Volume 14 Issue 1
Journal for Clinical Studies PEER REVIEWED
Using Machine Learning to Identify At-Risk Sites in Acute Schizophrenia Clinical Trials Optimising Early Clinical Development Strategies in Oncology 2022 Cold Chain Predictions: Creating a new Normal Advancing Data from the Real World Novel Approaches to Clinical Studies
www.journalforclinicalstudies.com
Listen to our podcast
TriloTalk! Available on
Apple Podcasts,
Google Podcasts, and Spotify
MEET THE MASTERS OF MEDICAL WRITING
Dr. Douglas Fiebig, Senior Partner Master Skills - Listening with intent, synthesizing data, a steadfast commitment to excellence
Successful medical writing requires determination and tenacity. Our expert
writers are able to move the process forward. Break through team dysfunctions. And challenge assumptions. At Trilogy, conquering broken processes is what
we do best. And it’s just one more way we’re leading the fight against mediocre medical writing.
Visit us at TrilogyWriting.com writers@TrilogyWriting.com www.TrilogyWriting.com/TriloTalk
Contents 4 FOREWORD WATCH PAGES 6
Journal for Clinical Studies MANAGING DIRECTOR Mark A. Barker BUSINESS DEVELOPMENT Clement Brown clement@senglobalcoms.com EDITORIAL MANAGER Beatriz Romao beatriz@senglobalcoms.com DESIGNER Jana Sukenikova www.fanahshapeless.com RESEARCH & CIRCULATION MANAGER Jessica Chapman jessica@senglobalcoms.com ADMINISTRATOR Barbara Lasco FRONT COVER istockphoto
Cell And Gene Start-Ups Need "CDMO 2.0"
Louis Garguilo interviews Massimo Dominici a professional of eminent status in the CGT field, and founder of Rigenerand, about CDMOs. Over the years, Dominici and Mari have taken measure of the progress the contract development and manufacturing industry has made. 8
Advancing Data from the Real World – Novel Approaches to Clinical Studies
Technologies that gather and store health-related data are generating real-world data (RWD) and real-world evidence (RWE) that are increasingly being applied in healthcare decision-making. As noted by the US Food and Drug Administration (FDA), RWD and RWE are assisting the agency in monitoring post-market safety and adverse events (AEs) and in making regulatory decisions. Deborah Komlos at Clarivate explains more about those regulatory decisions to clinical trials. REGULATORY 10 Key Clinical Trial Considerations for the New Normal and the Future A great many new and varied approaches to clinical trial management are being adopted during the COVID-19 pandemic through the help of virtualisation tools, strong partnerships, and regulatory guidance. Despite the upheaval this year and last, there appears to be a silver lining largely due to the systemic changes leading to the remarkably quick development in adapting trials to accommodate different
PUBLISHED BY Senglobal Ltd. Unite 5.02, E1 Studios, 7 Whitechapel Road, E1 1DU, United Kingdom Tel: +44 (0) 2045417569 Email: info@senglobalcoms.com www.journalforclinicalstudies.com Journal by Clinical Studies – ISSN 1758-5678 is published bi-monthly by Senglobal Ltd.
The opinions and views expressed by the authors in this magazine are not neccessarily those of the Editor or the Publisher. Please note that athough care is taken in preparaion of this publication, the Editor and the Publisher are not responsible for opinions, views and inccuracies in the articles. Great care is taken with regards to artwork supplied the Publisher cannot be held responsible for any less or damaged incurred. This publication is protected by copyright. Volume 14 Issue 1 February 2022 Senglobal Ltd.
www.journalforclinicalstudies.com
Journal for Clinical Studies 1
Contents environments and the incredible speed at which COVID-19 vaccines have been developed and administered. Regulatory guidance has accommodated this abrupt shift. Jennifer Bradford and Sheelagh Aird at PHASTAR, will cover the key factors needed prior to adapting to patient-centric clinical trials.
stages of development – reflected in a shift towards later clinical trials. Inevitably this is resulting in more formal requirements for manufacturing process consistency, GMP standards and so on. Elena Meurer at Biopharma Excellence discusses the manufacturing and quality implications of novel biopharma treatments.
12 Operating a Clinical Trial in the Cloud: The Fundamental Aspects of Creating an Agile, Scalable and Flexible Solution Capable of Delivering Better Results than Previous Study Models.
24 Using Machine Learning to Identify At-Risk Sites in Acute Schizophrenia Clinical Trials
Bogged down by recruiting challenges and impaired by inconvenient travel distances, almost half of clinical trial sites miss enrolment targets, with nearly half of patients dropping out before study completion. Greg Killian at EPAM Systems, Inc. shows the fundamental aspects of creating an agile, scalable and flexible solution capable of delivering better results than previous study models. 16 How a Defined Informatics Entity Is the Key to Bridging Pharma’s Data Divide When the health authorities began placing further emphasis on medicinal product data standards, the overlap between regulatory operations and systems beyond the regulatory scope became increasingly apparent. Timm Pauli at PharmaLex explores this rapidly changing regulatory environment and the role of a defined cross-functional informatics entity – within R&D or beyond – in bridging the divide between the various functions and the technology and data that supports these functions. MARKET REPORT 18 Simplifying the Process of Managing Complex Country-byCountry Regulations in Clinical Trials Clinical trials themselves have their own complications. Regulatory compliance and safety reporting are of the utmost importance in managing clinical trials. When you add to this how the regulations change in various countries across Europe – and the world – the complexities are exponentially greater. Karin van Dort at Pharmasol reviews how a central hub can simplify tracking country rules for compliance. This approach can feature dynamic templates, which can be modified to meet local requirements, help to set a standard for tracking each country’s regulations and upon distribution to clinical trial investigators, can be blinded to avoid putting clinical trial information in jeopardy. THERAPEUTICS 20 Optimising Early Clinical Development Strategies in Oncology According to the largest analysis of almost 7,500 clinical and regulatory phase transitions, the likelihood of an oncology drug progressing from Phase I clinical testing is 5%, the lowest of the 14 major disease areas analysed. Of the oncology drugs that do progress, twice as many are for haematological malignancies than for solid tumours. Dr. Steve McConchie at Aptus Clinical shows more about the strategies in the clinical development in Oncology. 22 The Biotech Revolution: The Manufacturing & Quality Implications of Cell and Gene Therapies The market for Advanced Therapies has come a long way in the last decade, to the point that some cell and gene therapies have already been approved, and others have now entered the later 2 Journal for Clinical Studies
Data quality concerns are frequent in schizophrenia clinical trials, causing many to suffer from drug placebo separation. Machine learning offers the opportunity to proactively identify raters and sites at risk of developing data quality concerns for early intervention. Xingmei Wang, Emanuel Pintilii, Andrei Iacob and Dr. Alan Kott at Signant Health explore more about using machine learning to identify at-risk sites in acute schizophrenia clinical trials. TECHNOLOGY 28 AI/ML to Generate Medical Insights... While Maintaining Patient Data Security and Privacy Machine Learning (ML) and Artificial Intelligence (AI) have come to the forefront of data analytics with the promise of generating new medical insights. However, for healthcare data, patient data security is paramount due to the GDPR and similar regulations. Traditional methods of data consolidation for Machine Learning/AI modelling into a single warehouse or a data lake are often not possible even with anonymised data due to data protection rules. Douglas Drake at Clinerion Ltd. demonstrates how this ensures data security within the original data domain while allowing analytics for modelling and AI to be applied in a federated fashion. 34 Advanced Technologies Are Harnessing Patient Information to Drive Better Safety Data on patient experiences with new drugs and therapies promises to play a prominent role in drug reviews and approval. Analysis of this information could help identify and address adverse drug reactions and safety problems before they occur. Currently, these problems cost the industry nearly $30 billion per year. Saba Darvesh at Veeva Systems talks about the safety of patient information. LOGISTICS & SUPPLY CHAIN 36 2022 Cold Chain Predictions: Creating a new Normal At this time last year, the world was eagerly anticipating 2021. COVID-19 vaccines delivered hope for normalcy and excitement grew over convenience-related changes to healthcare, work and more. But the pandemic had other plans. New COVID-19 variants emerged, supply chain issues deepened and much of the day-to-day still looks different than pre-pandemic. Adam Tetz at Peli BioThermal expands the new predictions for 2022. 38 Getting a Grip on Covid-19 Test Samples In early 2020, the world of pathology and infectious disease testing was thrown into chaos by the Covid-19 pandemic. One of the key challenges in the global testing programmes for Covid-19 is how to track very large numbers of patient samples passing through inexperienced or under-staffed laboratories that have been asked to increase their daily throughput by as much as 10 times their normal workload. Neil Benn and Stephen Knight at Ziath Ltd explain more about the Covid-19 testing protocol. Volume 14 Issue 1
A Full Service Clinical CRO as Advanced as your Therapy
Full Service Delivery Early Phase Oncology Specialists Technology Driven Clinical Consultancy contact@aptusclinical.com +44 (0) 1625 238662
Foreword Welcome to the first issue of the Journal for clinical studies. This year we decided to give a new face to the journal by having a new logo. We hope you like it! The goals and strategies of treatment in schizophrenia may vary according to the phase and severity of the illness. Antipsychotics remain the cornerstone in the acute phase treatment, in the long-term maintenance therapy and the prevention of relapse of schizophrenia. Contrary to popular belief, schizophrenia is not a split or multiple personalities. Schizophrenia involves psychosis, a type of mental illness in which a person cannot tell what’s real from what’s imagined. At times, people with psychotic disorders lose touch with reality. Data quality concerns are frequent in schizophrenia clinical trials, causing many to suffer from drug placebo separation. Machine learning offers the opportunity to proactively identify raters and sites at risk of developing data quality concerns for early intervention. Xingmei Wang, Emanuel Pintilii, Andrei Iacob and Dr. Alan Kott at Signant Health explore more about using machine learning to identify at-risk sites in acute schizophrenia clinical trials.
incredible speed at which COVID-19 vaccines have been developed and administered. Regulatory guidance has accommodated this abrupt shift. Jennifer Bradford and Sheelagh Aird at PHASTAR will cover the key factors needed before adapting to patient-centric clinical trials. At this time last year, the world was eagerly anticipating 2021. COVID-19 vaccines delivered hope for normalcy and excitement grew over convenience-related changes to healthcare, work and more. But the pandemic had other plans. New COVID-19 variants emerged, supply chain issues deepened and much of the day-today still looks different than pre-pandemic. Adam Tetz at Peli BioThermal expands the new predictions for 2022. I hope you all enjoy the 1st edition of JCS in the year 2022, and I look forward to featuring more enlightening articles in the next issue of March. Beatriz Romao, Editorial Manager Journal for Clinical Studies
A very informative article by Douglas Drake at Clinerion Ltd. demonstrates how this ensures data security within the original data domain while allowing analytics for modelling and AI to be applied in a federated fashion. Machine Learning (ML) and Artificial Intelligence (AI) have come to the forefront of data analytics with the promise of generating new medical insights. However, for healthcare data, patient data security is paramount due to the GDPR and similar regulations. Traditional methods of data consolidation for Machine Learning/AI modelling into a single warehouse or a data lake are often not possible even with anonymised data due to data protection rules. In this journal, you will find some articles that approach how clinical trial management is being adopted during the COVID-19 pandemic through the help of virtualisation tools, strong partnerships, and regulatory guidance. Despite the upheaval this year and last, there appears to be a silver lining largely due to the systemic changes leading to the remarkably quick development in adapting trials to accommodate different environments and the JCS – Editorial Advisory Board
•
Hermann Schulz, MD, Founder, PresseKontext
•
Ashok K. Ghone, PhD, VP, Global Services MakroCare, USA
•
Jeffrey W. Sherman, Chief Medical Officer and Senior Vice President, IDM Pharma.
•
Bakhyt Sarymsakova – Head of Department of International Cooperation, National Research Center of MCH, Astana, Kazakhstan
•
Jim James DeSantihas, Chief Executive Officer, PharmaVigilant
•
Catherine Lund, Vice Chairman, OnQ Consulting
•
•
Mark Goldberg, Chief Operating Officer, PAREXEL International Corporation
Cellia K. Habita, President & CEO, Arianne Corporation
•
•
Maha Al-Farhan, Chair of the GCC Chapter of the ACRP
Chris Tait, Life Science Account Manager, CHUBB Insurance Company of Europe
•
•
Deborah A. Komlos, Senior Medical & Regulatory Writer, Clarivate Analytics
Rick Turner, Senior Scientific Director, Quintiles Cardiac Safety Services & Affiliate Clinical Associate Professor, University of Florida College of Pharmacy
•
Robert Reekie, Snr. Executive Vice President Operations, Europe, AsiaPacific at PharmaNet Development Group
•
Stanley Tam, General Manager, Eurofins MEDINET (Singapore, Shanghai)
•
Stefan Astrom, Founder and CEO of Astrom Research International HB
•
Steve Heath, Head of EMEA – Medidata Solutions, Inc
• •
Elizabeth Moench, President and CEO of Bioclinica – Patient Recruitment & Retention
•
Francis Crawley, Executive Director of the Good Clinical Practice Alliance – Europe (GCPA) and a World Health Organization (WHO) Expert in ethics
•
Georg Mathis, Founder and Managing Director, Appletree AG
4 Journal for Clinical Studies
Volume 14 Issue 1
Insights from the real world, in real time Clinerion enables faster access to innovative medical treatments for patients and physicians by unlocking the full scope of longitudinal information from patient electronic health records to generate productive real-world data.
Develop cohort models to understand and map patient risk factors, follow individual treatment paths or demographic groupings.
Enable metrics for personalized patient care and stratification, with instantaneous access to live data on patient journey and outcomes.
Image by Bill Oxford, on Unsplash
Understand pandemic epidemiology: follow infection rates and treatment patterns in near-real-time, as well as identify patient risk factors.
www.journalforclinicalstudies.com
Journal for Clinical Studies 5
Watch Pages
Cell And Gene Start-Ups Need "CDMO 2.0" Try as I may, I can’t get Massimo Dominici to use the term “CDMO,” although that’s what we’re talking about. At least I think that’s what we’re talking about. Finally, he relents; instead of saying “partnerships” again, he references “CDMO 2.0” as a different breed of external organisation, established to support cell and gene therapy (CGT) companies – a prelude partner before potentially utilising larger, established CDMOs. New Support Spin Dominici is a professional of eminent status in the CGT field, and founder of Rigenerand, a CGT company in Modena, Italy, managed by CEO Giorgio Mari. Along with internal product pursuits, Rigenerand has established itself as this new wave “CDMO 2.0” to serve similar CGT companies. Over the years, Dominici and Mari have taken measure of the progress the contract development and manufacturing industry has made. Impressive it has been, yes, but that flows predominantly to what’s been accomplished for the small-molecule and then biologics industries. Not so much (at least not yet) for the CGT space. Outsourced Pharma has documented how CDMOs today may not mesh with CGT pursuits, where practically every program is “novel” to some degree; there’s limited a priori knowledge or experienced professionals; differing schedules, equipment, and pricing needs; batches are minimal and intermittent; and CDMOs have grown larger while biopharma customers are more virtual and lean. In one sense, to overcome these challenges, Dominici (who maintains his position as fulltime professor at University of Modena and Reggio Emilia) wants to backpedal to a simpler time when smaller shops focused on specific skill sets and capabilities and weren’t in “one-stop-shop” mentalities. But his new iteration of that focused CDMO eclipses those earlier service providers in the areas of a peer-to-peer mentoring and sharing of developing knowledge and knowhow.
He believes if you are out in front of the field of CGT development, why not establish a path to help bring others in the industry along with you? It’s a business model Dominici presents in terms of “everyone pitching in to help the field mature faster, stronger, and more productively than at current trajectories, and ultimately on behalf of patients.” In the meantime: Why not help fund your own development programs with the proceeds from providing such “partnering”? Back To The Need Dominici explains that emerging companies most of all need knowledge support, and thus a “new sort of consultancy partnership focused on process development and advice-giving, not just CDMO contracts for services.” Today for CGT, neither the knowledgeable, skilled workers nor business models, which we’ll get to in a moment, actually reside at CDMOs, as they do for small molecule and biologics. As for the growing expertise base, it still resides within the CGT companies leading the way in the sector. Thus, a company like Rigenerand, having gained insights while pursuing its own programs since its inception in 2016, can chisel out an authentic support model in which like-minded companies can access that knowledge and tutoring. “Current CDMO contracts and pricing assigned to standard RFPs for services for small molecule or biologic production cannot be applied to emerging gene and cell therapy,” says Dominici. “Instead,” he says, “you need partnerships providing access to current expertise that can better translate your early ideas into early clinical phases, accomplish things such as consolidating your processes, so they are faster, more robust and cheaper. This is the real expertise required to jump start the industry today.” “A company like ours can provide this expertise because we have been challenged by our own products,” he continues. “That’s the type of expertise we are willing to use to help others. We come before relationships with bigger CDMOs.
This new-breed service provider can materialise in a familiar form to readers – a drug-developing and service-providing hybrid organisation. I’ve even taken to naming this, as a “Biopharma CDMO.”
“Once you receive this consulting on the process, the focus on problem solving, and you fix the type of challenges that have not been addressed in the past even with commercial products, you can go on to take full advantage of that large-scale manufacturing capacity.”
But here’s what’s new: A more vigorous, utopian vision assigned to this model by Dominici.
Dominici believes that in the past, big-company cell-and-gene pioneers – “the Novartis’s and others” – started out looking for large-
6 Journal for Clinical Studies
Volume 14 Issue 1
Watch Pages
scale manufacturing. “But their technology needed evolving, even after products appeared on markets. They needed more efficiency in manufacturing.” “This is where a CDMO 2.0 could have added value,” says Dominici. “I am not talking strictly about ‘business opportunity’ here. We want to help. I'm talking about a new way to create industry partnerships to improve CGT technology, to move us safely and efficiently into GMP environments.” Final Comradery I’ll buy into the industry bonhomie and cooperative comradery Dominici (and some others) say the cell-and-gene industry needs to turn dreams into future drugs and therapies. It does appear today’s CDMOs are of less help than required. But is the Biopharma CDMO – or CDMO 2.0 – the model for accomplishing this? Readers, what do you think? For one thing, financial and operational focus needs to be maintained when developing a drug or therapy. And competitive advantages need to be maintained as well. “I’ll answer by saying there is a need for this type of communication and assistance within the industry,” says Dominici. www.journalforclinicalstudies.com
“If we are all aware of this need, we have at these early phases of development, then I think everybody will be more relaxed with these models, and in collectively facing challenges with a partner like a “CDMO 2.0’ to improve.” “So why don't we take advantage of this, and take on challenges to make our programs better, faster, more productively manufactured, and look for new ideas on tissue cultures, reagents, etc. “It would be the CDMO 2.0’s responsibility to take all these and put it together, like in a kitchen. This is the concept that we have.”
Louis Garguilo Louis Garguilo is chief editor and conference chair for Outsourced Pharma, and a contributing editor to Life Science Leader magazine. He studied public relations and journalism at Syracuse University. Among other positions, Garguilo spent a decade at a global pharmaceutical contract research, development, and manufacturing organisation, and served under the governor of New York in the state's economic development agency, as liaison to the pharmaceutical and biotechnology industries. Email: louis.garguilo@lifescienceconnect.com
Journal for Clinical Studies 7
Watch Pages
Advancing Data from the Real World – Novel Approaches to Clinical Studies Technologies that gather and store health-related data are generating real-world data (RWD) and real-world evidence (RWE) that are increasingly being applied in healthcare decision-making. As noted by the US Food and Drug Administration (FDA), RWD and RWE are assisting the agency in monitoring post-market safety and adverse events (AEs) and in making regulatory decisions. In the healthcare community, these data can support coverage decisions and the development of guidelines and decision tools for use in clinical practice. For medical product developers, RWD and RWE help to shape clinical trial designs (e.g., large simple trials, pragmatic clinical trials) and observational studies to generate novel treatment approaches. The use of these data for regulatory decision-making was brought into particular focus through passage of the 21st Century Cures Act (Cures Act) by the US Congress in December 2016. Among other provisions, the Cures Act under section 3022 added section 505F (“Utilizing Real World Evidence”) to the Federal Food, Drug, and Cosmetic Act (FD&C Act). In response to the section 505F requirements, the FDA published a framework1 in December 2018 for the agency’s Real-World Evidence (RWE) Program to evaluate the use of RWE in regulatory decision-making. As part of its RWE Program and in partial fulfilment of a mandate under section 505F, the FDA has issued guidance about the use of RWE in regulatory decision-making. The most recent guidance documents were published in the last quarter of 2021: • • • •
Considerations for the Use of Real-World Data and Real-World Evidence to Support Regulatory Decision-Making for Drug and Biological Products, draft guidance, December 2021 Real-World Data: Assessing Registries to Support Regulatory Decision-Making for Drug and Biological Products, draft guidance, November 2021 Data Standards for Drug and Biological Product Submissions Containing Real-World Data, draft guidance, October 2021 Real-World Data: Assessing Electronic Health Records and Medical Claims Data To Support Regulatory Decision-Making for Drug and Biological Products, draft guidance, September 2021
The FDA is also required under section 3022 of the Cures Act to establish a pilot program to improve the quality and acceptability of RWE-based approaches in support of new intended labelling claims, including approval of new indications of approved medical products or to satisfy post-approval study requirements. The Advancing RWE Program is to be established by no later than December 31, 2022. Additional requirements include hosting a public workshop or meeting to discuss RWE case studies and updating existing RWErelated guidances or issuing new draft guidance, as appropriate, to reflect experience gained in the pilot program. Improving RWE Generation A summary of the RWE pilot program was presented at a virtual 8 Journal for Clinical Studies
public meeting in September 2021 on the proposed recommendations for reauthorisation of the Prescription Drug User Fee Act (PDUFA) for fiscal years 2023 through 2027 (PDUFA VII). The current authorisation of the program (PDUFA VI) expires in September 2022. Exploring the use of RWE in regulatory decision-making is covered under section I.K of the PDUFA VII commitment letter (“Enhancing Regulatory Science and Expediting Drug Development”). During her opening remarks at the September meeting, Janet Woodcock, MD, FDA acting commissioner, highlighted areas of support provided by the PDUFA program. With the help of user fees, the FDA “is able to apply the best available science and most rigorous data to the decisions we make throughout the entire medical product lifecycle, from the earliest stages of product development to post-market safety,” Woodcock said. For this reason, she added, PDUFA VII “will continue advancing regulatory evidence generation and drug development tools in areas like realworld evidence, rare disease endpoint development, and complex innovative trial design.” Peter Stein, MD, director, Office of New Drugs, Centre for Drug Evaluation and Research (CDER), FDA, presented the PDUVA VII pre-market review proposed enhancements, which include the RWE pilot program. Stein described the program as “very exciting” but noted that the area of RWE is “certainly challenging.” It involves many issues, he said, including those related to the types of data, sources of data, how the data are to be analysed, and the development of comparative groups. Addressing these challenges, Stein said, requires a considerable amount of discussion between sponsors and the FDA on the appropriateness of the study design, and whether it is sufficient to be able to develop the data that the agency would need to support a regulatory decision. The RWE pilot program aims to tackle these challenges and seeks to have more interactions between the sponsor and the FDA, he said. Through the program, sponsors would receive feedback as their RWE protocols are being developed, improving the chance of developing more robust study protocols. At the meeting, invited industry and public stakeholders, as well as members of the public who had signed up to provide comments, shared their feedback about the PDUVA VII proposed enhancements. Relating to the RWE pilot program, Lucy Vereshchagina, PhD, with the Pharmaceutical Research and Manufacturers of America (PhRMA) emphasised the importance of the program’s public transparency element. With its requirement for the learnings to be publicly posted and the multiple opportunities for interactions with the FDA, there can be “meaningful” discussions, Vereshchagina said, about how to increase the regulatory acceptance of RWE. “That, arguably, can be used not just for safety purposes, but also can support efficacy indications,” she said. RWE and RWD “can make significant contributions in advancing all of our understanding of which patients will benefit the most from treatments, and this trend wasn’t slowed down by the COVID-19 pandemic,” said Cynthia Bens with the Personalized Medicine Coalition (PMC), a non-profit education and advocacy organisation. Volume 14 Issue 1
Watch Pages The COVID-19 public health emergency necessitated the use of nontraditional approaches to data gathering and clinical studies in order to facilitate patient participation, Bens noted. Further, analysis was needed on RWD sources to understand treatment patterns more quickly for hospitalised COVID patients. “Given the limitations on the size of the pilot, and uncertainty about how much information ultimately will be
REFERENCE 1.
Framework for FDA’s Real-World Evidence Program. December 2018. https:// www.fda.gov/media/120060/download
The FDA defines RWD and RWE as follows: • •
RWD are data relating to patient health status and/or the delivery of healthcare routinely collected from a variety of sources. RWE is the clinical evidence about the usage and potential benefits or risks of a medical product derived from analysis of RWD.
www.journalforclinicalstudies.com
disclosed, we’d also encourage FDA to provide additional venues for researchers, health data organisations, and other non-industry stakeholders to interact with the agency on RWE and RWD issues,” Bens said. Public comments were accepted by the FDA on the proposed PDUFA VII enhancements until October 28, 2021 (Docket No. FDA2021-N-0891).
Deborah Komlos Deborah Komlos, MS, is a Principal Content Writer for the Cortellis suite of life science intelligence solutions at Clarivate. In this role, her coverage centres on FDA advisory committee meetings, workshops, and product approvals. Her previous positions have included writing and editing for magazines, newspapers, online venues, and scientific journals, as well as publication layout and graphic design work. Email: deborah.komlos@clarivate.com
Journal for Clinical Studies 9
Regulatory
Key Clinical Trial Considerations for the New Normal and the Future Clinical trials are changing especially due to the COVID-19 pandemic – with patient-centric trials coming into the fore. Considerations are needed for the way these decentralised trials are managed as compared to traditional clinical trials, such as with telemedicine and data collection. Data format, frequency of collection, manner of collection, and validity are all concerns. This article will review the unique challenges and solutions to patientcentric trials with remote communication, how wearables devices are being used (including what questions needs to be asked), and the flow of data collection. Data collected from the patient needs to be standardised across sites and study visits and participants. A great many new and varied approaches to clinical trial management are being adopted during the COVID-19 pandemic through the help of virtualisation tools, strong partnerships, and regulatory guidance. Despite the upheaval this year and last, there appears to be a silver lining largely due to the systemic changes leading to the remarkably quick development in adapting trials to accommodate different environments and the incredible speed at which COVID-19 vaccines have been developed and administered. Regulatory guidance has accommodated this abrupt shift. In this article, we will cover the key factors needed prior to adapting to patient-centric clinical trials. For starters, there are quite a few differences regarding attaining and disseminating patient-level data in a decentralised or remote trial setting versus the traditional way an in-person study is designed. Telemedicine or remote visits, for instance, have traditionally only been used for patient-physician consultations in the health care setting. However, the value of telemedicine for use in clinical trials has grown ever more promising due to the greater access to research and reduced attrition it can deliver. Data collection methods tend to be the primary component that changes with decentralised clinical trials in comparison to in-person studies. With remote trials, the engagement with digital technology for data capture provides the potential to receive information at a higher frequency, which means more data is available. Challenges in Patient-Centric Trials Another major difference occurs with the effectiveness of communication among patients and clinicians who are no longer involved in face-to-face interactions. With remote trials, the process of sharing information with some patients can be impacted because the patients can become either more or less forthcoming. It depends on how comfortable they feel about having virtual interactions and their comfort levels with the use of different methods for data collection versus in-person visits. Additionally, with remote trials, methods such as traditional phone calls between clinician and patient are sometimes replaced by remote monitoring using virtual tools/dashboards. In the latter case, patients need to feel that their 10 Journal for Clinical Studies
voices are still being heard by the clinicians to be comfortable with the processes and methods being used. Dissemination of information is an additional important consideration. For instance, when using remote monitoring tools, clinicians should consider the quantity of information and details that are made accessible to their patient, especially if the patient can assess their information through dashboards. For example, considerations on whether to conduct in-person vs. remote should be made when returning lab measures to a patient because at the in-person visit, the clinician would likely explain to the patient and would be able to better read the patient’s body language to know whether additional clarifications would be helpful. Something as simple as details of the normal ranges could be concerning to the patient if, for instance, the normal for that patient happens to not be within the normal range for the general population. Whilst from a medical perspective there are no concerns for that patient, the relaying of such details to the patient requires careful management to avoid the patient misconstruing the information. Wearables in Patient-Centric Trials As wearables evolve, it appears that both consumer and clinical grade wearables could bridge some of healthcare’s most considerable gaps. However, it’s still uncertain if consumer wearables will ever gain the credibility or even functionality of medical-grade devices. One thing is for sure, both continue to contribute to the growing wave that is Big Data. With consumer-grade wearables, trial professionals often need to assess how they can ensure data collected from wearables is accurate, considering how difficult it can be to ensure patients keep their consumer wearables charged and in good working order. Additionally, they need to rely on the patients to wear their devices correctly and that the data is downloaded correctly and is secure. When using wearables, there are key questions to consider including: 1. What and why is the data being collected? 2. Which type of device will be used (consumer vs. medical grade or BYOD)? 3. How much data is used and what information is shared with the patient? 4. Who owns the data? 5. What is the frequency of data sharing and data flow? It is important that the trial sponsor think about what data they need and why the specific information is being collected. At this stage, considerations should be made regarding the format of the raw data that is provided from the device selected and how the information should be analysed. It is also essential to determine the frequency of data collection. As devices collect data continuously, the question is two-fold. First, assess the granularity of the data available from the device, such as minute-by-minute summaries vs. all raw data. Second, determine the expectations around wear time. Vetting this Volume 14 Issue 1
Regulatory out ahead of time helps to drive the device selection, e.g., BYOD, consumer devices, or medical-grade device options. Due to the pandemic, companies see patients as informed collaborators whose participation could be key to furthering the overall success of a trial. Hence the term and greater usage of “patient-centric trials.” There are challenges concerning access to the data, data format, frequency of data extraction, and validity of the data. Different types of devices have different challenges in accessing the data, e.g., BYOD vs. medical grade devices. BYOD is playing a more active role in patient-centric clinical trials. If BYOD simplifies the clinical trial process, sponsors may achieve improved patient compliance, higherquality data, and potentially lower operational costs. Data Flow and Collection Once a device is chosen, it is vital to understand the flow of data, from the extraction of data from the device to the storage of the raw data, of which there is likely to be a lot, through to the summarised data for the trial team. Certain measures should be put in place to ensure the flow is efficient, successful, and secure. Deciding who is responsible for different parts of the data flow is vital to ensure the information will disseminate correctly, and with consideration to upgrades and changes by the manufacturers, who will monitor and update the workflow. Therefore, it is important to factor in how this will this impact the final analysis. Generally, the trial sponsor would determine, with input from the clinicians, if the data collected from wearables will be used during the trial, and if so, how? The rationale for using wearables is assessed at the beginning of the trial and performance is monitored throughout the trial. It is important to factor in whether the trial will require all the data collected such as activity, steps, and location information or just certain aspects. How will all of this be controlled? Expectations for the patient’s wear time of the device must also be decided regarding the need to detect, monitor, and/or manage non-wear time during the trial. Such considerations as to whether the clinician should discuss the data from the device with the patient will need to be assessed, as well as how and what elements are to be included. If the patient is expecting to have a discussion, the patient may become disengaged if it doesn't happen. Additional areas may need to be addressed if device information doesn't match the information provided from the patient. For example, if a patient reports they are walking 10 miles a day, but the device only records 1,000 steps, this issue needs to be resolved. Accuracy and Standardisation One fundamental challenge for trial teams comes when they need to ensure their patient-level data collected remotely is accurate, complete, and in line with applicable regulatory considerations. Standardising remote data collection across participating sites, trial participants, and study visits can reduce variability in the data. Prior to deployment of remote data collection, trial sponsors need to evaluate the feasibility of the remote data collection method. It is vital to have trial sites and participants prepared and fully able to comply with the data collection methods chosen. Site training may also be required to ensure systematic data collection and the need to uphold patient confidentiality. In lieu of attaining all pertinent information that would normally have been collected during a faceto-face study assessment, a video assessment documented with date and time should be included in the study source documents. To preempt the potential of missing data in remote monitoring, notification technologies that remind and prompt the patients/participants to report their information should be deployed. www.journalforclinicalstudies.com
The Delicate Balance Research professionals continue to try to balance as best as they can the need for accuracy, efficiency, thoroughness, and other positive aspects in their trials, all while trying to reduce patient burden. The more unobtrusive the trial, the more favourable for patients and the clinical site. Spreading out the assessments, for example, and decreasing the quantity of assessments performed at each visit is something sponsors may want to consider. The implementation of verbal and/or electronic reminders in patient reporting can be helpful. It could likely reduce the amount of missed assessments/reporting and provide a better outcome. Training for clinical sites on how to efficiently capture study data remotely could be a worthwhile investment. Conclusion The advent of decentralised or remote clinical trials has created both challenges and opportunities for accurate data collection. With efficient processes in place, it has been proven possible to ensure data flow is accurate, successful, and secure. Clearly, the expansion of telemedicine approaches with patients offsite has created an opportunity for both mitigating patient risks in a face-to-face setting as well as and allowing the patients a more comfortable setting in which to deliver data to clinicians during a trial. As wearables (consumer, clinical grade, and BYOD) continue to gain traction they, along with telemedicine approaches, may become another important asset in closing some of healthcare’s considerable communications gaps. However, it’s still uncertain if consumer wearables will ever gain the credibility, or even functionality, of medical-grade devices. One thing is certain, both telemedicine and wearables will continue to contribute to widening opportunities for data collection. Continued adoption of digital technologies, to not only capture data at greater frequency and speeds, may also lead to improved patient collaboration and the potential for lower overall operating investments.
Jennifer Bradford Jennifer Bradford, PhD is Director of Data Science for the CRO PHASTAR. She previously worked for the Advanced Analytics Group at AstraZeneca, leading the development of the REACT clinical trial monitoring tool, which she later customised and delivered to other sponsors as part of Cancer Research UK (CRUK). Within CRUK and in close collaboration with the Christie hospital she worked on EDC, app development and wearables data analytics in the context of clinical trials. She has a degree in Biomedical Sciences from Keele University and a bioinformatics Masters and PhD from Leeds University.
Sheelagh Aird Sheelagh Aird, PhD is Senior Director, Data Operations for the CRO PHASTAR (www. phastar.com). She has more than 30 years of experience in clinical data management, Sheelagh has directed and delivered projects in all phases of clinical trials across numerous therapeutic areas and data collection platforms. Sheelagh holds a BSc in pharmacology and doctorate in pharmacokinetics from the University of Bath. She has led PHASTAR’s Data Operations group since 2016.
Journal for Clinical Studies 11
Regulatory
Operating a Clinical Trial in the Cloud:
The Fundamental Aspects of Creating an Agile, Scalable and Flexible Solution Capable of Delivering Better Results than Previous Study Models. Bogged down by recruiting challenges and impaired by inconvenient travel distances, almost half of clinical trial sites miss enrollment targets, with nearly half of patients dropping out before study completion. Because of these issues, clinical trials incur $600K to $8M in potential daily losses.1 Accordingly, the pharmaceutical industry is quickly pivoting away from “site-centric” to “patient-centric” clinical trials, not only to curb losses but to improve results and outcomes. Through process and technical innovation including remote monitoring, advanced analytics, and agile ways of working, digital alternatives have the potential to help reduce time to market by 500 days and reduce development costs by 25%.2 More accessible, affordable, and faster than traditional models, digital and decentralised trials can allow for greater clinical diversity, seamless coordination, better trial process control and enhanced patient progress tracking. When building and deploying platforms to support digital trials, they must be decentralised and focus on connecting patients, Clinical Research Organizations (CRO), and healthcare provider teams through a virtual, user-friendly, outcome-driven experience. Additionally, it must be modular and integrated, following a well-defined capability map and patient end-toend journey. Underpinning the future of digital clinical trials are these five components: digital enrollment, trial journey/ roadmap, omnichannel communications, increased compliance and connected apps. Ideally, an optimal cloud platform should permit life sciences companies to leverage existing technology investments while also helping them integrate best-of-breed capabilities tailored toward their therapeutic area and patient population needs. Life sciences organisations hoping to improve their clinical trials should find a partner capable of building solutions in the cloud. And the results of switching will have benefits across multiple stakeholders – sponsors, investigators, and patients together with their families and caregivers. Through secure omnichannel communication channels, enterprises can access real-time data to simplify patient recruitment and enrollment. Plus, organisations will enable unified incorporation of patient outcomes, improve operational efficiencies, reduce study timelines and increase patient engagement. Furthermore, these trials will effectively improve enrollment, streamline administrative processes, increase retention, boost patient outcomes and encourage better medication adherence. Clinical trials are research studies performed in people to evaluate a medical, surgical, or behavioral intervention. It is the primary way that researchers find out if a new treatment, such as a new drug or diet or medical device is safe and effective. Often a clinical trial3 is used to learn if a new treatment is more effective and/or has less harmful side effects than the standard treatment. While close adherence to process and protocol is a necessary aspect of any testing methodology, the programs, procedures and systems that uphold clinical trials are based on decades old processes and are perceived 12 Journal for Clinical Studies
to lack flexibility and have not yet adapted to current expectations and capabilities based on remote and digital engagement. The Impact of COVID-19 on Clinical Trials As society shut down, so too did most clinical trials. Enrollment plummeted4 and prospective participants noted that they were fearful of visiting hospitals that had become COVID-19 treatment centers. Trials were short-staffed and low on resources as everything got reallocated. Some were too risky to continue, but ones classified as “lifesaving” were kept open. Nevertheless, the SWOG Cancer Research Network reported that enrollment in large clinical trials dropped by half. Small trials – intended to establish new medicine safety – were also paused. Newly launched studies were most impacted by the pandemic – with one study discovering that of 62,000 trials that began before the pandemic, only 57% of the expected number of studies to get initiated came to fruition. Findings from one very comprehensive study,3 which analysed data across the globe (but primarily the US), revealed further effects the pandemic had on clinical trials. For example, there was a significant increase in delayed subject enrollment as well as operational gaps in most ongoing clinical trials, which negatively impacted the trials themselves and their data integrity. From a global perspective, outside of the sites overseeing clinical trials relating to COVID-19, all trial sites experienced timeline delays or a total stoppage of operations. More specifically, 69% of the study’s respondents indicated that the pandemic affected their ability to perform ongoing trials; 78% believed that the current global crises prevented the initiation of new trials. The study also discovered that (based on the weighted average of their answers) the chief four concerns of staff were patient enrollment, patient recruitment, financial implications from canceled studies and financial implications from delayed milestones. A deeper examination of the issues troubling clinical trials showed that 54.8% of respondents saw a decline in patients’ willingness to go to physical sites, but conducting telehealth visits posed a considerable challenge in terms of time restraints. Plus, 51.6% noted concern over the time it took to discuss modifying trial procedures to accommodate patients unwilling to come physically. The same percentage also specified that their limited ancillary services were problematic. Because of these difficulties, the US Food and Drug Administration created a series of guidelines for clinical trials to follow in order to continue despite the pandemic. The intuition permitted4 the delivery of experimental medicines to participants’ homes instead of having the patient go to the medical center themselves. As online platforms got introduced, people could give consent virtually to participate in clinical trials. The time between doctors’ visits got lengthened, doctors made remote visits, and trial staff leveraged phone and video to complete questionnaires. Some participants could visit their local doctor for basic procedures. While these policies got trial enrollment back up to near-normal levels – the longer intervals between assessments meant less data could be collected from patients. And the quality of the trial data decreased as well. Likewise, the efforts to make clinical trials more convenient could not prevent cancer survivors from becoming less likely to enroll in a clinical trial postrecovery. Because of the decrease in clinical trials (particularly for cancer), there is a shortage4 of tumor samples for even the most Volume 14 Issue 1
Regulatory basic research and testing. Recognising that these hybrid solutions were more of a temporary solution, third-party companies worked to consolidate hybrid clinical trials into a more unified experience. Creating an Integrated, Cloud-Based Platform for Clinical Trials Bogged down by recruiting challenges and impaired by inconvenient travel distances, almost half of clinical trial sites miss enrollment targets, with nearly half of patients dropping out before study completion. Because of these issues, clinical trials incur $600K to $8M in potential daily losses. Accordingly, the pharmaceutical industry is quickly pivoting away from “site-centric” to “patientcentric” clinical trials, not only to curb losses but to improve results and outcomes. Although switching to a hybrid model served as a suitable substitute, the eventual development of unified, cloudbased digital platforms for clinical trials rectified the lingering questions of data integrity4 and challenges of conducting telehealth visits.3 Through process and technical innovation including remote monitoring, advanced analytics, and agile ways of working, digital alternatives have the potential to help reduce time to market by 500 days and reduce development costs by 25%.2 Likewise, digital trials allowed for greater clinical diversity, seamless coordination, better trial process control and enhanced patient progress tracking. Ideally, when building a cloud-based platform to deliver the future of hybrid and decentralised clinical trials, it must be supported by a virtual, effortless, outcome-driven experience that connects patients, Clinical Research Organizations (CROs), and healthcare providers. Modular and integrated, a digital clinical trial solution expands patient outreach and optimises the study start-up timeline while decreasing the burden on study teams and patients. Moreover, by analysing the exemplar platforms built thus far for the pharmaceutical industry, others can find similar success in this post-pandemic era. The Five Components Underpinning the Creation of a Clinical Trial in the Cloud When creating any clinical trial in the cloud, it is foundational to adhere to these five components: digital enrollment, trial journey/ roadmap, omnichannel communication, remote monitoring and connected apps. •
•
•
•
Digital Enrollment – The ability to recruit patients from the comfort of their own homes is essential to a digital clinical trial. By equipping CROs, investigators and local HCPs (Healthcare Providers) with tools such as eConsent (Electronic Consent) and TeleConsent, they can remotely screen patients at scale, meet study enrollment targets faster, and accommodate for study needs and patient preferences. And with machine learning and data mining, the enrollment process becomes even more convenient. Trial Journey/Roadmap – Keeping the patient engaged is critical, and a roadmap can do just that as it provides the participant with an on-demand overview of their trial progress. Furthermore, an organised journey outfitted with tasks, intelligent reminders, and behavioral nudges will ensure that the patient follows necessary procedures and maintains medication adherence while meeting milestones. Omnichannel Communication – The more convenient the clinical trial experience is, the more likely patients will see it through to completion. And by offering secure, mobile and reliable communication channels, patients can schedule visits, contact support or chat with the click of a button. Of course, all telehealth channels must be HIPAA-regulated and encrypted. Remote Monitoring – While there must be transparency for the patient via the trial roadmap, the healthcare side also needs a 360 degree view of information for compliance and safety purposes. Through easy-to-pair IoMT (Internet of Medical Things)
www.journalforclinicalstudies.com
•
integration, eCOA (Electronic Clinical Outcome Assessments), ePRO (Electronic Patient-Reported Outcomes) and AE (Adverse Events) reporting and real-time updates, medical teams can guarantee patient safety. Connected Apps – To enable faster study start-up timelines and smooth document exchange between sponsor teams and trial site teams, it is vital that a digital clinical trial platform leverage connected apps. Better access and visibility will allow for quicker eCRF verification, query resolution and turnaround times while empowering the seamless exchange of information between all involved parties.
One consideration is that the top digital clinical trial solutions allow clients to leverage existing technology investments, simultaneously helping them build frameworks with best-of-breed capabilities to improve operational efficiencies at sites and increase patient engagement.
Capability Map and The Patient End-to-End Journey With the five core practices established for the ideal digital clinical trial, it is also noteworthy to closely examine the capability map and patient end-to-end journey implemented within solutions currently available. Capability maps can be split into two sections: the digital clinical trials platform and the data factory. The former divides further into patient engagement on mobile apps, site web apps, mobile nurse apps and sponsored infrastructure. The latter is separated it into analytics, reporting, data ingestion, data lake and IoMT devices. By observing these divisions and segmentations, life sciences companies will glean best practices for their own digital clinical trials. The patient journey begins with (unsurprisingly) the patient and all the initial interactions the life sciences company would have with that individual. At the outset, the platform will use social media and trial matching to provide visibility of the clinical trial to various patient communities. Once the patient volunteers, the screening and enrollment phase begins. During this time frame, the app uses decision aid features and virtual communication to receive consent from the patient to participate in the trial. The participant will complete a virtual onsite visit as well as an eConsent/screening. Then, the onboarding process, where the patient sets up their profile. The next phase of the patient journey is the onsite/offsite/patient direct data capture process. This stage is the most involved section and includes the patient completing trial tasks, telehealth visits and surveys. Participants will also get supported by a health chatbot equipped with a symptom tracker which passively captures their data in real-time integrations through connected devices and IoMT. The fourth and final section of the patient journey is the companion stage; the trial participant will receive behavioral nudges to encourage patient retention using alerts, reminders and goals based on health metrics and relevant therapeutic areas. Additionally, to promote reenrollment, maintain post-trial support and alumni health insights, patients will receive updates and new trial availability based on collected health metrics. The Benefits of Digital Solutions As mentioned in the previous sections, a digital clinical trial built in the cloud can benefit all involved parties, including sponsors, investigators and patients. For sponsors, a modular, cloud-based platform can increase recruitment opportunities because it reduces barriers to travel and location via telemedicine so that more eligible patients can enroll from a broader and more diverse area. Sponsors will also benefit from enhanced visibility and enrollment data, that can lead to accelerated study start-up timelines and reduced overall study timelines. Similarly, data and insights from connected Journal for Clinical Studies 13
Regulatory
devices can also improve intelligence allowing for richer datasets. Additionally, real-time indicators of patient clinical progress can help sponsors validate digital clinical endpoints for tracking purposes. Investigators and CROs profit from increased efficiencies that smooth and simplify administrative processes. Moreover, investigators are more likely to see increased patient retention and patient safety from leveraging digital clinical trials as patients are monitored remotely, allowing for early interventions, tailored notifications and round-the-clock support to strengthen engagement and medication adherence. For the instances where patients are immune-compromised, the digital clinical trial can safeguard individuals by minimising their exposure to waiting rooms and/ or transportation. Finally, the digitally enabled model provides investigators with a single collection point for internal and partnered platforms, improving protocol structure, streamlining trial execution and enabling more seamless coordination. As for the patients, it is significantly more convenient than traditional clinical trials since recruitment and trial coordinators come straight to their homes. And regardless of where the patient lives, investigators and HCP can help them through virtual appointments or at-home visits. Now, an even greater clinical diversity is possible, meaning patients of a much greater regional and economic variety can access clinical trials should they volunteer to participate. Plus, patients are more likely to have increased medication adherence enabled by easy-to-use progress trackers, virtual support and early digital interventions. An Era of Hybridity and Virtualisation The pandemic has showed that the systems upholding our tightly interconnected world were much more fragile than we hoped. When factories halted in China, the unintended domino effect reached as
14 Journal for Clinical Studies
far as our supermarkets. Nevertheless, we also experienced that the ingenuity and adaptability of people should not be underestimated, as seen in the rapid devolvement and deployment of digital solutions including clinical trials. And although the challenges of data integrity and patient engagement are legitimate, having a dynamic, proactive strategy for risk assessment built into a digital cloud-based solution will ensure that clinical studies not only continue but can actually improve as a consequence. The post-pandemic era values hybridity and virtualisation, from classrooms and offices to concerts and hospital visits. It is likely that many more practices and activities, such as clinical trials, will also convert to digital for convenience, simplicity, and improved outcomes. REFERENCE 1.
2.
3.
4.
IS Reports. Tech + Patient Recruitment. IS Reports. https://isrreports.com/ wp-content/uploads/2016/06/ISR-PatientRecruitmentTrends-InfographicJuly2016.pdf. Accessed Jan. 6, 2022. G. Agrawal, H. Keane, M. Prabhakaran, M. Steinmann. The pursuit of excellence in new-drug development. McKinsey & Company. Published 1, 2019. Accessed Jan. 6, 2022. National Institute on Aging. What Are Clinical Trials and Studies. National Institure of Health. https://www.nia.nih.gov/health/what-are-clinical-trialsand-studies. Last Reviewed April 09, 2020. Accessed Jan. 6, 2022. H.Ledford. The COVID pandemic’s lingering impact on clinical trials. Nature. https://www.nature.com/articles/d41586-021-01569-9. Published on June 28, 2021. Acessed on Jan. 6, 2022.
Greg Killian Greg Killian is VP, Business Unit Head for Life Sciences at EPAM Systems. In this role, he is responsible for P&L and business management in the sector as well as market strategy and execution. His background is based on more than 25-years’ experience in life sciences and healthcare, spanning business management, product management, and software development. His recent roles and responsibilities have included global leadership of digital and analytics business units in the US, EU, CIS, and APAC with multiple accomplishments including high-growth business outcomes as well as entry into new markets with next-gen analytics and clinical solutions, manufacturing automation, and process analytical technology. Greg graduated from Villanova University with a BS in Mechanical Engineering and an MBA from University of Massachusetts, Boston.
Volume 14 Issue 1
Corporate Profile
Ramus Corporate Group is a union between Ramus Medical, Medical Diagnostic Laboratory Ramus and Medical Centre Ramus. All the companies are situated in the Ramus building in Sofia, Bulgaria. They are certified in compliance with the requirements of ISO 9001:2015.
Ramus Medical is full service CRO, working CTs in a variety of therapeutic areas and medical device.
• •
• • • • • •
Medical Centre Ramus with Phase I Unit
Medical writing for drugs and devices Scientific review of documentation Clinical trial management Monitoring Data management Regulatory advising and services during clinical trial
• • • •
Total laboratory automation with Abbott GLP-System Bioanalytical laboratory – ISO/IEC 17025:2017 accredited
PK/PD studies Medical devices investigations Phase I–IV Non-interventional studies
Medical Diagnostic Laboratory Ramus (SMDL-Ramus)
Others:
• • •
• • •
• •
30 clinical laboratories in Bulgaria and North Macedonia 325 affiliates for sampling in Bulgaria and North Macedonia More than 20 years’ experience in the CT field as central and safety laboratory; Largest PCR laboratory in Bulgaria Laboratory System integrates cluster generation, sequencing, and data analysis
, fast, correc t! Safe
Readability user testing Bridging report Carriage and storage of dangerous goods in compliance with ADR principles
Medical Diagnostic Laboratory Ramus Ltd
26 Kapitan Dimitar Spisarevski Street, 1592 Sofia, Bulgaria Tel/Fax: +359 2 944 82 06 www.ramuslab.com email: info@ramuslab.com
Ramus Medical Ltd Tu
to Cito
www.journalforclinicalstudies.com www.journalforclinicalstudies.com
V
e re
26 Kapitan Dimitar Spisarevski Street, 1592 Sofia, Bulgaria Tel./Fax: +359 2 841 23 69 www.ramusmedical.com email: office@ramusmedical.com
Dimitar Mihaylov Marketing Director
Journal for Clinical Studies 15 15
Regulatory
How a Defined Informatics Entity is the Key to Bridging Pharma’s Data Divide When the health authorities began placing further emphasis on medicinal product data standards, the overlap between regulatory operations and systems beyond the regulatory scope became increasingly apparent. Interoperability with other key functions – clinical, pharmacovigilance, quality, supply chain and others – was needed to implement xEVMPD and IDMP, which in turn led to the development of suites of software to manage cross-functional processes. However, technology solutions don’t address the knowledge gaps between regulatory and other functions and the informatics that are integral to managing product authorisation and life cycle maintenance. Timm Pauli, Head of R&D Informatics at PharmaLex, explores this rapidly changing regulatory environment and the role of a defined cross-functional informatics entity – within R&D or beyond – in bridging the divide between the various functions and the technology and data that supports these functions. For many years, pharmaceutical companies have operated in silos. Clinical, regulatory, pharmacovigilance, quality and manufacturing all held their data separately, each with their own product dictionaries. As regulators have sought to streamline the exchange of product information through standardisation and harmonisation – most notably with the plans to implement Identification of Medicinal Product (IDMP) standards – companies have needed to find their own ways to build cross-functional knowledge sharing. What makes these functional separations all the more problematic is that not only is there a barrier in terms of processes, but there are also complex information technology requirements to address, which has been exacerbated with the advance of digitisation. I would argue that the most effective way to harmonise and break down the divide is through a defined entity within the organisation that traverses these different functions to bring processes and systems together across the product lifecycle. Such an entity might best be defined as an R&D informatics function, whose primary objective is to provide the architectural backbone or foundation to support and enable harmonised decision making and to build effective interfaces across interconnected functions. Protecting the Core Value To understand why a defined R&D informatics capability is the link that organisations are crying out for, consider the overarching purpose of a pharmaceutical company – to develop and bring drugs to market. As such, I would argue that a company’s core value is its marketing authorisations and marketed products. It’s about developing and getting those products to market, keeping them on the market through efficient product maintenance and maximising their potential with regards to value and geography through appropriate market access programs and marketing authorisation programs to more regions and market. The marketing authorisation consists of information from different areas, each created and maintained with separate processes, tools and systems. Non-clinical, clinical, quality assurance, regulatory operations, pharmacovigilance and manufacturing all work to support and maintain the marketing authorisation. 16 Journal for Clinical Studies
The challenge has been that these various functions have historically been poorly harmonised – a situation that is still the case in many companies. Despite many attempts at automation (or semiautomation), use of simple tools as well as the introduction of new digital technologies, creating the dossier and maintaining the marketing application still requires many manual steps, built on unstructured or semi-structured documents rather than structured data. Inevitably, this increases the risk of human error and inconsistency across those different processes and systems. After authorisation, product maintenance was further separated from product development, creating additional divides. But the move to digital systems means this separation is starting to dissipate, for example, with recognition that clinical as well as CMC-related information (Chemistry, Manufacturing and Controls) that is key to the marketing authorisation is also important for authorised products. The European Medicines Agency has made its own changes in response to growing recognition of the overlap across the product lifecycle. While there has been a distinct divide between defining and processing the same or similar information in clinical trials and postauthorisation (i.e. approved marketing authorisations), the agency has made efforts to merge these in one master data approach, known as SPOR data management services (substance, product, organisation and referential). Combining Expertise Given the divide that has existed – between functions and between different parts of the lifecycle of a product – it is clear that a person or, preferably, entity, is needed to bridge the silos. Such a function would need to understand the processes and problems that various parts of the R&D organisation encounter, be able to translate these challenges into potential technology solutions and align those solutions with the way they are implemented by IT experts. An R&D informatics entity should be staffed by people who can traverse various disciplines – from the clinical phase to development of the marketing authorisation to distribution and beyond – and translate the information process requirements for the rest of the organisation. It’s important to specify that this expertise is not about IT infrastructure but rather about informatics across R&D. This defined entity would not be the ultimate decision-maker but rather an advisory function to enable other departments to be compliant and collaborative in an efficient and sustainable way. Experts within this R&D informatics function would support colleagues in other parts of the business with recommendations on innovative solutions that fit both the needs of the department and the broader corporate informatics strategy. An R&D informatics entity could ensure that any technology that one function decides to incorporate would be able to interface with solutions in other functions, not only in regards to technology aspects – which might be the domain of the IT department -- but also in regards to process support and information management. They would also be able to help map an integrated and harmonised data model with standardised terminologies, so when pharmacovigilance talks about a product, they speak the same language as regulatory affairs. It’s about guiding the different functions to ensure consistency from a technology and data perspective so that each function is processing and using information in a coherent way. Finally, and most importantly, R&D informatics must ensure that the combination of technology, data Volume 14 Issue 1
Regulatory management approach and processes comply with the regulatory and legal requirements. Perhaps a good comparison with the R&D informatics entity would be an architect – a person, or group of people, who can support the entire process, who understand the material (or technologies) and can share expertise on what would be most effective when building the complete structure. The experts within the defined entity therefore need to have a specific set of resources and capabilities to effectively bridge the divide between the functions and required informatics. Finding Synergies To address issues of data inconsistency and informatics barriers, the R&D informatics team might start by identifying synergies and similarities in each function. While each function is focused on their own issues and requirements, there are inevitably similarities with some approaches or ways challenges are addressed. By monitoring and understanding what is happening in various functions, the informatics team can identify seemingly independent changes and find those overlaps or synergies to create greater efficiency. These differences, yet synergies, are evident in how each department talks about and understands a simple term such as a “product”. What does a “product” mean? For regulatory, the term “product” is often used as a synonym for a marketing authorisation or a registration – even though this is not entirely correct. For manufacturing and supply chain, it generally means a manufactured item which is delivered to various markets under different marketing authorisations. In pharmacovigilance processing of adverse events typically focuses on the active substance, which plays the main role in the medical assessment. All ways in which the term “product” is used are understandable as they each emphasise different aspects of a “product”, but they aren’t all the same thing, and the term “product” can be confusing. With the ISO standards of IDMP, especially those that refer to the regulated pharmaceutical product information (ISO 11616) and regulated medicinal product information (ISO 11615), we finally have a clear definition and terminology of many terms in this domain. For the informatics defined entity, it’s about understanding those differences and coming up with the proper data model or terminology to cover those differences. The informatics team also needs to understand how changes in one department affect other functions. That’s because if one department changes its systems or how they process information, it could impact processes in another department, such as the use of pharmacovigilance data in regulatory submissions or leveraging regulatory data for adverse event monitoring. It is about understanding how data is used in various functions and therefore how change might impact seemingly unrelated processes in other departments. Finding these synergies and bridging the gap between functions is important when it comes to interacting with the regulatory authorities, where that separation is irrelevant. To regulators, it doesn’t matter whether information or data comes from pharmacovigilance, regulatory or manufacturing – it is all part and parcel of the same product. It’s important, therefore, that companies start to mimic that in their organisations and think the way the regulators think. And the only way this is likely to be achieved is through process and informatics harmonisation enabled by a defined entity that understands the regulatory space, including how changes and developments from the health authorities will impact decisions over solutions and processes. Making the Right Changes Harmonisation has been a priority for the health authorities for many years and EMA has been looking at how it can address its own issues with disconnected regulations and technical requirements. Initially, the agency sought to address inconsistencies through a telematics management board, but that endeavour only oversaw about half of the agency’s IT and informatics projects – always with www.journalforclinicalstudies.com
the additional complexity of aligning EMA’s strategy with those of the EU’s National Competent Authorities. A few years ago, the agency adopted an information management strategy,1 focused on achieving interconnected IT systems for managing and sharing information on medicines. The information management strategy has several overarching objectives that aim to harmonise the agency, both in terms of digital adoption as well as business process. These include:2 • •
• •
enabling the business to benefit from process optimisation by putting in place platforms on which to bring together business processes and related data including scientific knowledge enabling digital ways of working, better collaboration and information security through putting in place a single collaboration platform that integrates what is delivered today via multiple solutions fostering more agile ways of working by promoting a culture of collaboration among all EMA’s services without compromising IT governance practices putting in place Master Data Management to support standardisation, data consistency and data quality when used in different initiatives and business cases, by EMA and Telematics systems and by other stakeholders
The EMA, despite its many challenges as a body made up of 26 member states, has been moving in the right direction in terms of bridging its informatics divides; the pharmaceutical industry now needs to catch up and do the same. Some companies are on the right track, but many still have extremely disconnected systems and processes, which mean those inter-function gaps continue to be a problem. These gaps will become glaringly obvious with the connection between the Product Management Services (PMS) of SPOR and the submission of variations. Going forward, any variations for a product will only work if the product datasets in the SPOR database are correct and complete. The benefit of such an approach is companies applying for a change to their authorisation won’t have to start from scratch in terms of filling out forms and providing product details to the authorities. Companies will only be required to provide the ID number of the product in the SPOR database. In the long run, this process will make the daily business of regulatory, which is submitting variations, much easier, while accelerating the review process and, most importantly, improving safety oversight for patients. Harmonising cross-functional processes will not only be key to meeting this changing regulatory environment but will also reduce the risk of non-compliance and improve overall business efficiency. REFERENCES 1. 2.
https://www.ema.europa.eu/en/about-us/how-we-work/informationmanagement https://www.ema.europa.eu/en/documents/report/informationmanagement-strategy-2020-2022_en.pdf
Timm Pauli Timm Pauli is senior director and head of R&D informatics at PharmaLex, where he combines his informatics and systems expertise with deep knowledge of regulatory affairs. Timm has been with PharmaLex for 7 ½ years in a series of senior roles. Before joining PharmaLex, Timm worked in various technology-driven R&D roles at leading pharmaceutical and healthcare companies. Timm has a degree in medical informatics.
Journal for Clinical Studies 17
Market Report
Simplifying the Process of Managing Complex Country-by-Country Regulations in Clinical Trials Clinical trials themselves have their own complications. Regulatory compliance and safety reporting are of the utmost importance in managing clinical trials. When you add to this how the regulations change in various countries across Europe – and the world – the complexities are exponentially greater. The onus is on the clinical trial sponsor to ensure that all regulations have been met throughout the process of managing a clinical trial. They must also be able to trace and defend each step in the reporting process. Essentially laws, these regulations require interpretation by experienced and qualified individuals to gain a full understanding of what exactly the requirements are and how they can best be applied in the context of drug development. However, due to the complexity of the task, this effort is commonly supported by a network of professionals with different backgrounds: Clinical trial teams, Regulatory Affairs and Pharmacovigilance. Automated technology has many befits for managing the process for safety and compliance information. This article will review how a central hub can simplify tracking country rules for compliance. This approach can feature dynamic templates, which can be modified to meet local requirements, help to set a standard for tracking each country’s regulations and upon distribution to clinical trial investigators, can be blinded to avoid putting clinical trial information in jeopardy. Managing Cross-Border Regulations and Languages Regulations are generally high-level, which means that guidelines are provided but do not offer the specific details and processes that are necessary to attain compliance. This is where interpretations become exceedingly important. The country-level guidelines need to be translated into practical actions that can be applied across trial sites and institutions. In many countries, local competent authorities can apply their own interpretations of these guidelines, creating complexities for the clinical trial management team in safety reporting to authorities. Additionally, local and central ethics committees often add their own subtleties and don’t always distribute notifications as to updates or changes. This causes even greater challenges to the sponsor in maintaining up-to-date information on regulatory compliance and to apply this information correctly during the course of a trial. We also witness many countries that publish their regulations in local languages. This requires knowledgeable interpreters who can grasp nuances of the language and deliver accurate and transparent regulatory updates and alerts to clinical trial managers and sponsors. Thus, for a sponsor that is evaluating a compound with clinical trials in various global geographies, management of 18 Journal for Clinical Studies
regulatory compliance becomes even more complex. For example, if you are conducting a trial in Poland for a drug that will be marketed across Europe and North America, you need to not only consider regulations that apply to the country trial site but also to regulations in all of the countries where the drug will be available for distribution. Compliance with multiple countries’ requirements obviously requires significant resource allocations and standardisation of data capture and reporting. This is where a centralised hub becomes a tremendous opportunity to take away some of these complexities and deliver data to customers who only have to focus on just a ‘yes or no’ acceptance of the standards. Tracking Countries’ Regulatory Alterations with Humans Generally, in larger countries such as the U.S. and most European countries, there is ample notice when a regulation is changing. However, it’s not the same in every country in which trials are being conducted. To remain in compliance, clinical trial regulators need to be certain that a mistake hasn’t been made in reporting. This also means devising methods to capture new requirements or delayed information regarding regulations to the sponsor. When there are automated processes in place, this can be documented and explained. But it’s when issues in reporting are deliberate, i.e., specifically not reporting something critical like safety documents, it can become a significant problem that could seriously impact the progression of the trial. It’s particularly challenging when someone is not advised of changes if there are a number of countries participating in the trial. Subscribing to a regulatory database may be helpful but due to the issues occurring when countries do not issue notice of changes, there still needs to be human oversight, a quality check of sorts, plus of course the interpretation into a practical ‘what needs to be reported, how and when.’ Again, this takes time and experienced resources, particularly when trials are being conducted in multiple locations and in multiple languages. While databases may be helpful, complexities still exist that require human intervention. Simplifying the Process The clinical trial sponsor always has the ultimate responsibility of adhering to regulatory guidelines during the trial. However, the management of regulatory compliance documentation can vary from a sponsor’s in-house team to a CRO or other vendor that supports safety regulatory affairs. For example, a vendor specialising in pharmacovigilance can support the trial sponsor with an automated system, i.e., a dynamic data hub that is designed to manage contact information for recipients at all clinical trial sites, distribution of safety documents uploaded or linked from existing data sources, provide audit trails as well as manage reporting requirements per country and per recipient type. For example, in Volume 14 Issue 1
Market Report a typical safety document distribution scenario, ethics committee members would receive specific reports, i.e., periodic reports, line listings, SUSARS, defined for each role and document type.
Templates can be very helpful in supporting this distribution process – with designs based on role and document type. These are based on local values and will apply to the vast majority of all the rules but are able to be modified if necessary. Recipients are identified as to whether they would receive blinded or unblinded reports, critical when an unblinded report is inadvertently delivered to an unvetted recipient. This kind of mistake can seriously jeopardise a trial by creating bias. The data hub can be updated, when new contacts are onboarded, or new safety documents created – all of which are automatically entered into the system. The messages that are sent out are then tracked, with a detailed audit trail, allowing you to drill down from a document to see who it was sent to, and when it was acknowledged. Most countries accept the standard report formats: MedWatch, CIOMS-I and CIOMS-II. The forms are generally distributed using English as a single language. However, specific processes or local forms are often translated into local languages, again which require translation and transparent interpretation. Another interpretation challenge may come from CROs operating in different countries where processes may be similar but not exactly the same. Therefore, there is still human intervention involved for a quality check on accuracy and ‘clean-up’ of any errors. Regulatory bodies in both the U.S. and Europe are offering directives to streamline reporting processes. For example, in the EU the regulations for drugs and medical devices have become far more aligned than they had been in the past. But while regulations for safety document distribution often remain in place for several years, the reporting processes may change slightly, particularly with countries just learning to manage clinical trials where laws may change more often. In Summary Technology advances are supporting the automated reporting of clinical trial efforts. This will vastly simplify the tasks required to coordinate regulatory changes. This will positively affect the transparency in data reporting and greatly reduce labour, manual errors, and their associated costs. The key to this journey forward is a basis in which the information flows are interactive and dynamic via a central dashboard and where humans and machines work in tandem to assure prompt reporting with accurate results predetermined to meet global guidelines.
Karin van Dort Karin van Dort is Product Owner at pharmasol, which provides pharmacovigilance hosting services, and software, to pharmaceutical and CRO customers. She has been instrumental in developing their psiXchange platform for safety information distribution and associated process changes and regulatory intelligence. Karin started her career at Pharmachemie as a Medical Affairs Officer and subsequently moved to Parke-Davis and then to Pfizer with a focus on clinical trials, specializing in HIV and oncology. At Pfizer, she was the recipient of the Pfizer Global Health Fellowship for her work in India combatting leishmaniasis. Her previous position was with Grünenthal to implement key performance indicators (KPIs) and it was there that she codeveloped psiXchange. Karin has a degree in Bio-Pharmaceutical Sciences. She is a co-author on a patent for inhaled morphine and has certifications as Data Protection Officer and SCRUM Product Owner. Web: www.pharmasol.de
www.journalforclinicalstudies.com
Journal for Clinical Studies 19
Therapeutics
Optimising Early Clinical Development Strategies in Oncology According to the largest analysis of almost 7,500 clinical and regulatory phase transitions, the likelihood of an oncology drug progressing from Phase I clinical testing is 5%, the lowest of the 14 major disease areas analysed. Of the oncology drugs that do progress, twice as many are for haematological malignancies than for solid tumours.1 Despite this, drugs with cancer indications are approved at a faster rate than for any other major disease. A 2018 review of the 58 new cancer drugs approved by the FDA between 2012 and 2017 found that 95% of them qualified for an expedited development or approval pathway, including 79% for priority review, 45% for accelerated approval, 48% for fast-track approval, and 43% for breakthrough therapy status.2 While these new regulatory paradigms offer the potential to significantly reduce drug development time, this is only possible if a clear, highly effective, and regulatory acceptable strategic development programme is in place. So, what are the challenges in Oncology Drug Development? The traditional regulatory development route for oncology drugs, where discreet testing phases culminate in a large, randomised superiority trial, has evolved into multiple development pathways. Advances in cancer biology understanding and molecular diagnostic technologies have led to ever-smaller patient subgroups being identified for molecularly targeted therapy. Many of these have shown unprecedented responses in early phase trials, leading regulators to approve them without the need for large scale studies. FDA regulations enable the rapid review and accelerated approval of certain drugs in the absence of survival data. These regulatory approvals, and those based on large-cohort trials with surrogate or intermediate clinical end points or on non-inferiority trials, as well as new tumour-agnostic indications, set important precedents in the field. There are high levels of uncertainty in the transition from preclinical evaluations to human studies. A thorough understanding of the pre-clinical programme requirements for new therapeutic treatments and approaches in oncology is a must. The aim of the preclinical programme should be to support the decision for entering the clinical phase by providing a robust data package. This would include tests in various experimental settings that allow for an adequate initial benefit versus risk assessment. With respect to safety, it should be realised that toxicology study requirements for oncology products may differ considerably from other pharmaceuticals; they need to focus on schedule dependency where substantial savings could be obtained with the correct and optimal study design. 20 Journal for Clinical Studies
Broad experience in the design and implementation of translational science programmes is a great help with the above. Including due diligence for late stage pre-clinical and early clinical opportunities, assessment of pre-clinical results, and innovative trial design, to maximise the chance of success. There is a high failure rate after entering Phase III and missing regulatory requirements. Only one third of drugs completing Phase II development go on to Phase III trials, of these, many more will fail to reach approval. The challenges of drug development in oncology include complexity of the disease and underlying physiological processes, biomarker development and patient stratification, clinical methodology, defining dose and dosing schedules, selection of endpoint(s), and balancing regulatory and clinical requirements. Key to successful development is a flexible approach which allows integration of advances in the understanding of the specific type of cancer, evolving data of the investigational product and changes in regulatory environment. Early clinical development ideally should involve partnering with oncology experts who have experience with integrated development planning and systematic approaches. This will maximise the efficiency of the development process, data quality and regulatory acceptability. Furthermore, the use of targeted therapies and patient subpopulations will support the objective. With more targeted treatments, both agencies and payers are increasingly demanding efficient patient selection and outcomes data in order to obtain regulatory approval and rapid reimbursement decisions. Biomarkers are an important tool in the development of targeted therapies. The identification of predictive biomarkers helps to select patients who might benefit from the treatment or those who will not respond. It is essential to validate the biomarker thoroughly when developing as a companion diagnostic. Validated biomarkers can help to refine the development programme and the target product profile. Targeted therapies can provide added value to a specific subset of patients. Case Study – Design for innovation and success A UK Biotech needed assistance in the design of their First in Human (FiH) study for their lead project, a small molecule inhibitor of p300/ CBP. This required a multi-disciplinary panel of medical, scientific, and technical experts to ensure the Biotech received industry-leading input into the development of an innovative clinical plan to support the development of their lead oncology asset. They were provided with all the tools to make informed decisions about the next steps for their vital oncology asset. A comprehensive FiH clinical plan was produced which involved the creation of a Study Design Concept document. Volume 14 Issue 1
Therapeutics
This included: • endpoints • patient population/numbers • outcome variables and statistical considerations • go/no-go triggers for starting additional modules • Phase II study options • including combination strategies and additional patient segments • high level time and study budget estimates • competitor landscape reports for mode of action • tumour types of interest The result gives this biotech the best possible chance to be one of the 5% who make it past Phase I. Having chosen to partner with an experienced oncology provider to gather all the information this plan enabled the small molecule inhibitor developers to confidently enter discussions with their key stakeholders and investors. Now there is an additional factor, COVID-19. When the pandemic began over 2003 interventional oncology studies were suspended in the first 3 months. Halting drug development across all therapy areas in an unprecedented manner. Oncology seemed to be heavily affected by site closures leaving an alarming number of patients and in limbo and missing treatments. A year on COVID-19 has transformed the clinical trials landscape with contingency plans becoming standard procedure and we have seen many new innovations and approaches to ensure vital research can continue. Remote options such as telemedicine or wearable technology, have become the norm where the study protocol will allow. What does seem a certainty is that investigators, sites, and other clinical providers will have to continue to be flexible to ensure the best possible outcomes for patients and medicines. www.journalforclinicalstudies.com
In conclusion, oncology trials are incredibly challenging with a relatively low number of studies progressing. However, it appears the success of each molecule is partly determined before the study begins in the planning and design stage. The clinical trial landscape is continuing to shift at an unprecedented speed, embracing this change is likely to prove key in delivering new medicines. REFERENCES 1.
2.
3.
Clinical development success rates. Available at: https://www.bio.org/ sites/default/files/legacy/bioorg/docs/Clinical%20Development%20 Success%20Rates%202006-2015%20-%20BIO,%2Biomedtracker,%20 Amplion%202016.pdf. Accessed July 2020. Hwang, T. J. et al. Efficacy, safety, and regulatory approval of Food and Drug Administration-designated breakthrough and non-breakthrough cancer medicines. J. Clin. Oncol. 36, 1805–1812 (2018). Impact of COVID-19 on oncology clinical trials, National Library of Medicine https://pubmed.ncbi.nlm.nih.gov/32424343/
Dr. Steve McConchie Steve has worked in clinical development for nearly 25 years with AZ where he successfully delivered numerous oncology and haematology clinical development programmes. Since forming Aptus Clinical, he and his expert team continue to support the design, conduct and delivery of innovative early phase clinical studies in oncology and rare diseases for a broad range of clients. He is a member of the Scientific Innovation and Advisory Committee of the Bioindustry Association.
Journal for Clinical Studies 21
Therapeutics
The Biotech Revolution:
The Manufacturing & Quality Implications of Cell and Gene Therapies The market for Advanced Therapies has come a long way in the last decade, to the point that some cell and gene therapies have already been approved, and others have now entered the later stages of development -reflected in a shift towards later clinical trials. Inevitably this is resulting in more formal requirements for manufacturing process consistency, GMP standards and so on. Yet this presents some challenges given that the products, and the development and manufacturing processes involved don’t typically fit with traditional models and approaches. In this timely article, Elena Meurer, Principal Consultant and CMC expert at Biopharma Excellence, discusses the manufacturing and quality implications of novel biopharma treatments. These include not only CAT-T therapies, CRISPR technologies and induced pluripotent stem cell therapies, but also innovations from new market entrants involving 3D printing of organs and products to enable ‘bedside manufacturing’. The novel or advanced therapies field has been propelled into the spotlight. Ten or 15 years ago, this was a very young field with the majority of products only just entering early clinical research. Now the market is more evolved and there is now a shift towards later clinical trials as a number of cell and gene therapies enter the later stages of development. Indeed, we already have a number of approved therapies now. We’re seeing several examples of approved CAR-T therapies. Meanwhile CRISPR technologies, and induced pluripotent stem cell therapies, have made substantial progress too. At the same time, new players are coming in and we’re seeing new types of products emerge, including 3D printing, organs and products for “bedside manufacturing’. All of these developments are enormously positive, of course, but they also mean that the bar is being lifted now in terms of the formal requirements for manufacturing process consistency, GMP standards and so on. This is the natural course of things as the knowledge base grows. Yet it also presents some challenges, since this involves products, as well as development and manufacturing processes, that are unlikely to fit readily with traditional models and approaches. Manufacturing Challenges In many cases, in an advanced therapies context, Chemistry, Manufacturing and Controls (CMC) provision lags behind clinical development. There are clear reasons for this, including the need early on for a clinical proof-of-concept, and accelerated programmes that promote clinical development. But this typically means that the time allowed for CMC development is squeezed, presenting challenges in later development phases. Other practical factors include the small batch sizes associated with advanced therapies, which make it harder to reach product consistency. The use of biological materials and autologous manufacturing approaches add to the variability, too. 22 Journal for Clinical Studies
In addition, for many of the new players, novel therapy types require new solutions – around the manufacturing approach and quality control, as well as around regulatory strategy. These strategies and approaches will differ for each individual case, too. ‘Bedside’ Manufacturing Take the example of manufacturing at the patient bedside. Typically this won’t involve a facility separate from the hospital, or even a dedicated hospital room, where the product is manufactured. Instead, the product will be generated with the help of a particular device beside the patient’s bed. This presents a range of challenges. For one thing, this is not a controlled environment. Additionally, cell-based therapeutics and gene therapies currently require a very long product release time: at least several days, more typically even several weeks before a product is released. Those protracted timeframes are not viable if the goal is to establish highly efficient manufacturing at the patient bedside. This means we need to find another approach to timely deliver the product for the patient while still ensuring its safety and efficacy. Addressing these kinds of issues means much earlier consideration of all of the moving parts. It also requires a multi-pronged strategy that combines Regulatory, Manufacturing, Quality, and Development perspectives so that nothing is missed in the planning. Going back to the inherent variability of advanced therapies, the potency assays (the quantitative measure of biological activity) are mostly biologic assays and are by definition prone to higher variability. This must be factored in early in development and addressed in a way that’s appropriate, especially since the potency assays are indispensable as a measure of process changes. Other Considerations to Bear in Mind Early on To avoid being caught out, it is important to consider global clinical development from the outset. This is not just because of regulatory differences across and between regions, but also because this could influence the choice of materials used in manufacture, including the starting materials of biological origin. There will be differing standards for facility qualification between regions such as the US and the EU, for instance, as well as viral safety requirements. Take the establishment of standards for iPS cell lines for manufacture: viral and transmissible spongiform encephalopathy (TSE) safety has to be very carefully considered from the beginning if the cell line is going to serve product manufacture across the globe. Also, we often see that research-level materials are used or relevant quality information is provided by the material manufacturer in the Drug Master File in the US, so that the manufacturer of advanced therapies often lacks sufficient insight into the quality of critical materials. All of this needs to be carefully considered, then: otherwise, these omissions could trigger change later. Of course some manufacturing changes are inevitable, but by being prepared – by understanding the critical quality attributes of Volume 14 Issue 1
Therapeutics
the product and working out comprehensive comparability plans – manufacturers can have control over the fallout. Taking care over this early groundwork involves engaging the right blend of experts from the outset: those who understand the product well from a development point of view; and who have a strong grasp of all the various regulatory aspects. They’ll also need expertise and experience around the requirements linked to establishing a GMP manufacturing process – or the specifics of transferring the process into a clean room (with a proper readout, tracing the influence of those manufacturing changes), to ensure product quality. 2022 & Beyond: What Lies Ahead More advanced therapies are progressing through the later stages of development now, which means we can expect to see more product approvals soon. The competition is undoubtedly growing and an interdisciplinary approach to planning is becoming more essential – to ensure from the outset that nothing is missed. What is so exciting about cell and gene therapy is the genuine potential to cure disease, not just alleviate the symptoms. We’ve seen the example of CAR T-cell therapy, and the potential which gene therapies have, in fighting disease very efficiently and we can be confident of more ground-breaking examples coming to market in the upcoming years. To maximise the potential of the scientific knowledge, which is increasing at pace, and of novel manufacturing technologies, www.journalforclinicalstudies.com
there will need to be strategies and approaches that foster and support this innovation and which allow products to be developed, manufactured and brought to market efficiently. The good news is that stakeholders from scientists and regulatory authorities to biotech companies and experienced consultants are working together to develop suitable strategies to cover all requirements, so the right kind of help should be close at hand as development cycles advance.
Elena Meurer Elena Meurer is Principal Consultant and CMC expert at Biopharma Excellence, which helps dynamic biopharma companies bring innovative new therapies to market by providing strategic, proactive practical support across their product journey. The Biopharma Excellence team is a fusion of three scientific powerhouses that came together under the PharmaLex brand. It combines more than 35 years of empirical experience and respected regulator relationships. Elena’s background is in quality, manufacturing and regulatory, with cell and gene therapies as core area of expertise. Email: elena.meurer@biopharma-excellence.com Web: www.biopharma-excellence.com
Journal for Clinical Studies 23
Therapeutics
Using Machine Learning to Identify At-Risk Sites in Acute Schizophrenia Clinical Trials Data quality concerns are frequent in schizophrenia clinical trials, causing many to suffer from decreased drug placebo separation. Machine learning offers the opportunity to proactively identify raters and sites at risk of developing data quality concerns for early intervention. CNS clinical trials tend to have a low success rate. In fact, only 1 in 14 molecules entering phase 1 clinical testing reaches the market.1 Given the high risk associated with CNS drug development, it’s no surprise that many pharmaceutical companies avoid researching this therapeutic area altogether. Schizophrenia clinical trials are no exception. These studies are particularly vulnerable to failure because they utilise relatively complex, subjective endpoints, the patientreported outcomes are often inconsistent or unreliable, and drug placebo differences are modest at best. Successful development of new antipsychotics has been made even more difficult by increasing placebo response and diminishing effect sizes over the last two decades.2 These trends are widely acknowledged to be multifactorial and have been attributed to causes including industry sponsorship, the growing number of sites involved, a higher probability of receiving medication over a placebo, and inadequate data quality. Of these issues, data quality requires continuous attention for the clinical trial’s best hope of success. To better understand how data quality frequently affects datasets, let’s look at an example. Based on an internal analysis of 45,000 clinic visits collected from 6,500 patients in 17 acute schizophrenia clinical trials, the amount of clinically meaningful data quality concerns impact roughly 27% of all study visits. The presence of these data quality concerns has been shown to additionally decrease drugplacebo separation. For example, the presence of outlying variability led to a loss of signal in affected patients in an otherwise successful phase 2 clinical trial.3 Similarly, the presence of erratic ratings in the negative factor of the Positive and Negative Syndrome Scale (PANSS) increased placebo response and decreased drug-placebo separation in a global, phase 3 negative symptom clinical program not only in affected patients but at sites, too.4 Often, data quality concerns accumulate at a small number of research sites.5 Identification of these sites and timely intervention is paramount in maintaining the integrity of the trial, especially should any of these concerning sites recruit a disproportionally large number of patients. Data analytics implementing smart algorithms allow such an early identification of concerning sites and a timely targeted intervention in a form of retraining, remediation, rater replacement, or site closure. Given the detrimental effect of data quality concerns on study outcomes, sponsors often implement a battery of solutions to identify and prevent these data concerns in addition to their data analytical programs. In acute schizophrenia clinical trials, these solutions typically consist of: • site selection based on previous performance; • pre-study calibration of interview and symptom severity measurement technique; 24 Journal for Clinical Studies
• • • • • •
placebo response mitigation training; operationalisation and monitoring of acuity criteria; enhanced instructions and data quality checks embedded in eCOA; recording and independent expert review of audio recorded PANSS interviews; rapid remediation of rating and interview errors; and site enrolment continually tied to data quality.
The application of all these measures has repeatedly shown to improve the overall quality of collected data,6 yet a non-negligible proportion of data concerns remains in the datasets. Ideally, an effective solution should offer not only a retrospective identification and resolution of data quality concerns but also a prospective identification of raters and sites at risk of developing these data quality concerns in the future. With the advances of machine learning this has now become possible. (Figure 1)
Figure 1: Paradigm Shift in Data Analytics Leveraging Machine Learning In clinical research especially, it is critical to apply machine learning models with extreme caution and realistic expectations. Machine learning models should strive to be generalisable, offer actionable results, and address only the clinically meaningful issues. To maximise the benefits machine learning offers over traditional statistical methods, the technology should be generalisable and allow seamless implementation across clinical trials from the first patient visit. This proves especially challenging in CNS clinical trials due to the variability in study design, treatment allocation, efficacy outcomes, and visit schedules between clinical trials. Researchers are continually refining study designs for the best chance at providing valid readouts. Thus, two clinical trials are rarely alike and the number of useful commonalities between trials is limited, even within the same indication. This makes the development and implementation of generalisable complex models challenging, especially since they rely on a plethora of data types and features (variables) unique to the trial. While various data manipulation and feature engineering techniques could possibly overcome some of these issues, these approaches would unnecessarily increase the complexity of the models and the risk of inaccurate results. In addition, machine learning models should offer actionable outcomes. If a rater or a site is identified as a data quality risk, there should be a clear set of actions defined upfront with customised Volume 14 Issue 1
Therapeutics
Figure 2: Building the Pipeline interventions automatically carried out. Such an approach leaves no room for interpretation and allows a uniform response to identified risks. Finally, and arguably most importantly, the considered models should address only those data quality concerns that are clinically relevant and have the largest potential to significantly affect study outcomes. While correct identification of minor clinically insignificant issues could theoretically improve the performance characteristics of the overall machine learning pipeline, the consequences of such an approach could be risky. Sites and raters would likely become overburdened and alienated by frequent, unnecessary contacts, while sponsors would experience frustration over the program’s inability to discern important from unimportant issues, ultimately resulting in the analytical solution being removed from the study and negatively impacting the data. It is thus critical to carefully select the issues that have repeatedly shown to strongly impact placebo response and drug placebo separation, or the issues that are clinically improbable and are associated with data tampering or fraud. Successful Implementation for PANSS Inconsistencies The Positive and Negative Syndrome Scale (PANSS) is one of the most frequently used primary efficacy outcomes in schizophrenia research. The scale consists of 30 items, each of which is rated from 1 (absent) to 7 (extreme). Many items measure similar symptoms and are therefore expected to show a strong degree of association. For example, a patient suffering from severe persecutory delusions should have a high score on both, generic Delusions (P1) and specific Suspiciousness/Persecution (P6) items. A failure to do so is an indication of an error (logical within PANSS inconsistency) that could originate in incorrect or idiosyncratic application of PANSS scoring guidelines, inadequate understanding of underlying
psychopathology, rating sloppiness, or, in the worst-case scenario, data fabrication. The presence of these logical inconsistencies within PANSS has been previously shown not only to significantly increase placebo response in both affected patients and sites but also eliminate the drugplacebo separation.7 These findings thus make logical inconsistencies a good target for the application of machine learning pipelines. Given the obstacles and goals mentioned previously, Signant Health’s team built a layered machine learning pipeline. (Figure 2) The first defensive layer is a parsimonious model that uses only the most common screening data available (in this case just the PANSS scale), so that as the patient is screened the data is initially processed by this screening model only. As the patient progresses to baseline, a second more complex model will be utilised that includes both screening and baseline data. If audio review of the study visits and PANSS administration were available, once the reviewer data become available, a third model would be implemented. Lastly, with critical mass of data within the study available, a fourth model could be implemented that would rely not only on the data commonly collected across trials but also the study-specific measures. You can appreciate that with each layered model the complexity increases, and the generalisability of the models decreases. Once the unique study data is used, these models become study specific, and thus, not reusable. If we assess the performance of the pipeline (Figure 3), it can be noted that the performance improves as expected with the more complex models, however the gains in the performance decrease with the additional complexity of the models.
Figure 3: How Does the Pipeline Perform? www.journalforclinicalstudies.com
Journal for Clinical Studies 25
Therapeutics
2.
3.
4.
5.
6.
7.
Innovation Organization (BIO); Informa Pharma Intelligence; QLS. Available online at https://pharmaintelligence.informa.com/~/media/ informa-shop-window/pharma/2021/files/reports/2021-clinicaldevelopment-success-rates-2011-2020-v17.pdf, checked on 1/3/2022. Leucht, Stefan; Leucht, Claudia; Huhn, Maximilian; Chaimani, Anna; Mavridis, Dimitris; Helfer, Bartosz et al. (2017): Sixty Years of PlaceboControlled Antipsychotic Drug Trials in Acute Schizophrenia. Systematic Review, Bayesian Meta-Analysis, and Meta-Regression of Efficacy Predictors. In The American journal of psychiatry 174 (10), pp. 927–942. DOI: 10.1176/appi.ajp.2017.16121358. Kott, Alan; Brannan, Stephen; Wang, Xingmei; Daniel, David (2021): The Impact of Aberrant Data Variability on Drug–Placebo Separation and Drug/Placebo Response in an Acute Schizophrenia Clinical Trial. In Schizophrenia Bulletin Open 2 (1), p. 295. DOI: 10.1093/schizbullopen/ sgab037. Umbricht, Daniel; Kott, Alan; Daniel, David G. (2020): The Effects of Erratic Ratings on Placebo Response and Signal Detection in the Roche Bitopertin Phase 3 Negative Symptom Studies—A Post Hoc Analysis. In Schizophrenia Bulletin Open 1 (1), p. 203. DOI: 10.1093/schizbullopen/ sgaa040. Daniel, David G.; Kott, Alan (2014): Risk Based Data Quality Monitoring Utilizing Data Analytics and Recorded PANSS Interviews in Global Schizophrenia Trials. Poster presentation at the 10th Anniversary International Society of Clinical Trials Methodology (ISCTM) Meeting, Philadelphia, PA, 18–20 February 2014. Kott, Alan; Brannan, Steven K.; Wang, Xingmei; Murphy, Christopher; Targum, Steven D.; Daniel, David G. (2020): Procedures to Optimize Endpoint Data Quality in an Acute Schizophrenia Study. Presented at the ISCTM 2020 Autumn Virtual Conference, September 21 – 25, 2020., 9/21/2020. Kott, Alan; Lee, Jeniffer; Forbes, Andy; Pfister, Stephanie; Ouyang, John; Wang, Xingmei; Daniel, David G. (2016): Logical inconsistencies among PANSS items are associated with greater placebo response in acute schizophrenia trials – A post-hoc analysis. Poster presentation. Philadelphia, PA, USA, 9/26/2016.
Dr. Alan Kott
This does not mean that the more complex models should be discounted; it merely informs us that the simpler models are perfectly capable of identifying risks without a huge loss in the model performance. In the final step, the outcomes of the machine learning pipeline are statistically analysed, and at-risk raters and sites are retrained. This step further improves the performance of the system and decreases the risk of possible over contacting. In conclusion, machine learning offers a viable analytical solution allowing researchers to predict and address future data quality concerns throughout a study. However, one needs to be realistic and cautious about the implementation of machine learning. Only highly accurate and clinically relevant models should be considered, as the consequences of inaccurate or irrelevant models may ultimately result in data quality deterioration. A layered, onion-like structure allows studies to implement parsimonious models collecting from the clinical trial’s first patient visit and add more complex, study-specific models as subsequent data becomes available to further improve the pipeline performance. REFERENCES 1.
Dr. Kott oversees the design and reporting of Signant Health's data analytics in large schizophrenia studies. For the past seven years, he has also lent his valuable expertise to training investigators based on best practices.
Andrei Iacob As Associate Clinical Data Scientist, Andrei supports Signant Health's Blinded Data Analytics and related initiatives. He continues to lend his years of experience to support various machine learning projects.
Emanuel Pintilii Emanuel is serves as a Clinical Data Scientist for Signant Health and has been instrumental in the company's development of different machine learning systems.
Xingmei Wang Xingmei Wang is a Senior Statistician on Signant Health's Science & Medicine team, providing company-wide, statistical support.
Thomas, David; Chancellor, Daniel; Micklus, Amanda; LaFever, Sara; Hay, Michael; Chaudhuri, Shomesh et al. (2021): Clinical Development Success Rates and Contributing Factors 2011–2020. Biotechnology
26 Journal for Clinical Studies
Volume 14 Issue 1
INSIGHT / KNOWLEDGE / FORESIGHT
SUPER PUBLICATIONS FOR SUPER PHARMACEUTICALS IPI
Peer Reviewed, IPI looks into the best practice in outsourcing management for the Pharmaceutical and BioPharmaceutical industry.
www.international-pharma.com
JCS
Peer Reviewed, JCS provides you with the best practice guidelines for conducting global Clinical Trials. JCS is the specialist journal providing you with relevant articles which will help you to navigate emerging markets.
PHARMA’S DNA
www.journalforclinicalstudies.com
Listen to industry experts on the latest in drug discovery, development, research, industry regulations and much more at Pharma,s DNA, the podcast channel by Senglobal Ltd., available on Sound Cloud, Spotify, iTunes and YouTube.
IAHJ
IBI
www.international-animalhealth.com
www.international-biopharma.com
Peer Reviewed, IAHJ looks into the entire outsourcing management of the Veterinary Drug, Veterinary Devices & Animal Food Development Industry. www.journalforclinicalstudies.com
Peer reviewed, IBI provides the biopharmaceutical industry with practical advice on managing bioprocessing and technology, upstream and downstream processing, manufacturing, regulations, formulation, scale-up/technology transfer, drug delivery, analytical testing and more. Journal for Clinical Studies 27
Technology
AI/ML to Generate Medical Insights… While Maintaining Patient Data Security and Privacy Machine Learning (ML) and Artificial Intelligence (AI) have come to the forefront of data analytics with the promise of generating new medical insights. However, for healthcare data, patient data security is paramount due to the GDPR and similar regulations. Traditional methods of data consolidation for Machine Learning/ AI modelling into a single warehouse or a data lake are often not possible even with anonymised data due to data protection rules. The new, emerging alternative is a federated in situ data analytics construct in which healthcare data are anonymised within the care facility and accessed via a secure cloud application. This paper demonstrates how this ensures data security within the original data domain, while allowing analytics for modelling and AI to be applied in a federated fashion. Introduction – The Goal of Healthcare RWD is to Gain Insights in the Patient’s Clinical Journey Increasingly, understanding the patient journey through their healthcare system is being recognised as critical to modeling and understanding different patient outcomes and care metrics. Patients are complex, with not just a single factor between diagnosis, treatment, and outcome. Especially for chronic diseases, the social demographics, the pharmacogenomics , the time between diagnosis and treatment, the distance to a healthcare provider, existing comorbidities and health risks play in the longitudinal metrics of care and probable outcomes.
Healthcare has become multifactorial and is by nature an open system, with dynamic, sometimes unpredictable, and often chaotic behaviour. The recent and ongoing COVID-19 pandemic is a perfect example of this: the evolution and spread of the vaccine has not been predictable or anticipated, nor has the uptake of vaccine treatments been global or sufficient on a voluntary basis to prevent the further spread of the virus. A doctor’s first assessment of a patient often refers to any previous records possible with the first review looking for any immediate changes or reported conditions. However, this retrospective review, must now become more detailed and more extensive, especially considering multiple conditions or comorbidities the patient may present. This is often more than a single physician in a single visit can manage but is increasingly available through digital data sources (provided below) that physicians, caregivers, epidemiologists, and healthcare researchers can use to model, test, and validate various healthcare questions. In their article on understanding the care pathway/patient journey, the authors point out that “Quantitative analyses carried out to generate deeper insights into any unmet needs and patient subpopulations that may benefit most from the new treatment... may involve: • • • • •
pharmacy claims data analyses electronic medical record database analyses retrospective patient chart reviews analysis of registry data cohorts or longitudinal studies
Figure 11 A typical care pathway starting with initial symptoms, diagnosis and progressing through treatment methodologies into various outcome metrics. The overall patient journey can be predictive to outcomes and overall patient care metrics. 28 Journal for Clinical Studies
Volume 14 Issue 1
Technology • •
epidemiology or treatment pattern studies burden of illness studies.”1
The advent of the digitalisation of patient care data has much improved the prospects of showing a physician a unified overview of a patient’s medical care journey. However, the relatively recent General Data Protection Regulation (GDPR) in the EU is a herald for tighter personal data protections than before, and this has now to be accommodated for us to achieve this goal. Two Regions, Two Models – How GDPR Has Changed What’s Needed With over 330 million citizens, the US block of healthcare data largely exceeds any other country, except China and India, and is the single largest high reimbursement country in the world. The EU plus 1 countries, which traditionally means the UK, France, Germany, Italy, and Spain, are also considered as high healthcare reimbursement markets, but both individually and collectively their size is still slightly less than the USA.
Table 1: Population of the EU5 and the USA2 Although together slightly smaller than the USA, the disparate healthcare systems, languages, regulations, and standards of care in the EU, tend toward more fragmented healthcare data sources and RWD availability compared to the USA. Therefore, the use of healthcare data across these two large blocks of high reimbursement healthcare patient populations, that are largely separate and independently regulated, has become highly challenging. When it comes to the use of real-world patient data for the generation of medical insights, HIPAA healthcare data regulations (in the USA) allows greater secondary use including implied consent, while GDPR regulations (in the EU) require explicit patient permission for data use and have strict provisions for personal and private data protections. Digital Healthcare as the Next Forefront While digital data is clearly important to healthcare, what comes into question is its application across disparate healthcare systems and regulations. There are increasingly more data and more capabilities unlocked by them. However, technological enablement across one healthcare system, such as the US, does not immediately mean global enablement across separate, disparate, and differently regulated healthcare systems. GDPR has a significant impact on data access and usage, not only within the EU but regions adopting similar regulations, and even within the US. The globalisation of digital healthcare data does not necessarily mean equal access to healthcare data. Data access and use permissions differ across different parts of the world, and even country to country, disparate systems of healthcare are still the rule, and are just as fragmented as the general accessibility of healthcare itself on a global scale. www.journalforclinicalstudies.com
Our purpose is to examine the disparate systems for high reimbursement healthcare in the US and the EU, both of which have their pros and cons. HIPAA and GDPR Differences in a Nutshell In 2012, a now infamous story appeared of how the retailer Target identified from shopping patterns that a family’s teenage daughter was pregnant.3 Shopping habits and consumer data being used to segment and target clients based upon their purchasing patterns are now common not only in retail sales.4 However, these purchasing patterns are also a key metric in credit card fraud protection programs,5 in the same way that patient prescription patterns can be used to understand a patient's care journey. Social media companies such as Facebook, Google, and others readily use their consumer data to develop targeted consumer demographics for their own messaging and to sell advertising. One of the key differences between HIPAA and GDPR is that the latter requires explicit permission from an individual for their personal data to be collected and used (unless covered by specific GDPR exemptions), as ownership intrinsically belongs to the individual. GDPR imposes significant penalties for collection and use of personal and private data without proper consent. Whereas in the US and other countries, the use of a data application on your smart device includes an inherent agreement in the Service-Level Agreement (SLA) code for the application that data can be collected and re-used, in the EU, the philosophy is that the consumer controls third party use of their own data and can decide to retract their permission at any time. The impact on our story is that patients always have implicit rights to and ownership of their data. This impacts the use of personal and private information, as well as pseudonymised data, as these carry a potential risk of potential re-identification to protect the security of healthcare data. Data Harmonisation versus Data Interoperability US healthcare data are extremely harmonised and readily available through primary resources and a variety of data resellers as claims and EMR data. There is much more variation in the data from the big five European countries based upon local language and coding specifics, and subject to local and national data use restrictions in addition to the GDPR data privacy and protection provisions. US data use provisions allow sale and direct ownership transfer of data, whereas under the GDPR provisions and patients’ rights, the actual transfer of ownership rights is not possible. This requires the creation of a structure for interoperability of data where harmonisation is not possible. Data Sharing without Sharing In the US, large-scale data aggregation is readily possible and common; however, in the EU, as we have seen, this is significantly more difficult due to tighter patient privacy regulations and has further challenges in data harmonisation across multiple languages, coding, and reimbursement practices. Increasingly, the best option is to maintain the original healthcare data in situ and develop searches and integration models across a federated network of originating data sites, regardless of whether they are government or healthcare institutional sites. The result is that a federated network capability in the EU holds more potential for success (this aligns with the federated structure of the EU, itself) than the harmonised data and common data structure of the US. The EU is a hybrid system of negotiated agreements between governments and a supranational union and Journal for Clinical Studies 29
Technology therefore more like a federated network. The United States is a single constitutional federal republic. In some sense this mimics Alexis de Tocqueville’s comments on the tyranny of the majority in the US.6 The US healthcare system is the largest, has the most available data, and is the most easily accessible, and therefore sets the global benchmark. By comparison, the EU’s structure and privacy provisions do not allow it similar capabilities or clout. The Application of AI to Healthcare Data The inherently differential approaches to patient rights and data protections, which lead to different approaches in both technology and data application use cases between the US and EU, are also evident in the application of ML/AI in Healthcare in the two regions. In the United States, the large population living under limited privacy controls and a harmonised healthcare system means that healthcare data is plentiful and readily available, so traditional AI methods of data aggregation, pooling and sampling are possible. In the EU, multifarious coding, languages, ontologies, and data movement restrictions make traditional AI methods impractical. The solution in the region would have to be in the various alternatives to the transfer of data, such as Data Fabric technology and Federated Learning. The remainder of this article will discuss these approaches. Data Fabric Data Fabric is a data architecture methodology that creates relationships across various metadata points within disparate or even unconnected data sources, and thereby allows specific relationship maps to be created and followed. An example in the context of healthcare data would be the ability to link patient care patterns in localised electronic healthcare records (EHRs) with geographically specific pharmacy claims data, and then with overlapping physician registry information by Zip code, thereby enabling the mapping of specific physician–patient diagnoses and treatments with the direct costs of care by treatment centercentre and location. The specific connectivity between the disparate data sources does not have to be common data elements, or primary keys, such as in a data warehouse, but rather common metadata patterns that can be mapped or linked into a relational “fabric” across the differential data landscape.
An example of this would be a federated EHR network across different countries and languages where a composite phenotypic cohort model could be searched for across the differential metadata linkages. Data Fabric Rare Disease Use Case Rare diseases are by nature often undiagnosed, misdiagnosed, and untreated. In addition, rare diseases are often heterogenous in symptoms and therefore hard to diagnose, leaving room for ambiguity in diagnoses. The delay in diagnosis that arises makes it very difficult for patients, their families, and caregivers to manage their medical journey. Studies show that the impact of a rare disease is much wider than on the affected individual and represents a significant challenge for the healthcare system itself.7 In a survey of patients and caregivers in the USA and the UK, patients reported that it took on average of 7.6 years in the USA and 5.6 years in the UK to get a proper diagnosis, during which patients typically visited eight physicians (four primary care and four specialist) and received two to three misdiagnoses.8 Of the 7,000 known rare diseases, 90% do not have an FDA-approved medication, which means patients must live with no treatment or go with off-label use of existing medicines to treat their symptoms.9 Patients with rare diseases can live up to 20 to 30 years before diagnosis, or even entirely undiagnosed during their lifetime.10,11,12 The Data Fabric composite cohort model, which may represent several different phenotypic expressions of, for example, a rare disease, will link different symptoms, treatments, and combinations that an undiagnosed rare disease patient has had but not responded to. This can be used as a predictive and relational data fabric to potentially find rare disease patients that have never been correctly identified. The authors of a recent white paper explain that “Data fabrics are particularly useful for deep learning use cases because they reduce “fuzziness” that often results from algorithmic training across numerous types of data.”13
Figure 2: 14 A schematic of federated learning where (a) a model is distributed across various nodes from the central cloud, to individual institutional incidences, with (b) as an example institution, wherein the model learns from the local data source. In step (c), the model improvements from all the various individual (b) nodes are shared back to the federated cloud architecture. In (d), the various models are consolidated before being redistributed again (a) in a repetitive cyclical pattern. 30 Journal for Clinical Studies
Volume 14 Issue 1
Technology
Figure 3: We applied these approaches to a sponsored search for Fabry, Pompe, Gaucher and Mucopolysaccharidosis Type 1 diseases in Turkey and in the United Arab Emirates, using an existing real-world data platform linking an international network of hospitals.18 The figure above shows traditional direct identification through diagnostic codes, and bottom right, triangulation and predictive outcome metrics using related symptom or phenotype models to highlight underdiagnosed Pompe patients. Federated Learning Federated Learning is the architectural framework based upon a single global server with decentralised data across many differential client servers. The goal of Federated Learning is to apply discrete models on multiple client servers and allow them to iterate and learn across the disparate data centers and learn collectively through the central global server. The advantage of this methodology is that data are not aggregated or pooled but stay locally in the original host server and all that is transacted is the model outputs or learning from the federated framework. At the same time, models can adapt across disparate data centerscentres and iteratively learn across the aggregate without data pooling.14 The advantages to healthcare institutions as well as research sponsors for federated learning are significant, as highlighted below: Healthcare Institutions Data stays secure with no transfer rights. Personal and Private information remain secure. Data access is selectable. Encourage cooperation. Sponsors Access to data that is otherwise not accessible. Strict access to only anonymised records. Data access is for specific purpose. Creates collaboration. Federated Learning Rare Disease Use Case Typically, you need to look across around 10,000 features in a full EHR to find a relevant patient, reliably. With typical ML methods and standard toolsets, you expect to have to have around 50,000 “labels” (examples of patients with the disease in question) to allow the machine learning to generate a reliable model. As can immediately be gathered, this is an impossible threshold, as, by definition, the rare disease patients are very rare and often misdiagnosed, so “hidden” within the system. The solution is to use the federated nature of the partner hospital network as the backbone for a federated learning layer over a cloud infrastructure. This is a breakthrough for patients that are likely to be held up in a lengthy “diagnostic odyssey,” since the models can be adapted to the healthcare systems now on our own platform through our partnerships with hospitals around the world. www.journalforclinicalstudies.com
It is critical to construct prediction models which are both accurate and interpretable. As in all medical applications, it is essential that clinicians understand the basis for the predictions and recommendations of decision-support systems. One way to increase interpretability of the complex models produced by modern ML algorithms (e.g., deep learning, ensembles) is to identify which predictors/features are ‘important’ to the model’s predictions and to quantify this importance. Within rare disease, this means looking at the patient clinical journey and identifying cognitive biomarkers, digital biomarkers and medical biomarkers that drive a mechanistically predictive rare disease model: • • •
Cognitive biomarkers are objective measurements that can be used to track the progression of a disease or the outcome of a treatment.15 Digital biomarkers are where the actual data are informative in some way about the disease. Physical biomarkers are phenotypic features of the patients that are predictive of the disease.
These biomarkers are discovered by the model learning process, and we often find them out only as the model improves, so we can subsequently derive clinically interpretable models. The overall metric is to develop phenotypic models based upon actual patient journey that represent all the possible presentations, symptoms, and medical conditions, separately and in various combinations. The advantage of this approach over a federated network is that these models can be developed, shared, and learn (evolve) in a collaborative or dataprivate process for collaborative learning, institutional incremental learning, or cyclical institutional incremental learning.16 The point is that federated learning enables incremental and progressive modelling and model learning across discrete datasets rather than nodes securely and effectively. By working with and across smaller datasets, the federated network creates a greater networked database.17 Interestingly, Federated Learning has been shown to reach performances comparable to traditional centralised data model analytics across diverse therapeutic research topics (heart failure, diabetes, MIMIC-III, SARS-CoV-2, Avian Influenza, Bacteremia, Azithromycin, and Tuberculosis), while preserving privacy. As in the case of Federated Learning to identify rare disease patients that have been underdiagnosed, we have applied these approaches in sponsored research for lysosomal storage disorders including Fabry, Gaucher, POMPE, and Mucopolysaccharidosis Journal for Clinical Studies 31
Technology
Figure 4: AI/ML predictive federated learning models developed by EvidNet across a Korean hospital network to predict chronic disease onset based upon healthcare screening metrics, with modelingmodelling and global AI optimisation based upon individual site model learning. Type 1 (MPS1) in Turkey and the United Arab Emirates. The iterative models are being tested and now being clinically validated in studies, in which selected patients are tested for diagnosis as part of a clinical outreach study. The triangulation of potentially underdiagnosed patients using related symptoms/phenotypes/biomarkers, represents the effectiveness of learning to understand digital signals within the patient journey to better triage and flag patients who may not 32 Journal for Clinical Studies
normally be identified as part of traditional patient screening for diagnostic review. In a further example of Federated Learning, the Korean company EvidNet has developed a federated learning capability across their hospital EHR network. In this use case, the AI/ML algorithm is being used to develop disease prediction models based on health Volume 14 Issue 1
Technology screening inputs to predict the likely onset of chronic diseases. Figure 5 shows an example of the data flow, global optimisation of the AI/ML model and high-level model metrics. Conclusions This article has discussed how disparate healthcare systems and access to data often dictate different strategic approaches to analysis and modelling. Circumventing these constraints can be achieved with Data Fabric and Federated Learning, in cases where data cannot be readily extracted or consolidated. These methods are proving to be effective and comparable to traditional centralised models, meaning that different approaches to data access and modeling can be effective and comparable. Likely, not one approach necessarily can work on a global scale, but each in concert can enable perspective and insight that contribute to our global vision and understanding of healthcare and patient care, specifically. The larger aim of this review is to create the understanding that no single methodology is necessarily better, and that any solution approach needs to take into account access to data, heterogeneity within the data and the extent of harmonisation possible. The global need is for digital enablement of healthcare, better insights, patient treatments and outcomes. The patient clinical journey is available across many EHR systems in a SMART hospital reference (https://healthcareglobal.com/hospitals/what-smarthospital). The key is making this accessible and useful for patient care and stratification, not solely care reimbursement.
13.
REFERENCES 1.
2. 3.
4.
5.
6. 7. 8.
9.
10.
11.
12.
Joyeux A, Olivaris R, Understand the care pathway/patient journey, available at https://rwe-navigator.eu/using-the-navigator-decisionsupport-tool/clarify-the-issues/understanding-the-patient-journey/ Statistics taken from https://www.worldometers.info/population/ countries-in-europe-by-population/ Hill K, How Target Figured Out A Teen Girl Was Pregnant Before Her Father Did, February 2012, available at https://www.forbes.com/sites/ kashmirhill/2012/02/16/how-target-figured-out-a-teen-girl-waspregnant-before-her-father-did/?sh=413c93b16668 Duhigg C, How Companies Learn Your Secrets, February 2012, available at https://www.nytimes.com/2012/02/19/magazine/shopping-habits. html Baboo SS, Preetha N, Analysis of Spending Pattern on Credit Card Fraud Detection, IOSR Journal of Computer Engineering (IOSR-JCE), e-ISSN: 2278-0661, p-ISSN: 2278-8727, Volume 17, Issue 2, Ver. 1 (Mar–Apr. 2015), PP 61-64, available at https://www.iosrjournals.org/iosr-jce/papers/ Vol17-issue2/Version-1/J017216164.pdf Mansfield HC, Winthrop D, Alexis de Tocqueville, Democracy in America. Chicago: University of Chicago Press; 2000 Drake D, Finding and Treating Rare Disease Patients in a Global Digital Haybale, Journal for Clinical Studies, Volume 12 Issue 4, September, 2020. Shire Report 2013, Rare Disease Impact Report: Insights from patients and the medical community, 2013, available at: https://globalgenes.org/ wp-content/uploads/2013/04/ShireReport-1.pdf Toth Stub, S, Conquering Rare Disease – Should taxpayers keep paying to develop drugs for unusual disorders?, 2020, available at: https://library.cqpress.com/cqresearcher/document.php?id=cqresrre2020 012400&type=hitlist&num=0) Mehta A et al., Fabry disease defined: baseline clinical manifestations of 366 patients in the Fabry Outcome Survey, European Journal of Clinical Investigation (2004), 34, 236–242 Mistry PK et al., Timing of initiation of enzyme replacement therapy after diagnosis of type 1 Gaucher disease: effect on incidence of avascular necrosis, British Journal of Haematology, 147, 561–570 Muenzer J et al, Ten years of the Hunter Outcome Survey (HOS): insights, achievements, and lessons learned from a global patient registry, Orphanet Journal of Rare Diseases (2017) 12:82
www.journalforclinicalstudies.com
14.
15.
16.
17.
18.
How Healthcare Organizations are Improving Time to Insights with Data Fabrics, TechTarget Custom Media White Paper, available at How Healthcare Organizations are Improving Time to Insights with Data Fabrics, available at https://www.delltechnologies.com/asset/ en-us/solutions/industry-solutions/industry-market/how-healthcareorganizations-are-improving-time-to-insights-with-data-fabrics.pdf Xu J, Glicksberg BS, Su C, Walker P, Bian J, Wang F, Federated Learning for Healthcare Informatics, Journal of Healthcare Informatics Research, November 2020, available at https://doi.org/10.1007/s41666-020-00082-4 Torous J, Keshavan M, A new window into psychosis: The rise digital phenotyping, smartphone assessment, and mobile monitoring, Schizophr Res. 2018;197:67-68. doi:10.1016/j.schres.2018.01.005. Sheller MJ, Edwards B, Reina GA, Martin J, Pati S, Kotrotsou A, Milchenko M, Xu W, Marcus D, Colen RR, Bakas S, Federated learning in medicine: facilitating multi-institutional collaborations without sharing patient data, July 2020, available at https://doi.org/10.1038/s41598-02069250-1 Sadilek A, Liu L, Nguyen D, Kamruzzaman M, Serghiou S, Rader B, Ingerman A, Mellem S, Kairouz P, Nsoesie EO, MacFarlane J, Vullikanti A, Marathe M, Eastham P, Brownstein JS, Aguera y. Arcas B, Howell MD, Hernandez J, Privacy-first health research with federated learning, npj Digital Medicine, September 2021, 4:132; available at https://doi. org/10.1038/s41746-021-00489-2 Drake D, Rudolf C, Strategies Toward Identifying Undiagnosed Rare Disease Patients, poster presented at Frontiers of Pediatric Genomic Medicine, April, 2021.
Douglas Drake Douglas Drake, MS, MBA, is originally a life science researcher with a passion for digital enablement of better patient care. For over 30 years, Douglas has worked in various aspects of diagnostics, therapeutic research, drug discovery and global business development. He has broad experience in transformative technologies, data sciences, digital healthcare and applying these to improving patient engagement and the patient journey.
Journal for Clinical Studies 33
Technology
Moving to Patient-Centric Pharmacovigilance
Advanced Technologies are Harnessing Patient Information to Drive Better Safety Data on patient experiences with new drugs and therapies promises to play a prominent role in drug reviews and approval.1 Analysis of this information could help identify and address adverse drug reactions and safety problems before they occur. Currently, these problems cost the industry nearly $30 billion per year.2 The increase in patient-reported safety data has led to growth in pharmacovigilance workloads. Safety leaders have more data sources and values to analyse, making identifying true safety signals and trends harder. The lack of data standardisation within the industry3 brings additional complexity and requires more time and resources to aggregate information for analysis. The industry is looking for better ways to leverage patient safety data effectively and efficiently. Modern pharmacovigilance solutions bring together information while streamlining scientific and regulatory processes. These new tools are helping safety leaders to drive patient focus, use new sources of data, and improve collaboration with stakeholders. Modernizing Pharmacovigilance to Increase Patient Focus Growing case volumes and new data sources are leading to an increase in outsourcing of safety to contract research organisations (CROs) generally low risk, high volume functions such as processing nonserious cases. By 2026, demand for outsourced pharmacovigilance services is expected to reach $10.6 billion,4 increasing the need for solutions that improve collaboration, provide greater visibility into safety data, aid analysis, and make compliance easier. “Many of the systems used for pharmacovigilance dated back to the 1990s,” said Martijn van de Leur, head of global pharmacovigilance for Biomapas, a CRO that offers clinical trial, regulatory, and pharmacovigilance services. “Given the strict regulations affecting pharma, many have been afraid of adopting new technologies. There is a shift happening as more companies openly embrace change because they see the larger benefits to help meet compliance standards and reduce cost without affecting quality.” Cloud-based systems play a significant part in bringing together stakeholders and supporting end-to-end processes. They also help drive standardisation of data and reporting to aid the monitoring, analysis, and sharing of safety information. In the end, if leaders have the tools to streamline pharmacovigilance processes and easily collaborate with partners, they can enable a greater focus on patient safety. Leveraging New Data Sources Artificial intelligence (AI) can be used to sift through terabytes of data and automate manual, error-prone processes. Streamlining the analysis of information leads to faster determinations of risk levels and drug safety signals. The use of AI, including machine learning and natural language processing, promises to enable predictive pharmacovigilance by automating the identification of adverse event trends and potential signals. 34 Journal for Clinical Studies
AI can leverage real-world data from patients’ electronic health records and insurance bills to improve drug safety monitoring. Exploring AI to mine other data sources, such as social media, online patient communities, and call centre data can provide further insights into adverse events. AI also allows users to perform more extensive literature searches, improving their ability to identify or predict drug safety trends. Automating Case Safety Reports Automating administrative work, particularly the intake, data entry, and processing of individual case safety reports (ICSRs), is a major potential benefit of AI. Today, many companies still rely on manual processes and paper to handle these crucial functions, increasing the risk of non-compliance and errors and preventing them from keeping up with rising caseloads. According to the FDA, the number of adverse events found in the US tripled from 2010 to 2019, leading to a sharp increase in pharmacovigilance team workloads. By 2017, one study found that a normal large biotech’s pharmacovigilance team was processing more than 200,000 ICSRs, up from 84,960 in 2007.5 Using AI to process these volumes of ICSRs can significantly streamline safety reporting and analysis and allow teams to focus on more severe cases. Deciding which steps to automate first requires an understanding of risks balanced with the time and effort needed to complete a task. A study by the non-profit group TransCelerate found the ICSR to be the most labour-intensive and crucial step in safety because it feeds subsequent processes such as reporting, signal detection, benefitrisk analysis, and risk management. Breaking the process down, researchers found the top priorities for automation to be language translation, case verification, in-line quality control, and case prioritisation and triage.6 Improving Collaboration Through Better Data Access Given the state of the safety data landscape, true cloud-based multitenant solutions are needed to bring together data and content and support end-to-end processes. These solutions should be designed for ease of use and feature interactive dashboards, intelligent automation, and notification prompts. They should also streamline the electronic submission of ICSRs to regulatory authorities and license partners, with workflows that help with routing, escalation, and task completion. NAMSA has found that its cloud safety solution is more intuitive and has simplified training, data access, and report generation, allowing them to deliver a higher level of service. “We can easily set up new studies, add or modify workflows and fields, and create or run reports to support our clients as they move from pre-clinical to commercial,” said Jennifer Kratz, senior product development strategist for pharmacovigilance at NAMSA. “We’re delivering a better level of service because we can align with their processes and provide greater transparency.” For Biomapas, the solution is enabling greater collaboration with clients by providing different levels of data access, depending on needs and preferences. “Some customers want to outsource pharmacovigilance functions entirely and have little to no Volume 14 Issue 1
Technology involvement in day-to-day activities, while others want to be informed regularly or closely involved. Our cloud safety solution gives us options to meet customer requirements, such as granting read-only access to data,” says van de Leur.
“Pharmacovigilance and quality must go hand in hand, and strong quality management systems must be in place to maintain multinational oversight and compliance, along with tracking key performance metrics,” Kohut says.
In the future, van de Leur expects to see better connections between safety, clinical, regulatory, and quality, which are important in supporting pharmacovigilance processes and understanding drug safety from different perspectives. “Connectivity between pharmacovigilance and other cross-functional systems is becoming a key requirement,” he notes.
Pharma companies will need to automate QMS systems to manage the updates coming in from local regulators more effectively and to exchange drug safety information more efficiently, Kohut adds. There may be challenges based on the level of automation at CRO and other contract partners.
Using Technology to Cut Through the Noise Even though the industry is studying AI and applying it to existing operations,7 we are years away from routine use of predictive pharmacovigilance tools. Efforts today focus on automation, aiding processing, better data access and management, and earlier signal detection to protect patients and improve their quality of life. As the science and practice of pharmacovigilance continue to evolve, cloud-based applications will play a crucial role in helping more companies to improve patient safety and reduce risk. “Drug safety has historically been a labour-intensive field. The cloud will change the scope of case processing so that pharmacovigilance departments can focus on higher value-add activities rather than filling out paperwork,” says van de Leur. Keeping Up with Safety Regulatory Change The past decade has brought increasingly stringent requirements designed to protect patients. One of the firsts is the EU’s Product Safety Master Files (PSMFs), which are now required by many countries and by local regulatory authorities within the EU. Cloud-based systems are helping pharmacovigilance teams to standardise data and processes and keep a closer watch on revisions and updates to documents that must be sent to QPPVs and filed with regulatory authorities.
“Some CROs and CDMOs may have fully automated systems, while others won’t,” he adds. Using technologies that allow data to be summarised, shared, and reported via dashboards, a hallmark of cloud-based systems will make it easier for companies to meet requirements. REFERENCES 1.
2. 3. 4. 5.
6. 7.
P. Kruger and C Gasperin, “The Value of Direct Patient Reporting in Pharmacovigilance,” Therapeutic Advances in Drug Safety, October 26, 2020. J. Sultana, Pl Cutroneo, et al., “Clinical and Economic Burden of Adverse Drug Reactions,” J Pharmacol Pharmacother.12(4)(Suppl1): S73–S77, 2013. “How to Use Pharmacovigilance Methods to Detect Safety Issues,” Clinical Leader, December 2015 “Pharmacovigilance Outsourcing Market Share, 2026 Forecast,” Global Market Insights, December 2019. S. Stergiopoulos, M. Fehrle, et al., “Adverse Drug Reaction Case Study Practices in Large Biopharmaceutical Organizations from 2007 to 2017,” Pharmaceut. Med 33(6) 499-510 (2019). R. Ghosh et al., “Automation Opportunities in Pharmacovigilance: An Industry Survey,” Pharmaceutical Med. 35(2), February 6, 2020. J. Kratz and K. Traverso, “Modernizing Pharmacovigilance Outsourcing,”veeva.com, February 2, 2020.
Standardisation is essential because changing regulations, particularly on the local level, have led to an explosion in the amount of drug safety data that companies must monitor and process from different regions and languages. “This data comes in different formats (e.g., Microsoft Excel and ASCII), so companies must expend more effort to deal with the noise surrounding it,” said Peter Kohut, formerly director of drug safety at the Dublin-based pharmacovigilance services firm Arriello, s.r.o. “Avoiding duplicate reports and data is especially important when companies have multiple markets in the pharmacovigilance system and various authorisations across the same territory.” Business changes (including mergers and acquisitions) and a lack of proper change management can also affect pharmacovigilance systems and departments. A Need for Seamless Processes Between Safety and Quality An increasingly complex regulatory picture also mandates stronger connections between pharmacovigilance and quality management systems for improved operational efficiencies and compliance. These connections, which cloud-based systems can help enable, offer greater visibility into SOPs, CAPAs, change control, and training and allow more reliable tracking of key performance indicators.
www.journalforclinicalstudies.com
Kelly Traverso Kelly Traverso, vice president, Vault Safety strategy, Veeva Systems, is responsible for strategy for the Vault Safety suite of applications. She has extensive knowledge and experience with the US FDA and EMA safety and quality regulations. Kelly has over 20 years of PV experience and applied specialized knowledge of regulatory and compliance aspects of business processes across life sciences to define and implement strategy, design/re-design processes, support technology implementations, and define governance across the spectrum of PV activities.
Journal for Clinical Studies 35
Logistics & Supply Chain
2022 Cold Chain Predictions: Creating a new Normal At this time last year, the world was eagerly anticipating 2021. COVID-19 vaccines delivered hope for normalcy and excitement grew over convenience-related changes to healthcare, work and more. But the pandemic had other plans. New COVID-19 variants emerged, supply chain issues deepened and much of the day-to-day still looks different than pre-pandemic. The cold chain industry experienced rapid growth in 2020 and continues to experience both growth and change for the foreseeable future. The pandemic’s influence remains, which creates opportunities to innovate and better serve pharmaceutical and healthcare customers delivering medicine in new ways. This sets the stage for a few of our predictions this year. While COVID-19’s influence persists, we also see renewed interest in sustainability and the evolving impact of Brexit’s export regulations in the United Kingdom (UK). These will also shape how the pharmaceutical and cold chain industries operate. Let’s take a look at what all of this means for 2022. Outsourcing the Cold Chain In our predictions last year, we anticipated that more pharmaceutical companies would outsource capabilities to contract manufacturing organisations (CMOs) and contract development and manufacturing organisations (CDMOs). Pharmaceutical companies already engage these organisations in manufacturing and development of therapies, but adding additional services allows pharmaceutical companies to focus valuable time and resources on areas where they have the most expertise. We did see a shift last year to CMOs and CDMOs adding additional services, like cold chain logistics. We expect to see outsourcing grow again this year. Offering end-to-end expertise will help reduce additional supply chain complexities by standardising more of the supply chain during a time when raw materials are scarce and transportation is unpredictable, necessitating dedicated and seasoned professional resources.
contract research organisations expected to engage in a high volume of virtual research trials with 48 percent of those expecting to run a trial with most activity conducted in participants’ homes. These numbers increased to 100 percent and 89 percent respectively in December 2020.1 Many research organisations initially piloted this new model of clinical trials with smaller Phase I and Phase II trials. In the past year, we saw organisations pilot fully home-based and hybrid clinical trials in Phase III trials.1 This year we expect to see trials using home-based care grow with continued focus on improving the experience for patients and physicians. Logistics remain a challenge, especially given the narrow timeframe for deliveries and pick up of biologics or sample materials. All timing must coordinate with homecare visits and ensure temperature-sensitive materials arrive at their final destination still within the required temperature range. Services like phlebotomy, drug administration and sample collection that require refrigeration will require cold chain solutions. We anticipate an ongoing drive toward solutions that require little training and are easy for home healthcare professionals and patients to operate. We should also see even more assessment and evaluation of the cold chain for homebased care in 2022. Brexit Runs Smoothly Brexit, or the UK’s exit from the European Union (EU), officially began on January 31, 2020. However, nothing changed until a new trade deal was reached nearly one year later. Implemented in January 2021, the new deal outlined how the UK and EU would live, work and trade together. The most significant concern: new paperwork for export businesses. As new trade rules began, export businesses did in fact experience significant issues with increased paperwork. As a result, a large number of shipments were delayed, held at customs points or cancelled altogether. The impact of Brexit was further affected by the global shortage of shipping containers and lack of drivers to transport goods.
Additionally, we expect companies not yet ready to fully outsource their supply chains to increase their use of services that make cold chain operations easier and eliminate the challenges associated with unforeseen circumstances. These include services like offsite conditioning of coolants or onsite conditioning with coolants inventoried to their unique needs.
Over time, exporting companies, shipping agents and logistics companies began to understand the nuances of paperwork required to enable shipments to the EU to take place without issue. This was a learning process for companies that hadn’t dealt with the complexities of customs clearance for decades. To enable this, companies recruited for new staff, which further slowed down the processes required.
Direct-to-Patient and Direct-from-Patient Growth Also on our list last year was a new focus on direct-to-patient and direct-from-patient care, including significant growth in home-based clinical trials. In December 2019, 38 percent of pharmaceutical and
It is still too early to draw concrete conclusions about the overall impact of Brexit on trade with the EU. The Office of National Statistics (ONS) noted survey data suggesting businesses’ trading activities were being held back by Brexit frictions, such as extra paperwork
36 Journal for Clinical Studies
Volume 14 Issue 1
Logistics & Supply Chain
and higher transportation costs. However, the UK economy showed improvement during the course of the year. Shipments to the EU began to return to pre-Brexit levels as companies became more confident with the complexities of the new rules. Further growth is expected into 2022, and with the general improvement with trading conditions as COVID-19 restrictions are relaxed, we can expect a surge in UK – EU trade during the course of the next year. Sustainability Initiatives Move Forward Only a few short years ago, pharmaceutical manufacturers viewed supplier sustainability initiatives as a nice-to-have, but not a deciding factor in their decision to do business together. That shifted in 2020 with companies focusing on how they could impact the United Nations’ Sustainability Development Goals (SDGs), as well as building plans to execute and measure their efforts. What began as internal initiatives quickly turned to a focus on how vendor and supplier sustainability initiatives impact a company’s sustainability goals. Pharmaceutical companies soon required vendors like Peli BioThermal to demonstrate that temperature-controlled packaging and modes of transportation minimise negative effects on the environment. Additionally, vendors were also asked to provide tools that help pharmaceutical manufacturers measure their actual environmental impact so they can monitor and demonstrate reductions in carbon footprint and waste. We expect to see this trend continue in 2022. However, we also expect pharmaceutical companies to dig deeper into the supply chain. Vendors and suppliers like Peli BioThermal will be asked to also look at their supply chains and begin monitoring how their www.journalforclinicalstudies.com
own vendors and suppliers contribute to their environmental impact. Each year brings hope and anticipation of new challenges. This upcoming year is no different. Though COVID-19 continues to influence supply chains, companies are learning to operate in an unpredictable world. Overall, we anticipate this year will bring more stability and renewed focus on pre-COVID-19 priorities. REFERENCES 1.
McKinsey, No place like home? Stepping up the decentralization of clinical trials, June 2021
Adam Tetz Adam Tetz is the Director of Worldwide Marketing at Peli BioThermal and has more than 25 years of marketing experience. He is responsible for telling the story of Peli BioThermal to our worldwide audiences. His areas of responsibility include brand identity, product launch and communication strategy. Prior to Peli BioThermal, Tetz held positions in product management and marketing communication across a variety of industries, including medical software, financial software, information services and professional consulting services. He holds an MBA in Marketing from the University of Saint Thomas, a BA in Advertising from the University of Minnesota and is a veteran of the United States Coast Guard. Email: adam.tetz@pelican.com
Journal for Clinical Studies 37
Logistics & Supply Chain
Getting a Grip on Covid-19 Test Samples
In early 2020, the world of pathology and infectious disease testing was thrown into chaos by the Covid-19 pandemic. One of the key challenges in the global testing programmes for Covid-19 is how to track very large numbers of patient samples passing through inexperienced or under-staffed laboratories that have been asked to increase their daily throughput by as much as 10 times their normal workload. Over the last 20 years sample management has totally changed from its origins in the laboratory that were quite basic and rudimentary. In the beginning, clinical sample management worked by storing the samples in tubes. At this time many labs, even in large hospitals & pharmaceutical companies, stored their tubes in standard chest freezers either unlabelled or at best with handwritten labels on the rack or on the outside of the storage tube. If a clinical lab was particularly advanced, their samples might have a barcode, a 1D barcode, labelled on to the side of the rack. The databases of samples stored, if they existed, were quite often just Word documents or Excel documents, although patient samples in Pathology were tracked more effectively from the start. More advanced sample management really began when 2D barcoded tubes were introduced. We have evolved today to sophisticated tube and sample tracking applications. Sample management is now thought of as a genuine, serious discipline, as organisations realised that their samples were highly valuable. The biggest issue for sample tracking during the current crisis is the availability of suitable consumables, followed by the available instruments for testing. With some countries in Europe looking to test more than 100,000 samples per day it is not hard to see why. If a lab uses 96-well PCR plates for this testing they will need over a 1000 plates per day. Typically, these are sold in cases of 100 plates so that’s 10 cases per day, 50 cases per week. If all 27 EU states did that it would be 1350 cases of PCR plates per week. Now add in the USA & Canada, Australia, South East Asia... trying to mould, sterilise, pack and ship that many plates, even distributed across, say, 10 different manufacturers is a major logistical problem. Add to that the liquid handling tips and the RT-PCR and RNA extraction reagents which are also in short supply and the scale of the problem becomes apparent. There are now shortages of the very 2D-barcoded tubes which are so desperately needed to help track large numbers of samples. In the short-term, if 2D-barcoded tubes are not available, the next best option is to use linear barcodes. These can be laser printed or inkjet printed onto tubes for short-term disposable use, but perhaps the quickest way to use them is a print-and-apply self-adhesive label. Suppliers in the lab field can offer the printers software to design the label and the consumables. Typically, a sample receipt form should be generated and attached to this should be multiple copies of a unique barcode on labels. This allows for both the documents and the tubes to carry identical copies of the bar code and for samples to be split further downstream.
each holding 96 wells. These can have a linear barcode label applied to one end or side that can be read on most available automation platforms, but it is important that ALL the plates in a batch have the barcodes applied to the SAME end or side panel relative to the A1 position. This will prevent the plate being loaded the wrong way around on the robot deck. Because of their wide availability and use in agri-bio, compound storage, biosynthesis and sample storage, there is currently no shortage of 2ml deep well storage plates. These can be used for the RNA clean-up step in conjunction with 96 well filter plate technology. Alternatively, magnetic bead separation can be used if the liquid handling robot is equipped for this technique. Transfer to a 96 well PCR plate is still required, although 384-well PCR plates can also be used to speed up the throughput if the liquid-handling in use can support such small aliquots. All of these blocks and plates can carry 1D linear barcode labels. Tracking the samples within the plates relies on the A-H, 1-8 well location co-ordinates being accurately recorded; less error-proof than 2D barcoded vials, but cheaper and currently easier to set up. The issue of RT-PCR reagents is a thorny one, as is the RNA extraction kit. Currently, the Francis Crick Institute in the UK is testing using home-made reagents reverse-engineered from proprietary solutions by Qiagen, Merck and Roche. This can be risky and obviously, there are no quality controls or manufacturers’ guarantees that it will work in the same way that an off-the-shelf kit should. However, the Crick Covid-19 Consortium have published their SOPs for these tests on their website https://www.crick.ac.uk/ research/covid-19/covid19-consortium which are optimised for Hamilton Star and Starlet robots for the clean up and Beckman FX robots for the extraction. These would need to be adapted to any other or smaller pipetting station in order to replicate the Crick Covid-19 Consortium protocol. Although the Francis Crick Institute already had 2D-barcode rack scanners from Ziath and others, they were hampered by the lack of available tubes. The Covid-19 testing protocol followed by the Francis Crick Institute involves collection of sample swabs at hospital sites in 15ml Falcon tubes. The swabs are then transferred to uncoded 2ml screw cap plastic vials and linear barcodes received with the sample from each hospital are applied to these, such that the robot’s 1D linear bar code readers can scan them on the deck. For less sophisticated automation, it should be possible to use commercially available wired, wireless or Bluetooth linear scanners such as those from Opticon for this step. The robot then transfers the digested contents of each 2ml vial to one well of a 2ml deep well plate which is already bar-coded on the short edge. Of course, all that vital sample location data needs to be entered into a LIMS or database of some sort so that the RT-PCR results
To achieve some level of high-throughput, samples could be grouped into solid polypropylene blocks or deep well microplates, 38 Journal for Clinical Studies
Volume 14 Issue 1
Logistics & Supply Chain
can be tied directly to the correct patient samples. There are many systems and software available for this. One of the simplest and most cost-effective is the Samples software from Ziath. This basic “What is it? Where is it?” programme is an easy-to-use relational database that can be customised to track samples by any number of user-defined tracking tags, making it easy to find. For example, “all patients with positive RT-PCR result, over 50, who live in Cambridge, UK”. These are some of the ways that Sample Management issues are and will be tackled during the Covid-19 crisis and into the future. You can keep up to date with Neil Benn’s thoughts on this subject via his blog posts on the Ziath website at www.ziath.com/index.php/blog www.journalforclinicalstudies.com
Neil Benn Neil Benn graduated from Leeds with a BSc in Biotechnology and then an MSc from Hertfordshire University in Computer Science. His distinguished career in Laboratory Automation encompasses GlaxoSmithKline, CAT and the Max Plank Institute before setting up Ziath in 2005. He is a recognized authority on sample management in the laboratory.
Journal for Clinical Studies 39
Ad Index
Page 3
Aptus Clinical
Page 5
Clinerion Ltd.
Page 15
Ramus Medical Ltd.
Page 27
Senglobal Ltd
OBC SGS
IBC
Temperature Control and Logistics (By IQPC)
IFC
Trilogy Writing & Consulting GmbH
Journal for Clinical Studies
I hope this journal guides you progressively, through the maze of activities and changes taking place in the pharmaceutical industry JCS is also now active on social media. Follow us on:
Subscribe today at
Email: info@senglobalcoms.com www.journalforclinicalstudies.com
40 Journal for Clinical Studies
www.twitter.com/jforcsjournal www.facebook.com/Journal-for-Clinical-Studies www.plus.google.com/+Jforcsjournal www.journalforclinicalstudies.com
Volume 14 Issue 1
Clinical Research Solutions Health Inspired, Quality Driven.
SGS provides clinical research and bioanalytical testing, with a focus on early stage development and biometrics. Delivering solutions in Europe and North America, SGS offers clinical trial services. SGS has its own clinical unit in Belgium including a viral challenge testing facility and two phase I patient units based in Belgium and Hungary. SGS has a wealth of expertise in FIH studies, viral challenge testing, biosimilars and complex PK/PD studies with a high therapeutic focus in Infectious Diseases, Vaccines, and Respiratory. We offer a variety of tests that are bespoke, client-specific and support the full clinical development, from Phase I
Contact us healthscience@sgs.com sgs.com/healthscience sgs.com/healthcommunity
First-in-Human trials in our Clinical Pharmacology Units, to Phase II and Phase III studies in patients: • • • • • • •
Full Service CRO Drug Development Consultancy Clinical Pharmacology Unit Clinical Trial Management Biometrics Medical Safety & Regulatory Clinical Trial Analytics