Commentaries| Volume 40, ISSUE 12, P1967-1972, December 01, 2018

# Pharmacovigilance is Everyone's Concern: Let's Work It Out Together

Open AccessPublished:October 26, 2018

## Abstract

In recent decades, the field of drug safety/pharmacovigilance (PV) has advanced dramatically in some ways and yet has remained stagnant or progressed slowly in others. One way to assess the PV landscape is to view it through both a regulatory lens and a science and technology lens. This commentary highlights some of the current PV issues that can be resolved by sustained collaboration among all relevant stakeholders.

## Introduction

There has been recent attention in the literature on the history, ongoing developments, and current challenges in the field of drug safety/pharmacovigilance (PV).
• Dal Pan G.J.
Ongoing challenges in pharmacovigilance.
• Beninger P.
• Ibara M.A.
Pharmacovigilance and biomedical informatics: a model for future development.
• Price J.
Pharmacovigilance in crisis: drug safety at a crossroads.
• Pitts P.J.
• LeLouet H.
Advancing drug safety through prospective pharmacovigilance.
The focus of the present commentary is specifically on those areas that, from regulatory or scientific/technology viewpoints, I believe are in immediate need of being addressed: the lack of globally harmonized PV regulations and guidance documents that has led to confusion and unnecessary complications about what and how a company should report certain adverse events to regulatory authorities. Furthermore, the dearth of viable, clinically predictive drug safety models continues to greatly hinder 2 lofty goals that both industry and regulatory health authorities desire: attainment of a more cost-effective drug development process and fruition of the concept of personalized (precision) medicine. Through more structured and sustainable collaborations with all key stakeholders, there is good reason to expect that these lingering PV challenges can be met in the near future.

## Regulatory Viewpoint: A Trending De-harmonization

Health authorities, such as the US Food and Drug Administration (FDA) and the European Union's European Medicines Agency (EMA), are steeped in the regulatory science of drug safety. The FDA has the longest standing set of safety reporting regulations, dating back to the Federal Food, Drug, and Cosmetic Act of 1938. The US Code of Federal Regulations and accompanying guidances for both Investigational New Drug and New Drug Applications cover all aspects of drug safety/PV requirements throughout the drug development process and after approval. While not as long-standing, the EMA has experienced tremendous growth over the past 20 years, including a well-documented series of Good Pharmacovigilance modules and other PV guidances that similarly cover all aspects of regulatory expectations of a manufacturer in developing and maintaining effective drug safety oversight.
Although the specific regulations and guidances may differ between the FDA and EMA, both agencies share the same goal of protecting patients from the harmful effects of drugs, both are long-standing members of the International Council for Harmonisation (ICH), and both communicate on a regular basis to discuss PV matters.
• Dal Pan G.J.
• Arlett P.R.
The US Food and drug administration-european medicines agency collaboration in pharmacovigilance: common objectives and common challenges.
Thus, one would expect that these 2 health authorities are essentially aligned with regard to drug safety/PV requirements and expectations. A closer look, however, reveals challenging differences and trends. In general, there seems to be a basic philosophical difference between the agencies with regard to identifying and/or confirming safety signals: the FDA emphasizes the assessment and reporting of only the highest quality data, whereas the EMA approach appears to be one of looking at everything because one never knows what might be missed. Although both philosophies are valid and have merit, these differences have led to downstream inefficiencies for the manufacturer in staff utilization, prioritization of workload, and concerns about inconsistent global safety reporting.
Three such examples follow that have created significant discussion with regard to the trending de-harmonization over the past several years.

### Causality Assessment

The FDA 2010 Final Rule for Investigational New Drug safety reporting requirements clarifies that the determination of a serious adverse drug reaction can be based on the opinion of either the investigator or the sponsor,

Investigational New Drug Safety Reporting Requirements for Human Drug and Biological Products and Safety Reporting Requirements for Bioavailability and Bioequivalence Studies in Humans, https://www.fda.gov/Drugs/DevelopmentApprovalProcess/HowDrugsareDevelopedandApproved/ApprovalApplications/InvestigationalNewDrugINDApplication/ucm226358.htm.

meaning that a sponsor can downgrade an investigator's causality assessment from related to unrelated. This approach is inconsistent with the EMA 2006 guidance in which the sponsor should not downgrade the causality assessment provided by the investigator. If the sponsor disagrees with the investigator's causality assessment, then both the opinion of the investigator and the sponsor should be provided with the report. This difference has led to significant operational issues for global companies, such as deciding what is to be reported to each agency, one independent of the other, or taking the more conservative approach and simply reporting more than what is necessary. This latter approach causes other downstream effects, including creation of more unnecessary work for the staffs of both industry and health authorities and the addition of unnecessary information (ie, “noise”) to the safety databases.

### Adverse Event Reporting From Solicited, Noninterventional Programs

The hotly debated area of adverse event reporting has created a lot of angst during PV inspections over at least the last 5 years. As a data source, the format and volume of information can vary greatly due to the various program types and designs that fall under this broad category, including patient support programs, patient assistance programs, disease management programs, compliance and persistency programs, and market research programs. Of greater importance is that rarely are these programs designed to collect safety information, and there is the added risk of adding significant noise to the safety database, a risk that increases the potential of missing a true safety signal. This situation was first recognized in 2001 by the Council for International Organizations of Medical Sciences Working Group V.
• Report of CIOMS Working Group V
Current Challenges in Pharmacovigilance: Pragmatic Approaches.
Unfortunately, the views of regulatory agencies on this topic are not aligned. The FDA expects only suspected unexpected serious adverse reactions to be reported from such programs

FDA 1997 Guidance for Industryreport. Postmarketing Adverse Experience Reporting for Human Drug and Licensed Biological Products: Clarification of what to Report.

; however, the EMA GVP Module VI has a much broader view and, thus, a stricter requirement to report all suspected adverse events.

EMA GVP Module VI 2017—Collectionreport, Management and Submission of Reports of Suspected Adverse Reactions to Medicinal Products (Rev 2).

As for causality assessments, these types of discrepancies have also led to significant operational challenges, with the potential for each agency receiving different safety data. Various solutions have been proposed and presented to the health authorities over the past few years, including risk-based approaches to assess information only from program types known to have potentially valuable safety content.
• Biotechnology Innovation Organization (BIO)
A Proposed Risk-based Approach for Assessing Adverse Event Reporting from Patient Support and Market Research Programs.
• Portnoff J.M.
• Lewis D.J.
The enigma of pharmacovigilance of patient support programs: a survey of marketing authorization holders in europe.

### Risk Management

Both agencies have risk management as a priority, but the approaches and requirements are different for each. EMA risk management plans are much more extensive regarding content, are required for all newly approved products in the European Union, and represent a more proactive approach to the detection, assessment, and prevention of potential risk.

Approved Risk Evaluation and Mitigation Strategies (REMS), https://www.accessdata.fda.gov/scripts/cder/rems/.

The FDA risk evaluation and mitigation strategies are briefer in content and are not required for all newly approved products but can be required anytime postapproval, should safety issues warrant it. Thus, a sponsor could have both a risk management plan and a risk evaluation and mitigation strategy on the same product collecting or monitoring different information. Of more immediate concern for both approaches has been the issue of quantifying actual impact versus effort being made to implement each of them. A standard global risk management approach that has been validated would benefit all.
Such de-harmonized trends also come at a time when many other regions, such as Latin America, Asia–Pacific, and Africa, are making strides to establish or enhance their national and regional PV capabilities, seeking advice from the more PV-experienced regions such as the FDA and EMA for harmonized, best-practice approaches.

## Science and Technology Viewpoint

There is no doubt that, in recent years, we have witnessed significant technological advancements to support both operations management and signal management of drug safety/PV data.
• Beninger P.
• Ibara M.A.
Pharmacovigilance and biomedical informatics: a model for future development.
• Pitts P.J.
• LeLouet H.
Advancing drug safety through prospective pharmacovigilance.
Despite the many breakthroughs, 2 areas need to be addressed with a sense of urgency.

### Too Much Data Is Not Necessarily a Good Thing

Traditional data sources such as a national spontaneous reporting system (eg, the FDA Adverse Event Reporting System in the United States), medical literature, and clinical trial outputs have been used as the main sources to obtain patient safety information on new investigational products or established products. Unfortunately, none of these alone is sufficient to serve as a gold standard by which a molecule's complete safety profile can be established, nor new or rare safety signals can be identified or confirmed. Thus, we enter the world of big data and real-world evidence, which has dramatically increased our arsenal of data sources to include administrative claims data, electronic health records, and a variety of social media and wearable platforms. Although these new data sources may provide valuable information to identify new or rare safety signals, it comes at a price in that it is very time-consuming to analyze the massive volumes of information, of which the majority seems to be of little-to-no value or contains many informational gaps that make routine follow-up with the patient or reporter impractical. Thus, this scenario begs the question, is it worth the effort? One could argue that such an undertaking could greatly affect the efficiency of safety operations and safety signal assessments for no meaningful gain. In addition, there is the risk of adding significant noise to the safety database, thus masking the detection of true safety signals, especially rare events.
A potential solution is the use of artificial intelligence technologies, specifically through machine learning and deep learning approaches to support all PV activities. Machine learning has already made contributions in PV in the area of single case assessment for identification of adverse events.
• Comfort S.
• Perera S.
• Hudson Z.
• et al.
Sorting through the safety data haystack: using machine learning to identify individual case safety reports in social-digital media.
Both the pharma/biotech industry and the FDA are already exploring how machine learning algorithms can be utilized to enhance the identification of valid adverse events from >1 million individual reports per year that come to their attention, a currently painstaking manual and/or semi-automated (non–artificial intelligence) process. Technology companies

IBM. Scaling Safety Expertise in Life Sciences: A Turning Point in Pharmacovigilance. https://www-935.ibm.com/services/us/gbs/thoughtleadership/scalingsafety/.

are also exploring the application of artificial intelligence to support PV activities. The surface is just beginning to be scratched as this technology is still very early in its development.

### A Systems Approach for Predictive Drug Safety: What Happened?

Drug development costs continue to escalate, with estimates ranging from $500 million to$2.6 billion, depending on the therapeutic area, and these estimates do not include costs for the majority of drug failures during the development process.
• DiMasi J.A.
• Grabowski H.G.
• Hansen R.W.
Innovation in the pharmaceutical industry: new estimates of R&D costs.
Such costs make it vitally important that the selection of drug candidates be as accurate as possible to avoid delays or even cancellation of an entire drug development program. Adverse drug reactions identified during clinical trials is one of the main causes of attrition,
• Tufts Center for the Study of Drug Development
Causes of clinical failures vary widely by therapeutic class, Phase of study.
with estimates ranging from the second most common reason for termination in Phase I (28%) and Phase III (30%), to the third most common reason for termination in Phase II (19%). In the postapproval setting, unexpected drug safety issues frequently arise, requiring continued monitoring throughout the product life cycle.
• Downing N.S.
• Shah N.D.
• Aminawung J.A.
• et al.
Postmarket safety events among novel therapeutics approved by the US Food and drug administration between 2001 and 2010.
To help mitigate these alarming figures, drug safety must move from a mostly observational (qualitative) science to a more predictive (quantitative) science to assess both the specificity and sensitivity of a drug effect to truly understand its benefit–risk balance and its overall clinical utility. Better predictive drug safety models are needed, especially at the stages of early drug discovery and preclinical research. Most early-stage predictive models have significant limitations and thus are unable to identify true safety signals before clinical testing or even after evaluation of Phase I data. This situation puts the sponsor in a difficult position about whether to continue evaluation to assure proactive identification of significant safety issues before investing in expensive clinical trials prior to approval, or to wait until the drug is on the market and large populations of patients can be evaluated for the occurrence of rare safety issues. It also puts the health authority in a difficult position should a product on the market eventually elicit unanticipated side effects, resulting in a product recall (as has happened in the past). Most importantly, it puts patients at risk that could have potentially been avoided or addressed at the onset of a trial.
One methodologic approach that has been touted for several years is systems pharmacology/toxicology. In theory, such an approach could reap many benefits, such as better predictions of a drug's potential to cause an adverse event (drug–event pairing), leading to better preclinical screens and clinical diagnostics (safety biomarkers) that would identify problems early on and provide insights into treatment strategies to mitigate potential adverse events. This scenario in turn would lead to a better understanding of individual patient risks (precision medicine) and gain biological/mechanistic insights to help validate potential postmarketing safety signals. However, this task is daunting in that it would require highly committed expertise from various scientific fields to collaborate for extended periods of time with the most innovative technologies. Also, although some aspects of a systems approach have been tested and validated, a full systems approach has not been seamlessly integrated into drug discovery and development. This topic must be given a high priority and expeditiously pursued so that drug safety can truly become a predictive science, contribute to a more efficient drug development process, and support the goal of personalized medicine. Several organizations, including the FDA,
• Abernethy D.R.
• Woodcock J.
• Lesko L.J.
Pharmacological mechanism-based drug safety assessment and prediction.
have started such work, and many articles have been written on the topic.
• Abernethy D.R.
• Woodcock J.
• Lesko L.J.
Pharmacological mechanism-based drug safety assessment and prediction.
• Garcia-Serna R.
• Vidal D.
• Remez N.
• Mestres J.
Large-scale predictive drug safety: from structural alerts to biological mechanisms.
• Wist A.D.
• Berger S.I.
• Iyengar R.
Systems pharmacology and genome medicine: a future perspective.
• Huang L.C.
• Wu X.
• Chen J.Y.
Predicting adverse side effects of drugs.
• Bai J.P.F.
• Abernethy D.R.
Systems pharmacology to predict drug toxicity: integration across levels of biological organization.
However, no well-described, standardized, and validated blueprint is available that can be implemented as part of early drug development, despite the fact that both the technology and biology exist to accomplish this goal.

## Call to Action

Given these examples, it is critical that we find ways to accelerate the development of updated and globally harmonized policies and procedures, applied predictive models, and data assessment technologies for drug safety/PV to play its part in meeting the oft-mentioned goal of improving the drug development process. Only in this way can we expedite the development of safer, more efficacious medicines and reduce costs while attaining the high expectations of innovative, personalized/precision medicines for patients. In what areas do we seem to be struggling the most? In my opinion, there is basically one, namely sustained collaboration when it comes to successfully following through on the many excellent initiatives and great think-tank ideas that necessarily require input from key stakeholders to be successful. I have witnessed more industry and regulator interactions to discuss PV issues over the past 5 years than in the previous 15 years, with examples of collaborative success through organizations such as the ICH. This is very encouraging, as we are all dealing with the same problems and have the same concerns regarding patient safety. We attend the same conferences and workshops, and participate in many of the same organizations to discuss these issues and accompanying innovative proposals. Unfortunately, these efforts often dissipate over time with little-to-no actionable outcomes. Why? There are the usual suspects: not enough time, a shortage of talent, insufficient resources, and equivocal support from upper management. And although these may be perfectly understandable and legitimate reasons, greater strides could be made by simply having a sustained cooperative mentality with a viable infrastructure.
Together, both industry and regulatory authorities need to develop a consensus approach that can deliver actionable results in reasonable time frames with the least additional burden to an already understaffed and over-stressed workforce where younger talent seems not to be in evident supply as in the past. In my opinion, the only way to achieve this goal would be for private sector organizations, particularly larger organizations, to have staff dedicated to such policy-heavy projects or focused, innovative initiatives; professionals could then spend the majority of their time participating in these efforts, especially the critically important development of long-term relationship-building through collaborative activities. Global expertise from numerous scientific fields, policy, and project management would be required at the onset of a project, with both industry and regulatory authorities working side-by-side from beginning to end.
Given the topics discussed here, the finished products would include the following: (1) harmonized ICH PV guidances on benefit–risk management reporting requirements and format; (2) unified global regulatory guidance on the requirements for reviewing all of the different sources of data for potential adverse events; and (3) a validated blueprint for a systems pharmacology/toxicology approach that would be acceptable as an integrated part of a drug development program. In addition, dedicated governance of any completed projects would have to be maintained to rapidly respond to changes in technology and new medical findings. Although such an approach may seem to be exceedingly difficult or impractical to achieve, especially given the lack of available experienced staff, the logistical challenges, and perceived barriers, I argue that in the long run it will put global PV in a much better place in short order for all, especially the patient.

## Conclusions

Drug safety/PV has made significant advances over the past few decades in regulatory policy, technology, and science. However, many of the most pressing issues, such as de-harmonization and the absence of predictive drug safety models, have led to a certain degree of stagnation. This situation can rapidly be addressed by use of a sustained collaborative approach involving all of the key stakeholders.

## Acknowledgments

The author acknowledges the insights and suggestions of Dr. Paul Beninger in the preparation of this commentary. The views expressed in this commentary are the personal views of the author and are not necessarily those of Genentech, Inc, A Member of the Roche Group.

## Conflicts of Interest

The author is a stockholder in Roche Holding AG, Pharmaceutical Company. The author has indicated that he has no other conflicts of interest regarding the content of this article.

## References

• Dal Pan G.J.
Ongoing challenges in pharmacovigilance.
Drug Saf. 2013; 37: 1-8
• Beninger P.
• Ibara M.A.
Pharmacovigilance and biomedical informatics: a model for future development.
Clin Ther. 2016; 38: 2514-2525
• Price J.
Pharmacovigilance in crisis: drug safety at a crossroads.
Clin Ther. 2018; 40: 790-797
• Pitts P.J.
• LeLouet H.
Advancing drug safety through prospective pharmacovigilance.
Ther Innov Regul Sci. 2018; 52: 400-402
• Dal Pan G.J.
• Arlett P.R.
The US Food and drug administration-european medicines agency collaboration in pharmacovigilance: common objectives and common challenges.
Drug Saf. 2015; 38: 13-15
1. Investigational New Drug Safety Reporting Requirements for Human Drug and Biological Products and Safety Reporting Requirements for Bioavailability and Bioequivalence Studies in Humans, https://www.fda.gov/Drugs/DevelopmentApprovalProcess/HowDrugsareDevelopedandApproved/ApprovalApplications/InvestigationalNewDrugINDApplication/ucm226358.htm.

• Report of CIOMS Working Group V
Current Challenges in Pharmacovigilance: Pragmatic Approaches.
2001
2. FDA 1997 Guidance for Industryreport. Postmarketing Adverse Experience Reporting for Human Drug and Licensed Biological Products: Clarification of what to Report.

3. EMA GVP Module VI 2017—Collectionreport, Management and Submission of Reports of Suspected Adverse Reactions to Medicinal Products (Rev 2).

• Biotechnology Innovation Organization (BIO)
A Proposed Risk-based Approach for Assessing Adverse Event Reporting from Patient Support and Market Research Programs.
Biotechnology Innovation Organization (White paper), 2017
• Portnoff J.M.
• Lewis D.J.
The enigma of pharmacovigilance of patient support programs: a survey of marketing authorization holders in europe.
Ther Innov Reg Sci. 2017; 51: 486-493
4. EMA Workshop on Patient Support and Market Research Programs. http://www.ema.europa.eu/docs/en_GB/document_library/Presentation/2013/06/WC500144667.pdf.

5. Approved Risk Evaluation and Mitigation Strategies (REMS), https://www.accessdata.fda.gov/scripts/cder/rems/.

• Comfort S.
• Perera S.
• Hudson Z.
• et al.
Sorting through the safety data haystack: using machine learning to identify individual case safety reports in social-digital media.
Drug Saf. 2018; 41: 579-590
6. IBM. Scaling Safety Expertise in Life Sciences: A Turning Point in Pharmacovigilance. https://www-935.ibm.com/services/us/gbs/thoughtleadership/scalingsafety/.

• DiMasi J.A.
• Grabowski H.G.
• Hansen R.W.
Innovation in the pharmaceutical industry: new estimates of R&D costs.
J Health Econ. 2016; 47: 20-33
• Tufts Center for the Study of Drug Development
Causes of clinical failures vary widely by therapeutic class, Phase of study.
Impact Rep. 2013; 15: 1-4
• Downing N.S.
• Shah N.D.
• Aminawung J.A.
• et al.
Postmarket safety events among novel therapeutics approved by the US Food and drug administration between 2001 and 2010.
JAMA. 2017; 317: 1854-1863
• Abernethy D.R.
• Woodcock J.
• Lesko L.J.
Pharmacological mechanism-based drug safety assessment and prediction.
Clin Phamacol Ther. 2011; 89: 793-797
• Garcia-Serna R.
• Vidal D.
• Remez N.
• Mestres J.
Large-scale predictive drug safety: from structural alerts to biological mechanisms.
Chem Res Toxicol. 2015; 28: 1875-1887
• Wist A.D.
• Berger S.I.
• Iyengar R.
Systems pharmacology and genome medicine: a future perspective.
Genome Med. 2009; 1: 11-11.8
• Huang L.C.
• Wu X.
• Chen J.Y.
Predicting adverse side effects of drugs.
BMC Genomics. 2011; 12: S1-S11
• Bai J.P.F.
• Abernethy D.R.
Systems pharmacology to predict drug toxicity: integration across levels of biological organization.
Annu Rev Pharmacol Toxicol. 2013; 53: 451-473