The CREST trial showed benefit driven by patients with carcinoma in situ (CIS), while the Potomac trial showed a lack of benefit in the same subgroup. This stark inconsistency demonstrates that subgroup analyses, even for stratified factors, can be unreliable and are a weak basis for regulatory decisions or label restrictions.

Novo Nordisk ran a nearly 4,000-patient Phase 3 Alzheimer's trial despite publicly stating it had a low probability of success. This strategy consumes valuable patient resources, raising ethical questions about whether a smaller, definitive Phase 2 study would have been a more responsible approach for the broader research ecosystem.

The failure of an adjuvant trial for the TKI pazopanib was likely caused by a protocol change that reduced the dose to manage transaminitis. While well-intentioned to improve tolerability and adherence, the lower dose was sub-therapeutic. This serves as a critical lesson that managing side effects by compromising dose can nullify a drug's potential efficacy.

Praxis Interactive's essential tremor drug succeeded in Phase 3 despite an earlier data monitoring committee (DMC) recommendation to stop for futility. This rare outcome shows that interim analyses on a small fraction of patients can be misleading due to high variance, and continuing a trial against DMC advice can be a winning strategy.

Developers often test novel agents in late-line settings because the control arm is weaker, increasing the statistical chance of success. However, this strategy may doom effective immunotherapies by testing them in biologically hostile, resistant tumors, masking their true potential.

Sepsis is not a monolithic condition. The failure of more than 100 immunomodulatory drug trials is likely because they treated all patients the same. The future of sepsis treatment mirrors oncology: subtyping patients based on their specific inflammatory profile to match them with a targeted therapy.

The TRILINX trial revealed Xevinapant's toxicity was so high that it forced reductions in standard, effective treatments like cisplatin and radiation. This compromised the foundational therapy, leading to worse patient outcomes and demonstrating a key risk in adding novel agents to established regimens.

Even when testing drugs in heavily pre-treated patients, clinical trials incorporate subtle biological selection criteria. For instance, the COMPASS trial excludes patients with visceral metastases, a tactic to enrich for a population more likely to respond and avoid the most aggressive disease subtypes.

The TRILINX trial found that adding Xevinapant to local chemoradiation did not improve local control but was associated with a higher rate of distant cancer failures. This counterintuitive outcome highlights the risk of unintended, detrimental systemic effects when developing combination therapies for localized disease.