2024 World Conference on Lung Cancer (WCLC) - ePosters-EP.11A.15 NSCLC-ClustOpt: A Semi-Supervised Optimized Clustering to Stratify Patients with NSCLC on Immune Checkpoint Blockers Therapy

EP.11A.15 NSCLC-ClustOpt: A Semi-Supervised Optimized Clustering to Stratify Patients with NSCLC on Immune Checkpoint Blockers Therapy

Back to course

Pdf Summary

The study titled "NSCLC-ClustOpt: A semi-supervised optimized clustering to stratify patients with NSCLC on immune checkpoint blockers therapy" by Francesco Paoloni and colleagues aims to improve the prognosis of patients with advanced non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors (ICIs). It introduces a semi-supervised K-means clustering algorithm called ClustOpt to effectively stratify patients based on their baseline clinicopathologic features, including the neutrophil-to-lymphocyte ratio (NLR), nutritional status (through the CONUT score), comorbidities, and body mass composition assessed using CT scans.<br /><br />The study retroactively analyzed data for 50 patients, identifying two distinct clusters with differing progression-free survival (PFS) and overall survival (OS) outcomes. Using ClustOpt, these patients were categorized into two groups. Cluster 1 showed significantly better clinical outcomes with a median PFS of 24.4 months and OS of 41.8 months, compared to 7.1 months PFS and 13.3 months OS in Cluster 2. Key factors influencing clustering included age, NLR, number of metastatic sites, and CONUT score.<br /><br />Further analysis revealed significant differences in body composition between the groups. Patients in Cluster 1 had higher median lean psoas muscle area and psoas muscle index, and lower intramuscular adipose tissue content, suggesting better muscle mass indicators contributing to longer survival.<br /><br />This research highlights the potential of using a semi-supervised machine learning approach with easily obtainable baseline features for prognostic stratification in NSCLC patients treated with ICIs, offering a method to potentially improve treatment plans based on the anticipated clinical outcomes associated with each cluster.

Asset Subtitle

Francesco Paoloni

Meta Tag

Speaker Francesco Paoloni

Topic Metastatic Non-small Cell Lung Cancer – Immunotherapy

Keywords

NSCLC

immune checkpoint inhibitors

ClustOpt

semi-supervised clustering

prognostic stratification

neutrophil-to-lymphocyte ratio

CONUT score

body mass composition

progression-free survival

overall survival