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
1260P - Efficacy and safety of sunvozertinib in prior platinum treated NSCLC patients with EGFR exon 20 insertion mutations: Primary analysis from the multinational WU-KONG1B pivotal study
Presenter: Ludovic Doucet
Session: Poster session 05
1261P - Efficacy of glecirasib in combination with JAB-3312 as a front-line treatment for patients with KRAS p.G12C mutated NSCLC with PD-L1 expression levels or co-mutations
Presenter: Jie Wang
Session: Poster session 05
1262P - Combined molecular analysis of circulating tumour DNA and tumour tissue to identify osimertinib resistance
Presenter: Tijmen van der Wel
Session: Poster session 05
1263P - Biomarker analysis of plasma samples in YAMATO study: A randomized phase II trial comparing switching treatment of osimertinib following 8 months of afatinib (A) and osimertinib alone (B) in untreated advanced NSCLC patients with common EGFR mutation (TORG1939/WJOG12919L)
Presenter: Hiroshige Yoshioka
Session: Poster session 05
1264P - Real-world evidence of treatment practices and therapeutic outcomes for newly diagnosed NSCLC patients with non-classical EGFR mutations demonstrates high unmet medical need
Presenter: John Heymach
Session: Poster session 05
1265P - A promising MET-EGFR bispecific nanobody-drug conjugate therapy for multiple solid tumours
Presenter: xianghai Cai
Session: Poster session 05
1266P - Interim analysis from the multicenter ROSE study: Radiation during osimertinib treatment safety and efficacy cohort
Presenter: Amanda Tufman
Session: Poster session 05
1267P - Sequential afatinib (AFA) to osimertinib (OSI) in EGFR-mutant NSCLC: Primary analysis of Gio-Tag Japan, a multicenter prospective observational study
Presenter: Naoto Takase
Session: Poster session 05
1268P - Concordances assessment between MET-positive circulating tumour cells and disease progression in patients with EGFR mutated NSCLC
Presenter: Jieun Park
Session: Poster session 05
1269P - Preventing infusion-related reactions with intravenous amivantamab: Updated results from SKIPPirr, a phase II study
Presenter: Luis Paz-Ares
Session: Poster session 05