Abstract 1271P
Background
Understanding the associates and causes of progression after first-line treatment in EGFR-mutant advanced NSCLC remains a critical yet elusive endeavor. In this study, we employed a comprehensive approach integrating machine learning classification and causal inference analysis to unveil the determinants of progression in this patient cohort.
Methods
Retrospective data from eleven academic centers were gathered to compile a cohort of EGFR mutant patients. Classical patient features, treatment characteristics, and the neutrophil to lymphocyte ratio (NLR) at the onset of first-line treatment were analyzed. Fifteen machine learning algorithms were employed to identify the optimal model, evaluated based on F1 score and area under the curve (AUC). Additionally, Linear Non-Gaussian Acyclic Model (LINGAM) was utilized for causal inference analysis to delineate the causal relationships driving progression.
Results
Among 333 patients initially evaluated, 205 cases were deemed suitable for machine learning analysis after accounting for missing data and follow-up duration. Linear discriminant analysis emerged as the most accurate model, achieving an AUC of 0.74 and an F1 score of 0.77. The top three influential variables identified by this model were smoking status, NLR category, and type of first-line treatment with variable importance scores of 1.2, 0.8, and 0.6, respectively. Moreover, LINGAM analysis revealed NLR category, smoking status, and gender as causal factors for progression, with causal estimate scores of -0.20, -0.17, and -0.04, respectively.
Conclusions
This study represents a pioneering effort in utilizing machine learning to explore the determinants of progression after first-line treatment in EGFR-mutant, advanced NSCLC. Our findings underscore the roles of smoking status, NLR category, and type of first-line treatment in driving progression. Furthermore, the causal inference analysis sheds light on smoking status, NLR category, and gender as causally significant factors. This understanding paves the way for elucidating the pathogenesis and developing novel therapeutic strategies for EGFR-mutant advanced NSCLC.
Clinical trial identification
Editorial acknowledgement
During the preparation of this work the authors used ChatGpt 3.5 in order to enhance readibility. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.
Resources from the same session
1334P - PET/CT-guided immune checkpoint blocker treatment discontinuation vs treatment continuation in lung cancer long-term responders: A National Network Genomic Medicine Lung Cancer Germany (nNGM) analysis
Presenter: Nikolaj Frost
Session: Poster session 05
1335P - The relationship between nivolumab pharmacokinetics and cancer cachexia biomarkers in patients with metastatic non-small cell lung cancer (NSCLC)
Presenter: Maaike Hofman
Session: Poster session 05
1338P - TP53 truncating and missense mutations are linked to differential response to checkpoint blockade in patients with advanced non-small cell lung cancer (NSCLC)
Presenter: Fabrizio Citarella
Session: Poster session 05
1339P - Impact of KRAS, STK11, and KEAP1 co-mutations on survival outcome and response to chemoimmunotherapy in patients with metastatic NSCLC
Presenter: Utsav Joshi
Session: Poster session 05
1341P - Exploring the role of the gut microbiome on the efficacy of ipilimumab and nivolumab in advanced non-small cell lung cancer: A prospective observational study
Presenter: Yuki Katayama
Session: Poster session 05
1342P - The efficacy of pembrolizumab vs nivolumab plus ipilimumab in metastatic NSCLC in relation to PD-L1 and TMB status
Presenter: Walid Shalata
Session: Poster session 05
1343P - Adding histology-driven chemotherapy (ChT) to overcome primary resistance to first-line immunotherapy (ICI) in patients (pts) with advanced non-small cell lung cancer (aNSCLC) with PD-L1 ≥50%
Presenter: Andrea De Giglio
Session: Poster session 05
1344P - Plasma proteomics indicated predictive biomarkers for immuno-chemotherapy in stage IIIB-IV non-small cell lung cancer without EGFR/ALK alterations
Presenter: Zhihuang Hu
Session: Poster session 05