Nivolumab exposure and efficacy in metastatic melanoma patients: A real-world study

Background: The PK/PD relationships of immune checkpoint inhibitors are not fully understood. In this real-world study, we monitored plasma concentrations of nivolumab and investigated to what extent they could predict clinical outcome in a cohort of unresectable stage III or IV melanoma patients. Methods: 52 adult patients (26M/26F), mean age 64.3 years (range 31-89), performance status 0 (79.2%) or 1 (20.8%) were treated with 1 mg/kg nivolumab + 3 mg/kg ipilimumab (35 patients) or single agent nivolumab 240 mg Q2W or 480 mg Q4W flat dose (17 patients). Radiological response was assessed every 3 months per clinical practice. Nivolumab plasma concentrations (Cmax: end of infusion and Cmin: trough levels) were measured at the first cycle (C1). Nivolumab was analysed using a validated mass spectrometry method. PK parameters were derived using a standard population PK approach on Monolix using a two-compartment model. Statistical analyses were performed using R. Results: confirmed response rates were 5.7% complete response, 32% partial response, 17% stable disease and 45.3% progressive disease. Patients with objective response or stable disease were next categorised as having clinical benefit (CB). There was no difference in efficacy between single-agent and combination therapy (odd ratio = 0.28, [0.06-1.087], Fisher test). After the first cycle (C1), nivolumab Cmax concentrations were 67.1 µg/mL ±64.1 (CV = 95%) and trough levels were 26.9 µg/mL ±23, (CV = 87%). In patients treated with the nivolumab + ipilimumab combination, there was no statistical difference in Cmax levels (37.9 µg/mL vs. 33.9 ug/mL p > 0.05) and trough levels (24.0 µg/mL vs. 15.4 µg/mL p > 0.05), between those achieving a CB and PD patients despite a +56% numerical difference in trough levels. Similarly, no statistical difference was seen in Cmax levels between CB and PD patients treated with nivolumab monotherapy (203.1 µg/mL vs. 159.6 µg/mL, p > 0.05) despite a numerical difference of +35%. There was a statistical difference in Cmin between CB and PD patients (54.2 µg/mL VS. 34 ug/mL p = 0.03). ROC analysis showed that a threshold of <42.6 µg/mL was significantly associated with treatment failure (p = 0.016). Conclusions: This proof-of-concept study suggests that when nivolumab is given as a single agent, trough levels after the first cycle may help predict clinical outcome, as patients with plasma concentrations <42.6 µg/mL are at significantly higher risk of experiencing PD. The marker inter-patient variability observed in nivolumab pharmacokinetics warrants the use of therapeutic drug monitoring to check that exposure levels are within the expected range since patients with trough levels < 42.6 µg/mL after the first cycle are at risk of treatment failure.