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.

In trials like ASCENT-4, where over 80% of the control arm received sacituzumab govitecan upon progression, the true overall survival (OS) benefit is obscured. This makes progression-free survival (PFS) a more reliable endpoint for evaluating the drug's first-line efficacy.

The control arm in the EMBARK study was blinded to PSA results, preventing physicians from intervening with standard-of-care AR antagonists at PSA progression. This design likely delayed subsequent effective therapies, making the control arm underperform and potentially exaggerating the overall survival benefit of the experimental arms.

In the SUNRISE 2 trial, 44% of patients had no detectable tumor after pre-treatment resection. This high baseline inflates the final clinical complete response (CR) rates (e.g., 59% in the control arm), making CR a misleading indicator of the actual therapeutic benefit, which was a much smaller improvement over baseline.

A significant criticism of the pivotal KEYNOTE-564 trial is that only half the patients in the control arm received standard-of-care immunotherapy upon relapse. This lack of subsequent optimal treatment complicates the interpretation of the overall survival benefit, raising questions about its true magnitude.

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.

Clinicians should avoid directly comparing the toxicity profiles of new ADCs, as the data often comes from different trial stages. A drug in a Phase 1 expansion cohort may appear more toxic than one with mature Phase 2 randomized data, making definitive safety assessments premature.

Experts believe the stark difference in complete response rates (5% vs 30%) between two major ADC trials is likely due to "noise"—variations in patient populations (e.g., more upper tract disease) and stricter central review criteria, rather than a fundamental difference in the therapies' effectiveness.

The control arm relapse rate in the SUNRISE 2 trial was only ~20%, while in the EV-303/KEYNOTE-905 trial it was ~60%. This huge discrepancy highlights that current clinical staging and selection criteria are poor at identifying patient risk, signaling an urgent need for better stratification tools like ctDNA for more effective clinical trials.