LOW-DOSE NIVOLUMAB IN ADVANCED SOLID TUMOURS: ACTIVITY, TOXICITY AND THE MISSING COMPARISON WITH STANDARD DOSE

Abstract

Background. Nivolumab shows a flat exposure–response relationship above about 0.3 mg/kg, and Indian and other low-resource centres have used fixed doses of 10 to 100 mg when the approved regimen is unaffordable. A systematic review of low-dose PD-(L)1 strategies did not pool data, and the only published meta-analysis covers Hodgkin lymphoma.
Methods. Cohorts treating advanced solid tumours with nivolumab monotherapy below the approved dose were eligible. Responders and treated patients were extracted from full texts, and each regimen was converted to dose intensity (mg per 4 weeks as a percentage of 480 mg). Proportions were pooled with random-intercept logistic models; comparisons with standard dose were summarised by disease with Mantel–Haenszel risk ratios. Randomised trials against non-immunotherapy control were reported narratively. All models were fitted in R (metafor).
Results. Nine solid-tumour cohorts (412 treated patients) were identified; six peer-reviewed cohorts formed the primary analysis. Dose intensity ranged from 4% to 34%. Objective response was 13.9% (95% CI 8.5 to 21.8; prediction interval 5.6 to 30.5), and 18.2% (12.9 to 25.1) with two conference abstracts and a preprint added. Response showed no increase with dose intensity, although tumour type and dose were confounded. Three cohorts compared low with standard dose in 137 patients (RR 0.78, 0.40 to 1.50). Grade ≥3 treatment- or immune-related events were 0.6% (0.0 to 12.4; one event in 137 patients) in peer-reviewed observational cohorts but 42.5% in the DELII trial at 8.3% dose intensity. DELII (HR 0.80, 0.66 to 0.97) and a head and neck trial (HR 0.545, 0.362 to 0.820) showed longer survival than non-immunotherapy control.
Conclusions. Low-dose nivolumab is active in pretreated solid tumours and has randomised evidence of benefit over chemotherapy-based control. Observational cohorts and the one randomised trial disagree by more than an order of magnitude on serious toxicity (0.6% against 42.5%), and neither estimate is reliable. Evidence comparing low with standard dose is too small and too confounded to estimate any efficacy loss; a randomised dose-comparison trial is the necessary next step.