Normalized administered atezolizumab dose and clinical outcomes in advanced solid tumors: an exploratory analysis of the phase I PCD4989g study

Background Atezolizumab dose selection was supported by linear pharmacokinetics, absence of a maximum tolerated dose, and trough concentrations exceeding the preclinical receptor saturation target. In the Project Optimus era, we explored whether higher body weight-normalized administered dose was associated with improved outcomes in the original phase I dataset. Patients and methods We assessed 660 patients from the first-in-human PCD4989g study (NCT01375842) separately in the dose-escalation (weight-adjusted) and dose-expansion (fixed 1200 mg) cohorts. Patients were grouped as ≤15 or >15 mg/kg. Sensitivity analyses modeled dose continuously and sequentially excluded individual tumor types. Outcomes were overall survival (OS), progression-free survival (PFS), objective response rate, and grouped immune-mediated adverse events. All analyses were exploratory; no equivalence or noninferiority hypothesis was tested. Results In the dose-escalation cohort ( N = 431), median OS was 15.4 compared with 16.1 months and median PFS was 2.1 compared with 2.7 months; adjusted models showed no improvement with higher normalized dose. In the fixed-dose cohort ( N = 229), median OS was 10.1 compared with 7.4 months and median PFS was 1.4 months in both groups. Higher normalized dose was associated with shorter OS in the multivariable model (hazard ratio 1.679, 95% confidence interval 1.186-2.377) but this attenuated after excluding breast cancer. Pneumonitis was numerically more frequent at >15 mg/kg, although events were sparse. Conclusions Higher normalized atezolizumab dose was not associated with improved outcomes. Substantial clinical and tumor-type imbalances preclude causal, equivalence, or noninferiority conclusions. These findings support prospective randomized dose-optimization studies rather than routine dose modification.