Model selection for prognostic time-to-event gene signature discovery with applications in early breast cancer data

Model selection between competing models is a key consideration in the discovery of prognostic multigene signatures. The use of appropriate statistical performance measures as well as verification of biological significance of the signatures is imperative to maximise the chance of external validation of the generated signatures. Current approaches in time-to-event studies often use only a single measure of performance in model selection, such as logrank test p-values, or dichotomise the follow-up times at some phase of the study to facilitate signature discovery. In this study we improve the prognostic signature discovery process through the application of the multivariate partial Cox model combined with the concordance index, hazard ratio of predictions, independence from available clinical covariates and biological enrichment as measures of signature performance. The proposed framework was applied to discover prognostic multigene signatures from early breast cancer data. The partial Cox model combined with the multiple performance measures were used in both guiding the selection of the optimal panel of prognostic genes and prediction of risk within cross validation without dichotomising the follow-up times at any stage. The signatures were successfully externally cross validated in independent breast cancer datasets, yielding a hazard ratio of 2.55 [1.44, 4.51] for the top ranking signature.

TBX2 represses CST6 resulting in uncontrolled legumain activity to sustain breast cancer proliferation: a novel cancer-selective target pathway with therapeutic opportunities.

TBX2 is an oncogenic transcription factor known to drive breast cancer proliferation. We have identified the cysteine protease inhibitor Cystatin 6 (CST6) as a consistently repressed TBX2 target gene, co-repressed through a mechanism involving Early Growth Response 1 (EGR1). Exogenous expression of CST6 in TBX2-expressing breast cancer cells resulted in significant apoptosis whilst non-tumorigenic breast cells remained unaffected. CST6 is an important tumor suppressor in multiple tissues, acting as a dual protease inhibitor of both papain-like cathepsins and asparaginyl endopeptidases (AEPs) such as Legumain (LGMN). Mutation of the CST6 LGMN-inhibitory domain completely abrogated its ability to induce apoptosis in TBX2-expressing breast cancer cells, whilst mutation of the cathepsin-inhibitory domain or treatment with a pan-cathepsin inhibitor had no effect, suggesting that LGMN is the key oncogenic driver enzyme. LGMN activity assays confirmed the observed growth inhibitory effects were consistent with CST6 inhibition of LGMN. Knockdown of LGMN and the only other known AEP enzyme (GPI8) by siRNA confirmed that LGMN was the enzyme responsible for maintaining breast cancer proliferation. CST6 did not require secretion or glycosylation to elicit its cell killing effects, suggesting an intracellular mode of action. Finally, we show that TBX2 and CST6 displayed reciprocal expression in a cohort of primary breast cancers with increased TBX2 expression associating with increased metastases. We have also noted that tumors with altered TBX2/CST6 expression show poor overall survival. This novel TBX2-CST6-LGMN signaling pathway, therefore, represents an exciting opportunity for the development of novel therapies to target TBX2 driven breast cancers.