Predictive impact of inflammation scores in non-small cell lung cancer patients treated with immunotherapy rechallenge

Most of Non-Small-Cell lung cancer (NSCLC) patients actually receive an immune checkpoint inhibitor (ICP) as first line treatment. Therapeutic alternatives at progression are limited and rechallenge with ICI may present a major therapeutic opportunity. Prior identification of patients who could benefit from this strategy remains an open question. It has previously been shown that inflammation scores can identify patients who respond to ICP. Their predictive role in rechallenged patients have never been explored. We explored the predictive role of NLR, dNLR, PRL and MLR scores in a cohort of 187 NSCLC patients that received at least one line of ICP. Patients were divided into rechallenged (Re = 81) or not rechallenged (NRe = 106). Biological data at C1D1 (+/- 15 days ) of rechallenge or first chemotherapy regimen post-ICP were collected. Inflammation scores cut-off were estimated by ROC curves. Cox proportional hazard models were used in the subgroup analysis. Patient in the Re group showed a higher overall survival (OS) since the first ICP compared to the NRe group (39.7 versus 13.3 months, p < 0.0001). We identified a prognostic cut-off able to stratify patient in responders and not-responders (time to treatment failure > or ≤ 3 months) to rechallenge (NLR: 4.2 [AUC 0.75, p = 0.0003]; dNLR: 2.14 [AUC 0.72, p =0.0013]; PLR: 188 [AUC 0.82, p < 0.0001]; MLR: 0.61 [AUC 0.73, p = 0.001]). We showed that patients scored as “low” had a better OS from rechallenge than patients scored as “high” regardless the score considered. The prognostic impact of the inflammation scores was confirmed at the multivariate analysis including smoking status, first ICP duration, best response obtained, reason for stopping, intercurrent treatments between the first ICP and the rechallenge and the ECOG status at the time of the rechallenge. While NLR and dNLR confirmed their prognostic value in the NRe group, MLR and PLR scores were not able to discriminate responder from not responder patients treated with chemotherapy. According to our results NLR and dNLR had a prognostic rather than predictive value in NSCLC treated with rechallenge. In contrast, PRL and MLR demonstrated a predictive value of response to rechallenge.