Identification of PIK3CA multigene mutation patterns associated with superior prognosis in stomach cancer

Background PIK3CA is the second most frequently mutated gene in cancers and is extensively studied for its role in promoting cancer cell resistance to chemotherapy or targeted therapy. However, PIK3CA functions have mostly been investigated at a lower-order genetic level, and therapeutic strategies targeting PIK3CA mutations have limited effects. Here, we explore crucial factors interacting with PIK3CA mutations to facilitate a significant marginal survival effect at the higher-order level and identify therapeutic strategies based on these marginal factors. Methods Mutations in stomach adenocarcinoma (STAD), breast adenocarcinoma (BRCA), and colon adenocarcinoma (COAD) samples from The Cancer Genome Atlas (TCGA) database were top-selected and combined for Cox proportional-hazards model analysis to calculate hazard ratios of mutation combinations according to overall survival data and define criteria to acquire mutation combinations with considerable marginal effects. We next analyzed the PIK3CA + HMCN1 + LRP1B mutation combination with marginal effects in STAD patients by Kaplan-Meier, transcriptomic differential, and KEGG integrated pathway enrichment analyses. Lastly, we adopted a connectivity map (CMap) to find potentially useful drugs specifically targeting LRP1B mutation in STAD patients. Results Factors interacting with PIK3CA mutations in a higher-order manner significantly influenced patient cohort survival curves (hazard ratio (HR) = 2.93, p-value = 2.63 × 10− 6). Moreover, PIK3CA mutations interacting with higher-order combination elements distinctly differentiated survival curves, with or without a marginal factor (HR = 0.26, p-value = 6.18 × 10− 8). Approximately 3238 PIK3CA-specific higher-order mutational combinations producing marginal survival effects were obtained. In STAD patients, PIK3CA + HMCN1 mutation yielded a substantial beneficial survival effect by interacting with LRP1B (HR = 3.78 × 10− 8, p-value = 0.0361) and AHNAK2 (HR = 3.86 × 10− 8, p-value = 0.0493) mutations. We next identified 208 differentially expressed genes (DEGs) induced by PIK3CA + HMCN1 compared with LRP1B mutation and mapped them to specific KEGG modules. Finally, small-molecule drugs such as geldanamycin (connectivity score = − 0.4011) and vemurafenib (connectivity score = − 0.4488) were selected as optimal therapeutic agents for targeting the STAD subtype with LRP1B mutation. Conclusions Overall, PIK3CA-induced marginal survival effects need to be analyzed. We established a framework to systematically identify crucial factors responsible for marginal survival effects, analyzed mechanisms underlying marginal effects, and identified related drugs.

Variable selection via penalized generalized estimating equations for a marginal survival model

Clustered and multivariate survival times, such as times to recurrent events, commonly arise in biomedical and health research, and marginal survival models are often used to model such data. When a large number of predictors are available, variable selection is always an important issue when modeling such data with a survival model. We consider a Cox’s proportional hazards model for a marginal survival model. Under the sparsity assumption, we propose a penalized generalized estimating equation approach to select important variables and to estimate regression coefficients simultaneously in the marginal model. The proposed method explicitly models the correlation structure within clusters or correlated variables by using a prespecified working correlation matrix. The asymptotic properties of the estimators from the penalized generalized estimating equations are established and the number of candidate covariates is allowed to increase in the same order as the number of clusters does. We evaluate the performance of the proposed method through a simulation study and analyze two real datasets for the application.