Background: The advanced lung cancer inflammation index [ALI: body mass index  serum albumin/neutrophil-tolymphocyte ratio (NLR)] reflects systemic host inflammation, and is easily reproducible. We hypothesized that ALI could assist guidance of non-small-cell lung cancer (NSCLC) treatment with immune checkpoint inhibitors (ICIs). Patients and methods: This retrospective study included 672 stage IV NSCLC patients treated with programmed deathligand 1 (PD-L1) inhibitors alone or in combination with chemotherapy in 25 centers in Greece and Germany, and a control cohort of 444 stage IV NSCLC patients treated with platinum-based chemotherapy without subsequent targeted or immunotherapy drugs. The association of clinical outcomes with biomarkers was analyzed with Cox regression models, including cross-validation by calculation of the Harrell's C-index. Results: High ALI values (>18) were significantly associated with longer overall survival (OS) for patients receiving ICI monotherapy [hazard ratio (HR) ¼ 0.402, P < 0.0001, n ¼ 460], but not chemo-immunotherapy (HR ¼ 0.624, P ¼ 0.111, n ¼ 212). Similar positive correlations for ALI were observed for objective response rate (36% versus 24%, P ¼ 0.008) and time-on-treatment (HR ¼ 0.52, P < 0.001), in case of ICI monotherapy only. In the control cohort of chemotherapy, the association between ALI and OS was weaker (HR ¼ 0.694, P ¼ 0.0002), and showed a significant interaction with the type of treatment (ICI monotherapy versus chemotherapy, P < 0.0001) upon combined analysis of the two cohorts. In multivariate analysis, ALI had a stronger predictive effect than NLR, PD-L1 tumor proportion score, lung immune prognostic index, and EPSILoN scores. Among patients with PD-L1 tumor

The low average error rates and high average F1-scores of each pipeline demonstrate the suitability of the investigated NPL methods. However, it was also shown that there is no best practice for an automatic classification of data elements from free-text diagnostic reports.

No abstract