Predicting immunotherapy outcomes under therapy in patients with advanced NSCLC using dNLR and its early dynamics.

ABSTRACT

BACKGROUND: dNLR at the baseline (B), defined by neutrophils/[leucocytes-neutrophils], correlates with immune-checkpoint inhibitor (ICI) outcomes in advanced non-small-cell lung cancer (aNSCLC). However, dNLR is dynamic under therapy and its longitudinal assessment may provide data predicting efficacy. We sought to examine the impact of dNLR dynamics on ICI efficacy and understand its biological significance. PATIENTS AND METHODS: aNSCLC patients receiving ICI at 17 EU/US centres were included [Feb/13-Jun/18]. As chemotherapy-only group was evaluated (NCT02105168). dNLR was determined at (B) and at cycle2 (C2) [dNLR≤3 = low]. B+C2 dNLR were combined in one score good = low (B+C2), poor = high (B+C2), intermediate = other situations. In 57 patients, we prospectively explored the immunophenotype of circulating neutrophils, particularly the CD15+CD244-CD16lowcells (immature) by flow cytometry. RESULTS: About 1485 patients treatment with ICI were analysed. In ICI-treated patients, high dNLR (B) (~1/3rd) associated with worse progression-free (PFS)/overall survival (OS) (HR 1.56/HR 2.02, P < 0.0001) but not with chemotherapy alone (N = 173). High dNLR at C2 was associated with worse PFS/OS (HR 1.64/HR 2.15, P < 0.0001). When dNLR at both time points were considered together, those with persistently high dNLR (23%) had poor survival (mOS = 5 months (mo)), compared with high dNLR at one time point (22%; mOS = 9.2mo) and persistently low dNLR (55%; mOS = 18.6mo) (P < 0.0001). The dNLR impact remained significant after PD-L1 adjustment. By cytometry, high rate of immature neutrophils (B) (30/57) correlated with poor PFS/OS (P = 0.04; P = 0.0007), with a 12-week death rate of 49%. CONCLUSION: The dNLR (B) and its dynamics (C2) under ICI associate with ICI outcomes in aNSCLC. Persistently high dNLR (B+C2) correlated with early ICI failure. Immature neutrophils may be a key subpopulation on ICI resistance.