Real-World Data on Pembrolizumab for Pretreated Non-Small-Cell Lung Cancer: Clinical Outcome and Relevance of the Lung Immune Prognostic Index.

Background: Pembrolizumab is licensed for the treatment of pre-treated and PD-L1 positive non-small cell lung cancer (NSCLC), but response is heterogeneous. In this context, the Lung Immune Prognostic Index (LIPI) has been proposed as tool to prognosticate outcome.

Objective: To investigate the real-world efficacy and safety of pembrolizumab in pre-treated NSCLC patients and the clinical utility of LIPI for patients' selection.

Patients And Methods: Patients with pre-treated NSCLC and PD-L1 ≥ 1% treated with pembrolizumab were included in this retrospective series. The LIPI was used to classify patients in 3 prognostics subgroups according to the pre-treatment dNLR (derived neutrophil to lymphocyte ratio) and LDH in blood. The prognostic impact of the LIPI on progression free survival (PFS) and overall survival (OS) was evaluated with Cox regression. The combined effect of LIPI and other relevant prognostic factors was explored with multivariate regression.

Results: In total, 113 consecutive patients were included. Median (mPFS) and mOS was 4.3 (2.6-6.7) and 13.5 (10.3-17.7) months, respectively. Good-, intermediate-, and poor-LIPI was found in 54 (47.8%), 45 (39.8%), and 8 (7.1%) patients, respectively. Median PFS was 5.1 (2.8-9.1), 3.0 (2.5-6.8), and 1.4 (0.5-18.7) months, and mOS was 17.2 (12.0-26.4), 11.8 (8.4-17.1), and 3.7 (0.5-not calculable) months for good-, intermediate-, and poor-LIPI group, respectively. Patients with intermediate-LIPI and poor-LIPI had worse PFS versus good-LIPI, and statistically significant worse OS (p = 0.030 and p = 0.013, respectively). In the multivariate analysis, intermediate- versus good-LIPI (p = 0.190) was not independently associated to PFS or OS. Patients with both good-LIPI and high (≥ 50%) PD-L1 had better OS than all other subgroups defined by LIPI and PD-L1. Immune-related adverse events (irAEs) occurred in 47 (41.6%) patients (12.4% grade ≥ 3). In a time-varying analysis, irAEs were statistically associated with longer OS (HR 0.51, 0.31-0.84; p = 0.008).

Conclusion: In our series, the outcome of pembrolizumab in pre-treated NSCLC is consistent with the registration trial. Lung Immune Prognostic Index is a readily available tool able to prognosticate outcome, also in PD-L1-high patients. The positive association between irAEs and OS might aid decision making.