CSL Behring

Lothar Tremmel

Lothar Tremmel is a biostatistics leader specializing in drug development. Starting as a statistician, he advanced to management roles, building and leading teams of statisticians, programmers, and data managers at CROs and biotech firms. A Ph.D. in Experimental Psychology, his expertise includes successful regulatory submissions and innovative data management. Lothar currently serves as VP of Clinical Sciences and Reporting (QCSR) at CSL Behring, focusing on non-traditional data and advanced analytics.

Aligning Non-Traditional Data Strategies

At CSL, I built a group we called ‘BMW’ for biostatistics and medical writing. Later, the functions of clinical epidemiology and patient-centric research were added to form QCSR. Our vision is to leverage non-traditional data sources to support clinical development, world-wide filings, and market access. Such data include real world data (RWD) and data from patient-centric research. We performed CSL’s first patient-centric social listing study, our first formal patient preference study, our first Mendelian randomization study, and are planning our first clinical trial with an external control arm.

While the functions are different for statisticians, medical writers, and programmers, all are quantitatively minded, creating synergies. As one example, medical writing and database programming have collaborated on digitizing protocols. For the technical group, digitizing protocols is an opportunity to exploit data standards, and for medical writing, the promise is to unleash GenAI on the protocol data to draft the clinical protocol. Talking about GenAI, QCSR launched a forum where staff could share their GenAI use cases. We had more than 20 sessions so far and investigated about 10 use cases, two of which will transform the way we perform business in the future.

The hurdle for using RWD for regulatory submissions remains high. At the same time, with careful planning and thoughtful regulatory interactions, successes can be obtained that can save hundreds of millions of dollars in development costs.

The key requirement for maintaining scientific integrity is pre-planning the analyses and housing the data in a controlled environment that can trace all manipulations performed. As the planning of the analysis must not be influenced by the data, prospective RWD-based studies are more palatable to regulators than retrospective studies.

“High-Impact” Quantitative Science

Quantitative science becomes impactful when it no longer focuses on analyzing a particular study right, but also deals with doing the right study. Quantitative insights can determine whether to conduct or continue a study based on probability of success and operational considerations, such as forecasting the time it will take based on enrolment patterns and event rates.

One important question concerns whole clinical development plans: When is it better to do a series of simple studies that add value incrementally and when is a more complex protocol justified that answers several questions at once? My vision is that our statistics group can give guidance on such questions by CDP-level simulations that go beyond a particular study.

Another high-impact area is the use of RWE as the secondary “pivotal” source of evidence in an initial filing, or even as the sole source for an additional indication. The new FDA leadership has indicated more openness to RWD for rare diseases. Each such step would typically reduce clinical trial development expenses by 50 million dollars or more.

Finally, quantitative sciences can support the successful marketing of a new treatment. We perform additional analyses issued by payors, which can be extensive. We also support medical affairs to produce evidence to optimize the use of our novel treatments. We often leverage some very low-cost real-world data sources to produce high impact insights.

Ensuring Methodological Rigor

The beauty of randomized clinical trials is that one doesn’t need to worry about third variables or “confounders” that influence both outcome and treatment assignment. That makes data analysis a relatively simple game. But randomized trials have become too expensive for a society that is no longer willing to fund exorbitant drug prices.

Therefore, we must learn to analyze non-randomized data from epidemiologic and real-world sources properly so that causal inference is methodologically rigorous. That is an area of active academic research, where some of the most advanced methods are based on AI and machine learning.

Building Strategic Partnerships in Clinical Development

The key factor in shaping outcomes is your ability to build relationships with people from different functions. They’ll learn that your way of looking at problems is different from theirs. That diversity of thinking makes you valuable.

As a “quant,” you already have a well-organized mind. You can turn that in your favor and approach the “science” of stakeholder management systematically by drawing up stakeholder maps, setting up proper regular meetings, and using technology to remind you when you haven’t seen a key stakeholder for too long. If you ever find a “master networker,” become their friend and ask for advice.