Clinical Trials: Costly Gambles

Clinical trials remain one of the riskiest and most expensive aspects of drug development. Each failure represents millions in sunk costs and substantial delays in bringing essential treatments to patients. Consider the recent failure of high-profile studies such as Biogen’s aducanumab in Alzheimer’s or Roche’s atezolizumab (IMvigor211) in bladder cancer—examples that underscore the magnitude of the problem. Instead of performing post-mortem analyses after costly failures, a smarter approach would be a “pre-mortem”—a structured, forward-looking analysis to anticipate risks and weaknesses before a trial even begins.

This blog marks the start of a series exploring the potential of Target Trial Emulation (TTE) as a powerful tool to de-risk clinical trials, improve trial design, and enhance regulatory approval pathways. Each aspect introduced here—bias management, patient selection, dosing precision, and endpoint optimization—will be examined in greater detail in subsequent posts.

Pre-Mortems Beat Post-Mortems

A pre-mortem approach using TTE could fundamentally shift how trials are designed. TTE harnesses observational data, often from routine clinical practice, to simulate randomized controlled trials (RCTs). This allows researchers to detect potential pitfalls in trial design before committing substantial resources.

Unlike post-mortem analyses—which merely explain why a trial failed—pre-mortems enable corrections to be made in advance, significantly improving the chances of trial success.

Using Real-World Data to Fix Clinical Trials

Yet, observational data are notoriously messy and susceptible to bias. Issues such as immortal time bias—where survival rates appear artificially inflated due to methodological errors—confounding, and selection bias routinely undermine findings. Addressing these biases rigorously is crucial.

When a drug has not yet reached clinical practice, TTE must rely on real-world data from other drugs with similar mechanisms. These "proxy drugs" serve as stand-ins to predict how the new treatment might perform. This method introduces complexity, requiring careful validation to ensure proxy drugs closely resemble the investigational treatment. Done poorly, TTE risks compounding bias. Done rigorously, however, it can significantly reduce trial risk.

Finding Patients, Refining Doses, Smarter Endpoints

Trial failure does not always imply drug failure. Drugs often fail trials because they are tested in the wrong populations or with inappropriate dosing. For instance, initial trials of aducanumab faltered until retrospective analyses of biomarker data suggested that patients with high amyloid-beta burden might indeed benefit—an insight that eventually informed a controversial FDA accelerated approval.

Similarly, incorrect dosing frequently contributes to ambiguity in trials. Real world data analyses can inform dose adjustments based on functional status or patient characteristics measured in actual clinical practice could mitigate future trial execution risks. Using these insights prospectively can enhance trial precision, potentially transforming a likely failure into a success.

Choosing appropriate endpoints is another critical but frequently overlooked aspect of trial design. Traditional trials often select endpoints—such as overall mortality—that may not fully reflect patient benefits or detect treatment effects quickly. Imagine a situation where a heart failure drug meets its mortality endpoint and is approved and launched, only to find out later from the analysis of real world data that a combined endpoint incorporating hospitalization rates and patient-reported outcomes actually provides an earlier signal of benefit.  If such analysis was done as a pre-mortem, and an endpoint was incorporated into the trial, it could have expedited regulatory approval.

Regulators Warm to Real-World Data, but There are No Shortcuts to Rigour

Regulatory and health technology assessment bodies such as the FDA, EMA, and NICE increasingly accept real-world evidence (RWE) in approvals and reimbursement decisions. Yet, poorly executed RWE studies can harm rather than help. TTE, rigorously applied, offers transparency and methodological robustness that align with regulators' standards, making it a powerful tool for navigating regulatory approval.

Implementing TTE is neither easy nor quick. It demands high-quality, structured real-world data, advanced causal inference techniques, and deep clinical and statistical expertise. Nonetheless, the potential benefits are enormous. Trials guided by TTE-based pre-mortems are more likely to enroll the right patients, choose appropriate doses, and select relevant endpoints, dramatically improving their likelihood of success.

Given the high stakes of clinical trials, adopting TTE as a pre-mortem strategy is both economically prudent and scientifically sound. Rather than conducting post-mortems after costly failures, the biopharmaceutical industry should embrace TTE, turning clinical trials from expensive gambles into calculated risks.  Yet, it remains an open question whether TTE could have prevented the high-profile setbacks faced by Biogen or Roche. This blog series will explore these cases and others in more detail, investigating how proactive use of TTE might have changed their outcomes.