A diversity of T helper (Th) subsets (Th1, Th2, Th17) has been identified in the human tumor microenvironment. In breast cancer, the role of Th subsets remains controversial, and a systematic study integrating Th subset diversity, T cell inflammation, breast cancer molecular subtypes, and patient prognosis, is lacking. In primary untreated breast cancer samples, we analyzed 19 Th cytokines at the protein level. Eight were T cell-specific, and subsequently measured in 106 prospectively-collected untreated samples. The dominant Th cytokines across all breast cancer samples were IFN-γ and IL-2. Th2 cytokines (IL-4, IL-5, IL-13) were expressed at low levels and not associated with any breast cancer subtype. Th17 cytokines (IL-17A and IL-17F) were up-regulated in triple negative breast cancer (TNBC), specifically in T cell non-inflamed tumors. In order to get insight into prognosis, we exploited the METABRIC transcriptomic dataset. We derived Th1, Th2, and Th17 metagenes based on manually curated Th signatures, and found that a high Th17 metagene was of good prognosis in T cell non-inflamed TNBC. Multivariate Cox modeling selected the Nottingham Prognostic Index (NPI), Th2 and Th17 metagenes as additive predictors of breast cancer-specific survival, which defined novel and highly distinct prognostic groups within TNBC. Our results reveal that Th17 is a novel prognostic composite biomarker in T cell non-inflamed TNBC. Integrating immune cell and tumor molecular diversity is an efficient strategy for prognostic stratification of cancer patients.

Omic data are characterized by the presence of strong dependence structures that result either from data acquisition or from some underlying biological processes. Applying statistical procedures that do not adjust the variable selection step to the dependence pattern may result in a loss of power and the selection of spurious variables. The goal of this paper is to propose a variable selection procedure within the multivariate linear model framework that accounts for the dependence between the multiple responses. We shall focus on a specific type of dependence which consists in assuming that the responses of a given individual can be modelled as a time series. We propose a novel Lasso-based approach within the framework of the multivariate linear model taking into account the dependence structure by using different types of stationary processes covariance structures for the random error matrix. Our numerical experiments show that including the estimation of the covariance matrix of the random error matrix in the Lasso criterion dramatically improves the variable selection performance. Our approach is successfully applied to an untargeted LC-MS (Liquid Chromatography-Mass Spectrometry) data set made of African copals samples. Our methodology is implemented in the R package MultiVarSel which is available from the Comprehensive R Archive Network (CRAN).