affairs) Dr J Phillips-Hughes (Publications) Honorary Treasurer Mr J Gunaratnam Honorary Secretaries Dr G Tudor Mrs E Hunt Chief Executive Ms J Fowler

Cover image credited as: Cervical spine MRI. © Deanaustinphotography (Dean Hoch) | Dreamstime.com

INTRODUCTION FROM THE HONORARY EDITORS Jane Phillip-Hughes, Honorary Editor (Medical) and David Bradley, Honorary Editor (Scientific) BJR is an international multidisciplinary journal which covers clinical and technical aspects of medical imaging, radiotherapy, oncology, medical physics and radiobiology.

As the oldest scientific journal in the field of radiology and related sciences, BJR has published a number of landmark articles, including the first description of computed tomography ‘Computerized transverse axial tomography’ by Godfrey Hounsfield in 1973 [1]. BJR has a well-established place in history, and digital archives are available from 1928 through to the current day.

We now need to ensure BJR continues to thrive and maintains its role as essential reading for all involved in radiological fields, well into the future.

With this in mind, the journal is being relaunched following a major review of publication processes, to bring BJR right up to date as a competitive, accessible, high-quality journal for the 21st century.

An international journal requires an international editorial board, and we are building on the expertise of our already well-respected board by recruiting new members who are experts in their fields from around the world.

Authors rightly expect their papers to be dealt with promptly; with the move to continuous publication as of January 2013, BJR authors can be assured that their accepted papers will be published online with their final citation within 8 weeks. In addition, measures have been put in place to ensure rapid, high-quality reviewer feedback for submitted articles.

Furthermore, our open access option allows all researchers to publish in the journal and ensures the work is freely available to the scientific community.

This collection contains a number of papers published in BJR over the last couple of years, selected to demonstrate the broad range of topics covered in the journal. Although the disciplines encompassed by BJR are diverse, there is common ground between the medical specialties and the medical sciences represented, enabling readers to keep up to date with advances in related specialties as well as their own.

With the expansion of our international board, stream-lined review and publication processes, open access facility and continuous publication model, we believe that this historically important journal can now evolve and move on into the future. We hope you will contribute to the continuing success of BJR by reading the journal articles, reviewing papers or submitting your work.

Reference 1. GN Hounsfield. Computerized transverse axial scanning (tomography): Part I. Description of system. Br J Radiol 1973;46:1016–22.

Submit to BJR HIGH QUALITY RAPID PUBLICATION INTERNATIONAL EDITORIAL BOARD GLOBAL READERSHIP OPEN ACCESS OPTION

birjournals.org

1 The British Journal of Radiology, 85 (2012), 697–708

REVIEW ARTICLE

Watching tumours gasp and die with MRI: the promise of hyperpolarised 13C MR spectroscopic imaging K BRINDLE,

MA, DPhil

Department of Biochemistry, University of Cambridge, Cambridge, UK

ABSTRACT. A better understanding of tumour biology has led to the development of ‘‘targeted therapies’’, in which a drug is designed to disrupt a specific biochemical pathway important for tumour cell survival or proliferation. The introduction of these drugs into the clinic has shown that patients can vary widely in their responses.

Molecular imaging is likely to play an increasingly important role in predicting and detecting these responses and thus in guiding treatment in individual patients: socalled ‘‘personalised medicine’’. The aim of this review is to discuss how hyperpolarised 13 C MR spectroscopic imaging might be used for treatment response monitoring. This technique, which increases the sensitivity of detection of injected 13C-labelled molecules by .10 000-fold, has allowed a new approach to metabolic imaging. The basic principles of the technique and its potential advantages over other imaging methods for detecting early evidence of treatment response will be discussed. Given that the technique is poised to translate to the clinic, I will also speculate on its likely applications.

The challenge in oncology will be to match the patient to the drug, selecting those patients who respond to a drug and switching the treatment of those who do not.

This has benefits from early phase clinical trials; for example, AvastinH (Genentech Inc., San Francisco, CA; bevacizumab) may have performed better in a carefully stratified breast cancer patient population [1] than in routine application in the clinic. In patients with nonsmall-cell lung cancer, sequencing of the epidermal growth factor receptor (EGFR) has identified activating mutations within the kinase domain that are associated with altered signalling and increased sensitivity to EGFR inhibitors [2]. Genome sequencing of tumour biopsies is likely, in the longer term, to play an important role in selection of the most appropriate drug or combination of drugs for individual patients [3]. However, there are some important limitations to this approach. Tumours can be genomically unstable and therefore intrinsically heterogeneous, which could bias the result of any analysis depending on when and where in the tumour the biopsy was taken [4]. Moreover, it may not always be possible to biopsy the tumour and its metastases routinely during the course of treatment, when new mutations may arise [3]. Therefore, while mutation analysis may be an important first step in patient stratification and therapy selection, imaging is likely to play an increasingly important role when treatment is under way, monitoring early regional responses of the primary tumour and its metastases and detecting Address correspondence to: Professor Kevin Brindle, Department of Biochemistry, University of Cambridge, Tennis Court Road, Cambridge CB2 1GA, UK. E-mail: kmb@mole.bio.cam.ac.uk

The British Journal of Radiology, June 2012

Received 3 October 2011 Revised 3 January 2012 Accepted 5 January 2012 DOI: 10.1259/bjr/81120511 ’ 2012 The British Institute of Radiology

possible relapse at later time points. Inevitably, cost– benefit must be considered. If a therapy costs $88 000 per annum [1], an imaging test that costs only $1000–2000 and which shows whether the treatment is working within weeks, days or even hours will clearly be cost-effective.

Currently, treatment response is still largely assessed by using imaging measurements to monitor reductions in tumour size, with the latest guidelines for Response Evaluation Criteria in Solid Tumours (RECIST, v. 1.1) defining partial therapy response as ‘‘at least a 30% decrease in the sum of diameters of target lesions, taking as reference the baseline sum diameters’’ [5]. However, this can be a slow method for detecting response, with reductions in tumour size not becoming evident for weeks or even months; in some cases, reductions may not occur at all, for example following treatment with anti-angiogenic drugs [6]. In these situations, ineffective treatments may be given, losing precious time and exposing the patient to harmful side-effects. Imaging tumour biology, however, may give a much earlier indication of treatment response than measurements of tumour size alone. This has been well demonstrated by [18F]fludeoxyglucose ([18F]FDG) positron emission tomography (PET) measurements of response to imatinib treatment in patients with gastrointestinal stromal tumours. PET measurements of the tumour uptake of this 18F-labelled glucose analogue can show a marked reduction in glucose uptake following drug treatment, indicating that the patient is responding to treatment, despite the fact that the tumour can continue to increase in size (Figure 1) [7].

In this review, I will describe how 13C MR spectroscopic imaging of tumour metabolism with hyperpolarised 697

2 K Brindle

Figure 1. A patient with a primary

(a)

(b)

(c)

(d)

13

C-labelled cell metabolites can be used to detect early evidence of treatment response. The intention is not to provide an encyclopaedic review of the substrates that have been hyperpolarised and how they have been used, since there are many recent reviews that have covered this area [8–13], but rather to critically evaluate these hyperpolarised methods for detecting treatment response in comparison with existing radionuclide and MRI methods, with a particular focus on the work of my laboratory.

Metabolic imaging with hyperpolarised 13 C-labelled cell metabolites Imaging tissue water protons, which are present in tissues at concentrations of ,80 M, can be used to produce relatively high-resolution three-dimensional images of tissue anatomy (in pre-clinical studies at high magnetic fields, isotropic image resolutions of ,10 mm are possible [14]). We have also known since the 1970s that it is possible to detect MR signals from tissue metabolites [15]. The problem is that these are present at ,10 000 times lower concentration than tissue water protons and therefore it is not possible to image them at clinical magnetic field strengths, except at relatively low spatial (1 cm3) and temporal (5–10 min) resolution [16].

Moreover, single spectra or spectroscopic images of tissue metabolites lack dynamic information about metabolic fluxes. The lack of sensitivity has inhibited application of MR spectroscopy in the clinic, so that while in some centres spectroscopy is used routinely, particularly 1H spectroscopy [16–18], it is not used both widely and routinely. Hyperpolarisation of 13C-labelled cell substrates increases their sensitivity to detection in the MR experiment by more than 10 000-fold, making it possible to image not only the location of a hyperpolarised 13C-labelled cell substrate in the body but also, 698

gastrointestinal stromal tumour in the colon. The pre-treatment CT scan (a) shows a peritoneal mass (arrows), corresponding to a lesion with markedly increased fludeoxyglucose (FDG) uptake on the positron emission tomography (PET) scan (b). The CT scan (c) obtained 2 months after treatment showed that the mass had become larger; however, there was no appreciable FDG uptake seen on the FDG PET scan (d), which corresponded to clinical improvement. Reproduced with permission from Choi et al [7].

more importantly, the kinetics of its conversion into other cell metabolites, with spatial resolutions of 2–5 mm and temporal resolutions in the subsecond range [19].

The technique promises unprecedented insights into tissue metabolism in vivo, which could have important clinical applications.

Physical principles The MR experiment is insensitive because it depends on a weak interaction between a nuclear spin and a magnetic field. For a spin one-half nucleus, such as 1H or 13 C, the populations of the two allowed energy levels are given by the Boltzmann distribution: . {DE Nupper Nlower ~e kT

ð1Þ

where DE is the energy difference between the spin states, k is the Boltzmann constant and T is the temperature in Kelvin. The conventional approach to increasing sensitivity has been to increase the magnetic field strength [i.e. increase the DE term in Equation (1)].

However, lowering the temperature can also increase polarisation. The problem is that to substantially increase the nuclear spin polarisation the temperature must be taken to almost absolute zero, which is difficult to do.

The solution to this problem, which was first suggested theoretically by Albert Overhauser in 1953 [20], is to transfer polarisation from an electron spin to a nuclear spin. The nuclear spins are mixed with electron spins (on a stable radical), rapidly frozen to form a glass and placed in a magnetic field. For some substrates a glassing agent, such as glycerol, must be used. At ,1 K, which can readily be achieved by boiling off helium under vacuum (at ,100 Pa), the electron spins become completely polarised. The electron spin polarisation can then The British Journal of Radiology, June 2012

3 Review article: Watching tumours gasp and die with MRI

be transferred to the nuclear spins by irradiating the electron spin resonance, which is in the microwave range (GHz). The key development, which has made possible the metabolic imaging experiments described here, was the realisation by Klaes Golman, Jan Henrik ArdenkjaerLarsen and their colleagues that the polarised nuclear spins can be warmed rapidly to room temperature with substantial retention of the polarisation that was present in the frozen solid state [21]. This is achieved by rapidly dissolving the frozen sample in pressurised superheated water at ,180 uC. The gain in sensitivity for the solution state experiment is dramatic (Figure 2). The major problem with the technique is that the polarisation is relatively short lived, typically 10–30 s for hyperpolarised 13C-labelled compounds in vivo, which means that injection and migration of the labelled compound to the target tissue and subsequent imaging must be accomplished within a few minutes. This also means that for any hyperpolarised 13C-labelled cell substrate to be useful the cell must take it up and metabolise it very

rapidly. The polarisation lifetime can be maximised by minimising 13C–1H intramolecular dipolar relaxation; this has usually been accomplished by placing the 13C label in a carboxyl or carbonyl group, although replacement of protons with deuterons has also met with some success and could widen the range of usable substrates [22, 23]. In principle, any nucleus can be polarised, although attention has been focused largely on 13Clabelled compounds because of their relatively long T1s (long polarisation lifetime) and the ready commercial availability of 13C-labelled cell metabolites that could be used to probe tissue metabolism. The inevitable dilution that accompanies the dissolution process requires that the 13C-labelled cell substrate is polarised at high concentration, which imposes a further limitation in that these substrates must be very soluble in water.

Nevertheless, despite these limitations, there are now numerous examples of hyperpolarised 13C-labelled cell substrates whose metabolism has been detected in vivo, including [1-13C]pyruvate, [1-13C]lactate, [1,4-13C2]fumarate,

Figure 2. (a) The dynamic nuclear polarisation polariser and parts. 1, polariser magnet (operating at 3.35 T); 2, vacuum pump; 3, variable temperature insert (VTI); 4, microwave source; 5, pressure transducer; 6, sample port; 7, microwave cavity; 8, sample holder; 9, sample cup; 10, dissolution wand. A frozen sample to be polarised is placed in the sample cup (9) and the sample holder (8) is lowered into the VTI (3). Liquid helium from the magnet cryostat (1) is bled onto the sample using a needle valve.

The sample temperature is lowered to ,1 K by applying a vacuum and is irradiated via the microwave cavity (7) using a microwave source (4) operating at 94 GHz. When the sample is fully polarised, which can be determined by monitoring the solid-state signal (a 13C tuned coil is built into the sample holder), the sample undergoes the dissolution process. After releasing the vacuum the dissolution wand (10) is inserted into the sample holder in the VTI (3), where it engages with the sample cup (9). The sample is then lifted out of the liquid helium and discharged from the sample cup using superheated water at ,1000 kPa (see flow arrows in 10). (b) (i) 13C spectrum of hyperpolarised urea (natural abundance 13C). The concentration of urea was 59.6 mM and the polarisation was 20%. (ii) Thermal equilibrium spectrum of the same sample at 9.4 T and room temperature. This spectrum was acquired under Ernst-angle conditions (pulse angle of 13.5u and repetition time of 1 s based on a T1 of 60 s) with 1 H decoupling. The spectrum is the sum of 232 128 transients collected over 65 h. Reproduced with permission from Ardenkjaer-Larsen et al [21].

The British Journal of Radiology, June 2012

699

4 K Brindle

Figure 3. Hyperpolarised [1-13C]pyruvate exchanges the 13C label with lactate in tumour cells in vitro and in tumours in vivo. (a)

Addition of non-hyperpolarised [3-13C]pyruvate to a tumour cell suspension containing added lactate, in which the 13C label was detected indirectly in the proton spectrum, demonstrated that there was exchange of label between lactate and pyruvate. The 1 H spectrum allows the concentrations of both the 13C-labelled and unlabelled species to be observed and shows that there is a decrease in the concentration of 13C-labelled pyruvate and a corresponding increase in the unlabelled form (12C). (b) Addition of lactate to a tumour cell suspension together with hyperpolarised [1-13C]pyruvate increases the rate of label flux between pyruvate and lactate. This is not what you would expect if there was net flux (see c), where addition of lactate inhibits flux between pyruvate and lactate, but is what you would expect to see if there is exchange of isotope between pyruvate and lactate. The filled symbols in (c) represent experiments with 20 mM pyruvate and the open circles with 2 mM pyruvate. These experiments also demonstrate that there is pyruvate inhibition of the enzyme in spectrophotometric measurements of net flux but not in the MR spectroscopic (MRS) measurements of hyperpolarised 13C label exchange. The lack of pyruvate inhibition in the latter experiments can be explained by the higher enzyme concentration (see Witney et al [29]). (d) Exchange of

700

The British Journal of Radiology, June 2012

5 Review article: Watching tumours gasp and die with MRI hyperpolarised 13C label between endogenous lactate and the injected [1-13C]pyruvate can also be demonstrated using magnetisation transfer measurements in vivo. Inversion of the lactate polarisation produces an increase in the rate of decrease of the pyruvate signal, demonstrating unequivocally that there is label exchange, with hyperpolarised label being transferred back to pyruvate from lactate. AU, arbitrary unit. Reproduced with permission from Day et al [24], Witney et al [29] and Kettunen et al [32].

13

C-bicarbonate, [2-13C]fructose, [1-13C]ketoisocaproate, [5-13C]glutamine and [1-13C]glutamate [13].

Detecting treatment response with hyperpolarised [1-13C]pyruvate Since [1-13C]pyruvate has been the most widely used hyperpolarised substrate to date, including for tumour response monitoring [24, 25], and was the first to be used in a clinical trial of the technique [12], it is perhaps worth considering what information it provides about tumour metabolism in vivo. Hyperpolarised [1-13C]pyruvate exchanges label with endogenous lactate and alanine, in the reactions catalysed by lactate dehydrogenase (LDH) and alanine aminotransferase, respectively.

13 CO2 can also be produced in the irreversible oxidative decarboxylation reaction catalysed by mitochondrial pyruvate dehydrogenase in those tissues with high levels of mitochondrial metabolism, such as the heart [26, 27]. Only very low or undetectable levels of 13CO2 and H13CO¯3, with which 13CO2 is in rapid exchange, have been observed in tumours. The experimental evidence that pyruvate exchanges hyperpolarised 13C label with endogenous unlabelled lactate is summarised in Figure 3. The reaction catalysed by LDH is near to equilibrium in the cell and the equilibrium constant is such that when pyruvate enters the cell the reaction will rapidly re-establish chemical equilibrium, with only a very small net conversion of pyruvate into lactate [13, 24]. What is then observed in the 13C MR measurements is predominantly the slower isotopic equilibration of hyperpolarised 13C label between the injected pyruvate and the endogenous lactate pool. Thus, hyperpolarised lactate will generally be observed in those tissues that have a large endogenous lactate pool, such as tumours.

This explains why the reverse experiment with hyperpolarised [1-13C]lactate has not been so successful [28] since endogenous pyruvate is present at much lower concentrations than lactate.

In a tumour, the observed rate of hyperpolarised 13C label exchange will depend on the rate of pyruvate delivery to the tumour, the rate of pyruvate transport into the cell and the activity of LDH. The relative importance of transport rate and LDH activity on the kinetics of the observed exchange has recently been considered, both experimentally and theoretically [29].

This showed that control of the exchange is shared between the transporter and LDH and that this varies according to the lactate and pyruvate concentrations. The observed exchange kinetics for LDH were well described by an ordered ternary complex mechanism, in which the coenzymes nicotinamide adenine dinucleotide (NAD+) and NADH (the reduced form of NAD+) bind first [Equation (2)], and by using rate constants that had been determined previously for the rabbit muscle enzyme using steady-state kinetic studies. This analysis showed that the apparent Km of LDH for pyruvate is 13 mM, The British Journal of Radiology, June 2012

where Km is the Michaelis constant, defined as the substrate concentration at which the enzyme shows half the maximal velocity. The analysis also showed that there is little pyruvate inhibition of the enzyme at the enzyme concentrations found in the cell and that the exchange rate is linearly dependent on the endogenous lactate concentration. The observed Km for pyruvate in cells is higher, depending on the extent to which the transporter limits the exchange [29, 30].

In many studies, with both tumour cells in vitro and tumours in vivo, the rate constant describing exchange of hyperpolarised 13C label between pyruvate and lactate (kP) has been determined by fitting the pyruvate and lactate peak intensities to the modified Bloch equations for two site exchange [24] [Equations (3)–(5)].

EzNADz uE:NADz zLactateuE:NADz :Lactateu ð2Þ E:NADH:PyruvateuE:NADHzPyruvateuEzNADH kL

P

L

ð3Þ

kP

dLz= ~{rL (Lz{L?)zkP Pz{kL Lz dt

ð4Þ

dPz= ~{rP (Pz{P?)zkL Lz{kP Pz dt

ð5Þ

Lz and Pz are the z magnetisations of the 13C nucleus in the lactate and pyruvate carboxyl carbons, rL and rP are the spin lattice relaxation rates and L‘ and P‘ are the equilibrium magnetisations (i.e. at t5‘). In some studies an arterial input function has also been included [31].

This simple analysis provides a robust estimate of kP that is relatively insensitive to the values of rL or rP, or to assumptions that may be made about their values in order to facilitate data fitting; for example, that rL5rP [32]. The data can often be fitted by setting kL to zero, with little effect on the estimated value for kP. However, this does not mean that there is no exchange (see above), only that the value for kL is poorly determined by an experiment with [1-13C]pyruvate. In many studies only a value for the first order rate constant, kP, has been reported rather than the biochemically relevant flux (in mM s21) since the latter requires an estimate of the pyruvate concentration in the image voxel. In the future considerable effort will be required to derive robust estimates of concentration and thus flux, particularly in a clinical setting.

Clearly any drug that affects the concentration of LDH or its substrates, or affects the levels of the monocarboxylate transporters (MCTs), which transport pyruvate and lactate across the cell plasma membrane, or affects pyruvate delivery to the tumour [33] will have an effect on the kinetics of lactate labelling by hyperpolarised [1-13C]pyruvate. The first study on response monitoring 701

6 K Brindle

with hyperpolarised [1-13C]pyruvate was in a murine lymphoma model, which showed that there was decreased lactate labelling within 24 h of treatment with the chemotherapeutic drug etoposide [24]. In cells this was shown to be due to a deoxyribonucleic acid (DNA) damage response, which led to activation of polyadenosine diphosphate ribose polymerase (PARP) and consequent depletion of the NAD+ pool; NAD+ is a substrate for PARP. Addition of a PARP inhibitor preserved the NAD(H) pool and delayed the loss of label exchange between pyruvate and lactate. In tumours, the loss of exchange post-etoposide treatment was shown to be due to a number of factors, including loss of NAD(H) and decreases in tumour lactate and LDH concentrations. A small molecule MCT inhibitor has been shown to inhibit the exchange [30], as have drugs that modulate LDH concentration through inhibition of the phosphatidylinositol 3-kinases (PI3K)-protein kinase B (PKB) (Akt) pathway [25, 29]. The hyperpolarised [1-13C]pyruvate experiment thus offers a novel way of monitoring PI3K–Akt pathway inhibition noninvasively in vivo, which has become an important drug target [34]. Since the central pathways of metabolism are so highly interconnected, in a so-called ‘‘scale-free’’ network, where perturbation of any part of the network is communicated as metabolite changes throughout the network [35], it is likely that the hyperpolarised [1-13C]pyruvate experiment will be exquisitely sensitive to the effects of many different drugs.

Monitoring treatment response with hyperpolarised [1-13C]pyruvate is analogous to monitoring treatment response with FDG PET, and the two methods have been compared directly [36]. In a murine lymphoma model the FDG PET experiment gave an earlier indication of treatment response than the pyruvate experiment, which was shown to be due to an early downregulation of the glucose transporters (GLUTs) 1 and 3 at the plasma membrane. However, although FDG PET gave an earlier indication of response, the amplitude of the changes in FDG uptake and lactate labelling were comparable, suggesting that the two techniques would have similar sensitivities for detecting treatment response in the clinic.

If this were the case then what would be the advantage(s)

of using the polarised pyruvate experiment? The FDG PET experiment does not work well in some tumour types; the prostate shows relatively low levels of FDG uptake, and high levels in the adjacent bladder make quantification of the signal more difficult and in the brain high levels of FDG uptake in normal surrounding brain tissue can mask tumour uptake. Both are tumour types in which the hyperpolarised pyruvate experiment has been shown to work well [37–39]. In a study of response to radiotherapy in a rat brain model of glioma, lactate labelling was shown to decrease following treatment, despite a continued increase in tumour size observed in contrast agent-enhanced MR images (Figure 4) [39]. An important advantage of the FDG PET experiment is that it is a whole body imaging technique and thus can be used to monitor several tumours simultaneously (e.g. the primary tumour and its metastases).

Detecting cell death with hyperpolarised [1,4-13C]fumarate Polarised pyruvate MR spectroscopy and FDG PET measurements can show whether a tumour is responding to treatment, and thus whether a drug has hit its target, but not necessarily whether the treatment has killed any tumour cells. A decrease in lactate labelling or FDG uptake might indicate a loss of cells within the tumour, but equally well could reflect some metabolic change; for example, a decrease in LDH activity through changes in its substrate concentrations or downregulation of the GLUTs at the plasma membrane. Because cell death soon after treatment can be a good prognostic indicator for treatment outcome, considerable effort has gone into the development of imaging methods that detect cell death more specifically. These include agents that bind the phospholipid phosphatidylserine (PS), which is exposed by dying cells [4], 18F-labelled molecules that bind to activated caspase-3/7 [40], which is increased in apoptotic cells, and diffusion-weighted MRI, which detects a loss of tumour cellularity through an increase in the apparent diffusion coefficient of tissue

Figure 4. Imaging response to radiotherapy in a rat brain glioma model. A

13

C chemical shift spectroscopic image obtained following the intravenous injection of hyperpolarised [1- C]pyruvate is shown in (a), superimposed on a 1H image of tissue anatomy. The chemical shift images were obtained from a 6-mm axial slice with an in-plane resolution of 2.5362.53 mm2. The tumour, which is readily observed in the contrast agent-enhanced image (b), was observed to increase in size following radiotherapy (lower image). However, the corresponding 13C images of pyruvate (c) and lactate (d), which were derived from spectroscopic images similar to those shown in (a), clearly show a relative decrease in lactate signal post therapy [compare signal intensities in the upper (before treatment) and lower images (after treatment)]. AU, arbitrary unit. Reproduced with permission from Day et al [39].

13

702

The British Journal of Radiology, June 2012

7 Review article: Watching tumours gasp and die with MRI

water [41]. The fundamental problem with the targeted agents that bind PS or activated caspase is non-specific background binding, which is a problem for any targeted molecular imaging agent. The difficulty is not to get the targeted agent to bind its specific target but rather to minimise the non-specific background binding, which will limit the generation of image contrast. This problem is illustrated by our experience with a PS-targeted agent based on the C2A domain of the protein synaptotagmin.

The protein has been labelled with superparamagnetic iron oxide nanoparticles and with Gd3+ chelates and has been shown to detect dead or dying tumour cells in vivo in T2 and T1 weighted images, respectively [42, 43], as illustrated for the Gd3+ chelate-labelled agent in Figure 5.

Although the accumulation of the agent in the tumour at 24 h after drug treatment is clear in these false colour images, the concentration of the agent in treated tumours was only twice that in the untreated controls. By labelling the agent with both Gd3+ chelates and with a fluorophore we were able to show by fluorescence microscopy measurements on tumour sections obtained following the MRI measurements that the MR agent was indeed binding to dead cells in the tumour. Fluorescence from labelled C2A was co-localised with a terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) stain, which detects DNA damage in dead cells (Figure 5). However, it is also evident from these fluorescence images that there was a low level of nonspecific background binding, in which C2A was apparently bound to viable cells. In these high-resolution microscopy images this background is not a problem and

the increased binding of C2A to the regions of dead, TUNEL-positive, cells is clearly evident. In a lower resolution image however, such as an MR image, the background binding will add up and reduce the contrast-to-background ratio. Analysing regions of interest, which is what is done routinely in analysing data of this sort, exacerbates the problem by effectively further reducing the resolution of the MR image and discarding important information on contrast agent distribution. We recaptured this information, and thus improved the sensitivity of cell death detection, by parameterising the distribution of C2A in a treated tumour using an image analysis technique that had been used previously by the astrophysics community to analyse images of galaxies [44]. The distribution of C2A in a treated tumour was much more heterogeneous than in an untreated tumour, reflecting the heterogeneous distribution of cell death that was evident in the microscopy images (Figure 5). By quantifying this heterogeneity we were able to substantially improve the sensitivity of detection of cell death.

Despite being able to improve C2A detection of dead cells by analysing its distribution, non-specific background binding remains a significant problem. This general problem in molecular imaging has been addressed previously by using ‘‘molecular beacons’’— molecules that generate a signal de novo, in the absence or with low levels of background signal. Well-known examples include firefly luciferase, which generates light in the presence of luciferin, oxygen and adenosine triphosphate [45], and protease-sensitive fluorescence agents, in which cleavage of a peptide bond within the

Figure 5. Non-specific binding limits the contrast that can be achieved using a targeted MR contrast agent. (a) Injection of a Gd3+ chelate-conjugated agent that binds to dead cells showed significant accumulation in a treated tumour at 24 h after injection (TA) (a), whereas a site-directed mutant of the protein, which was inactive, showed less accumulation (TI) (b). Neither the active nor inactive agent showed accumulation in the untreated tumours (UA and UI) (c, d).

Co-labelling of the active agent with a fluorescent probe showed that it was bound to dying cells. (c) A fluorescence image of a histological section obtained from a treated tumour following the MRI experiment.

Regions where the probe is bound, which are yellow in (c), were colocalised with regions of cell death, which are stained brown (terminal deoxynucleotidyl transferase dUTP nick end labelling stain) in (b). While the agent clearly binds to areas of cell death, it also appears to show nonspecific binding to regions of viable cells and it is this that reduces the contrast-to-background ratio in the lower resolution MR images.

The British Journal of Radiology, June 2012

703

8 K Brindle

agent leads to generation of fluorescence [46]. Detection of cell death in the absence of a background signal has been achieved using the hyperpolarised substrate, [1,4-13C]fumarate [47]. In viable cells there is no detectable uptake of hyperpolarised [1,4-13C]fumarate within the relatively short lifetime of the polarisation. However, in dying or necrotic cells, where the plasma membrane permeability barrier has been compromised, there is rapid uptake of labelled fumarate and subsequent hydration, in the reaction catalysed by the enzyme fumarase, to form hyperpolarised [1,4-13C]malate. We have demonstrated in tumour cells in vitro and tumours in vivo a linear correlation between the levels of cell necrosis and the rate of malate formation [33, 47, 48]. Given that fumarase is ubiquitous in biological systems and the only other substrate required is water, this could be a general method for detecting cell death in vivo. It has the advantage over other methods of detecting cell death in that there is no background. If labelled malate is observed then current evidence suggests that there must be dead cells present within the image voxel. The technique has already been demonstrated in different tumour types and with different types of drugs, including antivascular drugs, such as combretastatin A4 phosphate, in which we showed that measurements of hyperpolarised pyruvate and fumarate metabolism could provide a more sustained and sensitive indicator of response to this vascular disrupting agent than dynamic contrast agent-enhanced or diffusionweighted MRI, respectively [33]. There is also evidence that this technique could be used to detect necrosis in other tissues [49], including the kidney [50]. Potential disadvantages of using hyperpolarised [1,4-13C]fumarate to detect cell death include possibly low levels of fumarase in some tumour types and washout of the enzyme from the necrotic cell, which would limit the time period over which necrosis could be detected. The latter may, however, also be an advantage in the context of detecting early tumour responses to treatment in that only recently killed cells should be detected.

Hyperpolarised glutamate

13

C-labelled glutamine and

Another hyperpolarised substrate that could be valuable in detecting tumour treatment response is [5-13C]glutamine. It has long been known that tumours show upregulated glutamine utilisation, the amino acid being used for the generation of metabolic intermediates involved in macromolecular biosynthesis, which supports the increased growth rate of tumour cells, and also the synthesis of glutathione, which is an important cellular anti-oxidant [51]. The rapid growth rate of tumours leads to increased production of reactive oxygen species (ROS) and it is becoming increasingly clear that the metabolism of tumours is adapted to cope with this oxidative load by upregulation of anti-oxidant systems, such as increased glutathione production and pentose phosphate pathway activity [51]. The cellular oncogene MYC, which encodes a transcription factor that promotes cell proliferation, is directly involved in glutamine utilisation, upregulating the expression of the glutamine transporters and of a glutaminase isoform, in the latter case by repressing the expression of a 704

micro-ribonucleic acid [52]. We have shown that the glutaminase activity in human hepatoma cells can be measured by monitoring the conversion of hyperpolarised [5-13C]glutamine into glutamate [53]. The problem with this substrate was that the polarisation lifetime of the label in the C-5 position was relatively short, ,16 s, which precluded its use in vivo. However, a recent study has shown that deuteration of the molecule (to produce L[5-13C-4-2H2]glutamine) results in a significant extension of the polarisation life time [23], raising the possibility that this substrate could be used to monitor tumour glutaminase activity in vivo. Here it might be used as a surrogate for Myc expression and to assess inhibition of tumour cell proliferation, in much the same way as 39-deoxy-39[18F]fluorothymidine uptake has been used as a cell proliferation marker in PET [54, 55].

Mammalian cells express .60 dioxygenases that utilise a-ketoglutarate (a-KG), an intermediate in the tricarboxylic acid cycle, as a co-substrate, including the prolyl hydroxylases that control the stability of hypoxia-inducible factor (HIF1), and the histone demethylases and the ten–eleven translocation family of 5-methylcytosine hydroxylases. Many of these a-KG-dependent dioxygenases have a Km for a-KG near physiological concentrations, suggesting that its concentration may influence HIF1 stability and gene expression [56, 57]. We have shown recently, using hyperpolarised [1-13C]glutamate, that we can detect, for the first time, a-KG in a tumour in vivo [58], raising the possibility that we may be able to interrogate the potential role of a-KG in controlling gene expression and HIF1 stability in a tumour.

Monitoring the tumour microenvironment: pH and redox state The tumour microenvironment can have a profound influence on tumour aggressiveness, treatment response and metastatic spread. We have demonstrated that tumour extracellular pH can be imaged from the ratio of the signal intensities of hyperpolarised H13COÂŻ3 and 13 CO2 following intravenous injection of hyperpolarised H13COÂŻ3 [59] and, more recently, that the tumour redox state can be determined by monitoring the oxidation and reduction of hyperpolarised [1-13C]ascorbate and [1-13C]dehydroascorbate, the reduced and oxidised forms of vitamin C, respectively [60] (Figure 6). The polarised bicarbonate experiment showed, as expected, that the tumour extracellular space has a low pH, as has been observed using other MR probes in pre-clinical animal models [61]. Similarly, the hyperpolarised ascorbate experiment demonstrated, as has become apparent from recent work, that tumours maintain a highly reduced microenvironment [51]. In tumour cell suspensions, in which there is production of ROS, there was observable oxidation of [1-13C]ascorbate. However, when these same cells were implanted to form subcutaneous tumours and the [1-13C]ascorbate was injected intravenously there was no detectable oxidation in the tumours.

When [1-13C]dehydroascorbate was injected into these animals, which is rapidly transported into tumour cells on GLUT1, then very rapid reduction to ascorbate was observed. This would explain why we failed to see ascorbate oxidation since any dehydroascorbate will The British Journal of Radiology, June 2012

9 Review article: Watching tumours gasp and die with MRI

Figure 6. Monitoring the tumour microenvironment. (a) The false colour images in (iii) and (iv) show the distribution of labelled HCO3 and CO2 following intravenous injection of hyperpolarised 13C-labelled HCO¯3 into a tumour-bearing mouse. The image resolution was 16616 voxels, each of which measured 26266 mm. The images were smoothed by multiplying by a cosine function and zero filling to 128 points in both spatial directions, line broadening to 30 Hz and then zero filling to 512 points in the spectral direction before Fourier transformation. HCO¯3 and CO2 peaks were fitted in the frequency domain and only pixels with a frequency separation between the two peaks of 36¡1 ppm were included. These 13C spectroscopic images are superimposed on a 1H image of tissue anatomy (i). The tumour margin is outlined in red in (i) and in white in (ii–iv). The ratio of the CO2 (iv) and HCO¯3 (iii) images can be used to calculate an image of extracellular pH (ii). (b) 13C MR spectra acquired from tumours following intravenous injection of [1-13C]-ascorbic acid (i) and [1-13C]-dehydroascorbic acid (ii). Sequential spectra were collected over a period of 16 s (i) and 32 s (ii). While there was no observable flux of hyperpolarised 13C label from [1-13C]ascorbic acid (179 ppm) to [1-13C]-dehydroascorbic acid (175 ppm) (i), significant label flux was observed from [1-13C]dehydroascorbic acid to [1-13C]-ascorbic acid, indicating rapid reduction of the injected [1-13C]-dehydroascorbic acid in the tumour. Reproduced with permission from Gallagher et al [59] and Bohndiek et al [60].

be rapidly taken up and re-reduced and is entirely consistent with the current view that tumours have upregulated anti-oxidant systems to cope with the increased ROS load.

Prospects for clinical translation The hyperpolarised 13C MR experiment has already taken its first steps to clinical translation with a trial of hyperpolarised [1-13C]pyruvate in prostate cancer, which commenced at the University of California at San Francisco, CA, in November 2010 and where the aim was to use it to detect treatment response (ClinicalTrials.gov identifier: NCT01229618). A 250 mM pyruvate solution was injected at up to 0.43 ml kg–1, which equates to a blood concentration of ,1.5 mM, and spectroscopic images were acquired using a custom-designed 13C transmit volume coil and a 1H/13C endorectal receive coil in conjunction with a 3 T whole body scanner. Such a coil arrangement is also likely to be used for superficial tumours in other abdominal regions such as the breast and lymph nodes. Despite the enormous gain in sensitivity obtained through hyperpolarisation, its transient nature, which precludes signal averaging, means that detectable signal is still limited and places a premium on getting the receiver coil as close as possible to the target tissue. Although such hardware is currently not widely available in hospital radiology departments, the coils are relatively inexpensive and the experiments could reasonably be implemented on any modern scanner with spectroscopy capability. This first trial employed an adapted pre-clinical polariser, The British Journal of Radiology, June 2012

which had to be used in a clean room and which would not be appropriate for routine clinical use.

However, a hyperpolariser intended for clinical use has been described recently [62].

The relatively short lifetime of the polarisation and the dynamic nature of the metabolic labelling that results from injection of a hyperpolarised 13C-labelled molecule has necessitated the development of very fast MR spectroscopic imaging techniques that make efficient use of the short-lived polarisation (reviewed recently in [13]). Much of the information in these experiments is in the kinetics of metabolite labelling and therefore ideally one should collect multiple images over the lifetime of the polarisation (2–3 min), in which fitting of individual pixel intensities to a kinetic model can be used to produce metabolic rate maps. The majority of the preclinical studies have used pulse sequences based on gradient refocusing, such as echoplanar (spectroscopic) imaging [63, 64], spiral chemical shift imaging [65, 66] and rosette chemical shift imaging [67], which have produced nominal image resolutions of 5x5x10 mm3 in under 1 s [19]. The likely spatial resolution in the clinic is open to debate but will probably be between 2 and 5 mm.

In some instances, however, spatial resolution may not be that important. For example, when detecting cell death in a tumour with hyperpolarised [1,4-13C]fumarate post treatment, acquisition of signal from the entire tumour volume may allow detection of diffuse cellular necrosis. A disadvantage of metabolic imaging with hyperpolarised 13 C, when compared with radionuclide imaging, is the relatively limited field of view that is imposed by the short lifetime of the polarisation. Thus, it is unlikely that hyperpolarised 13C MR spectroscopic imaging will be able 705

10 K Brindle

to monitor simultaneously the metabolism of a primary tumour and its distant metastases.

Conclusion In this review, I have taken a biochemical view of the field, reflecting my own interests and expertise.

However, I believe that in the future radiology will move increasingly from a discipline that images tissue anatomy to one that images tissue function as well.

Hyperpolarised 13C MR spectroscopic imaging is a very promising tool for imaging tissue biochemistry, particularly in tumours, where it can be used to ask quite sophisticated questions about tumour biology and how this changes in response to treatment. If radiologists are to fully exploit the potential of this technique in the clinic then they will need to know more about the underlying biochemistry.

Metabolic imaging with hyperpolarised 13C-labelled cell substrates has, with the first clinical trial of pyruvate in prostate cancer, taken its first steps on the road to clinical translation. Given that fumarate is used in some drug preparations and bicarbonate, glutamine and ascorbate have already been infused into patients at the concentrations needed for hyperpolarised MR studies, there is a reasonable expectation that these substrates will also translate to the clinic in the future. However, if this technique is to become widely used in radiology then we will need to find compelling applications: things which it can do much better than existing imaging technologies, which provide robust, readily interpretable and clinically meaningful results at a reasonable cost.

Based on a growing body of pre-clinical data, it is perhaps reasonable to speculate now on what some of these might be. In the case of [1-13C]pyruvate, which has been the most widely used hyperpolarised substrate to date, there are already some strong indications as to where it might be useful in oncology, for example for staging tumours [37] and monitoring early evidence of treatment response [24]. The important question with pyruvate is what advantages it might have over existing imaging techniques. In the case of treatment response monitoring, it would appear to have several advantages over the current gold standard in the clinic, FDG PET. In addition to being usable in those tumours where FDG PET does not work well, such as prostate and brain, it does not involve the use of ionising radiation. Although the radiation dose from FDG is low, current evidence suggests that the risk of excess cancer and heritable effects from PET/CT exposure is ,5% per sievert [68]; this may become an issue for repeat measurements, such as those that would be needed to guide treatment, and when examining children or females of childbearing age.

With hyperpolarised 13C-labelled substrates there are no obvious hazards associated with injecting molecules that are endogenous. Although injected at relatively high concentrations they clear quickly and thus it is possible to contemplate guiding treatment using these substrates in an image–treat–image–treat paradigm. Hyperpolarised bicarbonate might find use as a generic marker for detecting the presence of disease since almost all pathological states are accompanied by a low extracellular tissue pH, including tumours, ischaemia, infection, 706

hypoxia and inflammation. However, the relatively short lifetime of the polarisation in this molecule might necessitate finding ways to inject the substrate closer to the site of disease. Since sites of inflammation are often hypoxic, they can also be detected using hyperpolarised [1-13C]pyruvate [69]. Hypoxia can lead to increased lactate concentration and LDH activity, both of which will lead to proportional increases in the rate of lactate labelling [29].

In tumours a low extracellular pH has been correlated with tumour aggressiveness and metastatic potential and so hyperpolarised bicarbonate might also be used for staging tumours, as has been done with pyruvate. In the latter case increased lactate labelling was observed in the more aggressive tumours, presumably reflecting a higher endogenous lactate concentration [37]. Measuring the rate of dehydroascorbate reduction has never been possible before and, in the absence of information about what controls this rate, it is difficult to predict how it might be used clinically. However, since we have evidence that it is faster in tumours than in surrounding normal tissue it may be possible to use it to distinguish, for example, between benign and malignant lesions. It might also be used to detect the action of drugs targeted at the pathways responsible for maintaining the highly reduced state that seems to be a characteristic of tumours [51]. Hyperpolarised glutamine may have utility in detecting the activity of cytostatic drugs. However, since the aim of any tumour therapy is to kill tumour cells, then possibly the most promising agent for detecting treatment response is hyperpolarised [1,4-13C]fumarate. Uniquely among the methods that my laboratory has used to detect cell death in vivo it seems to provide an unequivocal measure of cell death, which has no background signal.

It seems likely that the important clinical applications for hyperpolarised 13C imaging will only be found through widespread utilisation of the technology at multiple research centres and could take some time, perhaps 10–15 years. However, the exciting developments in the pre-clinical arena, which have driven the rapid translation of the technique into the clinic, give reason to be optimistic that the technique will have an impact on the practice of radiology in the future.

Acknowledgments The author acknowledges Cancer Research UK, the Leukemia and Lymphoma Society, the Wellcome Trust, the National Institute for Health Research Cambridge Biomedical Research Centre and GE Healthcare for support. I would also like to thank members of my laboratory, both past and present, for their efforts, insights and dedication that made much of the work described here possible.

References 1. Twombly R. Avastin’s uncertain future in breast cancer treatment. J Natl Cancer Inst 2001;103:458–60.

2. Sequist LV, Bell DW, Lynch TJ, Haber DA. Molecular predictors of response to epidermal growth factor receptor antagonists in non-small-cell lung cancer. J Clin Oncol 2007;25:587–95.

3. Sawyers CL. The cancer biomarker problem. Nature 2008;452:548–52.

The British Journal of Radiology, June 2012

11 Review article: Watching tumours gasp and die with MRI 4. Brindle K. New approaches for imaging tumour responses to treatment. Nat Rev Cancer 2008;8:94–107.

5. Eisenhauer E, Therasse P, Bogaerts J, Schwartz L, Sargent D, Ford R, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228–47.

6. Miller JC, Pien HH, Sahani D, Sorensen AG, Thrall JH. Imaging angiogenesis: applications and potential for drug development. J Natl Cancer Inst 2005;97:172–87.

7. Choi H, Charnsangavej C, Faria SD, Tamm EP, Benjamin RS, Johnson MM, et al. CT evaluation of the response of gastrointestinal stromal tumors after imatinib mesylate treatment: a quantitative analysis correlated with FDG PET findings. AJR Am J Roentgenol 2004;183:1619–28.

8. Golman K, Petersson JS. Metabolic imaging and other applications of hyperpolarized 13C. Acad Radiol 2006;13: 932–42.

9. Gallagher F, Kettunen M, Brindle K. Biomedical applications of hyperpolarized 13C magnetic resonance imaging.

Prog NMR Spectrosc 2009;55:285–95.

10. Ardenkjaer-Larsen JH, Golman K, Brindle KM. Hyperpolarized 13C magnetic resonance imaging: principles and applications. In: Weissleder R, Gambhir SS, Ross BD, Rehemtulla A, eds. Molecular imaging: principles and practice. Hamilton, Canada: BC Decker Inc.; 2009.

11. Viale A, Aime S. Current concepts on hyperpolarized molecules in MRI. Curr Opin Chem Biol 2010;14:90–6.

12. Kurhanewicz J, Vigneron D, Brindle K, Chekmenev E, Comment A, Cunningham C, et al. Analysis of cancer metabolism by imaging hyperpolarized nuclei: prospects for translation to clinical research. Neoplasia 2011;13:81–97.

13. Brindle KM, Bohndiek SE, Gallagher FA, Kettunen MI. Tumor imaging using hyperpolarized 13C magnetic resonance spectroscopy. Magn Reson Med 2011;66:505–19.

14. Johnson GA, Benveniste H, Black RD, Hedlund LW, Maronpot RR, Smith BR. Histology by magnetic-resonance microscopy. Magn Reson Q 1993;9:1–30.

15. Gadian DG, Radda GK. NMR studies of tissue metabolism.

Annu Rev Biochem 1981;50:69–83.

16. Nelson SJ. Assessment of therapeutic response and treatment planning for brain tumors using metabolic and physiological MRI. NMR Biomed 2011;24:734–49.

17. Kurhanewicz J, Swanson MG, Nelson SJ, Vigneron DB. Combined magnetic resonance imaging and spectroscopic imaging approach to molecular imaging of prostate cancer.

J Magn Reson Imaging 2002;16:451–63.

18. Kwock L, Smith JK, Castillo M, Ewend MG, Collichio F, Morris DE, et al. Clinical role of proton magnetic resonance spectroscopy in oncology: brain, breast, and prostate cancer. Lancet Oncol 2006;7:859–68.

19. Mayer D, Yen YF, Tropp J, Pfefferbaum A, Hurd RE, Spielman DM. Application of subsecond spiral chemical shift imaging to real-time multislice metabolic imaging of the rat in vivo after injection of hyperpolarized 13C1pyruvate. Magn Reson Med 2009;62:557–64.

20. Overhauser AW. Polarization of nuclei in metals. Phys Rev 1953;92:411–15.

21. Ardenkjaer-Larsen JH, Fridlund B, Gram A, Hansson G, Hansson L, Lerche MH, et al. Increase in signal-to-noise ratio of .10,000 times in liquid-state NMR. Proc Natl Acad Sci USA 2003;100:10158–63.

22. Allouche-Arnon H, Gamliel A, Barzilay C, Nalbandian R, Gomori J, Karlsson M, et al. A hyperpolarized choline molecular probe for monitoring acetylcholine synthesis.

Contrast Media Mol Imaging 2011;6:139–47.

23. Qu WC, Zha ZH, Lieberman BP, Mancuso A, Stetz M, Rizzi R, et al. Facile synthesis [5-(13)C-4-(2)H(2)]-L-glutamine for hyperpolarized MRS imaging of cancer cell metabolism.

Acad Radiol 2011;18:932–9.

The British Journal of Radiology, June 2012

24. Day SE, Kettunen MI, Gallagher FA, Hu DE, Lerche M, Wolber J, et al. Detecting tumor response to treatment using hyperpolarized 13C magnetic resonance imaging and spectroscopy. Nat Med 2007;13:1382–7.

25. Ward CS, Venkatesh HS, Chaumeil MM, Brandes AH, Vancriekinge M, Dafni H, et al. Noninvasive detection of target modulation following phosphatidylinositol 3-kinase inhibition using hyperpolarized 13C magnetic resonance spectroscopy. Cancer Res 2010;70:1296–305.

26. Schroeder MA, Cochlin LE, Heather LC, Clarke K, Radda GK, Tyler DJ. In vivo, assessment of pyruvate dehydrogenase flux in the heart using hyperpolarized carbon-13 magnetic resonance. Proc Natl Acad Sci USA 2008;105: 12051–6.

27. Merritt ME, Harrison C, Storey C, Jeffrey FM, Sherry AD, Malloy CR. Hyperpolarized 13C allows a direct measure of flux through a single enzyme-catalyzed step by NMR. Proc Natl Acad Sci USA 2007;104:19773–7.

28. Chen AP, Kurhanewicz J, Bok R, Xua D, Joun D, Zhang V, et al. Feasibility of using hyperpolarized [1-13C]lactate as a substrate for in vivo metabolic 13C MRSI studies. Magn Reson Imaging 2008;26:721–6.

29. Witney TH, Kettunen MI, Brindle KM. Kinetic modeling of hyperpolarized 13C label exchange between pyruvate and lactate in tumor cells. J Biol Chem 2011;286:24572–80.

30. Harris T, Eliyahu G, Frydman L, Degani H. Kinetics of hyperpolarized 13C1-pyruvate transport and metabolism in living human breast cancer cells. Proc Natl Acad Sci USA 2009;106:18131–6.

31. Zierhut ML, Yen YF, Chen AP, Bok R, Albers MJ, Zhang V, et al. Kinetic modeling of hyperpolarized C-13(1)-pyruvate metabolism in normal rats and TRAMP mice. J Magn Reson 2009;202:85–92.

32. Kettunen MI, Hu D-E, Witney TH, McLaughlin R, Gallagher FA, Bohndiek SE, et al. Magnetization transfer measurements of exchange between hyperpolarized [1-13C]pyruvate and [1-13C]lactate in a murine lymphoma.

Magn Reson Med 2010;63:872–80.

33. Bohndiek SE, Kettunen MI, Hu D-E, Witney TH, Kennedy BWC, Gallagher FA, et al. Detecting tumor response to a vascular disrupting agent using 13C magnetic resonance spectroscopy and hyperpolarized [1-13C]pyruvate and [1,4-13C2]fumarate. Mol Cancer Ther 2010;9:3278–88.

34. Vivanco I, Sawyers CL. The phosphatidylinositol 3-kinaseAKT pathway in human cancer. Nat Rev Cancer 2002;2: 489–501.

35. Brindle K. Metabolomics: Pandora’s box or Aladdin’s cave?

Biochemist 2003;25:15–17.

36. Witney T, Kettunen M, Day S, Hu D, Neves A, Gallagher F, et al. A comparison between radiolabeled fluorodeoxyglucose uptake and hyperpolarized C-13-labeled pyruvate utilization as methods for detecting tumor response to treatment. Neoplasia 2009;11:574–82.

37. Albers MJ, Bok R, Chen AP, Cunningham CH, Zierhut ML, Zhang VY, et al. Hyperpolarized 13C lactate, pyruvate, and alanine: noninvasive biomarkers for prostate cancer detection and grading. Cancer Res 2008;68:8607–15.

38. Park I, Larson PEZ, Zierhut ML, Hu S, Bok R, Ozawa T, et al. Hyperpolarized 13C magnetic resonance metabolic imaging: application to brain tumors. Neurooncology 2010;12:133–44.

39. Day SE, Kettunen MI, Cherukuri MK, Mitchell JB, Lizak MJ, Morris HD, et al. Detecting response of rat C6 glioma tumors to radiotherapy using hyperpolarized [1-13C]pyruvate and 13 C magnetic resonance spectroscopic imaging. Magn Reson Med 2011;65:557–63. ˚ rstad E, 40. Nguyen QD, Smith G, Glaser M, Perumal M, A Aboagye EO. Positron emission tomography imaging of drug-induced tumor apoptosis with a caspase-3/7 specific

707

12 K Brindle

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

54.

55.

[18F]-labeled isatin sulfonamide. Proc Natl Acad Sci USA 2009;106:16375–80.

Moffat BA, Chenevert TL, Lawrence TS, Meyer CR, Johnson TD, Dong Q, et al. Functional diffusion map: a noninvasive MRI biomarker for early stratification of clinical brain tumor response. Proc Natl Acad Sci USA 2005;102:5524–9.

Zhao M, Beauregard DA, Loizou L, Davletov B, Brindle KM. Non-invasive detection of apoptosis using magnetic resonance imaging and a targeted contrast agent. Nat Med 2001;7:1241–4.

Krishnan AS, Neves AA, de Backer MM, Hu D-E, Davletov B, Kettunen MI, et al. Detection of cell death in tumors using MRI and a gadolinium-based targeted contrast agent.

Radiology 2008;246:854–62.

Canuto HC, McLachlan C, Kettunen MI, Velic M, Krishnan AS, Neves AA, et al. Characterization of image heterogeneity using 2D Minkowski functionals increases the sensitivity of detection of a targeted MRI contrast agent.

Magn Reson Med 2009;61:1218–24.

Prescher JA, Contag CH. Guided by the light: visualizing biomolecular processes in living animals with bioluminescence. Curr Opin Chem Biol 2010;14:80–9.

Weissleder R, Tung CH, Mahmood U, Bogdanov A. In vivo imaging of tumors with protease-activated near-infrared fluorescent probes. Nat Biotech 1999;17:375–8.

Gallagher FA, Kettunen MI, Hu DE, Jensen PR, in’t Zandt R, Karlsson M, et al. Production of hyperpolarized [1,4-13C)]malate from [1,4-13C)]fumarate is a marker of cell necrosis and treatment response in tumors. Proc Natl Acad Sci USA 2009;106:19801–6.

Witney TH, Kettunen MI, Hu D-E, Gallagher FA, Bohndiek SE, Napolitano R, et al. Detecting treatment response in a model of human breast adenocarcinoma using hyperpolarised [1-13C]pyruvate and [1,4-13C2]fumarate. Br J Cancer 2010;103:1400–6. in’t Zandt R. Metabolic reactions studied with 13C-DNPMR in vitro and in vivo. Proc Int Soc Magn Reson Med 2008;16:827.

Kettunen M, Clatworthy M, Witney T, Hu D-E, Kennedy B, Bohndiek S, et al. Detection of acute kidney injury using hyperpolarized [1,4-13C2]fumarate. Proc Int Soc Magn Reson Med 2011;19:3513.

Cairns RA, Harris IS, Mak TW. Regulation of cancer cell metabolism. Nat Rev Cancer 2011;11:85–95.

Gao P, Tchernyshyov I, Chang TC, Lee YS, Kita K, Ochi T, et al. c-Myc suppression of miR-23a/b enhances mitochondrial glutaminase expression and glutamine metabolism.

Nature 2009;458:762–5.

Gallagher F, Kettunen M, Day S, Lerche M, Brindle K. 13C Magnetic resonance spectroscopy measurements of glutaminase activity in human hepatocellular carcinoma cells using hyperpolarized 13C-labeled glutamine. Magn Reson Med 2008;60:253–7.

Shields AF, Grierson JR, Dohmen BM, Machula H-J, Stayanoff JC, Lawhorn-Crews JM, et al. Imaging proliferation in vivo with [F-18]FLT and positron emission tomography. Nature Med 1998;4:1334–6.

Vesselle H, Grierson J, Muzi M, Pugsley JM, Schmidt RA, Rabinowitz P, et al. In vivo validation of 39deoxy-39-[F18]fluorothymidine ([F-18]FLT) as a proliferation imaging

708

56.

57.

58.

59.

60.

61.

62.

63.

64.

65.

66.

67.

68.

69.

tracer in humans: correlation of [F-18]FLT uptake by positron emission tomography with Ki-67 immunohistochemistry and flow cytometry in human lung tumors. Clin Cancer Res 2002;8:3315–23.

Zhao SM, Lin Y, Xu W, Jiang WQ, Zha ZY, Wang P, et al.

Glioma-derived mutations in IDH1 dominantly inhibit IDH1 catalytic activity and induce HIF-1 alpha. Science 2009;324:261–5.

Xu W, Yang H, Liu Y, Yang Y, Wang P, Kim SH, et al.

Oncometabolite 2-hydroxyglutarate is a competitive inhibitor of alpha-ketoglutarate-dependent dioxygenases. Cancer Cell 2011;19:17–30.

Gallagher F, Kettunen M, Day S, Hu D-E, Karlsson M, Gisselsson A, et al. Detection of tumor glutamate metabolism in vivo using 13C magnetic resonance spectroscopy and hyperpolarized [1-13C]glutamate. Magn Reson Med 2011;66:18–23.

Gallagher F, Kettunen M, Day S, Hu D-E, Ardenkjær-Larsen J, in’t Zandt R, et al. Magnetic resonance imaging of pH in vivo using hyperpolarized 13C-labeled bicarbonate. Nature 2008;453:940–3.

Bohndiek SE, Kettunen MI, Hu D-E, Kennedy BWC, Boren J, Gallagher FA, et al. Hyperpolarized [1-13C]-ascorbic and dehydroascorbic acid: vitamin C as a probe for imaging redox status in vivo. J Am Chem Soc 2011;133:11795–801.

Gillies RJ, Raghunand N, Garcia-Martin ML, Gatenby RA. pH imaging. A review of pH measurement methods and applications in cancers. IEEE Eng Med Biol Mag 2004;23: 57–64.

Ardenkjaer-Larsen JH, Leach AM, Clarke N, Urbahn J, Anderson D, Skloss TW. Dynamic nuclear polarization polarizer for sterile use intent. NMR Biomed 2011;24: 927–32.

Cunningham CH, Chen AP, Albers MJ, Kurhanewicz J, Hurd RE, Yen YF, et al. Double spin-echo sequence for rapid spectroscopic imaging of hyperpolarized 13C. J Magn Reson 2007;187:357–62.

Yen YF, Kohler SJ, Chen AP, Tropp J, Bok R, Wolber J, et al.

Imaging considerations for in vivo 13C metabolic mapping using hyperpolarized 13C-pyruvate. Magn Reson Med 2009;62:1–10.

Mayer D, Levin Y, Hurd R, Glover G, Spielman D. Fast metabolic imaging of systems with sparse spectra: application for hyperpolarized 13C imaging. Magn Reson Med 2006;56:932–7.

Mayer D, Yen YF, Levin YS, Tropp J, Pfefferbaum A, Hurd RE, et al. In vivo application of sub-second spiral chemical shift imaging (CSI) to hyperpolarized 13C metabolic imaging: comparison with phase-encoded CSI. J Magn Reson 2010;204:340–5.

Schulte R, Wiesinger F, Fish K, Whitt D, Hancu I. Efficient hyperpolarised 13C metabolic imaging with rosette spectroscopic imaging. Proc Int Soc Magn Reson Med 2009;17:2433.

Valentin J. The 2007 recommendations of the International Commission on Radiological Protection. ICRP Publication no. 103. Ottawa, Canada: ICRP; 2007.

MacKenzie JD, Yen YF, Mayer D, Tropp JS, Hurd RE, Spielman DM. Detection of inflammatory arthritis by using hyperpolarized (13)C-pyruvate with MR imaging and spectroscopy. Radiology 2011;259:414–20.

The British Journal of Radiology, June 2012

13 The British Journal of Radiology, 84 (2011), 31–37

Diffusion-weighted magnetic resonance imaging for monitoring prostate cancer progression in patients managed by active surveillance 1

V A MORGAN, MSc, 1S F RICHES, MSc, 2K THOMAS, S GILES, BSc and 1N M DESOUZA, MD, FRCR

MSc,

3

N VANAS,

MRCP, FRCR,

3

C PARKER,

MD, FRCR,

1

1

Cancer Research UK Clinical Magnetic Resonance Research Group, 2Department of Statistics, and 3Academic Urology Unit, Institute of Cancer Research and Royal Marsden NHS Foundation Trust, Downs Road, Sutton SM2 5PT, UK

Objectives: We studied patients managed by active surveillance to determine whether there was a difference over time in apparent diffusion coefficients (ADCs) derived from diffusion-weighted MRI in those who progressed to radical treatment (progressors, n517) compared with those who did not (non-progressors, n533).

Methods: 50 consecutive patients (Stage T1/2a, Gleason grade # 3+4, prostate-specific antigen (PSA) ,15 ng ml–1, ,50% cores positive) were imaged endorectally (baseline and 1–3 years follow-up) with T2 weighted (T2W) and echo-planar diffusion-weighted MRI sequences. Regions of interest drawn on ADC maps with reference to the T2W images yielded ADCall (b50–800), ADCfast (b50–300) and ADCslow (b5300–800) for whole prostate (minus tumour) and tumour (low signal-intensity peripheral zone lesion in biopsy-positive octant).

Results: Tumour and whole prostate ADCall and ADCfast were significantly reduced over time in progressors (p50.03 and 0.03 for tumours, respectively; p50.02 and 0.007 for the whole prostate, respectively). There were no significant changes in ADC over time in non-progressors. A 10% reduction in tumour ADCall indicated progression with a 93% sensitivity and 40% specificity (Az of receiver operating characteristic (ROC) curve 5 0.68). Percentage reductions in whole prostate ADCall, ADCfast and ADCslow were also significantly greater in progressors than in non-progressors (p50.01, 0.03 and 0.008, respectively).

Conclusion: This pilot study shows that DW-MRI has potential for monitoring patients with early prostate cancer who opt for active surveillance.

Patients with early-stage prostate cancer may be offered active surveillance because of the indolent nature of the disease in many cases. This involves regular monitoring with prostate-specific antigen (PSA) levels and repeat biopsy. Repeat biopsy is invasive, sometimes poorly tolerated and carries a risk of morbidity. Non-invasive imaging methods are therefore being explored increasingly to provide biomarkers of prostate cancer behaviour.

Although T2 weighted (T2W) MRI is the best way of visualising anatomical detail within the prostate, it has a sensitivity of just 60–76% for disease detection within the gland, with a specificity of around 55% [1, 2].

An increasingly useful addition to conventional T2W MRI is the use of ‘‘apparent diffusivity’’ (tissue water incoherent displacement over distances of 1–20 mm) to develop image contrast. Diffusion-weighted (DW) MRI has been used in both clinical and research settings to detect and evaluate a variety of tumour types [3–7].

In prostate cancer, DW-MRI is proving useful in tumour detection [8]. The apparent diffusion coefficients

Address correspondence to: Profesor N M deSouza, MRI Unit, Royal Marsden Hospital, Downs Road, Sutton, Surrey SM2 5PT, UK.E-mail: nandita.desouza@icr.ac.uk

The British Journal of Radiology, January 2011

Received 27 October 2009 Revised 8 December 2009 Accepted 16 December 2009 DOI: 10.1259/bjr/14556365 ’ 2011 The British Institute of Radiology

(ADCs) that are derived by this technique provide quantitative information on the degree to which water diffusion, including the contributions made by microcapillary perfusion and true diffusion within the extracellular space, is restricted within tissues. ADCs are therefore directly associated with coherent microvessel density and cellularity [9] and with microcapillary perfusion (which contribute to a ‘‘fast’’ diffusion component), and with water movement within the extracellular or intracellular space over a shorter diffusion path (which contributes to a ‘‘slow’’ diffusion component). We have shown previously that significant differences in tumour ADCs exist between patients with low-risk and higher-risk localised prostate cancer, and that this is the case for both the fast and the slow components [10]. Changes in ADC components in the tumour and in the surrounding normal prostate tissue have not, however, been studied previously in relation to disease progression in low-risk patients managed by active surveillance. The aim of this pilot study of patients managed by active surveillance was therefore to use DW-MRI to establish whether the changes in tumour and whole prostate ADCs in patients who progressed to radical treatment differed over time from those in patients whose disease did not progress.

31

14 V A Morgan, S F Riches, K Thomas et al

Methods and materials Patient population

Following endorectal MRI, an external pelvic phased array coil was used to acquire axial T1 weighted (T1W) (TSE 650/7 ms (TR/TE) and T2W (TSE 5396/80 ms (TR/ effective TE)) images through the pelvis as part of the routine clinical staging scan. The B0 maps and T1W images confirmed the absence of any significant postbiopsy haemorrhage in this patient cohort.

This was a single-institution, longitudinal, retrospective pilot study with institutional approval from the local research ethics committee. The study included 50 consecutive patients with localised prostate cancer (Stage T1 or 2a disease; Gleason 3 + 3 (n541), 3 + 4 (n58), 4 + 3 (n51); median PSA 7.2 ng ml–1 (minimum 0.4 ng ml–1, maximum 15.0 ng ml–1); ,50% cores positive) who had had MRI for clinical staging within 4 months of diagnostic biopsy and who had opted for management within an active surveillance research programme. Patients underwent DW-MRI in addition to their standard T2W MRI at baseline (time-point 1). All patients had a follow-up MRI as part of the active surveillance monitoring schedule (time-point 2) after a mean of 24 months (median 24 months, minimum 7 months, maximum 41 months). The patients were classified at the time of this second MRI as non-progressors (who continued on active surveillance) or as progressors to radical treatment (those requiring treatment within 6 months). The time between baseline and follow-up scan was not significantly different in nonprogressors compared with progressors (25¡7 months for non-progressors; 23¡8 months for progressors). All untreated patients had at least 12 months of follow-up.

Radical treatment was recommended to patients with a PSA velocity .1 ng ml–1 per year or with adverse features on repeat biopsy (defined as Gleason score .3 + 4 or .50% of cores involved) [11]. The clinicians responsible for treatment (CP and NVA) were blinded to the MRI results.

Axial T2W and DW-MRI images and ADC maps were transferred offline for analysis. Regions of interest (ROIs) were drawn around the whole prostate (peripheral zone and central gland) and around the tumour (low-signalintensity lesion in a sextant biopsy positive for tumour) on the scanner-generated ADC maps. As the results were quantitative, the ROI placement was done by a single observer with 10 years’ experience of prostate MRI. Mean ADC values (6 1023 mm2 s–1) from tumour and whole prostate (minus tumour) regions were calculated from the ADC-derived ROIs using a weighted monoexponential model that included all b-values (ADCall) for both baseline and follow-up imaging (in-house software developed using Interactive Data Language IDL; RSI Ltd, Boulder, CO). In addition, data obtained using just the low (0–300) or high (300–800) b-values were also analysed separately to reflect fast (ADCfast) and slow (ADCslow) diffusion components [12]. Values for ADCall, ADCfast and ADCslow were calculated in patients who progressed to radical treatment (n517) and compared with values from those that did not (n533).

MRI data acquisition

Statistical analysis

MR studies were done on a 1.5 T Intera system (Philips Medical Systems, the Netherlands) using a balloon design endorectal coil (Philips Medical Systems) inflated with 55 ml of air. Hyoscine butyl bromide (20 mg) was administered intramuscularly immediately prior to centring the patient in the scanner in order to reduce peristalsis: this is routine at our institution for abdominopelvic MRI and this drug is preferred to glucagon because it provides more effective antiperistalsis. None of our patients had a history of urinary retention. Hyoscine butyl bromide is contraindicated in patients with large prostates and urinary retention, but when given intramuscularly at this dose, we have had no cases of urinary retention in .500 prostate examinations. Conventional T2W fast-spinecho images were obtained in three orthogonal planes (turbo spin echo (TSE) 2000/90 ms (repetition time (TR)/ effective time to echo (TE), echo train length 16, 2 signal averages) with a 256 6 512 matrix re-sampled to 512 6 512, 3 mm slice thickness, no gap and a 14 cm field of view (FOV) (total imaging time 12 min). Echo-planar DW images (DWI 2500/69 ms (TR/TE)) with b-values of 0, 300, 500 and 800 s/mm2 were obtained transverse to the prostate and parallel to the corresponding set of T2W images. The phase-encoding gradient was from left to right in order to minimise motion artefacts in the prostate.

Either 12 or 18 3–4 mm thick slices (no gap, 20 cm FOV, matrix 128) provided coverage of the prostate with an image acquisition time of 1 min 40 s. ADC maps were generated using the system software and all b-values.

Statistical analysis of the data was performed using SPSS, version 15.0 (Chicago, IL) for Windows. ADC data for whole prostate and tumour at both time points were tested for normality using a Shapiro–Francia normality test. All data were normally distributed. ADC values for the tumour and whole prostate (minus tumour) regions were compared between time-points 1 (baseline) and 2 (follow-up) using paired t-tests. Means and standard deviations of the absolute change and mean percentage changes were calculated for the whole group. Percentage ADC changes in patients who progressed to radical treatment were compared with those in patients who continued on active surveillance after time-point 2 using an unpaired t-test (unequal variance assumed). All significance tests were two-sided and a p-value of ,0.05 was chosen as the criterion for statistical significance. A receiver operating characteristic (ROC) curve was used to determine the sensitivity and specificity of a cut-off value for change in a parameter for indicating progression to radical treatment.

32

Data analysis

Results Changes in tumour and gland volumes In non-progressors, tumour and whole gland volumes (mean¡SD) determined from ROI sizes were 0.33¡0.38 cm3 and 53.7¡27.0 cm3, respectively, at The British Journal of Radiology, January 2011

15 Diffusion weighted MRI for monitoring progression of prostate cancer

time-point 1 and 0.47¡0.45 cm3 and 59.4¡31.1 cm3, respectively, at time-point 2. In progressors, these data were 1.1¡1.9 cm3 and 44.6¡17.8 cm3, respectively, at time-point 1 and 1.1¡1.2 cm3 and 46.0¡21.1 cm3, respectively, at time-point 2. PSA concentrations rose from 6.4¡3.6 ng ml–1 at time-point 1 to 7.9¡4.6 ng ml–1 at time-point 2 in non-progressors, compared with a rise from 8.4¡2.4 ng ml–1 to 10.8¡3.8 ng ml–1 in progressors.

There was no significant change over time in tumour or whole gland volume in either group. The change in PSA concentration over time approached significance (p50.059), which is to be expected as this parameter is used as a criterion for defining progression.

No tumour was visible in 7 of the 33 non-progessors and in 1 of the 17 progressors. Tumour ADC values were therefore obtained from 26 non-progressors and 16 progressors, whereas whole prostate ADC values were obtained from the entire cohort. Table 1 shows the mean ADCall, ADCfast and ADCslow values for the entire patient cohort, for those that did not progress to radical treatment (non-progressors) and for those that did (progressors).

Representative T2W images with corresponding ADC maps from a non-progressor and a progressor are illustrated in Figures 1 and 2, respectively. When tumour regions for the entire cohort were considered together, ADCall and ADCfast were significantly lower at time-point 2 than at time-point 1, but there was no significant change over time in ADCslow (Table 1). This may be explained by the inherently lower signal to noise ratio (SNR), and thus greater variability of the data, at higher b-values.

Differences in whole prostate ADC values between the time-points were not significant.

At the outset (time-point 1), the ADCall, ADCfast and ADCslow in progressors were not significantly different from those in non-progressors. When data from the nonprogressors and progressors were analysed separately, it was evident however that the significant reduction in tumour ADCall and ADCfast seen over time could be

attributed to the progressors (p50.03) because there was no significant change in these components in non-progressors. The whole prostate also showed a significant reduction in ADCall (p50.02) and ADCfast (50.007) in progressors but not in non-progressors, indicating reduced microcapillary perfusion over both the tumour and the whole prostate over time in the progressors.

When comparing percentage change in ADCall, ADCfast and ADCslow in non-progressors and progressors, all whole prostate ADC values were significantly lower in progressors than in non-progressors (Figure 3a, Table 2). Tumour ADC values were not significantly different, however, possibly because of high variability between patients (Figure 3b). ROC curve analysis showed an area under curve of 0.68 for change in tumour ADCall. A 10% reduction in tumour ADC predicted progression with a 93% sensitivity and 40% specificity.

For ADCfast, a 10% reduction in value predicted progression with a 66.7% sensitivity and 40% specificity (Az5 0.45); for ADCslow, a 10% reduction in value predicted progression with a 66.7% sensitivity and 36% specificity (Az50.46).

Discussion This pilot study showed that ADC values for both tumours and whole prostate were significantly reduced at follow-up in patients who progressed to radical treatment, although the reduction in ADCslow was not significant, possibly because of low SNR and relatively high variability between patients. Although it was not possible to ‘‘test reproducibility’’ by including a second baseline measurement, the stability of values for whole prostate ADCall in non-progressors across the two timepoints indicates that this measurement is reproducible.

The reduction over time in tumour and whole prostate ADCfast in progressors was unexpected and indicates a reduction in microcapillary perfusion within both the

Table 1. Comparison of ADC components with time for whole prostate and tumour regions over the entire cohort and in the subsets of those who progressed to radical treatment and those who did not ADCall

Whole group TP1 Whole group TP2 p-value (TP1 vs TP2) Progressors TP1 Progressors TP2 p-value (TP1 vs TP2) Non-progressors TP1 Non-progressors TP2 p-value (TP1 vs TP2)

ADCfast

ADCslow

WP mean¡ SD 6 1023

*Tumour mean¡ SD 6 1023

WP mean¡ SD 6 1023

*Tumour mean¡ SD 6 1023

WP mean¡ SD 6 1023

*Tumour mean ¡SD 6 1023

1.73¡0.14

1.43¡0.29

1.99¡0.16

1.62¡0.32

1.35¡0.10

1.16¡0.24

1.74¡0.10

1.34¡0. 24

1.94¡0.11

1.52¡0.26

1.35¡0.10

1.11¡0.21

0.35

0.06

0.11

0.049

0.72

0.3

1.76¡0.10 1.684¡0.08 0.02

1.33¡0.20 1.20¡0.13 0.03

2.03¡113 1.92¡0.09 0.007

1.53¡0.23 1.37¡0.17 0.03

1.37¡0.09 1.33¡0.08 0.18

1.08¡0.18 1.00¡0.12 0.16

1.72¡0.16

1.49¡0.33

1.97¡0.18

1.68¡0.36

1.34¡0.10

1.21¡0.25

1.72¡0.12

1.41¡0.26

1.96¡0.12

1.60¡0.26

1.37¡0.11

1.17¡0.23

0.92

0.36

0.72

0.37

0.23

0.54

*Tumour values for n526 non-progressors and n516 progressors in whom a tumour was visible. Whole prostate (WP) values are given for the entire cohort. TP, time-point. ADC, apparent diffusion coefficient.

The British Journal of Radiology, January 2011

33

16 V A Morgan, S F Riches, K Thomas et al

(a)

(b)

(c)

(d)

Figure 1. Non-progressor to radical treatment showing tumour in the left midzone of the prostate (arrows) at baseline on (a) T2W (T2 weighted) axial (turbo spin-echo 2000/90 ms (repetition time (TR)/effective time to echo (TE), echo train length 16) and (b) the corresponding apparent diffusion coefficient (ADC) map (echo planar single shot diffusion-weighted image 2500/69 ms (TR/TE), b-values of 0, 300, 500 and 800 s mm–2) and after 24 months of follow-up on (c) T2W axial and (d) the corresponding ADC map. No change in appearance between the two time-points is seen.

tumour and the whole gland over a period of 2–3 years.

Previous studies that have used colour Doppler ultrasound to map the vascular anatomy of the prostate have demonstrated hypervascularisation of the peripheral zone of patients with prostate cancer [13, 14]. Conversely, hypoxia is known to promote proliferation of prostate stromal cells in culture, and when measured invasively using a polargraphic electrode has also been found to be a feature of prostate cancer [15]. To our knowledge, however, a reduction in vascularity within non-tumour regions of the prostate over time has not been described previously. It may be that this factor is an important driver of disease progression, and this is supported by ex vivo findings of increased hypoxia in more aggressive tumour types [16]. A larger series using dynamic contrastenhanced MRI to compare modelled vascular parameters of perfusion and permeability in patients on active surveillance programmes would be useful.

The mean area for a tumour ROI in this study, measured on both the T2W images and the ADC maps, was ,20 mm2. Pixel sizes of 0.27 6 0.27 mm on T2W images and 1.6 6 1.6 mm on the ADC maps meant that, on average, each tumour ROI encompassed ,274 pixels on a T2W image slice vs ,8 pixels on the ADC maps. Partial volume effects from tissue adjacent to the ADC-determined ROIs are therefore a significant problem. The image matrix size, 34

FOV and slice thickness were compromised in order to achieve a reasonable SNR within a reasonable timeframe that was clinically acceptable. Use of an external coil in conjunction with an endorectal coil or an endocavitary array would improve SNR and provide DW-MRI of greater spatial resolution and reduced partial volume effects when assessing small tumours within same timeframe.

The b-values chosen for our study enabled discrimination of fast and slow diffusion components. Although some researchers have advocated the use of b-values of 1000 s mm–2 to separate out the slow diffusion components adequately [17], our data show that values up to 800 s mm–2 are sufficient: use of higher b-values merely serves to increase the noise in the acquired data. Ex vivo data show that it is the slow diffusion component that is associated with cell density [18]. Thus, the ADC differences that discriminate progressors to radical treatment vs non-progressors are expected to be due to more dense cellular regions in the former group. An improvement in SNR would, however, be required to detect this.

For DW-MRI of the prostate, single-shot echo-planar (EPI) sequences are favoured over TSE sequences because of the need to freeze bulk motion. However, the susceptibility-induced distortion to which singleshot EPI is prone can be problematic in prostate imaging where air in the rectum or within the balloon of The British Journal of Radiology, January 2011

17 Diffusion weighted MRI for monitoring progression of prostate cancer

(a)

(c)

(b)

(d)

Figure 2. Images from a progressor to radical treatment showing tumour in the left mid-zone of the prostate (arrows) at baseline on (a) T2W (T2 weighted) axial (turbo spin-echo 2000/90 ms (repetition time (TR)/effective time to echo (TE), echo train length 16) and (b) the corresponding apparent diffusion coefficient (ADC) map (echo planar single shot diffusion-weighted image 2500/ 69 ms (TR/TE), b-values of 0, 300, 500 and 800 s mm–2) and after 35 months of follow-up on (c) T2W axial and (d) the corresponding ADC map. A change in the appearance of the tumour on the ADC maps between the two time-points is evident.

the endorectal coil causes significant local magnetic field inhomogeneity and susceptibility artefact. Our DWIs were acquired with an EPI readout and contained some distortion at tissue boundaries where there was a discontinuity in magnetic susceptibility. Advantages of an EPI sequence, however, are that it is possible to obtain

12 contiguous 4 mm-thick slices or 18 contiguous 3 mm slices, giving a supero-inferior coverage of 4.8 cm or 5.4 cm in less than 1.6 min. In all but two of our cases, this was sufficient to cover the prostate from apex to base.

Haemorrhage following biopsy can potentially alter ADC measurements. These effects can last for several

(a)

(b)

Figure 3. Percentage change in (a) whole prostate values and (b) tumour values for ADCall, ADCfast and ADCslow in progressors and non-progressors to radical treatment. Lower quartile, bottom line of box; median, middle line of box; upper quartile, top line of box; lower whisker, lower value; upper whisker, upper value; circles, outliers. ADC, apparent diffusion coefficient.

The British Journal of Radiology, January 2011

35

18 V A Morgan, S F Riches, K Thomas et al Table 2. Comparison of percentage change over time for those that progressed to radical treatment with those that did not ADCall

Progressor (n511) % change TP2–TP1 Non-progressor (n521) % change TP2–TP1 p-value

ADCfast

WP mean ¡SD

Tumour mean ¡SD

WP mean ¡SD

Tumour mean ¡SD

WP mean ¡SD

Tumour mean ¡SD

24.1¡5.4

27.1¡10.0

24.8¡5.9

27.3¡12.5

22.7¡5.8

24.4¡13.7

0.6¡6.5

23.0¡17.7

20.3¡7.0

23.2¡17.8

2.5¡6.8

21.2¡19.4

0.01

0.4

months, and time from biopsy to scan, even when done at the conventional 4 weeks post biopsy, may not be sufficient. In 11 cases, biopsy was performed for logistical reasons 3 weeks to 4 months before MRI at time-point 2.

However, in none of these cases was there evidence of intraprostatic haemorrhage on the T1W scans, nor was there any susceptibility artefact on the high b-value DWIs, indicating that the effects of haemorrhage on the ADC values were likely to be negligible.

In addition to the imaging limitations of poorer SNR when an external array is not used, ADC is affected by partial volume effects in the measurement of tumour regions and post-biopsy haemorrhage affecting ADC. This study is also limited by the definition of ‘progressors’, which must essentially be considered to be arbitrary because of the effects of sampling error on biopsy. The decision to institute radical treatment was based on the definitions of biochemical and histological disease progression, which are, of necessity, not evidence based.

However, pre-treatment PSA velocity has been shown to be an important determinant of fatal prostate cancer in several different settings [19–21], and Gleason score is an established predictor of prostate cancer mortality in localised disease. The policy for this patient cohort was to do repeat biopsies at 2 years and for the MRIs to be done immediately before the repeat biopsies. Unfortunately, there was considerable variation in the timing of the repeat biopsies because of a mixture of clinical concern and patient compliance. A future study comparing ADC values in those that are selected for active surveillance and those that opt for radical treatment at outset should also prove interesting.

Conclusion This pilot study shows that DW-MRI has potential for monitoring patients with early prostate cancer who opt for active surveillance. The quantitative data obtained indicate that the ADCs of tumours are lower than those of the whole prostate, and that microcapillary perfusion both within a tumour and in the whole prostate reduces significantly with time in progressors to radical treatment; percentage reductions in microcapillary perfusion and true diffusion in the whole prostate also are significantly greater in progressors compared with non-progressors. This suggests that the identification of individual tumour regions is not of paramount importance in the development of ADC as a prognostic biomarker in prostate cancer patients managed with active surveillance. Further work in larger-scale studies is needed to support this finding.

36

ADCslow

0.03

0.4

0.008

0.55

Acknowledgments Supported by Cancer Research UK and the ESPRC Cancer Imaging Centre; in association with MRC, Department of Health (CUK C1060/A10334) and NHS funding to the NIHR Biomedical Research Centre.

References 1. Casciani E, Polettini E, Bertini L, Emiliozzi P, Amini M, Pansadoro V, et al. Prostate cancer: evaluation with endorectal MR imaging and three-dimensional proton MR spectroscopic imaging. Radiol Med 2004;108:530–41.

2. Kirkham AP, Emberton M, Allen C. How good is MRI at detecting and characterising cancer within the prostate? Eur Urol 2006;50:1163–74.

3. Chenevert TL, Stegman LD, Taylor JM, Robertson PL, Greenberg HS, Rehemtulla A, et al. Diffusion magnetic resonance imaging: an early surrogate marker of therapeutic efficacy in brain tumors. J Natl Cancer Inst 2000; 92:2029–36.

4. Lo GG, Ai V, Chan JK, Li KW, Cheung PS, Wong TT, et al.

Diffusion-weighted magnetic resonance imaging of breast lesions: first experiences at 3 T. J Comput Assist Tomogr 2009;33:63–9.

5. Messiou C, Morgan VA, De Silva SS, Ind TE, deSouza NM. Diffusion weighted imaging of the uterus: regional ADC variation with oral contraceptive usage and comparison with cervical cancer. Acta Radiol 2009;50:696–701.

6. Somford DM, Futterer JJ, Hambrock T, Barentsz JO. Diffusion and perfusion MR imaging of the prostate.

Magn Reson Imaging Clin N Am 2008;16:685–95.

7. Uto T, Takehara Y, Nakamura Y, Naito T, Hashimoto D, Inui N, et al. Higher sensitivity and specificity for diffusionweighted imaging of malignant lung lesions without apparent diffusion coefficient quantification. Radiology 2009; 252:247–54.

8. Shimofusa R, Fujimoto H, Akamata H, Motoori K, Yamamoto S, Ueda T, et al. Diffusion-weighted imaging of prostate cancer. J Comput Assist Tomogr 2005;29:149–53.

9. Hayashida Y, Hirai T, Morishita S, Kitajima M, Murakami R, Korogi Y, et al. Diffusion-weighted imaging of metastatic brain tumors: comparison with histologic type and tumor cellularity. AJNR Am J Neuroradiol 2006;27:1419–25.

10. deSouza NM, Riches SF, Vanas NJ, Morgan VA, Ashley SA, Fisher C, et al. Diffusion-weighted magnetic resonance imaging: a potential non-invasive marker of tumour aggressiveness in localized prostate cancer. Clin Radiol 2008;63: 774–82.

11. van As NJ, deSouza NM, Riches SF, Morgan VA, Sohaib SA, Dearnaley DP, et al. A study of diffusion-weighted magnetic resonance imaging in men with untreated localised prostate cancer on active surveillance. Eur Urol 2008 [Epub ahead of print].

12. Riches SF, Hawtin K, Charles-Edwards EM, deSouza NM. Diffusion-weighted imaging of the prostate and rectal wall: comparison of biexponential and monoexponential

The British Journal of Radiology, January 2011

19 Diffusion weighted MRI for monitoring progression of prostate cancer

13.

14.

15.

16.

17.

modelled diffusion and associated perfusion coefficients.

NMR Biomed 2009;22:318–25.

Kelly IM, Lees WR, Rickards D. Prostate cancer and the role of color Doppler US. Radiology 1993;189:153–6.

Rifkin MD, Sudakoff GS, Alexander AA. Prostate: techniques, results and potential applications of color Doppler US scanning. Radiology 1993;186:509–13.

Parker C, Milosevic M, Toi A, Sweet J, Panzarella T, Bristow R, et al. Polarographic electrode study of tumor oxygenation in clinically localized prostate cancer. Int J Radiat Oncol Biol Phys 2004;58:750–7.

Butterworth KT, McCarthy HO, Devlin A, Ming L, Robson T, McKeown SR, et al. Hypoxia selects for androgen independent LNCaP cells with a more malignant genoand phenotype. Int J Cancer 2008;123:760–8.

Kitajima K, Kaji Y, Kuroda K, Sugimura K. High b-value diffusion-weighted imaging in normal and malignant

The British Journal of Radiology, January 2011

18.

19.

20.

21.

peripheral zone tissue of the prostate: effect of signal-tonoise ratio. Magn Reson Med Sci 2008;7:93–9.

Zelhof B, Pickles M, Liney G, Gibbs P, Rodrigues G, Kraus S, et al. Correlation of diffusion-weighted magnetic resonance data with cellularity in prostate cancer. BJU Int 2009;103:883–8.

Carter HB, Ferrucci L, Kettermann A, Landis P, Wright EJ, Epstein JI, et al. Detection of life-threatening prostate cancer with prostate-specific antigen velocity during a window of curability. J Natl Cancer Inst 2006;98:1521–7.

D’Amico AV, Chen MH, Roehl KA, Catalona WJ. Preoperative PSA velocity and the risk of death from prostate cancer after radical prostatectomy. N Engl J Med 2004;351: 125–35.

D’Amico AV, Renshaw AA, Sussman B, Chen MH. Pretreatment PSA velocity and risk of death from prostate cancer following external beam radiation therapy. JAMA 2005;294:440–7.

37

20 The British Journal of Radiology, 85 (2012), e906–e911

Evaluation of dose reduction and image quality in chest CT using adaptive statistical iterative reconstruction with the same group of patients L-P QI,

PhD,

Y LI, MD, L TANG, MD, Y-L LI, MD, X-T LI, MM, Y CUI,

PhD,

Y-S SUN, MD and X-P ZHANG, MD

Department of Radiology, Peking University School of Oncology, Beijing Cancer Hospital and Institute, Key Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Beijing, China

Objectives: The objective of this study was to compare the image quality and radiation dose of chest CT images reconstructed with a blend of adaptive statistical iterative reconstruction (ASIR) and filtered back-projection (FBP) with images generated using conventional FBP. Methods: Patients with chest CT re-examinations were alternately assigned to two scanners with different reconstruction techniques. The study groups included noise index (NI) 11 with 30% ASIR (A30), NI 13 with 40% ASIR (A40), NI 15 with 50% ASIR (A50) and NI 17 with 60% ASIR (A60), sequentially changed every 2 months. The control images were obtained using FBP and NI 11. All acquisitions were performed with automatic dose modulation. Paired t-test and non-parameter test were applied to compare the difference.

Results: The radiation doses were significantly lower in the examinations that used ASIR (p,0.001). The mean dose reduction rate was 27.7%, 45.2%, 57.1% and 71.8% for Groups A30, A40, A50 and A60, respectively. The image quality of Groups A30–A50 was not inferior to that of the control examinations. The image noise of Group A60 was greater and subjective image quality was inferior to that of the control.

Conclusions: ASIR enabled the use of a higher NI with automatic dose modulation.

With 50% ASIR and a NI of 15, the effective radiation dose was reduced by 57%, without compromising image quality.

The rapid development of imaging techniques has led to a remarkable increase in the use of CT. A 2007 report [1] estimated that more than 68.7 million CT examinations are performed each year in the USA, more than 20 times the number performed in 1980 (3 million).

Furthermore, CT is responsible for more than two-thirds of the total radiation dose associated with medical imaging [2, 3]. When a widely publicised article [4] claimed that the estimated cancer risk attributable to CT radiation in the USA had grown from 0.4% to 1.5–2.0% because of the substantial increase in the use of CT, radiation exposure became the focus of increased concern.

Many efforts have been made to investigate effective methods for minimising the radiation dose associated with CT. The adopted approaches include decreasing peak voltage, using low-current (mA) applications and using high pitch, among others [5–8]. Automated tube modulation techniques have produced radiation dose reductions of 40–60% without compromising image quality, and are now routinely used with most scanners [9–10].

Address correspondence to: Professor Xiao-Peng Zhang, Department of Radiology, Peking University School of Oncology, Beijing Cancer Hospital and Institute, Fu Cheng Road 52, Hai Dian District, Beijing 100142, China. E-mail: zxp@bjcancer.org This work was supported by the National Basic Research Program of China (973 Program) (grant no. 2011CB707705) and National Natural Science Foundation of China (grant no. 30970825).

e906

Received 24 August 2011 Revised 26 November 2011 Accepted 6 December 2011 DOI: 10.1259/bjr/66327067 ’ 2012 The British Institute of Radiology

Adaptive statistical iterative reconstruction (ASIR), another approach for minimising radiation, is a unique CT reconstruction algorithm based on raw pixels compared with the routine use of filtered back-projection (FBP).

Theoretically, ASIR is a potential reconstruction algorithm to reduce radiation dose with no effect on image quality [11, 12]. In a study on coronary CT angiography in a large multicentre cohort, ASIR enabled a reduced tube current and a lower radiation dose than FBP, and preserved signal, noise and study interpretability [13]. Preliminary studies of chest and abdomen CT examinations have shown that ASIR images had better image quality and less image noise at a lower radiation dose than images acquired with a conventional FBP reconstruction algorithm [14, 15]. In this study, we evaluated the subjective image quality, image noise and radiation dose of chest CT images with automated tube modulation techniques reconstructed with a blend of ASIR and FBP compared with conventional FBP for the same patient group.

Methods and materials Authors with no financial ties to GE Healthcare had complete, unrestricted access to the study data and unrestricted control over the data during the study. Our prospective clinical study was approved by the human research committee of our institutional review board.

The British Journal of Radiology, October 2012

21 Evaluation of dose reduction and image quality in chest CT using ASIR

Subjects and methods The local institutional review board approved this study with a waiver of informed consent. The participants were all adult patients with solid tumours who underwent restaging, evaluation of treatment effect or routine follow-up re-examinations with chest CT scans between 1 September 2010 and 9 April 2011.

The clerical booking team alternately scheduled patients to one of two 64-MDCT scanners. One scanner (DiscoveryH HD 750; GE Healthcare, Waukesha, WI) was equipped with both ASIR and FBP image reconstruction algorithms. This scanner was capable of generating images with a mix of ASIR and FBP reconstruction algorithms varying from 10% to 100% ASIR in 10% increments. The researcher prospectively designed the reconstruction parameters, including ASIR and noise index (NI; quantum noise level in the image data regulated and desired by the user). We adopted NI 11 with 30% ASIR (A30; based on recommendations from the vendor and the results of initial image quality assessments), NI 13 with 40% ASIR (A40), NI 15 with 50% ASIR (A50) and NI 17 with 60% ASIR (A60). The reconstruction parameters sequentially changed every 2 months, while the other parameters remained constant. For the first few days that the new parameters were adopted, two experienced radiologists reviewed the CT images to ensure their quality. When the quality of the images significantly decreased, the study was stopped. On the fourth day after A60 was adopted, we stopped the study.

The other CT scanner (LightSpeedH VCT; GE Healthcare) was equipped with only a conventional FBP reconstruction algorithm and NI 11 was used as the control.

CT data acquisition The scanning range of the two scanners went from the supraclavicular space to the upper abdomen, including the bilateral adrenals.

The image acquisition parameters were as follows: tube potential, 120 kV; variable tube current (10–300 mA for the Discovery HD 750 and 50–400 mA for the LightSpeed VCT) determined by x-, y- and z-axis dose modulation; pitch, 0.984:1; table speed, 39.37 mm per gantry rotation; detector configuration, 6460.625 mm; reconstructed slice thickness, 5 mm; reconstructed interval, 5 mm; gantry rotation time, 0.8 s; field of view appropriate to patient size; and standard (mediastinal) and bone (lung) reconstruction kernels.

For contrast-enhanced CT, a mechanical injector (StellantH; Medrad, Warrendale, PA) was used for the intravenous bolus injection of non-ionic contrast material (iohexol) with a concentration of 300 mg ml21 iodine.

60–70 ml of contrast material was injected at a flow rate of 2.5 ml s21 and a fixed start delay of 30 s.

examinations of the same patient had either employed contrast medium or had not. If more than one CT examination had the same parameters, then the experimental exam scanned closest to the control one was chosen for the study.

Thoracic CT image analysis Images obtained with a mediastinal window (window width, 450 HU; window level, 45 HU) and a lung window (window width, 1500 HU; window level, 2500 HU) were reviewed at a picture archiving and communication systems (PACS) workstation with all patient and scanner demographic data removed. Two radiologists who had 15 and 10 years of work experience assessed the image quality independently. The readers did not review any previous or subsequent images of any patient, regardless of indication or the presence or absence of a pathological condition. Mediastinal images were assessed with attention to visualisation of the mediastinum, aorta, pulmonary vasculature and chest wall. Lung images were assessed with attention to the bronchial walls and vessels. Image quality problems caused by respiratory motion, patient motion and degree of contrast opacification of the contrast-enhanced images were ignored.

The following three-point Likert scale was used to evaluate image quality [14]. A score of 1 (non-diagnostic) was defined as severely impaired image quality caused by excessive noise, poor delineation of the bronchiolar and/or mediastinal vessel margins, degradation that impeded parenchymal evaluation, and considerable image noise in the chest wall and mediastinal structures.

A score of 2 (suboptimal) indicated moderately reduced image quality with limitations in the bronchiolar and/or mediastinal vessel margin delineation, and image noise that limited the evaluation of the pulmonary parenchyma. A score of 3 indicated good image quality without degradation of the bronchiolar and/or mediastinal vessel margin delineation due to image noise and depiction of the pulmonary parenchyma and mediastinal structures that allowed evaluation without substantial noise-related

Recruiting criteria The following recruiting criteria were used: (1) the patient had undergone at least one CT examination on each of the two CT scanners; (2) all of the CT The British Journal of Radiology, October 2012

Figure 1. The method of placing a region of interest .1.0 cm2 in the centre of the descending thoracic aorta at the level of carina in the mediastinal image. e907

22 L-P Qi, Y Li, L Tang et al Table 1. Scan ranges of chest CT scans Group

Control (mm)

With ASIR (mm)

t

p-value

A30 A40 A50 A60

307¡22 311¡21 301¡25 306¡22

306¡20 316¡22 304¡28 312¡23

0.391 21.675 21.019 22.365

0.699 0.105 0.315 0.029a

A30, noise index 11 with 30% ASIR; A40, noise index 13 with 40% ASIR; A50, noise index 15 with 50% ASIR; A60, noise index 17 with 60% ASIR; ASIR, adaptive statistical iterative reconstruction. a Significant value.

degradation. The recorded image quality score for the examination was the lowest score after review of all mediastinal and lung images.

We measured the noise of the PACS images by placing an oval or circular region of interest .1.0 cm2 in the centre of the descending thoracic aorta at the level of carina in the mediastinal images (Figure 1). The mean CT value and the standard deviation (SD) from the region of interest were recorded. The mean value for this homogeneous soft tissue was interpreted as the signal, and the SD was interpreted as the noise.

According to department protocol, a screen capture of the scanner console protocol page was sent to the PACS system. This protocol page was reviewed in a separate reading session to record the scan length, the volume CT dose index (CTDIvol) and the dose–length product for each CT acquisition. We converted the dose–length product to an effective dose in millisieverts by multiplying it by the thoracic conversion factor of 0.017 mSv mGy21 cm21 [16].

Statistical analysis The paired Student’s t-test (SPSSH for Windows, v. 11.5; SPSS, Inc., Chicago, IL) was used to compare the range of the CT scans and the effective doses between groups. A non-parametric test of two related samples (the Wilcoxon signed-rank test) was applied to compare the image signal-to-noise ratio (SNR), noise and image quality grades. p,0.05 was considered statistically significant.

Results 81 patients (42 females and 39 males; age range 20–82 years; mean 56 years) with a total of 199 examinations were recruited into this study. Each patient had 2–4 examinations (48 participants had 2 examinations, 29 had 3 and 4 had 4). All of the patients had control examinations with FBP reconstruction. The number of cases in Groups A30,

A40, A50 and A60 was 34, 30, 34 and 20, respectively. 67 patients underwent contrast-enhanced CT examinations, and 14 received plain scans.

Radiation dose The mean scan lengths for each group and their corresponding control examinations are listed in Table 1.

The mean scan length for A60 was longer than that for the control examinations. There were no differences between the other groups and their corresponding controls. The mean radiation doses of the study groups with ASIR were all significantly lower than those of the control groups (paired t-test; p,0.001 for all). The mean effective doses (ED) for the controls and the study groups were 8.7 vs 6.3 for A30, 8.5 vs 4.7 for A40, 7.9 vs 3.4 for A50 and 9.5 vs 2.6 mSv for A60. The mean ED reduction rates for A30, A40, A50 and A60 were 27.7%, 45.2%, 57.1% and 71.8%, respectively (Table 2). The mean dose value of CTDIvol for each group and the corresponding control group are shown in Table 3.

Image characteristics The SNR and noise values of the CT examinations employed contrast medium are listed in Table 4. The mean SNR of the images of Groups A30, A40 and A50 was not different from that of the control group. The mean SNR of A60 was lower than that of the control group with a marginal significance (p50.07). The image noise of A60 was greater than that of the control group (p50.001).There were no significant differences between A30/A40/A50 and the control group (Figure 2).

Image scores Detailed scores of two readers are shown in Table 5.

The k-value is 0.214 (p50.001) with agreement in 88.9% (177/199) of cases. There were no cases scored as 1 by either reader. All of the images in the control group were scored as 3 by both observers. Detailed scores of Reader 2 are shown in Table 6. The study group of A50 and A60 showed significantly lower scores than their controls (p50.014, p50.005).

Discussion Strategies for CT radiation dose optimisation include tube current modulation, body mass index-based tube

Table 2. Radiation dose distribution between control and study group with different percentage ASIR Group

n

Control (mSv)

With ASIR (mSv)

Reduction rate (%)

t

p-value

A30 A40 A50 A60

34 30 34 20

8.7¡3.7 8.5¡3.0 7.9¡3.7 9.5¡2.8

6.3¡3.0 4.7¡2.2 3.4¡2.1 2.6¡0.6

27.7¡16.5 45.2¡14.6 57.1¡9.5 71.8¡5.4

8.182 10.722 12.213 12.941

,0.001a ,0.001a ,0.001a ,0.001a

A30, noise index 11 with 30% ASIR; A40, noise index 13 with 40% ASIR; A50, noise index 15 with 50% ASIR; A60, noise index 17 with 60% ASIR; ASIR, adaptive statistical iterative reconstruction; n, number of cases. a Significant value.

e908

The British Journal of Radiology, October 2012

23 Evaluation of dose reduction and image quality in chest CT using ASIR Table 3. CTDIvol between control and study group with different percentage ASIR Group

n

Control (mGy)

With ASIR (mGy)

Reduction rate (%)

t

p-value

A30 A40 A50 A60

34 30 34 20

14.5¡6.2 14.0¡4.9 13.4¡6.4 15.9¡5.1

10.6¡5.0 7.6¡3.7 5.7¡3.7 4.3¡1.1

28.2¡15.6 46.0¡14.4 57.3¡10.1 72.4¡5.3

7.334 11.309 11.875 12.128

,0.001a ,0.001a ,0.001a ,0.001a

A30, noise index 11 with 30% ASIR; A40, noise index 13 with 40% ASIR; A50, noise index 15 with 50% ASIR; A60, noise index 17 with 60% ASIR; ASIR, adaptive statistical iterative reconstruction; CTDIvol, volume CT dose index; n, number of cases. a Significant value.

and the maximum was 77.0%. The mean ED reduction for Group A30 was 27.7% of the conventional FBP reconstruction with the same NI. This result is consistent with the findings of Prakash et al [20], which showed that ASIR was associated with an overall mean decrease of 27.6% in ED. When the ASIR ratio increased from 30% to 50% and the NI correspondingly increased from 11 to 15, the image quality remained stable and diagnostically useful, and the total ED further decreased by approximately 30%. In total, more than a 50% ED reduction was attributable to ASIR. This study confirmed the important role that ASIR plays in ED reduction.

The SNR and noise for Groups A30–A50 showed no difference from the control groups (Figure 2). The noise of Group A30 was lower than that of the control, with the same NI, although no significance was obtained (p50.086). We noted that the noise of A60 was greater than the value of the control group. The SNR of A60 was 10% lower than that of the control group, with p50.07.

The subjective score showed that image quality of Groups A50 and A60 was inferior to the corresponding control groups, but for most (82.4%) cases in Group A50 image quality was good, while for 40.0% of cases in Group A60 image quality was suboptimal. Further analysis of images scored as 2 showed impaired image quality manifesting as the obscure delineation of the heart and vessels in the mediastinal window. However, the bronchioles and the depiction of the pulmonary parenchyma or vessel margins in the lung field did not degrade, which may be attributable to the obvious natural contrast between air and vessels (Figure 3).

This study validated ASIR as an effective dose reduction tool, which was also proved by previous studies on the chest [14, 20]. Mitsumori et al [21] have studied the usefulness of automated current modulation for liver imaging. To our knowledge, there were few articles studying the concurrent changes in ASIR and NI using automatic tube current modulation for the chest.

voltage reduction, decreased scan length and low tube current scanning [17]. Automated tube modulation techniques are now routinely used on most scanners to reduce radiation doses [9, 10, 18]. Owing to inherent density differences in the chest, chest imaging with automated tube modulation is advantageous and routinely used in clinical practice.

In recent years, several studies have shown that, in addition to other accepted methods, ASIR is a useful dose reduction tool for chest or abdomen CT [11–15, 19– 20]. Theoretically, with ASIR, it is not assumed that noise is evenly distributed across the entire image, unlike with FBP. ASIR consists of two steps. First, the iterative reconstruction algorithm synthesises forward projection for each X-ray projection angle, considering the actual CT scanning process, in which X-ray photons originate from a finite focal spot area and reach detectors after traversing through the object being scanned.

Subsequently, the error between this forward projection and the scanner-acquired raw projection data is backprojected to update the image. During this process, statistical information can be incorporated into the measurement, which results in a less noisy image.

NI is one of the important parameters related to image quality with automated tube modulation. With FBP, it is difficult to balance the radiation dose and image quality.

If a high NI is adopted, a lower radiation dose will be obtained, but image noise will increase accordingly. The results of this study showed that ASIR allowed the use of a higher NI, which reduced the tube current and radiation dose and allowed the generation of images with significantly reduced noise without compromising image quality.

This study showed the ED of each group with a combination of ASIR and FBP and corresponding increased NI was significantly lower than that of the control group, which had complete FBP reconstruction.

The mean ED reduction rate in Group A50 was 57.1%,

Table 4. SNR and noise of the control group and study groups SNR

Noise

Group

n

Control

With ASIR

p-value

Control

With ASIR

p-value

A30 A40 A50 A60

22 25 30 16

21.6¡5.0 23.5¡5.8 21.3¡4.1 21.0¡5.0

22.9¡4.2 23.1¡5.7 20.8¡6.7 18.8¡4.9

0.158 0.638 0.600 0.070

10.0¡1.5 9.8¡1.8 10.0¡1.7 10.4¡1.4

9.5¡1.3 10.4¡2.1 10.2¡1.4 11.6¡1.6

0.086 0.199 0.761 0.001a

A30, noise index 11 with 30% ASIR; A40, noise index 13 with 40% ASIR; A50, noise index 15 with 50% ASIR; A60, noise index 17 with 60% ASIR; ASIR, adaptive statistical iterative reconstruction; n, number of cases; SNR, signal to noise ratio. a Significant value.

The British Journal of Radiology, October 2012

e909

24 L-P Qi, Y Li, L Tang et al

(a)

(b)

Figure 2. A 74-year-old female followed up after treatment of left central small cell lung carcinoma. (a) and (b) are images of the same slice obtained under control and A50 conditions in the mediastinal window (width, 450 HU; level, 45 HU). Appearance of aorta and left atrium is similar in (a) and (b), without a difference in image quality despite use of different noise indexes. Both readers assigned Likert score of 3 to both examinations, and measured signal to noise ratio was 19.18 and 13.60 for (a) and (b) conditions, respectively. The calculated effective radiation doses of (a) and (b) conditions were 12.99 and 4.79 mSv, respectively.

Table 5. Detailed scores of two readers

Table 6. Scores of all the CT images of the control and study groups by Reader 2

Reader 2

Reader 1

Total

Score

1

2

3

Total

1 2 3

0 0 0 0

0 4 16 20

0 6 173 179

0 10 189 199

Control

With ASIR

Group

n

Score 3

Score 2 (%) Score 3 (%) p-value

A30 A40 A50 A60

34 30 34 20

34 30 34 20

3 3 6 8

(8.8) (10.0) (17.6) (40.0)

31 27 28 12

(91.2) (90.0) (82.4) (60.0)

0.083 0.083 0.014 0.005

A30, noise index 11 with 30% ASIR; A40, noise index 13 with 40% ASIR; A50, noise index 15 with 50% ASIR; A60, noise index 17 with 60% ASIR; ASIR, adaptive statistical iterative reconstruction; n, number of cases.

(a)

(b)

Figure 3. A 72-year-old male with squamous cell carcinoma received lobectomy of left lower lobe. (a) and (b) are the images under A60 conditions in the mediastinal and lung window. Both readers assigned a Likert score of 2 to the examination. (a) Increased noise and obscure margins in the mediastinal structures. (b) Lung window image, showing clear delineation of the bronchioles and pulmonary vessels.

e910

The British Journal of Radiology, October 2012

25 Evaluation of dose reduction and image quality in chest CT using ASIR

Our study demonstrated that mean ED acquired with NI 15 and 50% ASIR decreased by 57.1% compared with that obtained with NI 11 and conventional FBP, without obvious decreased image quality. These results provide useful references for ASIR applications with automatic tube current modulation. Because this study was conducted with the same group of patients, patients’ weight and body morphology could be ignored, and the results could be used to accurately evaluate the dose reduction of a blend of ASIR and FBP. Another advantage of this study is that it was conducted during clinical practice and did not expose patients to additional radiation.

There are limitations to our study. First, we did not investigate the dose with complete FBP for the Discovery CT750 HD scanner to examine the dose reduction attributable to the hardware and software. Another shortcoming of this work was that we did not investigate the optimal ratio of ASIR to FBP. Nevertheless, our study showed that ASIR allows the use of the higher NI of 15, which was helpful for significantly reducing the radiation dose without impairing image quality. In addition, this study did not consider patient weight change over the course of the study period.

In conclusion, when automatic tube current modulation is adopted, chest CT scans with ASIR allow the use of a higher NI. With an ASIR of 50% and an NI of 15, the effective dose was reduced dramatically, without significantly compromising image quality. This preliminary study validated ASIR as a useful tool to lower radiation doses with a low-dose acquisition mode. Much work remains to be done to optimise the blend of ASIR and FBP and further investigate the maximum noise index.

References 1. IMV Medical Information Division. CT market summary report. Des Plaines, IL: IMV Medical Information Division; 2007.

2. Kalra MK, Maher MM, Toth TL, Hamberg LM, Blake MA, Shepard JA, et al. Strategies for CT radiation dose optimization. Radiology 2004;230:619–28.

3. Linton OW, Mettler FA Jr. National conference on dose reduction in CT, with an emphasis on pediatric patients.

AJR Am J Roentgenol 2003;181:321–9.

4. Brenner DJ, Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med 2007;357: 2277–84.

5. Nakaura T, Awai K, Oda S, Yanaga Y, Namimoto T, Harada K, et al. A low-kilovolt (peak) high-tube current technique improves venous enhancement and reduces the radiation dose at indirect multidetector-row CT venography: initial experience. J Comput Assist Tomogr 2011;35:141–7.

6. Schueller-Weidekamm C, Schaefer-Prokop CM, Weber M, Herold CJ, Prokop M. CT angiography of pulmonary arteries to detect pulmonary embolism: improvement of vascular enhancement with low kilovoltage settings.

Radiology 2006;241:899–907.

7. O’Connor OJ, Vandeleur M, McGarrigle AM, Moore N, McWilliams SR, McSweeney SE, et al. Development of

The British Journal of Radiology, October 2012

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

low-dose protocols for thin-section CT assessment of cystic fibrosis in pediatric patients. Radiology 2010;257:820–9.

Lell M, Hinkmann F, Anders K, Deak P, Kalender WA, Uder M, et al. High-pitch electrocardiogram-triggered computed tomography of the chest. Invest Radiol 2009; 44:728–33.

Graser A, Wintersperger BJ, Suess C, Reiser MF, Becker CR. Dose reduction and image quality in MDCT colonography using tube current modulation. AJR Am J Roentgenol 2006;187:695–701.

Kalra MK, Maher MM, Toth TL, Kamath RS, Halpern EF, Saini S. Radiation from ‘‘extra’’ images acquired with abdominal and/or pelvic CT: effect of automatic tube current modulation. Radiology 2004;232:409–14.

Yanagawa M, Honda O, Yoshida S, Kikuyama A, Inoue A, Sumikawa H, et al. Adaptive statistical iterative reconstruction technique for pulmonary CT: image quality of the cadaveric lung on standard- and reduced-dose CT. Acad Radiol 2010;17:1259–66.

Silva AC, Lawder HJ, Hara A, Kujak J, Pavlicek W. Innovations in CT dose reduction strategy: application of the adaptive statistical iterative reconstruction algorithm.

AJR Am J Roentgenol 2010;194:191–9.

Leipsic J, Labounty TM, Heilbron B, Min JK, Mancini GB, Lin FY, et al. Estimated radiation dose reduction using adaptive statistical iterative reconstruction in coronary CT angiography: the ERASIR study. AJR Am J Roentgenol 2010;195:655–60.

Leipsic J, Nguyen G, Brown J, Sin D, Mayo JR. A prospective evaluation of dose reduction and image qualitY in chest CT using adaptive statistical iterative reconstruction. AJR Am J Roentgenol 2010;195:1095–9.

Sagara Y, Hara AK, Pavlicek W, Silva AC, Paden RG, Wu Q. Abdominal CT: comparison of low-dose CT with adaptive statistical iterative reconstruction and routine-dose CT with filtered back projection in 53 patients. AJR Am J Roentgenol 2010;195:713–9.

Bongartz G, Golding SJ, Jurik AG, Leonardi M, van Persijn van Meerten E, Rodriguez R, et al. European guidelines for multislice computed tomography: appendix C. Contract no.

FIGM-C T2000-20078-CT-TIP. Luxembourg, Luxembourg: European Commission; 2004.

Kubo T, Lin PJ, Stiller W, Takahashi M, Kauczor HU, Ohno Y, et al. Radiation dose reduction in chest CT: a review. AJR Am J Roentgenol 2008;190:335–43.

McCollough CH, Bruesewitz MR, Kofler JM Jr. CT dose reduction and dose management tools: overview of available options. Radiographics 2006;26:503–12.

Singh S, Kalra MK, Gilman MD, Hsieh J, Pien HH, Digumarthy SR, et al. Adaptive statistical iterative reconstruction technique for radiation dose reduction in chest CT: a pilot study. Radiology 2011;259:565–73.

Prakash P, Kalra MK, Digumarthy SR, Hsieh J, Pien H, Singh S, et al. Radiation dose reduction with chest computed tomography using adaptive statistical iterative reconstruction technique: initial experience. J Comput Assist Tomogr 2010;34:40–5.

Mitsumori LM, Shuman WP, Busey JM, Kolokythas O, Koprowicz KM. Adaptive statistical iterative reconstruction versus filtered back projection in the same patient: 64 channel liver CT image quality and patient radiation dose.

Eur Radiol 2012;22:138–43.

e911

26 The British Journal of Radiology, 85 (2012), 114–120

Bone age assessment by dual-energy X-ray absorptiometry in children: an alternative for X-ray?

1,2

D H M HEPPE, MSc, 1,2,3H R TAAL, MD, 1,2G D S ERNST, MSc, 3E L T VAN DEN AKKER, MD, PhD, 4 M M H LEQUIN, MD, PhD, 3A C S HOKKEN-KOELEGA, MD, PhD, 1,2,3J J M GEELHOED, MD, PhD and 1,2,3 V W V JADDOE, MD, PhD 1

The Generation R Study Group, and Departments of 2Epidemiology, 3Paediatrics and 4Radiology, Erasmus Medical Center, 3000 CA Rotterdam, The Netherlands

Objective: The aim of the study was to validate dual-energy X-ray absorptiometry (DXA) as a method to assess bone age in children.

Methods: Paired dual-energy X-ray absorptiometry (DXA) scans and X-rays of the left hand were performed in 95 children who attended the paediatric endocrinology outpatient clinic of University Hospital Rotterdam, the Netherlands. We compared bone age assessments by DXA scan with those performed by X-ray. Bone age assessment was performed by two blinded observers according to the reference method of Greulich and Pyle. Intra-observer and interobserver reproducibility were investigated using the intraclass correlation coefficient (ICC), and agreement was tested using Bland and Altman plots.

Results: The intra-observer ICCs for both observers were 0.997 and 0.991 for X-ray and 0.993 and 0.987 for DXA assessments. The interobserver ICC was 0.993 and 0.991 for X-ray and DXA assessments, respectively. The mean difference between bone age assessed by X-ray and DXA was 0.11 years. The limits of agreement ranged from 20.82 to 1.05 years, which means that 95% of all differences between the methods were covered by this range.

Conclusions: Results of bone age assessment by DXA scan are similar to those obtained by X-ray. The DXA method seems to be an alternative for assessing bone age in a paediatric hospital-based population.

Children with the same chronological age often have a different bone maturation as a consequence of various genetic and social factors [1–3]. Bone age is a useful indicator of children’s growth and biological maturation and is frequently assessed in paediatric endocrinology to determine delayed or advanced growth [4–7]. In children with growth disorders, regular hand X-rays are needed to follow skeletal development at an interval of once or twice per year [8–10]. The classical method to assess bone age is based on the recognition of changes in the maturity indicators in hand–wrist X-rays by comparison with a reference atlas (Greulich and Pyle method) [11].

The main problem with this method is the exposure to a certain amount of irradiation involved in X-ray procedures [12–14]. Although the precise risk estimate of paediatric cancers due to diagnostic X-ray exposure is not known [15–17], we know that the lifetime attributable risk of cancer due to one single X-ray exposure in childhood approximates 15% per sievert [18]. To avoid detrimental effects in later life as a result of cumulative radiation exposure, dose reduction is therefore particularly important in childhood [18, 19]. Consequently, methods involving less radiation would be preferable to assess bone age Address correspondence to: Dr Vincent Jaddoe, The Generation R Study Group (Room AE006), Erasmus MC, PO Box 2040, 3000 CA, Rotterdam, The Netherlands. E-mail: v.jaddoe@erasmusmc.nl Funding: This study was supported by the Erasmus Medical Center Rotterdam and the Netherlands Organization for Health Research and Development (ZonMw 21000074). The study sponsors had no role in study design, data analysis, interpretation of data or writing this report.

114

Received 1 March 2010 Revised 3 May 2010 Accepted 13 May 2010 DOI: 10.1259/bjr/23858213 ’ 2012 The British Institute of Radiology

in children. Dual-energy X-ray absorptiometry (DXA) has been suggested as a safer method to assess bone age [20].

In both children and adults, DXA is currently widely used to measure bone mineral density for the assessment of osteoporosis [21]. When applied to assess bone age, a hand–wrist scan by DXA (0.0001 mSv) produces a 10-fold lower effective dose than a hand–wrist X-ray (0.001 mSv) [22].

One previous study in a paediatric population of 60 Polish subjects (5–20 years old) suggested that results for bone age assessment by DXA are similar to those produced

by X-ray [20]. However, their results were presented as correlation coefficients and t-test analysis. For methods of comparison, Bland and Altman analysis is a more appropriate analysis, since it investigates agreement [23, 24]. Also, they used a reference method that applied to the Polish population [25], whereas the Greulich and Pyle method would be more generalisable [3].

Thus far, the accuracy of the assessment of bone age in children using DXA scans has not been properly validated.

Therefore, the aim of this study was to investigate whether hand–wrist bone age assessment by DXA produces similar results to the classical X-ray method.

Materials and methods Subjects Participants were selected from the outpatient clinic of the Department of Paediatric Endocrinology of the The British Journal of Radiology, February 2012

27 Bone age assessment by DXA: an alternative?

Erasmus Medical Center, Sophia Children’s Hospital, Rotterdam, the Netherlands. From April until September 2009, we invited children who were already planned to have a hand–wrist X-ray for medical reason to participate in this study. If written informed consent was obtained, participants had an additional hand–wrist DXA scan immediately after their planned X-ray examination. The medical ethics committee of the Erasmus Medical Center approved this study.

Data collection All hand–wrist X-rays were performed by qualified technicians at the Department of Radiology according to their usual ‘‘hand–wrist for bone age’’ guidelines: the patient was seated next to the X-ray table, placing his or her left hand and wrist on a double-layered phosphor cassette. The hand was positioned flat, with no radial or ulnar deviance. The X-ray tube was focused on the metacarpalia. The whole hand and wrist was required to be on the X-ray to assess all epiphyses. The settings of the X-ray machine differed according to age group. For children aged ,7 years, a standardised modus of 40 kV and 1.60 mAs was used, with a radiation time of 0.837 s, and for children aged >7 years, we used 42 kV tube voltage and 1.60 mAs, with a radiation time of 0.778 s.

The radiographs were uploaded to the electronic hospital information system for offline scoring afterwards.

Subsequently, each participant underwent a DXA scan (iDXA; General Electric, formerly Lunar Corp., Madison, WI) of the left hand on the same day. All scans were performed by one of the two involved investigators using a standardised modus of 100 kV and 0.188 mAs.

(a)

All DXA scans were obtained using the same device and software. Children were scanned in a supine position to enable even the younger children to keep their hand still for 66 s (actual scan duration). The left hand was placed in a flat position on the table. The scan was focused on the hand, using a starting point of two finger widths below the radiocarpal articulation, to obtain an image of all hand bones including the wrist and distal radius.

Prior to the analysis, we decided to exclude scans or radiographs with major interpretation difficulties due to movement artefacts. Examples of a radiograph and a DXA are given in Figure 1.

Bone age assessment To blind the assessments, all radiographs and DXA images were stored in a file from which patient characteristics (i.e. the subject’s name and calendar age) were deleted. Two well-trained observers (observers 1 and 2) independently assessed all patients’ bone ages, twice by X-ray and twice by DXA scan, to enable estimation of both intra- and interobserver variability. To avoid recall bias, we first assessed all radiographs followed by all DXA images before we started repeated measurements. An interval of .1 week was kept between the first and second assessments of one image.

A patient’s bone age was assessed by comparing the maturity indicators on the patient’s radiograph or DXA scan to the standardised reference atlas according to the Greulich and Pyle method. If the patient’s bone age was considered to be in between two adjacent standards, the intermediate value was appointed to the patient [11]. Bone age was assessed with a maximum precision of 0.5 years.

(b)

Figure 1. (a) Hand–wrist image derived by X-ray of a 12-year-old girl with hypophyseal insufficiency and growth hormone substitution therapy. (b) Hand–wrist image derived by dual-energy X-ray absorptiometry of the same 12–year-old girl with hypophyseal insufficiency and growth hormone substitution therapy.

The British Journal of Radiology, February 2012

115

28 D H M Heppe, H R Taal, G D S Ernst et al Table 1. Patient characteristics

Height (cm) Weight (kg) Chronological age (years) Age categories n (%) 0–4 years 4–8 years 8–12 years 12–16 years .16 years Bone age (years) Based on hand–wrist radiographs Based on DXA hand scans

Male (n548)

Female (n547)

139 (94–202) 34 (13–80) 10.4 (2.6–18.1)

144 (97–181) 42 (13–85) 10.3 (4.4–17.3)

2 (4.2) 14 (29.2) 13 (27.1) 15 (31.3) 4 (8.3)

0 (0) 6 (12.8) 23 (48.9) 16 (34.0) 2 (4.3)

11.6 (2.5–17.0) 11.3 (2.5–17.0)

11.4 (3.4–16.3) 11.1 (3.0–16.0)

Values reflect the median (total range) or absolute number (percentage).

Both the radiographs and the DXA scans were assessed using optimal brightness and contrast, which could be adjusted by the observers. In case a dissociated bone development was found between the carpals and the other regions, we focused more on the bone age assigned to the epiphyses of the radius, ulna, metacarpals and phalanges than that assigned to the carpals [26, 27]. We defined dissociated bone development as more than a 1-year difference between the epiphyses of the radius, ulna, metacarpals and phalanges compared with the carpals. We incorporated these adjustments to the Greulich and Pyle method, which has recently been shown to be a valid method in Dutch Caucasian children [28].

Statistical analysis To compare the two methods of assessing bone age, we used the statistical methods described by Bland and Altman [23, 24]. Firstly, we calculated the intra-observer and interobserver variability using the intraclass and interclass correlation coefficient (ICC) with a 95% confidence interval (CI). In advance, we decided to consider the assessments valid when ICC .0.90. Also, we used Bland and Altman analysis to investigate the difference within the observers and between the two observers.

Secondly, we plotted all X-ray and DXA assessments against the line of equality, which demonstrates the degree of agreement between the two methods. Then, we used Bland and Altman analyses to visualise the difference between the two methods, and its distribution.

We calculated the mean and the standard deviation (SD) of the difference to estimate the limits of agreement.

In advance, we decided to accept the mean difference between both methods to deviate a maximum of 5% from the mean of both methods. Furthermore, we decided to accept the limits of agreement to be within a range of ¡1 year. Statistical analysis was performed using the Statistical Package for the Social Sciences version 15.0 for Windows (SPSS, Chicago, IL).

Results In total, 97 (93%) patients agreed to participate in our study. Two patients were excluded because of major 116

movement artefacts on their DXA scan images. Thus, we were able to assess 95 patients. Patient characteristics and their medical indications are shown in Table 1 and Table 2, respectively. The mean chronological age was 10.4 years, similar for boys and girls. In total, 94% of all children were between the ages of 4 and 16 years.

Table 3 presents all intra-observer and interobserver ICCs for both observations. Intra-observer ICCs were 0.997 and 0.991 for the X-ray assessments and 0.993 and 0.987 for the DXA assessments for observers 1 and 2, respectively. The interobserver ICC was 0.993 for the X-ray and 0.991 for the DXA assessments. The intraobserver and interobserver variability of the second observation were similar.

The means for all the intra-observer and interobserver differences with the limits of agreement (mean¡1.96 SD) are listed in Table 4. Values are based on Bland and Altman analyses. The observed mean differences for intra-observer differences ranged from 0.03 (0.78%) to 0.06 (0.88%) years and for interobserver differences from 0.07 (0.91%) to 20.11 (1.51%) years.

Figure 2 shows a plot of the X-ray and DXA assessments against the line of equality. All paired assessment points, within a wide range, lie close to the line of equality, indicating good agreement and suggesting small differences between the methods. Moreover, all points seem to lie randomly around this line, indicating an apparent lack of systematic bias.

Table 2. Medical indications for bone age assessment (n595)

Abstinence syndrome Adrenogenital syndrome Short/tall stature Galactosaemia Growth hormone insensitivity Hypogonadism Hypochondroplasia Hypopituitarism Leri–Weill syndrome Obesity (morbid) Pubertas praecox/tarda Silver–Russell syndrome Turner syndrome

1 (1.1) 8 (8.4) 37 (39.0) 1 (1.1) 3 (3.2) 2 (2.1) 1 (1.1) 17 (17.9) 1 (1.1) 2 (2.2) 9 (9.5) 1 (1.1) 12 (12.6)

Values reflect the absolute number (percentage).

The British Journal of Radiology, February 2012

29 Bone age assessment by DXA: an alternative?

Table 3. Intra- and interobserver intraclass correlation coefficient and 95% confidence interval for all X-ray and dualenergy X-ray absorptiometry (DXA) assessments Observer 1 2 Observer 1 2

Method X-ray DXA X-ray DXA Method X-ray DXA X-ray DXA

Intraobserver ICC (95% CI) 0.997 (0.995, 0.998) 0.993 (0.989,0.995) 0.991 (0.987,0.994) 0.987 (0.981,0.991) Interobserver ICC (95% CI) 0.993 (0.989,0.995) 0.991 (0.983,0.992) 0.992 (0.987,0.994) 0.991 (0.987,0.994)

CI, confidence interval; ICC, infraclass correlation coefficient.

Figure 3 shows the Bland and Altman plot in which, for each subject, the difference between the X-ray and DXA assessments is plotted against the mean of these Xray and DXA assessments. This figure applies to the mean of all X-ray assessments and the mean of all DXA assessments. Separate analyses of both observers are shown in Figures 4 and 5. The limits of agreement (mean¡1.96 SD) are plotted in the figure. Differences between X-ray and DXA assessments were normally distributed.

Following this Bland and Altman plot, the mean difference between the X-ray and DXA assessments was 0.11 (1.95%) years with corresponding limits of agreement of 20.82 and 1.05 years (Table 4). The mean difference did not significantly differ from zero, indicating lack of systematic differences. Results for each observer are demonstrated in Table 5.

Discussion We observed high intra- and interobserver correlations for both the DXA and the X-ray method and high agreement between bone age assessments performed by DXA and X-ray. The Bland and Altman plots, as well as the simple plot of one method against the other, visualised very high agreement between both methods.

The mean difference between the methods did not deviate more than 5% from the mean of both methods, which we defined prior to the study to be the maximum

acceptable difference. The limits of agreement were around the defined 21 and 1-year limit. Our results suggest that both methods assess bone age with a very small difference and that 95% of all coupled assessments did not differ by more than 1 year. According to this level of agreement, the DXA method produces similar results to the common X-ray method.

To our knowledge, only one previous study compared bone age assessment performed by X-ray and DXA [20].

This study was conducted in 2004 in 50 Polish children (aged 5–18 years) free from any chronic diseases and 10 (aged 8–20 years) with multihormonal pituitary deficiency. The authors used a different type of DXA scan (Expert-XL Densitometer; General Electric, formerly Lunar Corp.) and another reference method to assess bone age, which was more applicable to the Polish population [25]. DXA hand scans and classical hand– wrist radiographs were evaluated by two independent observers. They described a high correlation and no significant difference between mean bone age based on radiographs and DXA hand scans. Likewise, they concluded that the DXA scan produces similar results to the classical method. However, their conclusion was based on correlation coefficients and t-test analyses, whereas in measurement studies a more accurate statistical method to apply is the Bland and Altman analysis [24]. Our study and their study used different statistical analyses, but both suggested good agreement.

Another strength of our study is that our study population covers a wide range of different ages and medical indications. We were able to compare the DXA method over a wide range of ages in children who have indications for an X-ray bone age assessment in clinical practice. By including children who were already planned to have a bone age assessment by X-ray for a medical indication, we avoided a substantial amount of extra exposure to radiation to healthy children. We do not consider the hospital-based population instead of a community-based population to be a major limitation, because in clinical practice only children with these medical indications have bone age assessments. Also, because our results were based on comparisons within each subject, we expect our results to be valid for other populations. However, this needs further study.

Table 4. Variation in intra-observer assessments and interobserver assessments and between methods in absolute terms and in proportions of the mean, with the limits of agreement Assessment

Mean difference Years

Intraobserver Observer 1 Observer 2 Interobserver Observation 1 Observation 2 Between methods Mean

95% limits of agreement (years) Percentage

Lower limit

Upper limit

X-ray DXA X-ray DXA

0.04 0.06 0.05 0.03

0.69 0.88 0.70 0.78

20.72 21.17 21.22 21.59

0.80 1.29 1.33 1.65

X-ray DXA X-ray DXA

20.11 0.09 20.09 0.07

1.51 1.01 1.50 0.91

21.29 21.42 21.35 21.26

1.07 1.61 1.16 1.40

0.11

1.95

20.82

1.05

DXA, dual-energy X-ray absorptiometry.

Values are based on Bland and Altman plots [23, 24].

The British Journal of Radiology, February 2012

117

30 D H M Heppe, H R Taal, G D S Ernst et al

Figure 2. Bone age assessed by Xray and dual-energy X-ray absorptiometry (DXA) scan, with the line of equality.

A drawback of using the DXA scan in bone age assessments is that it is a more time-consuming procedure.

The scan lasts 66 s whereas an X-ray examination takes less than 1 s. This might be relevant for movement artefacts in children. Although there might be time-saving opportunities in patients who also need a total-body DXA scan, overall, the DXA scan remains a more time-consuming method. Because assessment of logistics and cost-effectiveness was not part of our study, this needs to be further investigated. A limitation of this study may be the possibility of recall bias, a general issue in intra-observer studies. By first assessing bone age in all X-ray and DXAs before we started with the repeated assessment (interval .1 week), we avoided recall bias as much as possible. If present, recall bias would have affected only the intraobserver variability. We do not expect that this is the case.

For the bone age assessment, we used the Greulich and Pyle reference method. For that reason, we are unsure whether these study results will also apply to the Tanner and Whitehouse method. Since the introduction of the Tanner and Whitehouse standards, many studies have been accomplished to compare the validity of both reference methods [29–33]. Currently, there is no overall agreement on preference of method. It has been claimed that the Tanner and Whitehouse method produces slightly more precise results, but this was shown only in one of the studies [33]. The Greulich and Pyle method has considerable practical advantages.

Probably because this method is less time-consuming and requires less specific training, it is still the most commonly used reference method in clinical practice [33].

Figure 3. Bland and Altman plot of the variation between X-ray and dualenergy X-ray absorptiometry assessments. The mean indicates the mean difference, with the 95% limits of agreement. SD, standard deviation.

118

The British Journal of Radiology, February 2012

31 Bone age assessment by DXA: an alternative?

Figure 4. Bland and Altman plot of the variation between X-ray and dual-energy X-ray absorptiometry assessments: observer 1. SD, standard deviation.

Automatic assessment of bone age is a highly innovative method. It has recently been compared with manual Greulich and Pyle assessments and has shown to produce similar results [33]. A major advantage of this automatic assessment of bone age is absence of intra- and interobserver variability. Further research is needed to investigate whether this automatic assessment of bone age is also applicable to the DXA scan method.

Conclusion DXA seems to be an alternative method for assessing bone age in a common paediatric hospital-based population. The major advantage of this method compared with the classical method is lower exposure to radiation.

Results of this method are of similar accuracy to those obtained by X-ray. Further studies are needed to investigate the cost-effectiveness.

Figure 5. Bland and Altman plot [23, 24] of the variation between X-ray and dual-energy X-ray absorptiometry assessments: observer 2.

The British Journal of Radiology, February 2012

119

32 D H M Heppe, H R Taal, G D S Ernst et al Table 5. Variation between methods in absolute terms and in proportions of the mean, with the limits of agreement for each observer Assessment

Between methods Observer 1, mean Observer 2, mean

Mean difference

95% limits of agreement (years)

Years

Percentage

Lower limit

Upper limit

0.02 0.21

0.73 3.21

20.85 21.16

0.89 1.57

Values are based on Bland and Altman plots [23, 24].

Acknowledgments We gratefully acknowledge the contribution of all children who participated in the study. We also thank General Electric, formerly Lunar Corp.

References 1. Gilli G. The assessment of skeletal maturation. Horm Res 1996;45:49–52.

2. Mora S, Boechat MI, Pietka E, Huang HK, Gilsanz V. Skeletal age determinations in children of European and African descent: applicability of the Greulich and Pyle standards. Pediatr Res 2001;50:624–8.

3. Groell R, Lindbichler F, Riepl T, Gherra L, Roposch A, Fotter R. The reliability of bone age determination in central European children using the Greulich and Pyle method. Br J Radiol 1999;72:461–4.

4. Sproul A, Peritz E. Assessment of skeletal age in short and tall children. Am J Phys Anthropol 1971;35:433–9.

5. Bierich JR. [Significance of bone age estimation to the clinic (author’s transl)]. Die Bedeutung der radiologischen Skeletalterbestimmung fur die Klinik. Radiologe 1976;16: 381–90.

6. Heinrich UE. [Significance of radiologic skeletal age determination in clinical practice]. Die Bedeutung der radiologischen Skelettalterbestimmung fur die Klinik.

Radiologe 1986;26:212–15.

7. Bramswig JH, Fasse M, Holthoff ML, von Lengerke HJ, von Petrykowski W, Schellong G. Adult height in boys and girls with untreated short stature and constitutional delay of growth and puberty: accuracy of five different methods of height prediction. J Pediatr 1990;117:886–91.

8. Raiti S, Moore WV, Van Vliet G, Kaplan SL. Growthstimulating effects of human growth hormone therapy in patients with Turner syndrome. J Pediatr 1986;109:944–9.

9. Kaufman FR, Sy JP. Regular monitoring of bone age is useful in children treated with growth hormone. Pediatrics 1999;104:1039–42.

10. Fahmy JL, Kaminsky CK, Kaufman F, Nelson MD Jr, Parisi MT. The radiological approach to precocious puberty. Br J Radiol 2000;73:560–7.

11. Greulich WW, Pyle ID. Radiographic atlas of skeletal development of the hand and wrist, 2nd edn. Stanford, CA: Stanford University Press, 1959.

12. Shu XO, Potter JD, Linet MS, Severson RK, Han D, Kersey JH, et al. Diagnostic X-rays and ultrasound exposure and risk of childhood acute lymphoblastic leukemia by immunophenotype. Cancer Epidemiol Biomarkers Prev 2002; 11:177–85.

13. Berrington de Gonzalez A, Darby S. Risk of cancer from diagnostic X-rays: estimates for the UK and 14 other countries. Lancet 2004;363:345–51.

14. Hall EJ, Brenner DJ. Cancer risks from diagnostic radiology.

Br J Radiol 2008;81:362–78.

15. Wakeford R. Childhood leukaemia following medical diagnostic exposure to ionizing radiation in utero or after birth. Radiat Prot Dosimetry 2008;132:166–74.

120

16. Schulze-Rath R, Hammer GP, Blettner M. Are pre- or postnatal diagnostic X-rays a risk factor for childhood cancer? A systematic review. Radiat Environ Biophys 2008;47:301–12.

17. Linet MS, Kim KP, Rajaraman P. Children’s exposure to diagnostic medical radiation and cancer risk: epidemiologic and dosimetric considerations. Pediatr Radiol 2009;39:S4–26.

18. Hall EJ. Radiation biology for pediatric radiologists. Pediatr Radiol 2009;39:S57–64.

19. ICRP. The 2007 Recommendations of the International Commission on Radiological Protection. ICRP publication 103. Ann ICRP 2007;37:1–332.

20. Pludowski P, Lebiedowski M, Lorenc RS. Evaluation of the possibility to assess bone age on the basis of DXA derived hand scans—preliminary results. Osteoporos Int 2004;15:317–22.

21. Njeh CF, Fuerst T, Hans D, Blake GM, Genant HK. Radiation exposure in bone mineral density assessment.

Appl Radiat Isot 1999;50:215–36.

22. Mettler FA Jr, Huda W, Yoshizumi TT, Mahesh M. Effective doses in radiology and diagnostic nuclear medicine: a catalog. Radiology 2008;248:254–63.

23. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement.

Lancet 1986;1:307–10.

24. Bland JM, Altman DG. Applying the right statistics: analyses of measurement studies. Ultrasound Obstet Gynecol 2003;22:85–93.

25. Kopczynska-Sikorska J. Atlas radiologiczny rozwoju koscca dłoni i nadgarstka. Pan´stwowy Wydawnictw Lekarskich: Warsaw, Poland, 1969.

26. Acheson RM, Vicinus JH, Fowler GB. Studies in the reliability of assessing skeletal maturity from X-rays. Ii. The bone-specific approach. Hum Biol 1964;36:211–28.

27. Roche AF, Davila GH. The reliability of assessments of the maturity of individual hand–wrist bones. Hum Biol 1976; 48:585–97.

28. van Rijn RR, Lequin MH, Robben SG, Hop WC, van Kuijk C. Is the Greulich and Pyle atlas still valid for Dutch Caucasian children today? Pediatr Radiol 2001;31:748–52.

29. Acheson RM, Vicinus JH, Fowler GB. Studies in the reliability of assessing skeletal maturity from X-rays. 3.

Greulich–Pyle Atlas and Tanner–Whitehouse method contrasted. Hum Biol 1966;38:204–18.

30. Andersen E. Comparison of Tanner–Whitehouse and Greulich–Pyle methods in a large scale Danish Survey.

Am J Phys Anthropol 1971;35:373–6.

31. Milner GR, Levick RK, Kay R. Assessment of bone age: a comparison of the Greulich and Pyle, and the Tanner and Whitehouse methods. Clin Radiol 1986;37:119–21.

32. King DG, Steventon DM, O’Sullivan MP, Cook AM, Hornsby VP, Jefferson IG, et al. Reproducibility of bone ages when performed by radiology registrars: an audit of Tanner and Whitehouse II versus Greulich and Pyle methods. Br J Radiol 1994;67:848–51.

33. Bull RK, Edwards PD, Kemp PM, Fry S, Hughes IA. Bone age assessment: a large scale comparison of the Greulich and Pyle, and Tanner and Whitehouse (TW2) methods.

Arch Dis Child 1999;81:172–3.

The British Journal of Radiology, February 2012

33 The British Journal of Radiology, 85 (2012), e270–e273

Regional grey and white matter volumetric changes in myalgic encephalomyelitis (chronic fatigue syndrome): a voxel-based morphometry 3 T MRI study 1,2

B K PURI, PhD, FRCPsych, 3P M JAKEMAN, MSc, PhD, 4M AGOUR, MB, MRCPsych, , K D R GUNATILAKE, MD, MRCPsych, 6K A C FERNANDO, MBBS, MRCPsych, 7A I GURUSINGHE, MBBS, PGDPsych, 8 I H TREASADEN, MRCS, FRCPsych, 1,2,9A D WALDMAN, PhD, MRCP and 1,2P GISHEN, DMRD, FRCR 5

1

Department of Imaging, Hammersmith Hospital, London, UK, 2Imperial College London, London, UK, 3Department of Physical Education and Sport Sciences, University of Limerick, Limerick, Ireland, 4University of Hertfordshire, and Care Principles, Langley, UK, 5The Ridge Hill Centre, Dudley, UK, 6Brooklands Hospital, Birmingham, UK, 7Broadmoor Hospital, Crowthorne, UK, 8Three Bridges Unit, West London Mental Health Trust, Southall, UK, and 9National Hospital for Neurology and Neurosurgery, London, UK

Objective: It is not established whether myalgic encephalomyelitis/chronic fatigue syndrome (CFS) is associated with structural brain changes. The aim of this study was to investigate this by conducting the largest voxel-based morphometry study to date in CFS. Methods: High-resolution structural 3 T cerebral MRI scanning was carried out in 26 patients with CFS and 26 age- and gender-matched healthy volunteers. Voxel-wise generalised linear modelling was applied to the processed MR data using permutationbased non-parametric testing, forming clusters at t.2.3 and testing clusters for significance at p,0.05, corrected for multiple comparisons across space.

Results: Significant voxels (p,0.05, corrected for multiple comparisons) depicting reduced grey matter volume in the CFS group were noted in the occipital lobes (right and left occipital poles; left lateral occipital cortex, superior division; and left supracalcrine cortex), the right angular gyrus and the posterior division of the left parahippocampal gyrus. Significant voxels (p,0.05, corrected for multiple comparisons) depicting reduced white matter volume in the CFS group were also noted in the left occipital lobe.

Conclusion: These data support the hypothesis that significant neuroanatomical changes occur in CFS, and are consistent with the complaint of impaired memory that is common in this illness; they also suggest that subtle abnormalities in visual processing, and discrepancies between intended actions and consequent movements, may occur in CFS.

Myalgic encephalomyelitis, or chronic fatigue syndrome (CFS), as defined by the revised diagnostic criteria of the Centers for Disease Control and Prevention, is mainly characterised by persistent or relapsing fatigue lasting for at least 6 consecutive months [1]. As the aetiology of the disorder is currently unknown, it is important to establish whether it is associated with cerebral abnormalities; however, MRI has provided conflicting results when used to search for brain abnormalities in sufferers [2].

A recent, large British MRI study by Perrin et al [2] of 18 CFS patients and 9 healthy volunteers, in which the images were examined for abnormalities in brain atrophy, deep white matter hyperintensities, and cerebral blood and cerebrospinal fluid flow, reported no significant differences in brain structure between the 2 groups at either baseline or 1-year follow-up, with the authors concluding that ‘‘These results throw open the debate into whether MRI scanning can reveal diagnostic signs of CFS Address correspondence to: Professor Basant Puri, Department of Imaging, Hammersmith Hospital, Du Cane Road, London W12 0HS, UK. E-mail: basant.puri@imperial.ac.uk We should like to thank the MRC for funding this study.

e270

Received 14 November 2010 Revised 14 December 2010 Accepted 25 January 2011 DOI: 10.1259/bjr/93889091 ’ 2012 The British Institute of Radiology

and clinically questions the diagnoses of CFS made on the basis of previous research conclusions.’’

A small number of previous cerebral MRI studies have been conducted in CFS. A 1993 study involving the comparison by two radiologists of the scans of CFS patients and of controls whom had undergone imaging because of histories of head trauma or headache reported that the former had significantly more abnormal scans than controls (27% vs 2%) [3]; abnormalities included foci of increased white matter T2 signal in 17% of the CFS patients and ventricular or sulcal enlargement in 10%. On the other hand, a 1997 study of white matter abnormalities found no significant difference between CFS patients and controls [4]. A 1999 study involving the comparison of MR scans by two to three radiologists found, overall, no significant differences between CFS patients and healthy controls, although those CFS patients without a psychiatric diagnosis since illness onset had more brain abnormalities on T2 weighted images (mostly small, punctate, subcortical white matter hyperintensities, predominantly in the frontal lobes) than patients with such a diagnosis [5].

Brain MRI analysis using voxel-based morphometry offers advantages over the methodologies used in the The British Journal of Radiology, July 2012

34 A voxel-based morphometry study of myalgic encephalomyelitis

above studies. It is an objective method that is not operator dependent and that does not require a priori information about the location of possible differences between groups.

The technique involves spatially normalising all the MR images to the same stereotactic space (by registering each of the images to the same template image, by minimising the residual sum of squared differences between them), segmenting the grey matter from the normalised images, correcting for volume changes arising from spatial normalisation and, finally, carrying out a statistical analysis to localise differences between groups; the output from the method is a statistical parametric map that shows regions where grey matter concentration differs significantly between groups [6, 7].

Thus far, just one voxel-based morphometry study of CFS has been published. In this 2004 Japanese study of 16 CFS patients, reduced grey matter volume was reported in the bilateral prefrontal cortex [8]. This represents the first report of focal grey matter atrophy in the prefrontal cortex of CFS patients. There have been no attempts, until now, to replicate this finding.

Here, we report the largest voxel-based morphometry study of the brain in CFS.

Methods and materials Subjects

these rescaled structural images were brain-extracted using the brain extraction tool [11]. Next, tissue-type segmentation was carried out using The Oxford Centre for Functional MRI of the Brain automated segmentation tool [12]. The resulting grey matter partial volume images were then aligned to MNI152 standard space using affine registration [13]. The resulting images were averaged to create a study-specific template, to which the native grey matter images were then non-linearly reregistered. The registered partial volume images were then modulated (to correct for local expansion or contraction owing to the non-linear component of the transformation) by dividing by the Jacobian of the warp field. The modulated segmentated images were then smoothed with an isotropic gaussian kernel with a sigma of 3 mm.

Statistical analyses Statistical analyses were carried out using SPSS v. 16 statistical program (SPSS Inc., Chicago, IL). In the voxelbased morphometry analysis, randomised testing with 5000 permutations was used for statistical inference.

Voxel-wise generalised linear modelling was applied using permutation-based non-parametric testing, forming clusters at t.2.3 and testing clusters for significance at p,0.05, corrected for multiple comparisons across space. Significant clusters were then overlaid on the MNI152 template.

26 patients and 26 normal controls underwent cerebral structural MRI. All the patients met the revised diagnostic criteria for CFS of the Centers for Disease Control and Prevention [1], with a mean duration of symptoms of 10.9 [standard error (SE) 1.7] years; none of the healthy controls met the CFS criteria, nor did they suffer from undue fatigue or from any history of neurological or psychiatric disorder.

The study was carried out according to the Declaration of Helsinki. Each subject gave written informed consent. The study had research ethics committee approval.

The mean (ยกSE) age of the patients (42.9ยก2.2 years) did not differ significantly from that of the healthy controls (38.2ยก2.2 years, p.0.05). The male-to-female ratio of the patients (7:19) also did not differ significantly from that of the controls (13:13, p.0.05).

Imaging

Voxel-wise analyses

High-resolution three-dimensional T1 weighted turbo field echo anatomical images of the brain were acquired on all subjects using the same 3 T Philips Achieva system (Philips Healthcare, Best, the Netherlands) as a series of 150 sagittal slices [1.15 mm slice thickness, 2086208 matrix, repetition time (TR)50.7 ms, echo time (TE) 4.6 ms (inphase), flip angle58u, maximum water-fat shift, turbo field echo shot interval51779.5 ms, minimum time interval (TI) delay5562.3 ms] with a Cartesian acquisition mode, linear profile order and using cerebral grey matter as the reference tissue; specific absorption rate ,0.5 W kg21.

Significant voxels (p,0.05, corrected for multiple comparisons) depicting reduced grey matter volume in the CFS group compared with the control group were noted in the occipital lobes (right and left occipital poles; left lateral occipital cortex, superior division; and left supracalcrine cortex); the right angular gyrus; and the left parahippocampal gyrus, posterior division. Significant voxels (p,0.05, corrected for multiple comparisons) depicting reduced white matter volume in the CFS group were also noted in the left occipital lobe. All these anatomical locations were confirmed by an operator-independent electronic atlas, namely the Harvard-Oxford Cortical Structural Atlas. Figure 1 shows p-value maps in which some of these significant clusters, corrected for multiple comparisons, have been overlaid on the MNI152 template.

Voxel-based morphometry protocol The structural data were analysed with FSL-VBM (The Oxford Centre for Functional MRI of the Brain, Oxford, UK), a voxel-based morphometry-style analysis [6, 9] carried out with The Oxford Centre for Functional MRI of the Brain software library tools [10]. First, the values in the T1 images were scaled to lie between 0 and 10 000, and The British Journal of Radiology, July 2012

Results Subjects

Discussion This largest voxel-based morphometry study of CFS has shown evidence of reduced grey and white matter e271

35 B K Puri, P M Jakeman, M Agour et al

Figure 1. A p-value map in which significant clusters, corrected for multiple comparisons, have been overlaid on the MNI152 template. The upper left panel shows the coronal plane, the upper right panel the sagittal plane and the lower panel the transverse plane.

volume in the occipital lobes, as well as reduced grey matter in the right angular gyrus and the left parahippocampal gyrus. Thus, in contrast to the recent report in this journal by Perrin et al [2] in Manchester, our study confirms that, using voxel-based morphometry, there are indeed significant neuroanatomical changes in CFS. We have not replicated the findings of the first voxelbased morphometry study [8]; this may be (at least partly) a result of the fact that we studied 63% more CFS patients in our voxel-based morphometry analysis.

The occipital lobe is well established as being a part of the brain involved in visual processing. The angular gyrus is a circumscribed area strategically positioned between the parietal and temporal lobes, and close to the occipital lobe; its functions have been unclear until recently [14]. The right angular gyrus has now been shown to have a critical role in perceptual sequence learning [15]. It also computes action awareness representations; in particular, it is associated with both awareness of discrepancy between intended action and movement consequences, and awareness of action authorship [16]. Farrer et al [16] have proposed that the right angular gyrus is involved in higher-order aspects of motor control that allow one consciously to access different aspects of one’s own actions; specifically, it processes discrepancies between intended action and movement consequences in such a way that these will be consciously detected by the subject: this joint processing is at the core of experiences used to interpret one’s actions. On the basis of our results in the occipital lobes and right angular gyrus, we would suggest that subtle abnormalities in visual processing, and discrepancies between intended actions and consequent movements, should be investigated in CFS patients.

The parahippocampal gyrus is important in such mnemonic functions as encoding and retrieval. The fact that we found reduced grey matter in the posterior part of the left parahippocampal gyrus is of particular interest in light of a recent MRI study by Burgmans et al [17] showing that the posterior parahippocampal gyrus is e272

preferentially affected in age-related memory decline.

Impaired memory is indeed recognised as a clinical feature of CFS. The potential implications of these findings in respect of the aetiology and pathophysiology of myalgic encephalomyelitis/CFS is of interest. This disease is of unknown aetiology, but recently there has been renewed speculation that it may be related to persistent viral infection.

Such infections are likely to impair the ability of the body to biosynthesise n-3 and n-6 long-chain polyunsaturated fatty acids by inhibiting the D-6 desaturation of the precursor essential fatty acids a-linolenic acid and linoleic acid, which would in turn impair the proper functioning of cell membranes, including cell signalling, and have an adverse effect on the biosynthesis of eicosanoids from the long-chain polyunsaturated fatty acids dihomo-c-linolenic acid, arachidonic acid and eicosapentaenoic acid [18].

The resulting reduced availability of appropriate longchain polyunsaturated fatty acids at the Sn2 position of phospholipid molecules might be expected to be associated with a reduced rate of anabolism of membrane phospholipids, and therefore a relative increase in the level of free choline, which would otherwise form the polar head groups of some of these phospholipid molecules.

Interestingly, an increase in the choline resonance has been reported in systematic controlled proton MR neurospectroscopy studies in this patient group [19, 20]. Moreover, the first of these studies specifically reported an increased level in the occipital lobe, which is consistent with our present finding of reduced grey and white matter volume in this part of the brain.

The related issue also arises as to whether the present study would have produced similar results if the mean duration of symptoms in the patients had been much less than 10.9 years. Clearly, a longitudinal study would be an appropriate way of addressing this issue. In the absence of such data, however, it may still be possible to shed some light on the likely answer by examining the published results of a non-systematic proton neurospectroscopy study of choline resonances in myalgic encephalomyelitis/ The British Journal of Radiology, July 2012

36 A voxel-based morphometry study of myalgic encephalomyelitis

CFS of short duration. Tomoda et al [21] examined three cases (11-, 12- and 13-year-old children) with illness of relatively recent origin; all three children showed marked elevations in the choline resonances. These results would tend to suggest that the brain changes are of relatively early onset and that, accordingly, our study might have produced similar results if the mean duration of symptoms had been much less than 10.9 years.

In summary, the present study supports the hypothesis that CFS patients have structural abnormalities in the brain [22].

9.

10.

11.

12.

Acknowledgments We also wish to acknowledge the involvement of several myalgic encephalomyelitis/CFS charities, including ME Research UK (formerly known as MERGE) and ME Solutions. We are very grateful to the patients, their families and the healthy volunteers who took part in this study.

13.

14.

15.

References 1. Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff A, et al. The chronic fatigue syndrome: a comprehensive approach to its definition and study. Ann Intern Med 1994;121:953–9.

2. Perrin R, Embleton K, Pentreath VW, Jackson A. Longitudinal MRI shows no cerebral abnormality in chronic fatigue syndrome. Br J Radiol 2010;83:419–23.

3. Natelson BH, Cohen JM, Brassloff I, Lee HJ. A controlled study of brain magnetic resonance imaging in patients with the chronic fatigue syndrome. J Neurol Sci 1993;120:213–17.

4. Greco A, Tannock C, Brostoff J, Costa DC. Brain MR in chronic fatigue syndrome. AJNR Am J Neuroradiol 1997;18:1265–9.

5. Lange G, DeLuca J, Maldjian JA, Lee H, Tiersky LA, Natelson BH. Brain MRI abnormalities exist in a subset of patients with chronic fatigue syndrome. J Neurol Sci 1999;171:3–7.

6. Ashburner J, Friston KJ. Voxel-based morphometry–the methods. Neuroimage 2000;11:805–21.

7. May A, Ashburner J, Buchel C, McGonigle DJ, Friston KJ, Frackowiak RS, et al. Correlation between structural and functional changes in brain in an idiopathic headache syndrome. Nat Med 1999;5:836–8.

8. Okada T, Tanaka M, Kuratsune H, Watanabe Y, Sadato N. Mechanisms underlying fatigue: a voxel-based morphometric

The British Journal of Radiology, July 2012

16.

17.

18.

19.

20.

21.

22.

study of chronic fatigue syndrome. BMC Neurol 2004;4: 14.

Good CD, Johnsrude IS, Ashburner J, Henson RN, Friston KJ, Frackowiak RS. A voxel-based morphometric study of ageing in 465 normal adult human brains. Neuroimage 2001;14:21–36.

Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TE, Johansen-Berg H, et al. Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 2004;23:S208–19.

Smith SM. Fast robust automated brain extraction. Hum Brain Mapp 2002;17:143–55.

Zhang Y, Brady M, Smith S. Segmentation of brain MR images through a hidden Markov random field model and the expectation-maximization algorithm. IEEE Trans Med Imaging 2001;20:45–57.

Rueckert D, Sonoda LI, Hayes C, Hill DL, Leach MO, Hawkes DJ. Nonrigid registration using free-form deformations: application to breast MR images. IEEE Trans Med Imaging 1999;18:712–21.

Kombos T, Picht T, Suess O. Electrical excitability of the angular gyrus. J Clin Neurophysiol 2008;25:340–5.

Rosenthal CR, Roche-Kelly EE, Husain M, Kennard C. Response-dependent contributions of human primary motor cortex and angular gyrus to manual and perceptual sequence learning. J Neurosci 2009;29:15115–25.

Farrer C, Frey SH, Van Horn JD, Tunik E, Turk D, Inati S, et al. The angular gyrus computes action awareness representations. Cereb Cortex 2008;18:254–61.

Burgmans S, van Boxtel MP, van den Berg KE, Gronenschild EH, Jacobs HI, Jolles J, et al. The posterior parahippocampal gyrus is preferentially affected in agerelated memory decline. Neurobiol Aging 2009;32:1572–8.

Puri BK. Long-chain polyunsaturated fatty acids and the pathophysiology of myalgic encephalomyelitis (chronic fatigue syndrome). J Clin Pathol 2007;60:122–4.

Puri BK, Counsell SJ, Zaman R, Main J, Collins AG, Hajnal JV, et al. Relative increase in choline in the occipital cortex in chronic fatigue syndrome. Acta Psychiatr Scand 2002; 106:224–6.

Chaudhuri A, Condon BR, Gow JW, Brennan D, Hadley DM. Proton magnetic resonance spectroscopy of basal ganglia in chronic fatigue syndrome. Neuroreport 2003; 14:225–8.

Tomoda A, Miike T, Yamada E, Honda H, Moroi T, Ogawa M, et al. Chronic fatigue syndrome in childhood. Brain Dev 2000;22:60–4.

Chen R, Liang FX, Moriya J, Yamakawa J, Sumino H, Kanda T, et al. Chronic fatigue syndrome and the central nervous system. J Int Med Res 2008;36:867–74.

e273

37 The British Journal of Radiology, 85 (2012), e694–e701

Prognostic significance of SUV on PET/CT in patients with localised oesophagogastric junction cancer receiving neoadjuvant chemotherapy/chemoradiation: a systematic review and meta-analysis 1

W ZHU,

MD,

1

L XING,

MD,

1

J YUE,

MD,

2

X SUN,

MD,

1

X SUN,

MD,

1

H ZHAO,

MD

and 1J YU,

MD, PHD

1

Department of Radiation Oncology, Shandong Cancer Hospital, Shandong Academy of Medical Sciences, Provincial Key Laboratory of Rodiation Oncology, Jinan, China, and 2Department of Nuclear Medicine, Shandong Cancer Hospital, Shandong Academy of Medical Sciences, Provincial Key Laboratory of Radiation Oncology, Jinan, China

Objective: The objective of this study was to comprehensively review the evidence for use of pre-treatment, post-treatment and changes in tumour glucose uptake that were assessed by 18-fludeoxyglucose (18F-FDG) positron emission tomography (PET) early, during or immediately after neoadjuvant chemotherapy/chemoradiation to predict prognosis of localised oesophagogastric junction (AEG) cancer.

Methods: We searched for articles published in English; limited to AEG; 18F-FDG uptake on PET performed on a dedicated device; dealt with the impact of standard uptake value (SUV) on survival. We extracted an estimate of the log hazard ratios (HRs) and their variances and performed meta-analysis.

Results: 798 patients with AEG were included. And the scan time for 18F-FDG-PET was as follows: prior to therapy (PET1, n5646), exactly 2 weeks after initiation of neoadjuvant therapy (PET2, n5245), and pre-operatively (PET3, n5278). In the two meta-analyses for overall survival, including the studies that dealt with reduction of tumour maximum SUV (SUVmax) (from PET1 to PET2/PET3 and from PET1 to PET2), the results were similar, with the overall HR for non-responders being 1.83 [95% confidence interval (CI), 1.41–2.36] and 2.62 (95% CI, 1.61–4.26), respectively; as for disease-free survival, the combined HR was 2.92 (95% CI, 2.08–4.10) and 2.39 (95% CI, 1.57–3.64), respectively. The meta-analyses did not attribute significant prognostic values to SUVmax before and during therapy in localised AEG. Conclusion: Relative changes in FDG-uptake of AEG are better prognosticators. Early metabolic changes from PET1 to PET2 may provide the same accuracy for prediction of treatment outcome as late changes from PET1 to PET3.

It is known from several studies that the outcome of patients with adenocarcinomas of the oesophagogastric junction (AEG) treated with pre-operative chemotherapy/chemoradiation is heterogeneous. To the best of our knowledge, the reasons for this unpredictability in clinical outcome are not entirely clear but could be attributed to the differences in molecular compositions of cancers [1–5] and/or patient genetics [6, 7]. Two Phase III studies indicated that pre-operative chemotherapy improved survival in patients with oesophageal adenocarcinoma and AEG [8, 9]. However, a systematic review did show only marginal effects of pre-operative chemotherapy for resectable intrathoracic oesophageal cancer [10]. Of note, in non-responders, survival seems to be similar or even worse than after surgical resection alone [11]. Therefore, early identification of patients likely to have an unfavourable outcome after pre-operative therapy is highly important for the future use of neoadjuvant therapy in AEG. Address correspondence to: Dr Jinming Yu, Department of Radiation Oncology, Shandong Cancer Hospital, Shandong Academy of Medical Sciences, Provincial Key Laboratory of Radiation Oncology, Jiyan Road 440, Jinan 250117, Shandong Province, China. E-mail: yujmwin@126.com

e694

Received 30 March 2011 Revised 16 September 2011 Accepted 17 October 2011 DOI: 10.1259/bjr/29946900 ’ 2012 The British Institute of Radiology

The use of 18-fludeoxyglucose (18F-FDG) positron emission tomography (PET) as a metabolism-based imaging technique has been increased steadily during the last decade in most malignant tumours. The prognostic values of FDG uptake before and after chemotherapy/ chemoradiotherapy were reported, and other studies have been conducted to evaluate the prognostic value of a decrease in tumour metabolic activity on the survival of patients with AEG. However, the sample size of most studies was rather small and the results of the prognostic value of standardised uptake value (SUV) remained undetermined. Therefore, we performed a meta-analysis to assess the prognostic value of SUV for survival of patients with AEG receiving neoadjuvant chemotherapy/chemoradiation.

Methods and materials Literature search We searched the MEDLINE and Embase databases for articles published between January 1998 and January The British Journal of Radiology, September 2012

38 SUV and AEG in meta-analysis

2011, using the following terms: ‘‘esophagogastric/ oesophagogastric junction’’ or ‘‘esophagogastric/oesophagogastric junction cancer’’ or ‘‘esophagogastric/ oesophagogastric junction carcinoma’’ or ‘‘esophagogastric/oesophagogastric junction neoplasm’’; ‘‘PET imaging tomography’’ or ‘‘PET’’ or ‘‘positron emission tomography’’ or ‘‘FDG’’ or ‘‘18F-FDG’’ or ‘‘FDG-F18’’ or ‘‘fluorodeoxyglucose’’ or ‘‘18F-fluorodeoxyglucose’’; ‘‘standardized uptake value’’ or ‘‘SUV’’ or ‘‘uptake value’’ or ‘‘semiquantitation’’; and ‘‘outcome’’ or ‘‘survival’’ or ‘‘predict’’ or ‘‘prognosis’’ or ‘‘prognostic factor’’. A manual search of the cross-references for eligible articles was used to identify additional relevant articles.

Selection of studies Four investigators reviewed each article independently and scored them according to a quality scale (described in Appendix A). The methodology quality assessment consisted of four main dimensions modified on the basis of similar studies: the scientific design, the generalisability of the results, the analysis of the study data and the PET reports [12, 13]. A points value (0–2) was attributed to each item. Each dimension was worth 0–10 points up to a maximum of 40 points. As the scores were objective, a consensus was obtained in meetings with at least three-quarters of the investigators present.

Their participation guaranteed the correct interpretation of the publications. The final scores were expressed as percentages, with higher values reflecting a greater application of methodological standards. Because the scoring of quality is intrinsically subjective, the quality scores were not applied to exclude lower quality studies from the meta-analysis or to weigh the studies. The studies included in the systematic review were called ‘‘eligible’’ and those providing sufficient data for metaanalysis ‘‘evaluable’’.

Statistical methods SUV cut-off values to differentiate responders from non-responders were based on the definition used in each individual study. The correlation between the quality scores and the number of patients included in the studies was measured by the Spearman’s rank correlation coefficient, and testing a null hypothesis of equality to zero of the coefficient assessed its significance. Non-parametric tests (Mann–Whitney U-test or Kruskal–Wallis test) were applied to compare the distribution of the quality scores according to the value of a discrete variable. The effect of SUV on survival was measured by the hazard ratio (HR) between the survival distributions of two groups using the following methodology. For each study, the log HR and standard error can be calculated by extracting the unadjusted HR and confidence intervals (CIs) directly from each publication or from extracting cumulative survival data from published Kaplan–Meier plots as described by Parmar et al [14]. If authors reported survival of more than two groups, we pooled the results, making a comparison between two groups feasible.

The British Journal of Radiology, September 2012

The Q statistics were applied to test for heterogeneity among the evaluable studies. A fixed-effect model was applied to calculate the summary HRs, if there was no heterogeneity observed (i.e. Q-test p.0.05). When heterogeneity was observed, a random-effects model was used.

The I2 statistic was applied to estimate the percentage of variation across studies due to heterogeneity rather than chance. I2 can be calculated as following: I25100%6(Q d.f.)/Q

(1)

Q5Cochran’s heterogeneity statistics, d.f.5degree of freedom. We defined substantial heterogeneity within every meta-analysis as an I2 .50%. Publication bias was detected by performing the Egger test.

Studies about local control or recurrence were also included in the meta-analysis for disease-free survival (DFS). Survival rates on the graphical representation of the survival curves were read by Engauge Digitizer v. 2.5 (Trolltech, Oslo, Norway). HRs and their variations were calculated by Review Manager v. 5.0 (The Nordic Cochrane Centre, Copenhagen, Denmark). All reported p-values are two-sided and are performed at the 5% level of significance using SPSSH v. 13.0 (SPSS Inc., Chicago, IL).

Results Study selection and characteristics analysis The electronic and manual searches yielded 97 potentially eligible articles from all databases. Of these articles, 63 were eliminated because they were not about localised oesophagogastic junction carcinoma, without an outcome of interest, not full-text articles or non-English language.

The remaining 34 full-text articles were further analysed. 24 of these studies were excluded because the log HR and its variance of the oesophagogastric junction carcinoma from the total patients with oesophageal or gastric carcinoma could not be calculated (n512) [15–26], an author published two reports on the same population (n51) [27, 28], and irrelevant references including the studies without surgery such as Murthy et al (n511) [29]. Finally, a total of 10 studies were determined to be evaluable for the actual analysis [27, 30–37]. The principal characteristics of the 10 studies evaluable for the meta-analysis are described in Table 1. The scan time for 18F-FDG-PET was as follows: prior to therapy (PET1), exactly 2 weeks after initiation of neoadjuvant therapy (PET2) and pre-operatively (PET3).

Six studies dealt with the prognostic value of changes in glucose utilisation, measured by 18F-PDG-PET from PET1 to PET2 or PET3 for overall survival (OS) [30, 31, 33, 35, 37, 38]. Four studies dealt with the prognostic value of changes in maximum SUV (SUVmax) measured on 18F-PDG-PET for DFS at the same time [30, 34, 35, 37]. Moreover, the metaanalyses were performed for the studies that dealt only with changes in SUVmax from PET1 to PET2 for predictions of OS and DFS. Three studies dealt with the prognostic value of SUVmax measured on 18F-PDG-PET at PET1 [27, 32, 35] for OS, and three studies did it at PET2/PET3 for OS [31, 35, 36]. The meta-analyses that dealt only with SUV for DFS prediction at PET1 [35] and PET2/PET3 [35, 36] were excluded owing to too few articles. e695

39 W Zhu, L Xing, J Yue et al Table 1. Principal characteristics of the 10 studies included in the meta-analysis Study

Publication year

Number of patients

Weber et al [37]

2001

40

Ott et al [35] Wieder et al [33] Lordick et al [30] Roedl et al [38] Smith et al [34] Rizk et al [27] Javeri et al [32] Javeri et al [31] Patnana et al [36]

2006 2007 2007 2008 2009 2009 2009 2009 2010

56 24 104 51 21 189 161 151 152

A total of 798 patients, with a predominance of adenocarcinoma of the distal oesophagus, were included in survival analyses. SUVmax normalised by body weight was used in all studies and the mean SUV was not used.

Only five of the six studies for OS achieved definite statistical significance in addition to three of the four for DFS. Two of the studies with significant DFS results also had significant OS results [35, 39], while one study showed an undetermined effect on both OS and DFS [37]. Furthermore, in one study, the statistical result was different according to the threshold definition [33].

Table 2 shows the main SUVmax characteristics reported in each article. The response threshold in SUVmax chosen arbitrarily between patients with high and low survival was based on the SUVmax values decreasing by >35% from PET1 to PET2 in all studies. In the four studies that had the PET scan at PET3, a so-called ‘‘best cut-off’’ was used, which meant that the threshold maximised the log rank test statistic among several survival comparisons. It is known that this method may lead to a high risk of false-positive results, especially without adjustment of pvalues for multiplicity. Five studies indicated that patients with a high SUVmax had a worse OS than

Stage

Location

Methodology score (%)

Tumour stage T3/T4, NX, and M0 IIa–III cT3–4, N0/+, M1a cT3 or cT4 Not detailed Not detailed II–IVa T2N0–M1a T2N0–M1a II–IV

Types I and II

81.6

Types I and II Types I and II Types I and II Type I Not detailed Types I and II Types I, II and III Types I, II and III Types I and II

78.8 50 76.3 55.3 36.8 57.9 55.3 57.9 65.8

patients with a low SUVmax at PET1, PET2 or PET3.

However, none of the studies achieved definite statistical significance. Three studies used the median as the SUVmax threshold, and the best cut-off value was used in two studies in addition to analysing SUVmax as a continuous variable in one study [31].

Quality assessment Overall, the global quality score ranged from 36.8% to 81.6%, with a median of 57.9% (Table 1). An attempt was made to contact the authors, if necessary, to obtain missing details of methodological quality. There was a non-significant correlation between the global score and the number of patients included in the study (Spearman’s correlation coefficient, r50.250; p50.486).

Meta-analysis For OS, four meta-analyses were performed. In the first meta-analysis for OS, six studies that dealt with the

Table 2. Main SUV characteristics extracted from the 10 articles used for meta-analysis Study

Type of SUV

Weber et al [37] Ott et al [35]

Decreased SUVmax (from PET1 to PET2) Initial SUVmax (at PET1) FDG uptake at day 14 (at PET2) Decreased SUVmax (from PET1 to PET2) Decreased SUVmax (from PET1 to PET2) Decreased SUVmax (3–4 weeks after the completion of chemoradiotherapy, from PET1 to PET3) Decreased SUVmax (from PET1 to PET2) Decreased SUVmax (16.9 days¡6.8 after chemoradiotherapy, from PET1 to PET3) Decreased SUVmax (from PET1 to PET2) Initial SUVmax (at PET1) Initial SUVmax (at PET1) FDG uptake at day (12¡2 weeks, at PET3) Decreased SUVmax (12¡2 weeks after initiation of therapy, from PET1 to PET3) SUVmax (Approximately 5–6 weeks after the completion of chemoradiation, at PET3)

Wieder et al [33]

Lordick et al [30] Roedl et al [38] Smith et al [34] Rizk et al [27] Javeri et al [32] Javeri et al [31]

Patnana et al [36]

Correction of SUV

Threshold definition

SUV threshold

Weight Weight Weight Weight Weight Weight

Best cut-off Median Median Best cut-off Best cut-off Best cut-off

.35% 8.1 5.4 .35% .35% .63%

Weight Weight

Previous report Best cut-off

.35% .43%

Weight Weight Weight Weight Weight

Best cut-off Previous report Median Continuous variable Previous report

.50% 4.5 10.1 — .52%

Weight

Median

4.6

FDG, fludeoxyglucose; PET, positron emission tomography; PET1, PET scan prior to therapy; PET2, PET scan exactly 2 weeks after the initiation of neoadjuvant therapy; PET3, PET scan pre-operatively; SUV, standardised uptake value; SUVmax, maximum standardised uptake value.

— indicates that SUV was not used as a categorical variable in the article.

e696

The British Journal of Radiology, September 2012

40 SUV and AEG in meta-analysis

Figure 1. Review: meta-analyses of the studies dealt with the prognostic value of response in maximum standardised uptake value measured by fludeoxyglucose positron emission tomography (PET) (a) at least 2 weeks [from prior to therapy (PET1) to 2 weeks after initiation of neoadjuvant therapy (PET2) or preoperatively] and (b) exactly 2 weeks (from PET1 to PET2) after initiation of therapy for overall survival.

Results in (a) indicate that metabolic responders from PET1 to PET2/PET3 had a better overall survival than metabolic non-responders; results in (b) indicate that metabolic responders from PET1 to PET2 had a better overall survival than metabolic nonresponders. CI, confidence interval; HR, hazard ratio; SE, standard error.

response in SUVmax measured from PET1 to PET2 or PET3 were included (Figure 1a). The combined HR was 1.83 (95% CI, 1.41–2.36) with a fixed-effect model, meaning that metabolic responders had a better OS than metabolic non-responders. The test for heterogeneity gave no significant results (Q-test, p50.31, I2516%). The evaluation of publication bias showed that the Egger test was not significant (p50.237). The funnel plots for publication bias (Figure 2) also showed some symmetry.

The second meta-analysis for OS included the studies that dealt with the changes in tumour SUVmax measured from PET1 to PET2. The combined HR was with 2.62 (95% CI, 1.61–4.26); heterogeneity (Q-test, p50.55, I250%; Figure 1b). As for the prognostic value of SUVmax measured on 18F-PDG-PET at PET1 and PET2/PET3 for OS, the combined HRs were 1.52 (95% CI, 0.99–2.34) and 1.05 (95% CI, 0.96–1.16), respectively (Figure 3). This means that a high primary tumour SUVmax was not associated with a worse survival prognosis. Similarly, there were four studies that dealt with the prognostic value of response in primary tumour SUVmax for DFS [30, 34, 35, 37]. With the studies that dealt with the change in tumour SUVmax measured from PET1 to PET2 or PET3 (Figure 4a), the combined HR was 2.92 (95% CI, 2.08–4.10). The results of the test for heterogeneity were insignificant (Q-test, p50.30, I2518%). The evaluation of publication bias showed that the Egger test was insignificant (p50.371) in addition to the funnel plots. After excluding the studies that dealt with the response in SUVmax measured from PET1 to PET3 [30, 35, 37], the combined HR was 2.39 (95% CI, 1.57–3.64; Q-test, p50.57, I250%; Figure 4b).

Discussion Oesophagogastric carcinoma is frequently diagnosed at an advanced stage, and the number of deaths, as a result, has continued to rise [39, 40]. For localised adenocarcinomas, surgery following neoadjuvant The British Journal of Radiology, September 2012

therapy remains the mainstay of treatment [41].

However, only patients who responded to induction therapy gained survival benefits. Individual, responseguided treatment concepts in oncology are needed [42].

Despite intensive efforts to identify molecular markers, which are predictive for tumour response and prognosis, the value of each of these markers is currently not sufficiently validated to use them for the selection of patients for therapy [43–45]. Most previously reported studies showed that changes in tumour glucose uptake assessed by 18F-FDG PET during or immediately after neoadjuvant therapy were associated with response and prognosis in oesophageal and oesophagogastric cancer [15, 21, 25, 35, 37, 42, 46–50]. Therefore, a meta-analysis is useful because it reduces the effect that chance plays on the individual result, and quality assessments are important for reducing bias. The methodological quality of this study was moderate considering the median score of 57.9%. Sensitivity analysis was not performed owing to insufficient data.

The current meta-analysis provides two possible findings regarding the use of 18F-PDG-PET for monitoring chemotherapy of localised oesophagogastric cancer.

Firstly, early metabolic changes (14 days after the start of therapy, PET2) might provide the same accuracy for the prediction of treatment outcome as late changes (after the end of pre-operative therapy, PET3). Secondly, the predictive value of relative changes in tumour FDGuptake might be superior to measurements of tumour FDG-uptake at PET1, PET2 or PET3. In the meta-analyses of the studies that dealt with the response of SUVmax measured from PET1 to PET2, the results showed that the responses were associated with a significantly better OS and DFS, indicating that the pejorative impact of nonresponse in SUVmax on both OS and DFS was statistically significant. In all the meta-analyses, including the studies that dealt with the response of SUVmax regardless of the second scan time, the overall HR results were similar and positive. Therefore, PET might be used as a reference method for early response assessment in oesophagogastric e697

41 W Zhu, L Xing, J Yue et al

(a)

Figure 2. Funnel graph for the

(b)

assessment of potential publication bias in studies of response in standard uptake value in patients with oesophagogastic junction cancer. (a) The studies for overall survival and (b) the studies for disease-free survival. HR, hazard ratio; SE, standard error.

Figure 3. Review: meta-analyses of the studies dealt with the prognostic values of the maximum standardised uptake value measured by fludeoxyglucose positron emission tomography (a) prior to therapy and (b) at least 2 weeks after the initiation of therapy (at 2 weeks after the initiation of neoadjuvant therapy or preoperatively) for overall survival.

Results in (a) indicate that a high primary tumour maximum standard uptake value (SUVmax) at positron emission tomography 1 (PET1) was not associated with a worse overall survival prognosis. Results in (b) indicate that a high primary tumour SUVmax at PET2/PET3 was not associated with a worse overall survival prognosis. CI, confidence interval; HR, hazard ratio; SE, standard error. e698

The British Journal of Radiology, September 2012

42 SUV and AEG in meta-analysis Figure 4. Review: meta-analyses of the studies dealt with the prognostic value of response in maximum standardised uptake value measured by fludeoxyglucose positron emission tomography (PET) (a) at least 2 weeks [from prior to therapy (PET1) to exactly 2 weeks after initiation of neoadjuvant therapy (PET2) or pre-operatively] and (b) exactly 2 weeks after initiation of therapy (from PET1 to PET2) for disease-free survival. Results in (a) indicate that metabolic responders from PET1 to PET2/PET3 had a better disease-free survival (DFS) than metabolic non-responders. Results in (b) indicate that metabolic responders from PET1 to PET2 had a better DFS than metabolic nonresponders. CI, confidence interval; HR, hazard ratio; SE, standard error.

cancer. Furthermore, the meta-analyses did not attribute a significant prognostic value to SUVmax before and during therapy in localised AEG. It was encouraging for the clinical application of 18F-PDG-PET for prediction of treatment response that relative changes provided the same or higher accuracy for predicting or assessing tumour response as SUVmax before and during therapy [51].

However, the results of this systematic review might be criticised on several points. Firstly, the number and quality of studies included in each analysis might be different, which might affect the indirect comparison of results at the different PET scan time. Secondly, SUV estimates might suffer from poor reproducibility because of the lack of standardisation of the acquisition and processing protocols leading to its assessment. For example, Boellaard et al [51] have shown in a simulation study that differences in defining regions of interest can result in a difference of .50% of measured absolute SUVmax. In a clinical study, Stahl et al [52] found that the SUVmax of gastric carcinomas increased by 60% between 40 and 90 min post injection, indicating that absolute tumour SUVmax are highly dependent on the timing of the data acquisition. Fortunately, SUVmax ratios measured by different methods varied only minimally [51]. Therefore, relative changes represented more robust parameters and were preferable for establishing the response criteria that could be used at multiple institutions, e.g. in multicentre trials. Thirdly, the prognostic value was evidenced by the broad range of SUV threshold values that have been used in the literature to distinguish between patients with low and high survival (thresholds varying from 35% to 63%). Remarkably, all studies with the prognostic effect of decreased SUVmax from PET1 to PET2 for OS selected .35% as the SUV threshold definition [30, 33, 35, 37]. Indeed, a meta-analysis of the individual patient data (IPD) would be needed to compensate for the large heterogeneity of the reported SUV. Summary statistics meta-analyses have the advantage of including published studies which are immediately available for analysis and whose results can be checked by others. Although IPD meta-analysis results are usually similar to previous literature-based The British Journal of Radiology, September 2012

publications, they add some interest, such as including unpublished trials, updating results, and particularly allowing for multivariate analyses, adjusting for other variables, and subgroup analyses.

Conclusion Although this study had some drawbacks, including being restricted to articles published in English, most retrospective studies, HRs extrapolated from the survival curves and so on, our experience has been that 18F-PDGPET scanning is useful as a tool to help guide therapy for patients with this difficult and morbid AEG carcinoma.

We also found that the change in SUVmax on 18F-PDG-PET scans taken from PET1 to PET2 may identify patients with a better prognosis after surgery and may someday be able to select patients who would benefit more from additional chemotherapeutic approaches than from resection.

References 1. Chin L, Gray JW. Translating insights from the cancer genome into clinical practice. Nature 2008;452:553–63.

2. Sawyers CL. The cancer biomarker problem. Nature 2008;452:548–52.

3. van’t Veer LJ, Bernards R. Enabling personalized cancer medicine through analysis of gene-expression patterns.

Nature 2008;452:564–70.

4. Luthra MG, Ajani JA, Izzo J, Ensor J, Wu TT, Rashid A, et al.

Decreased expression of gene cluster at chromosome 1q21 defines molecular subgroups of chemoradiotherapy response in esophageal cancers. Clin Cancer Res 2007;13:912–19.

5. Luthra R, Wu TT, Luthra MG, Izzo J, Lopez-Alvarez E, Zhang L, et al. Gene expression profiling of localized esophageal carcinomas: association with pathologic response to preoperative chemoradiation. J Clin Oncol 2006;24:259–67.

6. Hildebrandt MA, Yang H, Hung MC, Izzo JG, Huang M, Lin J, et al. Genetic variations in the PI3K/PTEN/AKT/ mTOR pathway are associated with clinical outcomes in esophageal cancer patients treated with chemoradiotherapy. J Clin Oncol 2009;27:857–71.

7. Wu X, Gu J, Wu TT, Swisher SG, Liao Z, Correa AM, et al.

Genetic variations in radiation and chemotherapy drug

e699

43 W Zhu, L Xing, J Yue et al

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

action pathways predict clinical outcomes in esophageal cancer. J Clin Oncol 2006;24:3789–98.

Cunningham D, Allum WH, Stenning SP, Thompson JN, Van de Velde CJ, Nicolson M, et al. Perioperative chemotherapy versus surgery alone for resectable gastroesophageal cancer. N Engl J Med 2006;355:11–20.

Medical Research Council Oesophageal Cancer Working Group. Surgical resection with or without preoperative chemotherapy in oesophageal cancer: a randomised controlled trial. Lancet 2002;359:1727–33.

Malthaner RA, Collin S, Fenlon D. Preoperative chemotherapy for resectable thoracic esophageal cancer. Cochrane Database Syst Rev 2006;3:CD001556.

Kelsen DP, Ginsberg R, Pajak TF, Sheahan DG, Gunderson L, Mortimer J, et al. Chemotherapy followed by surgery compared with surgery alone for localized esophageal cancer. N Engl J Med 1998;339:1979–84.

Berghmans T, Dusart M, Paesmans M, Hossein-Foucher C, Buvat I, Castaigne C, et al. Primary tumor standardized uptake value (SUVmax) measured on fluorodeoxyglucose positron emission tomography (PDG-PET) is of prognostic value for survival in non-small cell lung cancer (NSCLC): a systematic review and meta-analysis (MA) by the European Lung Cancer Working Party for the IASLC Lung Cancer Staging Project. J Thorac Oncol 2008;3:6–12.

Pan L, Gu P, Huang G, Xue H, Wu S. Prognostic significance of SUV on PET/CT in patients with esophageal cancer: a systematic review and meta-analysis. Eur J Gastroenterol Hepatol 2009;21:1008–15.

Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med 1998;17:2815–34.

Levine EA, Farmer MR, Clark P, Mishra G, Ho C, Geisinger KR, et al. Predictive value of 18-fluoro-deoxy-glucosepositron emission tomography (18F-PDG-PET) in the identification of responders to chemoradiation therapy for the treatment of locally advanced esophageal cancer. Ann Surg 2006;243:472–8.

Swisher SG, Erasmus J, Maish M, Correa AM, Macapinlac H, Ajani JA, et al. 2-Fluoro-2-deoxy-D-glucose positron emission tomography imaging is predictive of pathologic response and survival after preoperative chemoradiation in patients with esophageal carcinoma. Cancer 2004;101: 1776–85.

Hong D, Lunagomez S, Kim EE, Lee JH, Bresalier RS, Swisher SG, et al. Value of baseline positron emission tomography for predicting overall survival in patient with nonmetastatic esophageal or gastroesophageal junction carcinoma. Cancer 2005;104:1620–6.

Smithers BM, Couper GC, Thomas JM, Wong D, Gotley DC, Martin I, et al. Positron emission tomography and pathological evidence of response to neoadjuvant therapy in adenocarcinoma of the esophagus. Dis Esophagus 2008; 21:151–8.

Swisher SG, Hofstetter W, Komaki R, Correa AM, Erasmus J, Lee JH, et al. Improved long-term outcome with chemoradiotherapy strategies in esophageal cancer. Ann Thorac Surg 2010;90:892–8.

Tian J, Chen L, Wei B, Shao M, Ding Y, Yin D, et al. The value of vesicant 18F-fluorodeoxyglucose positron emission tomography (18F-FDG PET) in gastric malignancies. Nucl Med Commun 2004;25:825–31.

Flamen P, Van Cutsem E, Lerut A, Cambier JP, Haustermans K, Bormans G, et al. Positron emission tomography for assessment of the response to induction radiochemotherapy in locally advanced oesophageal cancer. Ann Oncol 2002;13:361–8.

Kroep JR, Van Groeningen CJ, Cuesta MA, Craanen ME, Hoekstra OS, Comans EF, et al. Positron emission tomography using 2-deoxy-2-[18F]-fluoro-D-glucose for response

e700

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

monitoring in locally advanced gastroesophageal cancer; a comparison of different analytical methods. Mol Imaging Biol 2003;5:337–46.

Haley M, Konski A, Li T, Cheng JD, Maurer A, Haluszka O, et al. Influence of diabetes on the interpretation of PET scans in patients with esophageal cancer. Gastrointest Cancer Res 2009;3:149–52.

Stahl A, Ott K, Weber WA, Wieder H, Fink U, Schwaiger M, et al. FDG PET imaging of locally advanced gastric carcinomas: correlation with endoscopic and histopathological findings. Eur J Nucl Med Mol Imaging 2003;30:288–95.

Swisher SG, Maish M, Erasmus JJ, Correa AM, Ajani JA, Bresalier R, et al. Utility of PET, CT, and EUS to identify pathologic responders in esophageal cancer. Ann Thorac Surg 2004;78:1152–60.

Mochiki E, Kuwano H, Katoh H, Asao T, Oriuchi N, Endo K. Evaluation of 18F-2-deoxy-2-fluoro-D-glucose positron emission tomography for gastric cancer. World J Surg 2004;28:247–53.

Rizk NP, Tang L, Adusumilli PS, Bains MS, Akhurst TJ, Ilson D, et al. Predictive value of initial PET-SUVmax in patients with locally advanced esophageal and gastroesophageal junction adenocarcinoma. J Thorac Oncol 2009; 4:875–9.

Rizk N, Downey RJ, Akhurst T, Gonen M, Bains MS, Larson S, et al. Preoperative 18[F]-fluorodeoxyglucose positron emission tomography standardized uptake values predict survival after esophageal adenocarcinoma resection. Ann Thorac Surg 2006;81:1076–81.

Murthy SB, Patnana SV, Xiao L, Rohren E, Hofstetter WL, Swisher SG, et al. The standardized uptake value of 18fluorodeoxyglucose positron emission tomography after chemoradiation and clinical outcome in patients with localized gastroesophageal carcinoma. Oncology 2010;78: 316–22.

Lordick F, Ott K, Krause BJ, Weber WA, Becker K, Stein HJ, et al. PET to assess early metabolic response and to guide treatment of adenocarcinoma of the oesophagogastric junction: the MUNICON phase II trial. Lancet Oncol 2007;8:797–805.

Javeri H, Xiao L, Rohren E, Lee JH, Liao Z, Hofstetter W, et al. The higher the decrease in the standardized uptake value of positron emission tomography after chemoradiation, the better the survival of patients with gastroesophageal adenocarcinoma. Cancer 2009;115:5184–92.

Javeri H, Xiao L, Rohren E, Komaki R, Hofstetter W, Lee JH, et al. Influence of the baseline 18F-fluoro-2-deoxy-Dglucose positron emission tomography results on survival and pathologic response in patients with gastroesophageal cancer undergoing chemoradiation. Cancer 2009;115: 624–30.

Wieder HA, Ott K, Lordick F, Becker K, Stahl A, Herrmann K, et al. Prediction of tumor response by PDG-PET: comparison of the accuracy of single and sequential studies in patients with adenocarcinomas of the esophagogastric junction. Eur J Nucl Med Mol Imaging 2007;34:1925–32.

Smith JW, Moreira J, Abood G, Aranha GV, Nagda S, Wagner RH, et al. The influence of (18)flourodeoxyglucose positron emission tomography on the management of gastroesophageal junction carcinoma. Am J Surg 2009; 197:308–12.

Ott K, Weber WA, Lordick F, Hulshof MC, Hoekstra OS, Herrmann K, et al. Metabolic imaging predicts response, survival, and recurrence in adenocarcinomas of the esophagogastric junction. J Clin Oncol 2006;24:4692–8.

Patnana SV, Murthy SB, Xiao L, Rohren E, Hofstetter WL, Swisher SG, et al. Critical role of surgery in patients with gastroesophageal carcinoma with a poor prognosis after chemoradiation as defined by positron emission tomography. Cancer 2010;116:4487–94.

The British Journal of Radiology, September 2012

44 SUV and AEG in meta-analysis 37. Weber WA, Ott K, Becker K, Dittler HJ, Helmberger H, Avril NE, et al. Prediction of response to preoperative chemotherapy in adenocarcinomas of the esophagogastric junction by metabolic imaging. J Clin Oncol 2001;19: 3058–65.

38. Roedl JB, Colen RR, Holalkere NS, Fischman AJ, Choi NC, Blake MA. Adenocarcinomas of the esophagus: response to chemoradiotherapy is associated with decrease of metabolic tumor volume as measured on PET-CT. Comparison to histopathologic and clinical response evaluation. Radiother Oncol 2008;89:278–86.

39. Pohl H, Welch HG. The role of overdiagnosis and reclassification in the marked increase of esophageal adenocarcinoma incidence. J Natl Cancer Inst 2005;97:142–6.

40. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin 2010;60:277–300.

41. Mariette C, Piessen G, Triboulet JP. Therapeutic strategies in oesophageal carcinoma: role of surgery and other modalities. Lancet Oncol 2007;8:545–53.

42. Juweid ME, Cheson BD. Positron-emission tomography and assessment of cancer therapy. N Engl J Med 2006;354: 496–507.

43. Langer R, Specht K, Becker K, Ewald P, Bekesch M, Sarbia M, et al. Association of pretherapeutic expression of chemotherapy-related genes with response to neoadjuvant chemotherapy in Barrett carcinoma. Clin Cancer Res 2005;11:7462–9.

44. Zhu WQ, Yu JM, Sun XD, Xie P, Kong L. Serum CYFRA21-1 as a prognostic marker for patients with undifferentiated nasopharyngeal carcinoma. Biomarkers 2010;15:602–7.

45. Makdissi FB, Machado LV, Oliveira AG, Benvenuti TT, Katayama ML, Brentani MM, et al. Expression of Ecadherin, Snail and Hakai in epithelial cells isolated from the primary tumor and from peritumoral tissue of invasive ductal breast carcinomas. Braz J Med Biol Res 2009; 42:1128–37.

46. Brucher BL, Weber W, Bauer M, Fink U, Avril N, Stein HJ, et al. Neoadjuvant therapy of esophageal squamous cell carcinoma: response evaluation by positron emission tomography. Ann Surg 2001;233:300–9.

47. Downey RJ, Akhurst T, Ilson D, Ginsberg R, Bains MS, Gonen M, et al. Whole body 18FDG-PET and the response of esophageal cancer to induction therapy: results of a prospective trial. J Clin Oncol 2003;21:428–32.

48. Wieder HA, Brucher BL, Zimmermann F, Becker K, Lordick F, Beer A, et al. Time course of tumor metabolic activity during chemoradiotherapy of esophageal squamous cell carcinoma and response to treatment. J Clin Oncol 2004; 22:900–8.

49. Gillham CM, Lucey JA, Keogan M, Duffy GJ, Malik V, Raouf AA, et al. (18)FDG uptake during induction chemoradiation for oesophageal cancer fails to predict histomorphological tumour response. Br J Cancer 2006; 95:1174–9.

50. Westerterp M, Omloo JM, Sloof GW, Hulshof MC, Hoekstra OS, Crezee H, et al. Monitoring of response to pre-operative chemoradiation in combination with hyperthermia in oesophageal cancer by PDG-PET. Int J Hyperthermia 2006;22:149–60.

51. Boellaard R, Krak NC, Hoekstra OS, Lammertsma AA. Effects of noise, image resolution, and ROI definition on the accuracy of standard uptake values: a simulation study.

J Nucl Med 2004;45:1519–27.

52. Stahl A, Ott K, Schwaiger M, Weber WA. Comparison of different SUV-based methods for monitoring cytotoxic therapy with FDG PET. Eur J Nucl Med Mol Imaging 2004;31:1471–8.

The British Journal of Radiology, September 2012

Appendix A The quality scale used in this study Except when specified, the attributed value per item is 2 points if it is clearly defined in the article, 1 point if its description is incomplete or unclear, and 0 points if it is not defined or is inadequate.

Scientific design (1) Study objective definition. (2) Study design: prospective, 2 points; retrospective or retrolective, 1 point; not defined, 0 points. (3) Outcome definition. (4) Statistical considerations: fully reported with a preliminary assessment of the patient/sample number to be included and/or analysed, 2 points; patient/sample number to be included and/or analysed justified by the number of studied variables (minimum 10 patients per variable), 1 point; not defined, 0 points. (5) Statistical methods and tests description.

Generalisability (1) Patient selection criteria, including histological type, disease stage and treatment. (2) Patients’ characteristics, including histology type, disease stage and treatment. (3) Initial workup. (4) Treatment description. (5) Number of ineligible patients with exclusion causes.

Results analysis (1) Follow-up description, including the number of events. (2) Survival analysis according to the SUV. (3) Univariate analysis of the prognostic factors for survival: report of the relative risk with the CI, 2 points; results without evaluation of the relative risk and its CI, 1 point; not reported or inadequate, 0 points. (4) Multivariate analysis of the prognostic factors for survival: report the relative risk with the CI, 2 points; results without evaluation of the relative risk and its CI, 1 point; not reported or inadequate, 0 points.

The PET reports (1) Patients characteristics: weight/height; blood sugar level; histological subtype. (2) 18F-FDG-PET acquisition protocol characteristics: fasting duration; injected dose of 18 F-FDG; delay between injection and data acquisition.

(3) Technical parameters: investigation area; delay between CT thorax and PET acquisition; SUV formula; type of SUV; type of PET engine; duration of emission time; duration of transmission time; attenuation; and reconstruction parameters. (4) The analysis of the relationship between SUV was performed without knowledge of survival results and conversely (double blind). (5) SUV cut-off definition.

e701

Editorial Board

Publications

Available from bir.org.uk

HONORARY EDITORS Dr Jane Phillips-Hughes

Medical

Professor David Bradley

Scientific

Godfrey Hounsfield: Intuitive Genius of CT Written by Stephen Bates, Liz Beckmann, Adrian Thomas and Richard Waltham, and including many recollections from his family, friends and colleagues, this book provides a new insight into the life of one of science’s greatest pioneers.

DEPUTY EDITORS Dr Daniel Birchall Dr Nigel Hoggard Professor Alan Jackson Dr Simon Jackson Professor Paul Sidhu Professor Stuart Taylor Dr David Wilson Professor Alastair Munro Professor Kevin Prise Professor William Vennart Professor Habib Zaidi

Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Diagnostic Radiology Radiotherapy & Oncology Radiobiology Physics & Technology Physics & Technology

Radiation Shielding for Diagnostic Radiology This book is an update to the 2000 comprehensive report on the topic produced by a joint BIR/ IPEM Working Party. This report is designed to be a compendium of information for radiation physicists involved in specification of shielding requirements for X-ray facilities.

A scientifically based, realistic and straightforward approach to shielding specification. Sets out a framework for shielding calculations based on patient-dose area product and entrance surface dose information. Includes a review of the dose criteria employed in light of revised ionising radiation legislation and re-evaluates assumptions made in earlier methodologies.

ASSOCIATE EDITORS Abass Alavi USA Zahir Amin UK Richard Amos USA Diane Bergin Ireland David Brettle UK David Butteriss UK Fay Cafferty UK Mark Callaway UK Isabel Castellano UK Gary Cook UK Peter Corr Canada Justin Cross UK John Curtis UK Indra Das USA Marco Durante Germany Tarek El-Diasty Egypt Jonathan Eatough UK Phil Evans UK Sydney Evans USA Mark Farrugia UK Lesley Forsyth UK Nitin Ghonge India Rosalind Given-Wilson UK Stephen Harden UK Christopher Harvey UK John Holemans UK Peter Hoskins UK David Howlett UK Walter Huda USA

Charles Hutchinson Andrzej Kacperek Peter Keston Mayil Krishnam Dow-Mu Koh Iain Lyburn Ranald MacKay Alistair Mackenzie Lesley Malone David Manning Roger Martin Stewart Martin Soren Mattsson Michael Michell Andrew Morgan Shinichiro Mori Hilary Moss Ludvig Muren Kwan Hoong Ng Andrew Nisbet Christopher Njeh Alain Noel Philip O’Connor Allan Odurny Joerg Oestmann Oystein Olsen George Panayiotakis Tufail Patankar Niall Power

UK UK UK USA UK UK UK UK Ireland UK Australia UK Sweden UK UK Japan UK Denmark Malaysia UK USA France UK UK Germany UK Greece UK Ireland

Shonit Punwani Aloysius Raj Madan Rehani John Reynolds Rainer Rienmuller Peter Riley Michael Robbins Kevin Robson Andrea Rockall Carl Roobottom Nicola Russell Anju Sahdev Phillip Shorvon Andrew Slater Malcolm Sperrin David Sutton Katy Szczepura Alberto Tagliafico Nobuyuki Takahashi Paul Taylor Eliseo Vano Anthony Watkinson Elspeth Whitby John Winder Bernd Wittkop Conrad Wittram Wai-Lup Wong Hui-Xiong Xu

UK Malaysia Austria UK Austria UK USA UK UK UK Netherlands UK UK UK UK UK UK Italy Japan UK Spain UK UK UK UK USA UK China

BJR is essential reading for radiologists, medical physicists, radiotherapists, radiographers and radiobiologists. Articles included in BJR cover a wide range of subjects, including diagnostic radiology, radiotherapy, oncology, nuclear medicine, ultrasound, radiation physics, radiation protection and radiobiology.

Authors benefit from rapid decisions, constructive peer-review and global dissemination. For more information or to submit your next article to BJR, go to http://bjr.birjournals.org

Radiobiological Modelling in Radiation Oncology Radiobiological Modelling in Radiation Oncology, edited by Professors Roger Dale and Bleddyn Jones, will help to prepare you for the increasing use of modelling in practical situations, including treatment gap corrections, normal tissue tolerance predictions, optimisation of therapy determined by predictive assays, multi-modality schedule design, the simulation of clinical trials, testing contemporaneous medico-legal problems and teaching general principals of radiotherapy.

MCQ Pocketbook No. 2

MCQ Pocketbook No. 2 The best MCQ study guide from the British Institute of Radiology. Over 300 multiple choice questions. Based on recent journal articles and books, with the majority originating from the following key radiological journals: AJR, BJR, Clin Rad, and Radiology.

Multiple Choice Questions & Answers with References Adrian Mizzi Gavin Low Keh Oon Ong Ananth Shenoy Edited by Simon Ostlere

Published

by

AN INTERNATIONAL JOURNAL OF RADIOLOGY, RADIATION ONCOLOGY AND ALL RELATED SCIENCES