Employing Real-World Evidence to Improve Clinical Trial Strategies

Employing real-world evidence allows for the study of many aspects of diseases, such as natural history, patient populations, and outcomes, under everyday conditions. Therefore, real-world evidence has many applications in clinical research, ranging from optimising clinical trial design and population/outcome selection to reducing the burden of regulatory commitments.

Sumeet Bakshi, Vice President, Real World Data Solutions, Certara’s Evidence, Value & Access group Richard Tao, Associate Principal Regulatory Writer and Submission Leads Member, Synchrogenix, Certara’s regulatory science company

Under clinical trial conditions, a patient with diabetes who is prescribed an anti-diabetic drug must adhere to the study protocol and their outcome is measured based on that protocol. If the patient misses several doses, their data may need to be removed (or censored) from the final analysis dataset. Poor medication adherence is not typically factored into a clinical trial. However, in the real world, people frequently forget a dose of medicine, take the wrong dose, or take it at the wrong time. Medication adherence is just one real-world example; there are many others that need to be considered when trying to truly understand patient outcomes.

For example, inclusion and exclusion criteria for a clinical trial prevent people with specific comorbidities or prior treatments from participating in the trial. Many of these people are likely to receive the intervention in real life despite not qualifying for the RCT. A real-world study generally includes a wider variety of people and circumstances and better reflects everyday situations. However, the experimental design of traditional RCTs allows for easier isolation of the treatment effect of one therapy compared with another as it is less subject to the bias which can present challenges in analysis of Real-world Data (RWD). Thus, RWD are usually employed in conjunction with and in complement to RCTs rather than as a replacement for them.

Challenges Using Real-World Data

Despite their obvious advantages, the adoption of real-world studies involves challenges such as gaining access to and managing the heterogeneity and messiness of the data.

The Real-world Evidence (RWE) currently being generated is predominantly from structured data, i.e., electronic medical records, Electronic Health Records (EHRs), and healthcare claims. These data sources, although large and complex, form the tip of the iceberg, as many other data sources, such as free-text entries, paper records, telephonic and video consultations, images, or video camera footage are now available and could form valuable RWE input.

Increasing Adoption of RWD in Rare Diseases/Populations

Rare disease research is an area that can benefit greatly from using RWD. Many rare diseases are not well studied, their natural history is unknown, and the outcomes that should be targeted are unclear.

Traditional RCTs include treatment and control arms in which participants receive the current standard of care or placebo. Of course, in the case of rare and orphan disease populations, a sufficiently powered traditional RCT may be very difficult to carry out due to recruitment challenges and the potential absence of a current standard of care. As a result, single-arm trials are increasingly being relied upon as pharmaceutical companies focus on small populations in therapeutic areas of extremely high unmet medical need. This phenomenon is not only limited to rare and orphan disease populations, such as Batten disease and Lennox-Gastaut Syndrome, but also includes smaller underserved segments of more common diseases, such as HER2negative hormone-positive breast cancer or non-muscle invasive bladder cancer, where there is no current effective treatment available.

Similar scenarios occur with some types of advanced cancers where the disease is both rare and life threatening, making it impractical and unethical to recruit a control population for a clinical trial. In these instances where it is not possible to establish a control group, regulators and payers are increasingly accepting single-arm trials, but they prefer to see some comparative data as part of the marketing authorisation submission. With single-arm trials, RWD can be used to generate an external comparator arm; a practical approach that also saves time and lowers costs.

Creating an External Comparator Arm

Regulators are often interested in the use of so-called natural history studies to offer pure external comparators, especially in circumstances where there are no approved treatments or accepted standards of care. However, the term natural history is really a misnomer because all patients receive at least some kind of intervention in the real world. Even in cases where there is no standard of care, doctors always try to alleviate patients’ symptoms with some type of treatment.

Consider Batten disease, which refers to a group of rare, fatal, inherited nervous system disorders that affects about 50 children in the UK. These children have about 20 to 25 seizures a day and reduction in seizure frequency is the desired outcome. Although there is no approved treatment, doctors do prescribe different

types of anti-epileptics. Therefore, there cannot be a true natural history study offering a pure non-treated comparison. In such cases, the control group would be an arm that provides information about the treatments used and outcomes for these patients, without the interventional drug in question.

In addition to the use of natural history studies and historical controls, external control arms can leverage data synthesised from other clinical trials that are not part of the same protocol. A synthetic arm is designed by selecting patients from placebo groups in past clinical trials, matching them to participants in the current trial, and then studying the outcomes. Techniques (such as matching techniques) are often used to adjust outcomes for valid comparisons similar to those used in classical realworld studies.

New Trends in the Use of RWD in Clinical Trials

The development of COVID-19 vaccines and drugs provides some good examples of evolving RWD trends because they were created in response to a new disease where there was limited pre-existing clinical data that could be used for RCTs. Much of the COVID-19 disease knowledge and epidemiology was established using RWD and integrated into clinical trials. RWD is also being employed with clinical data to support prescribing decisions for patients with COVID-19.

For example, the World Health Organization (WHO) Solidarity Trial for COVID-19 treatments included RWD in its drug comparisons. It compared four types of existing antiviral or antiinflammatory drugs – remdesivir, hydroxychloroquine, lopinavir/ritonavir, and interferon beta – without a traditional control group. It was not a double-blind study, but a direct comparison study. The results were published by the WHO, and decisions were made by regulatory decision makers and policymakers about which drugs were useful for COVID-19 treatment and which were not recommended. The Solidarity Trial included local hospital standard-of-care procedures, which refer to real-world situations.

In addition to the changes in development and research pathways for COVID-19-related treatments, many non-COVID-19-related clinical trials are currently adversely impacted and progressing slowly because sites are busy caring for COVID-19 patients, physicians are not able to give trials the attention they need, and it is very difficult to recruit patients. As a result, researchers are looking for operational models that are less site dependent and which can leverage RWD. That impetus has further strengthened the trend toward democratising patient data, reinforcing patients’ ownership of their data. As a practical example, by obtaining consent from patients to access their EHRs, insurance claims, and administrative or social demographic data, the burden of collecting such information through the standard trial case report form (a duplicative effort) can be avoided. Blockchain technology is becoming one of the key tools that allows all those data to be brought together in a validated and secure environment, thereby reducing patient recruitment requirements, data collection needs, and dependence on personnel at the sites going forward.

Artificial Intelligence (AI) is also being used for real-world studies. Its earliest and most common uses were in epidemic modelling. AI is now being employed to screen patients and identify those at risk for a particular disease, predict clinical outcomes, and determine optimal drug doses for specific patient groups.

RWD can offer regulators and other decision-makers additional insights into the effectiveness of treatments in the ultimate setting in which they will be used by broadening the population for whom the evidence is assembled and by offering insights into how a drug is likely to perform under nonideal conditions.

The Asian regulatory context

In some Asian countries, pharmaceutical companies are using RWE to obtain marketing approval for traditional herbal medicines without the need for RCTs. As these herbal medicines have already been on market for many years, China’s National Medical Products Administration (NMPA) encourages companies to collect pertinent patient data and submit them for approval under its real-world study regulations.

NMPA released its “Guidelines for Real-World Evidence to Support Drug Development and Review (Interim)” in January 2020 and they were joined by the Center for Drug Evaluation’s “Technical Guidelines for Real-World Research Supporting Child Drug Development and Evaluation (Trial)” in November 2020.

Furthermore, when NMPA approved an Allergan glaucoma treatment product in March 2020, it became the first medical device approved in China using RWE. As the product was already marketed for glaucoma treatment in the US, Allergan could compare RWD from patients in China with the US clinical trial results to determine if there were any ethnic differences between the two patient populations. As no differences were detected, an additional double-blind RCT was not required in China. Using this approach, it took less than one year for Allergan to secure NMPA approval of its glaucoma

treatment system.

There are, however, challenges in comparing clinical data between countries. One difficulty is that most clinical trials are conducted from a western market perspective. For example, the comparator chosen in a clinical trial is often from the US, UK, or a European market.

Some diseases are classified differently in the Asian and European markets, presenting significant challenges for clinical trials. For example, certain tumours are defined differently in Asia and Europe depending on the prevalence of tumour sub-types. Another issue is that treatment pathways in some Asian countries may be very different from those in western countries due to drug availability and pricing, physician preferences, and other factors. The contexts can also vary greatly, not just in terms of disease definition, but also social demographics and the availability of traditional medicine methods. As a result, it can be challenging to overlay evidence from one market on the other. RWD can be leveraged to identify and address these differences as part of designing the clinical development program.

Looking Ahead

While post-marketing studies will likely remain the primary use for RWE for the next 10 years, the COVID pandemic could be a watershed that elicits real change. It has injected a sense of urgency and promoted a shift from the traditional way of doing things to a more innovative and proactive approach using RWE to access more data and accelerate timelines. The drive is toward getting better and broader data and not just clinical trial data.

Regulatory agencies were already becoming more accepting of the use of RWD in clinical trials, but COVID19 will likely accelerate their adoption. The US Food and Drug Administration (FDA) introduced draft guidelines for “Submitting Documents Using RealWorld Data and Real-World Evidence to FDA for Drugs and Biologics” in May 2019 and it plans to issue additional guidance in 2021.

In the interim, the FDA-funded RCT-DUPLICATE project has conducted 10 non-interventional, RWE studies designed to emulate RCTs and evaluate cardiovascular outcomes of anti-diabetic or anti-platelet medications. Initial results from the study, which is being conducted by Brigham and Women's Hospital and Harvard Medical School in close collaboration with the FDA and Aetion, were published in December 2020.

The researchers selected three activecontrolled and seven placebo-controlled RCTs for replication using patient claims data from US commercial and Medicare payers. Nine of the 10 RWE studies achieved at least two of the three agreement metrics. Six of the nine studies also achieved ‘regulatory agreement,’ i.e., interpretation of the results would have resulted in similar regulatory decisions. The results did highlight one significant challenge – as placebos are not administered in everyday clinical practice, they cannot be observed in RWD.

Pharmaceutical companies are also looking to incorporate RWD into decision making earlier in the drug-development process. This change is affecting some pharmaceutical companies’ organisational structures, which, in the past, were very clearly demarcated in terms of pre-launch and post-launch activities. RWD used to be the domain of the postlaunch team and often in the context of market access. Increasingly, the market access staff are joining drug development programs very early, much earlier than they did even as recently as two or three years ago, and bringing their expertise with RWD to address challenges across the development life cycle.

Conclusion

Use of RWD has the potential to help improve development program designs by enabling researchers to test hypotheses and define appropriate clinical trial endpoints for efficacy and safety. The insights gained can help avoid unnecessary clinical trials and improve the probability of success of a development program. RWD can offer regulators and other decision-makers additional insights into the effectiveness of treatments in the ultimate setting in which they will be used by broadening the population for whom the evidence is assembled and by offering insights into how a drug is likely to perform under non-ideal conditions. These data may be less structured and may require more handling and manipulation expertise to make them usable in conjunction with clinical trials, but they can provide a valuable and often more meaningful picture of a product’s potential.

References are available at www.pharmafocusasia.com

AUTHOR BIO

Sumeet Bakshi is Vice President, Real World Data Solutions in Certara’s Evidence, Value & Access group. Sumeet qualified as a physician at the University of Mumbai in India and holds an MBA from the Saïd Business School at Oxford University in the UK.

Richard Tao is Associate Principal Regulatory Writer and Submission Leads Member at Synchrogenix, Certara’s regulatory science company. Richard qualified as a medical doctor and public health research scientist from Nanjing Medical University in China and received an MS from Jiangsu Institute of Parasitic Diseases and postdoctoral training at the University of Massachusetts and Harvard School of Public Health.

PERSONALISING PRESCRIPTION

A laser sharp approach for complex disease indications

Personalising Prescription is a sub-set of the Personalised Medicine concept. Both the concepts were always in practise while the gene and cellular technological advances have made them sound fancier. There are common, complex diseases like cancers; the diseases for which there are no single underlying target like psychiatric conditions; the diseases that are rare like Orphan categories; the oral diseases that are at the helm of integrated treatment approaches. The personalisation happens both at the disease and patient level while the factors to be considered are genetic, environmental and personal variables for prescription.

Subhadra Dravida, Founder and CEO, Transcell Biologics Gargi Roy Goswami, Founder and Director, KROYNAS Private Limited Rajiv Gupta, Entrepreneur; Managing Partner, Lateral Consulting The finale of the 20th century brought fresh hopes of a revolution in medicine based on advancing knowledge of the human genome decoded. The Human Genome Project was possible due to swift advances in genetic technologies that made possible the parallel testing of many Single Nucleotide Polymorphisms (SNP) in a cost effective manner. The beginning of these technological advances led to a Science

journal editorial comment in 1997 that defined personalised prescription as tailoring drugs to a patient's genetic makeup and predicted that personalised prescription would become a reality in clinical practice. The discovery of induced Pluripotent Stem Cells (iPSC) has revolutionised some of the concepts in personalising and precision components of Medicine

The concepts of personalised or individualised medicine and prescription are not new to the medical jargon. However, genetic advances have made discussing 'personalised medicine' and 'personalised prescription' in genetic terms more appealing and practical while exploiting mainly genetic differences between patients. Physicians have traditionally practiced personalised medicine in their efforts to decide the best treatment for each of their patients and was based on subjective physician preferences and not on scientific knowledge. Personalised medicine is known as a global concept that may include personalised surgery, personalised rehabilitation, personalised nutrition and personalised prescription.

A personalised prescription includes not only the use of new tests, that may or may not be pharmacogenetic tests, but also the concerns of all scientific information valid for prescribing medication. For a comprehensive view of personalised prescription, clinicians are expected to consider genetic, environmental, and personal variables when prescribing any medication. Known important genetic variables in specific drug response can be explored using pharmacogenetics; environmental variables such as co-medication, supplements, foods, beverages, and smoking etc for some drugs; and personal factors such as age, gender or medical illnesses (renal or hepatic insufficiency) as crucial personal variables in the response to some other drugs.

Personalised Prescription in Clinical Practice of Psychiatry

To apply personalised prescription in clinical practice requires a thorough understanding of the pharmacokinetic and pharmacodynamic principles of psychiatric drugs and does not depend on new developments in pharmacogenomic or other biomarker testing. It appears to require only that sophisticated clinicians understand that genetic, environmental or personal variables influence pharmacokinetic and pharmacodynamic response; the therapeutic window of the drug to be taken into consideration. Blood levels, called therapeutic drug monitoring, have been used by psychiatrists to personalise dosing for lithium, tricyclic antidepressants and some antipsychotics including clozapine in the past. Risperidone prescription is the best known example with genetic, environmental and personal variations reviewed in clinical practice.

Personalised Prescription for Orphan diseases

For those with a rare disease and for which there is no targeted drug available, almost any medicine is personalised. Whatever drugs or treatments they take are to alleviate the symptoms with better or combination of drugs. In some cases genome sequencing, iPSC technology incorporated in personalising prescription are next generation in nature with scope to offer cures for patient segments who are neglected by the pharma research. Muscular Dystrophy (MD) is a neuromuscular degenerative disorder, is one such Orphan disease category that afflicts individuals of different age, race, and gender. MD is a group of rare hereditary muscle diseases characterised by progressive skeletal muscle weakness. There is no traditional small molecule targeted drug based treatment available to the Physicians to prescribe till date. Using some of the drug test endpoints like cell viability, apoptosis, creatine kinase secretion and dystrophin levels on patient’s sourced iPSC platforms in the labs, the drug’s suitability can be ascertained for

the particular patient. Another area of therapy that is considered in personalising is how the physiology of muscle pathology will interact with these therapies and a precision approach with tailored physical activity is likely to benefit individual patients.

Personalised Prescription In and For Oncology

The Genetic Connection and Use of Stem Cells in "Everyday" Medical Applications

There is significant ongoing effort by a large cross section of the medical and research fraternity, across the globe, to work with Cryopreserved Adult Healthy Stem Cells to develop highly Personalised Prescriptions, specifically tailored for, both, lifestyle enhancements and specific ailments/diseases, even the chronic ones for each individual. The individual's Whole Exome/Genome Map is becoming an essential detail to better plan for Personalised Prescriptions.

The increasing success of biobanks for preserving the Umbilical Cord Blood of the newborn is an example of awareness among the modern day parents of the need for their children to have access to their own cell and related genetic information for targeted health solutions. The same set of ‘aware’ parents are among the initial target audience which understands the need for having access to healthy stem cells for Personalised Prescriptions based on their own genetic map and hence, they are opting for not only the Whole Exome/ Genome Mapping but also going ahead to Cryopreserve their Healthy Stem Cells for any future requirement for themselves in combating cancers or anyone else from their genetic family tree. Genetic Mapping can definitely enable deeper understanding of the state of health, chances of potential diseases and inherent immunity for that individual. Personalised Prescriptions, based on Genome Mapping, can suggest a host of preventive therapies to mitigate or minimise the occurrence of debilitating diseases.

There is a distinct lowering of age limit at which chronic diseases have started to impact the general population. A range of diseases resulting from stressful lifestyle, pollution and general degradation of living standards are agnostic of age, geography, race and even social class and levels. Stem cells are integral to Personalised Prescriptions for quite a few of such diseases, such as Autologous and Allogeneic procedures for HSCT - Hematopoietic Stem Cell Treatment for cancers impacting the blood cells. The recent advancements enable Precise Cell Selection from the patient's own Peripheral Blood Stem Cells (PBSC) or Bone Marrow (BMT) and even Umbilical Cord Blood (UCBSC) for an autologous procedure, which has a much higher probability of the body accepting the infusion of healthy stem cell selection and minimised occurrence of GVHD - Graft vs Host Disease, post-transplant. In fact, there is a preference for cryostorage of a patient's healthy stem cells much earlier in the treatment regime, before chemotherapy or any other such procedure which may render the patient's own healthy stem cell unfit for use at a later date. An Autologous procedure also reduces the need to find a donor and definitely reduces the total cost and time for the treatment. In the event, Allogeneic procedure is necessitated, Personalised approach would prefer opting for PBSC of the Donor instead of bone marrow as a source, along with procedures to derive the necessary cell selection to reduce occurrence of GVHD post transplantation.

Personalised Prescription for Oral diseases

Most oro-dental pathologies such as dental caries, periodontal diseases, oral and pharyngeal cancers, chronic orofacial pain, etc. and craniofacial disorders, such as cleft lip/cleft palate, arise from a complex interaction of genetic, biological, behavioural, and environmental factors. Using high-throughput ‘omics’ approaches to assess disease susceptibility, prevent disease, and holistic treatment is slowly becoming a reality as our understanding of disease pathways, genomic interactions, and novel biomarkers of oral conditions continues to increase.

Below are a few examples to get a glimpse of how research is unravelling newer molecular targets that are promising personalised oral health care in the near future.

Head and Neck Squamous Cell Carcinoma

Head and Neck Squamous Cell Carcinoma (HNSCC) is a disease with complex gene alterations. These alterations result either in shutting down or amplification of regulatory signals within a cell that accelerates cellular growth giving rise to tumours. Current treatment options for HNSCC include surgery and cytotoxic therapies. Most of these treatment strategies result in drastic reduction of the quality of life of the patient. Better understanding of the biological heterogeneity of head and neck cancer will help customise treatment and optimise outcomes for this malignancy. Cancer management has long focused on care based on tumour stage, subtype, and histology. Knowledge obtained with the help of genomic technologies offers a scope for a more refined tumour classification based on signalling pathways that can be targeted more precisely. Molecularly targeted therapies developed and tested for use HNSCC are namely include EGFR -directed drugs like cetuximab and EGFR - tyrosine kinase inhibitors namely gefitinib, lapatinib, erlotinib is being used for targeted therapies for HNSCC.

Acute and Chronic Orofacial Pain

Individual genetic variation in Cytochrome P 450 superfamily of enzymes is known to be involved in the metabolism and bioactivation of most of the drugs. People with a certain allelic variation of CYPD26 gene are unable to convert codeine to morphine. So these groups of people experience insufficient analgesia but they are able to withstand many of the adverse side effects associated with opioids. On the other hand, morphine intoxication is

observed in individuals with multiple copies of CYP2D6 because of extremely rapid metabolism of codeine. Interindividual differences in response to anesthetics such as isofurane, halothane and fentanyl are produced due to variants in CYP2E1 and OPRM1 genes.

Thus, dentists would be able to customise safer and more effective peri-operative and post-operative pain management by identifying and monitoring such individual genetic variation.

Another important area of concern of chronic Orofacial pain management is TMD or temporo mandibular joint disorders. Identification of genetic variations in an individual’s pain perception can offer clues about susceptibility to TMD. This information will be beneficial to segregate individuals most susceptible to developing chronic TMD and early treatment strategies can be designed for them. One example of such an approach is the Orofacial Pain: Prospective Evaluation and Risk Assessment (OPPERA) study which is the most comprehensive analysis till date.

Oral Infectious Diseases

Genomic approaches have revealed several interesting information about genomes of oral pathogens involved in the progression of common oral infectious diseases such as dental caries and periodontal disease. A noteworthy example is the Human Oral Microbiome Database (HOMD) which is an assembly of almost 1000 predominant microorganisms that inhabit oral tissues. This has opened up a new arena of early intervention tactics to the development of novel strategies for oral polymicrobial disease diagnosis, prevention and treatment. For instance, a new bacterial species, Scardovia wiggsiae was identified through HOMD project. This bacteria is a potential pathogenic indicator for early childhood caries risk. Thus, information about this microorganism and similar findings about additional pathogens drive the decision of a successful treatment strategy that may call for a change in the diet pattern as well as microbiota. Several Genome Wide Association Studies (GWAS) findings from dental caries investigations have identified specific caries susceptibility loci and related information. An interesting example is about the TAS2R38 and TAS1R2 genes that mediate taste sensation. Some individuals with variations in these two genes have been shown to be more predisposed to eat cariogenic food choices that automatically makes them potential candidates for dental caries.

Dental signature of every individual is unique and while dentistry has also many oral treatments which are generic and applicable for the vast majority of the population, the advancements in the field have made the need for Personalised Prescriptions move up from being a value add to being inherent to the field of dentistry itself. One such example is the use of Stem Cells Extract (SCE) to enable regenerative recovery after a Tooth Implant procedure. A typical tooth implant is a Bio-inert external item, however, the use of SCE makes the same Tooth Implant to become Bio-Active offering regenerative capabilities, reducing the sense of pain and aiding recovery in much less time. Stem Cell Extract is fast on its way to become an integral part of every Personalised Prescription for Tooth Implant and it has the potential to be included in any dental surgical procedure because of its regenerative properties.

Challenges of Practising Personalised Prescription in Traditional Medicine

The integration of genomics and cellbased technologies into routine clinical practice comes with its own challenges. Few of the barriers which calls for a thought to find strategies to overcome are: • Skepticism by providers and payers about the usefulness and credibility of information • Added cost to both caregivers and patients • Lack of insurers • Lack of integration of data sets with patient history data • Lack of knowledge, resistance, risk aversion at the Practitioner’s end • Lack of public awareness.

To overcome present and future challenges, the authors feel that the first step should be to engage in a dialogue focused on preparing next generation clinicians, researchers and educators; upgrading the knowledge base of current health professionals through interdisciplinary training and skill sets; adopt, practise and build credibility in the space.

AUTHOR BIO

Subhadra Dravida is the Founder and CEO of Transcell Biologics (www.transcellbio.science) that is into translating adult stem cell technology prowess into real time applications

Gargi Roy Goswami is the Founder and Director of KROYNAS Private Limited, India (www. genedent.com) focused on education and training to support translation of research into clinical applications in the domain of Dental Genetics and Saliva Diagnostics.

Rajiv Gupta is an Entrepreneur; Managing Partner at Lateral Consulting.

Bioreactor Automation Enhances Productivity with Biologics

Driven by real-time sensing

Bioreactor performance for mammalian cell culture has been automated and improved with better soft-sensor inline analytics and model predictive process control. With simple customisation of CHO cell models using a novel software tool, precise and stable glucose feed-regulation specific to user cell lines and media requirements is enabled. This fully-integrated hardware and software system provides unprecedented custom automation and reproducibility for biologic product performance in CHO cell culture at the lab-scale.

Hiroaki Yamanaka, Yasuhito Murato, Paul E Cizdziel Members, Life Innovation Business Headquarters Division, Yokogawa Electric Corporation

In the quest for improved quality and productivity in drug manufacturing, the industry is moving toward increasing use of bioreactor systems with realtime integrated monitoring and advanced analytics that have the potential to enable automation, drive performance and improve data-rich quality control. However, there exist multiple options in sensors and technologies for monitoring important cell culture variables or Critical Process Parameters (CPP). Furthermore, cell culture vessel configurations can be disposable Single-use Bioreactors (SUBs), glass or even stainless-steel. They can be stirredtank in design, rocking platform bags or perfusion configurations. The sensors and monitoring technologies selected for these configurations need to be suitably designed and compatible with the bioreactor architecture. When properly selected, the Process Analytical Technologies (PAT) provide not only analytical insights into ongoing bioprocesses, but can be leveraged for real-time control and automation; especially in fed-batch or continuous culture.

Automation is a key trend driving improvements in manufacturing; especially for production of biologics in the biopharmaceutical industry. One big challenge is the integration of systems for effective automation. For example, recent uses of in-line sensors in mammalian cell culture can be used to monitor biomass production, key media nutrients, or metabolism indicators. Spectrophotometric techniques such as Near-infrared (NIR) and RAMAN technology have been employed to quantitate and correlate spectroscopic data with precise chemical signatures. As a result, the monitoring of glucose concentrations and lactate in real-time are becoming more common, reducing the need for off-line analysis and enabling the automated delivery of media supplements as required.

PAT Integration & Data Alignment

Comparing the capability of RAMAN and FT-NIR for spectroscopic applications in cell culture is frequently a debate among researchers. The ability to spectroscopically identify and differentiate cell culture media components is primarily affected by considerations of bandwidth, sensitivity, interference and software. And the latter (software), is quite capable of having a sizable impact on the prior three mentioned parameters as described in the examples below.

Figure 1 – Refinement of the Cell Culture Calibration Model in Three Steps

Illustrated in this figure is the three-step protocol for customising a calibration curve specific to a user cell line, media and process. Beginning with a baseline calibration curve (provided), it generally takes three cell culture runs on the BR1000 bioreactor to accurately align PAT sensor data with off-line reference values.

For instance, in bioreactor culture, two issues frequently cited and considered limitations of FT-NIR included baseline data variation due to cell density dependent optical scattering and interference by water in glucose detection. We have convincingly addressed the first issue largely by digital noise suppression software improvements. Whereas the second issue of signal interference by water or other molecules overlapping with glucose bond absorption profiles was reduced by the proper selection of, and integration of multiple wavelength micro-scan data to provide a more specific footprint of reliable molecule-specific bond signatures. Both of these NIR limitations were thus overcome by innovative data use and novel software approaches. Consequently, the determination of glucose concentration in cell culture media with FT-NIR can now be robustly correlated with actual reference concentrations or off-line cell culture glucose measurements using bench top chemical analyzers.

The derivation of cell models and the proper selection of micro-scan data for a particular cell line and cell culture medium are fairly complex tasks requiring significant resource time and effort to achieve the proper calibration. And although a universal cell model for any expression system would seem to be ideal, it is not an effective viable option. To this end, we have settled on a precondition and a calibration approach to simplify the alignment of NIR data with user-specific cells and media. The precondition is to focus on a specific cell type so that deviations from the standard model are minimised. And the calibration approach is one that leverages a baseline calibration curve to provide a suitable starting point for subsequent user data-derived refinements.

By limiting the developed FT-NIR application to specifically CHO cells, we were able to derive an effective lap-top software tool that uses a ‘datadriven’ model refinement algorithm. The tool enables the custom calibration of FT-NIR data to any CHO-cell based user expression system by smartly adjusting spectroscopic micro-scan data. It has been our experience that starting with the pre-set baseline calibration algorithm, and performing only three consecutive bioreactor runs, enough user data can be gathered to construct a precise custom cell model. By using this overall approach and the final cell model, NIR spectroscopic monitoring can provide highly accurate glucose and lactate concentration sensing in real-time specific to the user cell line and culture medium.

For some purposes, simple monitoring of CPP such as glucose & lactate may provide significant value for production and quality control. However, it can also be coupled with intelligent bioreactor software systems to enable automated feeding of glucose for maintaining constant concentrations or dynamic control. More will be discussed about this in a later section.

In-line biomass measurement is an important PAT to understand the culture growth phase and cell viability in bioreactors in real-time. Sometimes referred to as bio-capacitance, electrical impedance

Figure 2 – Precision Control of Glucose and Alignment with Off-line Analytical Data

Use of the refined calibration curve with Model Predictive Control (MPC) software results in strong alignment of glucose concentrations in fed-batch culture. DG44 CHO cells were seeded at 5 x 105 cells/ml in FortiCHOTM (ThermoFisher) medium containing 5 gm/L glucose. Cells were cultured for 14 days with a set-point concentration of 2 gm/L glucose. The delivery of glucose feed solution (450 gm/L) was automated via peristaltic pump action.

sensors monitor cumulative charge polarisation across intact plasma membranes to estimate total live cell biomass. This technology has been widely adopted in biopharma as a popular method to calculate Viable Cell Density (VCD) in real-time. Like many users, we had encountered issues with adopting this technology, including reliability issues at higher cell densities and interference from some media components. However, we have innovated around these issues in several ways. Firstly, a thorough investigation of the cell physical size and growth dynamics (in suspension) during the cell culture timeline was completed to properly relate biomass data to proper cell counts. This is critical information that was collected by using a wellknown immortalised strain of CHO (Chinese hamster ovary) cells. Secondly, electrical impedance data was selected at multiple frequencies to overcome some types of interference from media components or other factors. And thirdly, we derived growthphase-specific data conversion algorithms that recognise transitions, and switch or compensate (as appropriate) in the cell culture timeline. For instance, we have found that calculation of VCD using bio-capacitance data from early and exponential growth phase culture can be used to derive an algorithm that provides high reliability specific to those earlier growth phases. However, this algorithm becomes a poor predictor of actual VCD in stationary or later stage cultures. This fact is likely due to changes in cell morphology that are reflected in later stage culture perhaps due to physiological shifts and changing protein expression profiles. It has been shown that in recombinant cell lines, the biologic expression of monoclonal antibodies is generally induced to higher levels when cultures achieve peak cell densities. In bioreactor culture, as CHO cells approach peak cell densities, glycolytic pathways transition to more oxidative phosphorylation metabolism, cell sizes may change and membrane composition also likely changes.

These events appear to result in a deviation from the calibration parameters applied for determination of VCD in early stages of culture. Hence, an entirely different data correlation model seems to be appropriate once cultures reach the peak VCD. By integrating and switching from early stage to late stage models at the appropriate time, the generation of accurate VCD data from bio-capacitance measurements can be achieved with appropriate CHO cell models. In our hands, using this approach, bio-capacitance can provide high accuracy in calculation of VCD data throughout the entire bioreactor run, including up to 100 x 106 cells per ml in cell culture.

Model Predictive Control

As mentioned earlier, the employment of in-line PAT for real-time Process Control (PC) is an automation approach that has potential benefits for biotherapeutic drug manufacturing. A key example is to automate feed-control of nutrients in fed-batch culture. Spectroscopic technology for monitoring nutrient concentrations is limited in ways discussed earlier. However, it is universally agreed that glucose concentration is one CPP that has a direct impact on manufacturing performance. Hence, production cell lines are usually well characterised regarding glucose sensitivities and optimal concentration requirements for growth in bioreactor cell culture.

Glucose concentrations have been shown to affect the metabolic state and growth rate of cells in culture, in addition to the expression yields, and post-translational modification of recombinantly produced biologic drugs.

To control glucose concentration in bioreactors in real-time with PAT, both accurate detection and feeding strategy are important for process control. PC considerations are obviously not unique to biopharmaceutical production, but rather have been widely studied and employed in many other industries such as chemical plants and oil refineries since the 1980’s. Early strategies, especially in the chemical industry, employed Proportional-IntegralDerivative (PID) control, largely dependent on feedback mechanisms. These PID-based methods work primarily via input sensing and rapidly reiterative feedback adjustment cycles. However, in bioprocess, in-line sensor data collection is slower and living systems take longer time periods to adjust and equilibrate to environmental or process changes. Hence the effectiveness of new prediction-based approaches can exceed the traditional reactive feedback manner. In bioprocess, we believe that PID-methods will increasingly be supplanted with more sophisticated feed-forward predictive control strategies requiring defined constraints factored into decision outcomes.

We have developed an intelligent CHO cell Model Predictive Control (MPC) algorithm for glucose-feed control that accommodates multiple constraints including VCD, growth phase, future-state, current concentration, feed-volume dilution factors, and selfcorrection based on differences in measured versus predicted data values. PAT-driven use of this dataadaptive MPC for automated delivery of glucose by peristaltic pump action, has been shown to provide precision control of glucose in fed-batch bioreactor culture, even in low glucose concentration ranges such as 1 gram/liter. With flexible programming features the dynamic regulation of glucose within a single bioreactor run can be implemented, enabling growth phase-specific concentration control, among other unique applications. The effects of this strategy for bioproduction have largely been unexplored mostly due to prior limitations in precision control of glucose concentrations during cell culture.

Summary

The integration and calibration of all these technologies and software requires expertise in cell culture, model building, spectroscopy, programming and data

Figure 3 – A fully integrated bioreactor system with PAT sensing and precision glucose control for CHO cell culture

Shown is the BR1000 Advanced Control Bioreactor System from Yokogawa Electric Corporation. Capable of handling 1 to 5 liter stir tank bioreactor vessels, it uses near-infrared & bio-capacitance in-line spectroscopic sensors with model predictive control software for automated delivery of glucose with precision control. In addition to glucose and pH adjustment alkali, up to four other reagents can be automatically delivered by peristaltic pump action. The BR1000 is the only fully-integrated PAT sensing, automated glucose delivery bioreactor system for mammalian cell culture.

integration. Hence, it is less common for resourceconstrained organisations to explore sophisticated in-line bioreactor automation and process control technologies for biopharmaceutical production. Piecemeal assembly of the many components usually from many independent suppliers and software integrations are a challenging and time-consuming task.

In this article we have attempted to explain how we have addressed and solved many hurdles by creating a novel and fully-integrated bioreactor system that we call the BR1000, with unique software for operation and calibration. To maintain strong utility and precision in PAT sensor data but still allow customised use for CHO cell culture, the BR1000 bioreactor system contains programming with a significant degree of CHO cell application bias. Use of other cell lines or significantly divergent media compositions would likely not perform optimally due to these built-in biases. However, for applications with CHO-cell culture, the system we have assembled is likely the most accurate and efficient glucose control, lab-scale technology platform for recombinant monoclonal antibody process development. Currently we are working on next generation glucose sensing PAT control systems that provide similar process control benefits to CHO cell culture and automation for pilot and manufacturing scale bioproduction.

REFERENCES

E. Tamburini, M.G. Marchetti and P. Pedrini. “Monitoring key parameters in bioprocess using near-infrared technology”. Sensors (2014), Vol. 14, p.18941-18959.

T. Yardley, D. Kell, et al. “On-Line,RealTime Measurements ofCellular Biomass using DielectricSpectroscopy”. Biotechnology andGenetic Engineering Reviews, Vol. 17,2000, p.3-35.

B. Moore, A. Sanford and A. Zhang. “Case Study: The characterization and implementation of dielectric spectroscopy (biocapacitance) for process control in commercial GMP CHO manufacturing process”. Biotechnology Progress (2019), 35e2782. https:// doi.org/10.1002/btpr.2782

Danny Chee Furng Wong, Kathy Tin Kam Wonq, Lin Tang Goh, Chew Kiat Heng, and Miranda Gek Sim Yap. “Impact of dynamic online fed-batch strategies on metabolism, productivity and N-glycosylation quality in CHO cell cultures”. Biotechnology and Bioengineering (2005), Vol. 89, No. 2, p.64-177.

Y.-S. Tsao, A.G. Cardosos, R.G. Condon, M. Voloch, P. Lio, J.C. Lagos, B.G. Kearns, Z. Lui. “Monitoring Chinese hamster ovary cells culture by the analysis of glucose and lactate metabolism”. Journal of Biotechnology (2005), Vol. 118, Issue 3, p.316-327.

M. Buchsteiner, L-E Quek, P. Gray and L. K. Nielsen. “Improving culture performance and antibody production in CHO cell culture processes by reducing the Warburg effect”. Biotechnology and Bioengineering (2018), Vol.115, Issue 9, p.23152327.

N. Templeton, J. Dean, P. Reddy and J.D. Young. “Peak antibody production is associated with increased oxidative metabolism in an industrially relevant fed-batch CHO cell culture”. Biotechnology and Bioengineering (2013), Vol.110, Issue 7, p.20132024.

T. Namatame and P. Cizdziel. “Making antibody medicines manufacturing a reality”. Pharmaceutical Manufacturing (2019), September Issue, p.30-36.

AUTHOR BIO

Hiroaki Yamanaka is currently employed as a bioengineer at the Life Innovation Business HQ of Yokogawa Electric Corporation (Tokyo) working on bioreactor process control technologies and applications. A Ph.D. graduate of Kyoto University in Life Science.

Yasuhito Murato is the manager for bioprocess international sales at Yokogawa (Tokyo) corporate headquarters. Having prior international business experience with Novozymes and an advanced science degree in cell biology, Yasu is also an expert in Yokogawa process instrumentation.

Paul Cizdziel has nearly three decades of experience in market-leading global life-science research supply companies including ThermoFisher, Merck Millipore, and REPROCELL in executive positions of technology management, business development, marketing and sales. He has also held scientific positions at the NIEHS (NC, USA) and the Yokohama RIKEN Institute.

Future of Pharmaceutical Manufacturing

The role of Automation

The breakthrough in the search for a coronavirus vaccine highlights the speed with which the pharmaceutical industry can mobilise scientific ingenuity in the service of humankind’s core goals. Yet, industry leaders know that such speed is the exception to the norm. Here, John Young, APAC director at automation parts supplier EU Automation, looks at four areas where greater automation could have a significant impact in helping speed the process of bringing a new drug to market in the years ahead.

John Young, Sales Director, APAC region, EU Automation

The Asia-Pacific pharmaceutical sector is set to grow at a compound annual growth rate (CAGR) of 7.1 per cent in the period to 2027. Manufacturers wanting to take advantage of these opportunities should stay informed about the latest trends in automation technology. Unfortunately, many key decision-makers fear their companies will struggle to keep pace with technological innovation in this highly regulated sector. Here are four key areas where greater levels of automation could have a positive impact on the industry in the coming years:

(Ultra) High throughput screening In drug discovery, the chances of discovering a compound that produces the desired impact on the target is very low, so scientists typically check hundreds of thousands of potential compounds.