Abstract 116P
Background
Copy number alterations (CNAs) have emerged as promising biomarkers for predicting prognosis and therapeutic response in cancer. However, integrating CNA data from diverse gene panel-based sequencing platforms has posed a considerable challenge. This study aimed to develop a novel cytoband-mediated CNA algorithm capable of reliably predicting outcomes across multiple cancer types, no matter what gene panel was used.
Methods
This pan-cancer study included data from 16,093 patients across 6 datasets, comprising various tumor types profiled using whole-genome sequencing, whole-exome sequencing, or gene panel-based targeted sequencing. The CNA scoring algorithm leveraged the principle of linkage disequilibrium, using cytobands as an intermediary to integrate CNA data from different gene panels. The prognostic and predictive performance of the CNA score was extensively validated across the cohorts.
Results
Patients stratified into high-risk and low-risk groups based on the CNA prognosis score exhibited marked differences in survival outcomes, with the high-risk group demonstrating significantly poorer prognosis (with all p values less than 0.05). This prognostic power was maintained across diverse datasets. Moreover, the CNA response score outperformed traditional predictive biomarkers, such as homologous recombination deficiency score, tumor mutational burden, and microsatellite instability, in its ability to predict durable clinical benefit and response to immune checkpoint inhibitor therapy. Bootstrap analysis shows a stable predictive ability when the number of genes in panel exceeds 300.
Conclusions
This study established a predictive model that leverages CNA to forecast clinical outcomes, with cytobands serving as an intermediary. Irrespective of the input data source, whether it be whole genome sequencing or gene panel-based targeted sequencing, the model consistently yields stable predictive performance through the cytoband intermediary.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.