Margaret Gamalo is the vice president and statistics head of inflammation and immunology global biometrics and data management at Pfizer.
A few years after I began working for the Center for Drug Evaluation and Research at the US Food and Drug Administration, Ram Tiwari—then associate director of statistical policy at the Office of Biostatistics and someone I had been collaborating with on statistical research—invited me to join the DIA Bayesian Scientific Working Group to represent and contribute to noninferiority trials from a regulatory perspective. This small group included industry statisticians such as Fanni Natanegara of Eli Lilly, the late Frank Liu of Merck, Gouchen Song of Scholar Rock, and Heinz Schmidli of Novartis. Our collaboration led to the publication of two manuscripts on Bayesian noninferiority. Despite the time-consuming nature of the task due to our other commitments, I recognized the significance of assuming ownership and persisting in these voluntary efforts. Continuous progress on the project proved vital in sustaining motivation among all involved parties.
Real change, enduring change, happens one step at a time.
–Ruth Bader Ginsburg
Two years later, that work became even more pertinent with the emergence of public health threats posed by infections from multi-drug resistant bacteria. In fall 2013 and a subsequent follow-up in 2014, then Office of Biostatistics Director Lisa LaVange organized a think tank via the Clinical Trials Transformation Initiative to gather ideas about expediting clinical trials in areas with high unmet need. Tiwari and I were asked to provide two proposals—one centered on hierarchical models using Bayesian methodology and the other on augmented controls.
The Bayesian hierarchical model aimed to aggregate patients with infections in different organs to ascertain overall efficacy of antibacterial drugs, a practice not commonly undertaken at that time. On the other hand, augmented controls involved supplementing concurrent controls in a clinical trial with an external control. This marked the introduction of this methodology, offering feasibility and generalizability in clinical trials through a real-world application.
Recognizing the challenges facing the mainstream adoption of Bayesian methodology, including disagreement on priors and less-understood operating characteristics, I redirected my focus to diseases with high unmet medical need such as pediatrics and orphan diseases. I took the initiative to lead a small pediatric subgroup within the Bayesian Scientific Working Group, aiming to raise awareness about the suitability of Bayesian methodology for efficient pediatric trial design.
During that period, the concept of extrapolation was still in its infancy and the biopharmaceutical industry lacked a comprehensive understanding of its principles and methodologies. Nevertheless, given that extrapolation involves transferring conclusions from one population to another, the Bayesian methodology held significant appeal. Its ability to incorporate prior knowledge from the reference adult population made it particularly well suited for extrapolation.
That subgroup—which included Tiwari, Mathangi Gopalakrishnan of the University of Maryland, Laura Thompson of the Center for Devices and Radiological Health, Amy Xia of Amgen, Karen Price of Eli Lilly, and Brad Carlin of Pharmalex—collaborated to publish a review paper about Bayesian methodologies and their applications in the design and analysis of pediatric trials—one of the most-cited pediatric publications on extrapolation. My understanding of Bayesian methodology has evolved since then, leading me to a refined realization of its appropriate and scientifically sound use in pediatric trials.
The work on pediatrics propelled me into larger collaborative efforts with colleagues possessing diverse expertise beyond statistics. Upon joining Eli Lilly in 2016, I was introduced to AJ Allen, who led the Pediatric Center of Excellence at that time. His instrumental involvement led me to advocate for numerous innovative pediatric trials in all therapeutic areas at Lilly. This collaboration also allowed me to forge relationships with many individuals on the FDA Pediatric Review Committee who share a passion for advancing efficient pediatric trial designs and reducing the time lag for pediatric indication approval following initial adult approval.
Furthermore, Allen involved me in the Biotechnology Innovation Organization initiatives focusing on pediatric extrapolation, specifically in trials for pediatric Type II diabetes mellitus (T2DM). Between 2017 and 2018, there was mounting concern about the prolonged recruitment timelines for many trials related to this disease. This engagement with the Biotechnology Innovation Organization led to the following three significant contributions:
- Participation in an American Society for Clinical Pharmacology and Therapeutics panel on pediatric T2DM, during which I led the discussion on insights from failed trials
- Presentation alongside Matt Rotelli at an FDA workshop titled “Pediatric Trial Design and Modeling: Moving Into the Next Decade,” which focused on the application of systems pharmacology into Bayesian approaches
- Involvement in a multi-sponsor (Lilly, Novartis, Novo Nordisk, Boehringer Ingelheim) dialogue with key FDA experts and policymakers regarding the use of augmented control designs for pediatric T2DM trials
The collaboration within the Biotechnology Innovation Organization opened my eyes to a vast network of efforts aimed at advancing appropriate regulatory science in pediatric drug development. At that time, I assumed the role of co-chair for the Innovation Taskforce in Pediatric Drug Development and spearheaded the organization’s Workshop on the Use of Innovative Analytic Tools and Study Designs for Efficient and Feasible Pediatric Drug Development. This endeavor resulted in the publication of the Biotechnology Innovation Organization white paper titled “Extrapolation as a Default Strategy in Pediatric Drug Development,” written in collaboration with key pediatric and drug development experts in the field, including Christina Bucci-Rechtweg of Novartis, Robert “Skip” Nelson of J&J, Helen Thackray of BioCryst, and Ronald Portman of Novartis. The publication proved to be highly useful in crafting ICH E11A. In fact, many concepts from that publication were incorporated into the current draft of the ICH guidance.
While this was certainly influential, I recognized it was just one aspect of a broader collaboration involving initiatives such as iACT, IQ Consortium, Connect4Children, and the Children’s Medicine Working Party of the European Forum for Good Clinical Practice. Despite all these entities working toward the common goal of advancing medicines for children, interactions with stakeholders revealed divergent thoughts on implementation and perspectives.
Amid all the collaborative efforts I participated in, a central theme persisted: improving the efficiency of pediatric trials and reducing redundant data generated to establish efficacy and safety in children. Acknowledging the crucial role of statistics in addressing this challenge, Mark Rothman and James Travis of the FDA and I collaborated to establish the Statistics in Pediatric Drug Development Scientific Working Group under the auspices the ASA Biopharmaceutical Section. It attracted numerous individuals interested in pediatric drug development from the pharmaceutical industry, reviewers from multiple health authorities, and academia.
This initiative focused on disseminating innovative trial designs used in pediatric drug development through multiple conference presentations, short courses, and publications. It emphasized that numerous complex innovative trial designs disclosed by the FDA have been mostly applied toward expediting the development of medicines in children. Furthermore, upon closer examination of these complex innovative designs, it became apparent they rely on familiar methods such as Bayesian methods, hierarchical models, and external and augmented controls, which were introduced almost a decade earlier. Additionally, the working group elevated the issue of the extent of the pediatric safety database and pediatric safety analytics, highlighting concerns that an excessively large safety database could hinder progress achieved by efficient trial designs. This focus has prompted other groups, including Connect4Children, to adopt a multi-stakeholder approach to addressing the issue and proposing potential solutions.
Christina Bucci-Rechtweg of Novartis, with whom I collaborated previously, engaged me in another working group within the Children’s Medicine Working Party focused on age-inclusive trials. This diverse group includes European Medicines Agency regulators, drug developers, researchers, ethicists, and patient advocates. Together, we examined the barriers to the inclusion of adolescents in adult research, delving into all the disease state guidance issued by the FDA and European Medicines Agency—a comprehensive effort that yielded valuable insights. This endeavor culminated in the publication of “Strategies to Facilitate Adolescent Access to Medicines: Improving Regulatory Guidance,” which provided valuable recommendations to enhance regulatory guidance in this area.
Following this initial effort, broader work commenced to explore additional dimensions for evaluating the inclusion of adolescents in adult research. This culminated in the creation of a tool titled “Considerations for Adolescent Inclusion in Adult Research: A Decision Tree,” which was adopted by the Children’s Medicine Working Party. This tool serves to facilitate conversations across the research ecosystem, promoting the broader incorporation of adolescent populations within appropriate drug development trials.
The inquiry into pediatric safety, mentioned earlier, spurred my recent focus on safety analytics and quantitative benefit-risk assessment, in general. Present methods for characterizing a drug’s safety profile are inadequate, as they primarily focus on the incidence of the first event without considering factors such as onset, severity, duration, and recurrence. Moreover, there is a scarcity of methodology for incorporating correlations among events and for efficiently accounting for multiple testing in these outcomes.
This research on safety also ventured into two divergent areas on novel methods of signal detection in spontaneous adverse event reporting, as well as on less costly methods for quantitative benefit risk assessment. My partnership with academia and key technical experts in the industry became helpful, as they can provide novel solutions to these problems expediently. The remaining challenge is how to get this into mainstream analysis of medicinal safety profile and benefit risk.
Throughout my collaborative experiences, both during my tenure at the FDA and now within the industry, I have gleaned valuable insight on what it takes to affect regulatory policy and statistical innovation. Below are reflections based on my journey thus far.
Understand what is common interest or issues that strike accord from all parties. To drive progress in regulatory science and statistical innovation effectively, assembling a diverse group of stakeholders with requisite expertise is crucial. Academia’s leadership in fundamental research and cutting-edge statistical methodologies plays a pivotal role in advancing knowledge and cultivating skilled professionals. Leveraging the complementary roles of health authorities and industry, research findings and new statistical methodologies are translated into practical applications. Robust regulatory frameworks, on the other hand, provide the foundation for the development and sound implementation of emerging technologies, prioritizing public welfare. Industry serves as a crucial partner in driving scientific and statistical advancements by bringing innovation and domain expertise to the table. Operating within regulatory frameworks, industry drives progress while ensuring compliance with ethical and regulatory standards.
Recognizing and leveraging these complementary roles between academia, industry, and health authorities are essential for propelling responsible advancements forward, ensuring technological innovations prioritize ethical standards and promote societal well-being.
With a diverse stakeholder group, it is important to understand that each will actively advocate for policies and objectives aligned with their individual interests. However, collaboration thrives when common ground is identified, allowing parties to align goals. This alignment enhances willingness to engage in productive dialogue and collaboration. By acknowledging shared priorities, stakeholders increase the likelihood of achieving lasting outcomes, laying the foundation for enduring solutions to complex challenges. In any endeavor to improve regulatory science and statistical methodology, the guiding principle always remains the same and that is of the well-being and protection of patients. By prioritizing patient well-being, any group can navigate collaboration challenges effectively.
Tension has often arisen between academia and industry in think tanks I have participated in that focus on establishing safety databases for investigational pediatric drugs. Academics accuse industry of heavily influencing the agenda, while industry accuses academics of lacking understanding of pre-market safety complexities. It is crucial to continuously test our assumptions and challenge the accuracy of our biases. Progress can be hindered when we confine ourselves to the present context and perspective. Ultimately, our aim is to protect patients, which requires us to devise key principles that balance industry innovation with robust safety measures demanded by academics. Achieving this balance requires open dialogue and collaboration among the group. Prioritizing patient well-being while facilitating innovation in generating information and developing insights on pediatric safety profile helps groups move forward.
Implement a strategic approach to defining drug development or statistical problems. Understanding the landscape surrounding a drug development or statistical problem is crucial, as it provides context and accurate framing for collaborators, allowing them to grasp the broader public health or scientific issue. Moreover, knowing the landscape enables collaborators to assess the relevance and importance of the problem, ensuring alignment with their goals and priorities. This involves identifying specific aspects that require consideration and ensuring a well-defined and manageable problem statement.
In the Children’s Medicine Working Party working group I participated in, which was composed of a diverse group of people, we meticulously dissected barriers to the inclusion of adolescents in adult research that also addressed our domains of expertise. This led to a comprehensive analysis that encompassed understanding issues related to disease, product, statistical considerations, operational aspects, and legal and ethical dimensions. Additionally, the group examined the presence or absence of patient advocacy in various diseases as a factor for age-inclusive research.
In examining the regulatory landscape, the group recognized substantial scientific knowledge and regulatory precedence exist for the inclusion of adolescents within adult trials, which can inform research approaches. This led us to identify important opportunities for enhancing guidance. For instance, contextualizing developmental factors influencing adolescent disease progression provides valuable insight into the role of adolescent inclusion in research studies. Addressing these factors in guidance documents by health authorities can facilitate broader acceptance of age-inclusive trial methodologies and accelerate adolescents’ access to medicines. Indeed, conducting an exhaustive landscape search and questioning conventional wisdom and long-held assumptions enabled us to uncover new perspectives and alternative solutions to the problem.
In the domain of biostatistics, statistical methods must be clearly anchored in the landscape of science and practice. It requires meaningful translation of science. Furthermore, because most of the statistical methods attempt to solve real problems, it is essential to view the issue from a broader perspective, encompassing various stakeholders’ concerns beyond just statistical considerations. By addressing most stakeholders’ concerns, our solutions will be more comprehensive and applicable.
Expect nonlinear progress and embrace patience and persistence. It is worth highlighting that pharmaceutical companies—often with help from academia—commit significant resources to research and development endeavors, with a dedicated emphasis on swiftly introducing pioneering solutions to address pressing drug development challenges. In the domain of statistics, innovation holds equal significance, as it equips us with the methodologies required to address inquiries that drive forward our comprehension of medicinal efficacy and safety. A wealth of innovation is currently underway; however, the critical question remains how best to effectively harness and leverage this progress within the confines of a structured regulatory science.
Health authorities, on the other hand, actively foster collaborative research partnerships with pharmaceutical companies and academia. These collaborations are geared toward advancing regulatory science, refining drug development methodologies, and deepening our insight into safety and efficacy assessments.
With every change in regulatory science and improvement of statistical methodology, it is important to acknowledge their adoption often progresses in a nonlinear manner. My journey using Bayesian methodology and augmented controls was far from linear. I recall publishing the seminal paper on augmented control with Junjing Lin of Takeda and Tiwari was a prolonged process, marked by numerous rejections and lengthy journal review comments. At that time, the notion of combining an external control with a traditional randomized controlled trial seemed inconceivable, as it is tantamount to adding noise to a pristine methodology to obtain causal inference. After a decade, that strategy is gaining ground as the best way to benchmark external control given potential for unmeasured confounding and progress our understanding of real-world data.
In our contemporary landscape—characterized by a multitude of stakeholders—innovation, in general, demands adaptability and persistence to meet the diverse and evolving demands of our dynamic ecosystem. Innovative processes often involve iterations and feedback loops leading to necessary adjustments and shifts in direction, rather than following a linear progression. The understanding of a problem may significantly transform over time, with phases of consensus and progress in method development along with periods of stagnation.
Prioritize small wins while maintaining focus on long-term goals. Thinking big, acting small, and learning fast involves setting ambitious goals while systematically breaking down the process into manageable steps, fostering continuous learning, and adapting. Prioritizing flexibility and adaptability facilitates the translation of ambitious goals into practical advancements, ultimately benefiting the development of innovative solutions. Prioritizing solvable problems and achieving measurable progress through small wins sustains motivation and momentum within research teams.
In the collaborative working groups in which I have participated, the responsibilities usually start with modest goals that encompass promoting collaboration and knowledge-sharing. This involves sharing best practices and success stories and facilitating access to analyses, studies, and research. Additionally, fostering discussions about overarching challenges, offering technical assistance, and maintaining open communication on relevant issues are crucial. Furthermore, the role extends to coalition-building among stakeholders and providing valuable input and research to support informed decision-making and advancement within the field. Overall, these efforts aim to enhance cooperation, innovation, and progress in statistical endeavors. Some of these have led to statistical methodological work that was built through small coalitions with similar interests.
As I mentioned previously, it is important to take ownership in these volunteer efforts. Once the scope and stakeholders are defined, project planning is important. This involves outlining milestone steps, allocating resources, and establishing timelines to guide the problem-solving process systematically. This ensures alignment and progress toward common goals. Additionally, it facilitates benchmarking and evaluation by providing reference points for assessing the collaboration’s success.
Foster mutual respect and assume good intent. Collaboration often involves encountering disagreements and necessitates compromise. Mutual respect is key, requiring active listening and understanding of others’ perspectives. It is crucial to consider players’ risks and incentives to foster effective collaboration. One also must be aware that excessive collaboration can lead to project stagnation, highlighting the importance of balancing divergent perspectives to maintain progress. Negotiation prioritizes win-win outcomes through common ground and creative problem-solving, yet strong perspectives or unwillingness to cooperate may lead some to quit participating. Maintaining positive relationships is crucial for sustaining collaboration and resolving conflicts amicably. Building trust, showing respect, and maintaining transparency contribute to enduring partnerships. Celebrating small wins along the way further reinforces progress and momentum toward mutually beneficial outcomes.
In summary, innovation in statistics and regulatory science involves incremental progress, requires the engagement of various stakeholders, and often unfolds in a nonlinear manner. Patience, respect, persistence, adaptability, and flexibility are essential virtues in this process. Collaborative efforts among stakeholders contribute multiplicatively to broader innovation, generating numerous ripples of progress. My journey in drug development has provided valuable learning experiences and underscored the importance of these principles in driving meaningful advancements.
Editor’s note: A version of this article originally appeared in the Spring issue of the Biopharmaceutical Section’s Biopharm Report.
Leave a Reply