Established Prognostic Scores Predict Screen Failure But Not Post-Enrollment Outcomes in Phase 1 Oncology Trials: A...

Real-world safety and outcome data from Phase 1 trial populations-including screen failures-remain limited. We conducted a retrospective audit to evaluate the predictive utility of established Phase 1 prognostic scores for screen failure and post-enrollment outcomes, and to describe safety and clinical outcomes in a contemporary cohort.

A single-center retrospective audit was conducted of patients screened for Phase 1 trials between January 2020 and December 2024. Data were collected from electronic medical records, including demographics, comorbidities, enrollment outcomes, treatment regimens, toxicity, trial discontinuation, and survival.

Of 165 patients screened, 34 (20.6%) did not enroll. All 131 enrolled patients commenced treatment. Median age was 61 (range: 27-83) years, with predominantly good performance status (ECOG ≤ 1 = 99.2%). Most received immunotherapy-based regimens (74.0%), with combination regimens slightly more common than monotherapy. The overall objective response rate for patients who had an assessable response was 21.1% (27/128). For the 15 patients who received an antibody drug conjugate, the objective response rate was 46.7%. Unplanned admissions within 90 days of treatment occurred in 43 patients (32.8%), though only 8.3% were treatment-related. Unplanned admissions within 30 days occurred in 26/131 patients (19.8%). Disease progression was the main reason for trial discontinuation (79.8%), with toxicity accounting for 10.0%. Subsequent therapy was received by 45%. Median overall survival was 12.6 months (95% CI 9.5, 16.7), with 13.0% of patients dying within 90 days of treatment initiation and no deaths were attributed to treatment toxicity. Age was not associated with 90-day mortality, early admissions, or screen failure. Established prognostic scores in the Phase 1 trial setting (Royal Marsden, MD Anderson-ICI, and Gustave Roussy immune score) predicted screen failure but not post-enrollment outcomes.

Treatment-related toxicity and mortality were uncommon in this real-world Phase 1 cohort, and outcomes were driven by disease progression. Prognostic scores predicted screen failures but not clinical outcomes suggesting their utility lies in screening rather than stratification. These findings support the safety and feasibility of Phase 1 trials in appropriately selected patients.