Integrative Analysis of Prognosis Data on Multiple Cancer Subtypes using Penalization

TL;DR

This paper introduces a penalization-based integrative analysis method for prognosis data across multiple cancer subtypes, improving gene identification and outperforming traditional meta-analysis.

Contribution

It develops a novel compound penalization approach within an AFT model to analyze multiple cancer subtypes simultaneously, capturing heterogeneity in genetic associations.

Findings

The method outperforms meta-analysis in simulations.

It identifies subtype-specific prognostic genes.

Application to NHL data reveals new genetic insights.

Abstract

In cancer research, profiling studies have been extensively conducted, searching for genes/SNPs associated with prognosis. Cancer is a heterogeneous disease. Examining similarity and difference in the genetic basis of multiple subtypes of the same cancer can lead to better understanding of their connections and distinctions. Classic meta-analysis approaches analyze each subtype separately and then compare analysis results across subtypes. Integrative analysis approaches, in contrast, analyze the raw data on multiple subtypes simultaneously and can outperform meta-analysis. In this study, prognosis data on multiple subtypes of the same cancer are analyzed. An AFT (accelerated failure time) model is adopted to describe survival. The genetic basis of multiple subtypes is described using the heterogeneity model, which allows a gene/SNP to be associated with the prognosis of some subtypes but not the others. A compound penalization approach is developed to conduct gene-level analysis and identify genes that contain important SNPs associated with prognosis. The proposed approach has an intuitive formulation and can be realized using an iterative algorithm. Asymptotic properties are rigorously established. Simulation shows that the proposed approach has satisfactory performance and outperforms meta-analysis using penalization. An NHL (non-Hodgkin lymphoma) prognosis study with SNP measurements is analyzed. Genes associated with the three major subtypes, namely DLBCL, FL, and CLL/SLL, are identified. The proposed approach identifies genes different from alternative analysis and has reasonable prediction performance.