Abstract IA010: TP53-independent risk stratification in mantle cell lymphoma using a novel gene expression signature

Abstract

Mantle cell lymphoma (MCL) is an uncommon B-cell lymphoma with significant clinical heterogeneity. Despite improved prognostication in MCL, the impact of specific genomic changes is not well characterized beyond altered TP53, a high-risk marker commonly assessed at diagnosis, and the “proliferation signature” initially developed by Rosenwald et al. To bridge this knowledge gap, we applied tumor sequencing in a large MCL cohort to investigate the prognostic impact of genomic abnormalities in patients with MCL with and without TP53 alterations. The Atlas of Blood Cancer Genomes project is an international effort aimed at sequencing all hematologic malignancies. Its MCL component included 252 patients with detailed clinical data and tumors subjected to whole exome, whole transcriptome, and targeted sequencing. TP53-mutated or -deleted cases were identified via sequencing and clinical pathology reports. To develop a prognostic gene expression signature independent of the proliferation signature, the MCL cohort was divided into a training set and a test set, keeping the following factors constant among the two sets: age at diagnosis, stage, Ki67 by immunohistochemistry, TP53 status, MIPI score, histologic subtype, and overall survival duration and censoring. In the training set, bivariate survival analysis quantified the relationship of each gene’s log2-normalized expression value with survival. Subsequent filtering removed genes with low variance across the training set or high Pearson coefficient with the proliferation signature. This produced a 43-gene signature evaluated in the test set using a Cox proportional hazards model. We next analyzed the factors driving the novel prognostic signature’s scores. Gene set enrichment analysis and estimation of tumor microenvironment cell fractions using FARDEEP with the CIBERSORTx LM22 matrix showed that tumors designated as low-risk by the novel signature had higher proportions of CD8+ T cells and T follicular helper cells. The “immune signature” did not significantly associate with age, stage, or ECOG performance status, but its high-risk group was strongly associated with blastoid subtype and Ki67>50%. When applied to the entire cohort, both the proliferation and the immune signature effectively risk-stratified patients with MCL in terms of overall survival. The immune signature risk-stratified cases regardless of TP53 status (p<0.001 for low- vs. high-risk in both TP53-altered and TP53-wildtype cases), while the proliferation signature did not. Similar risk-stratification was observed when the immune signature was applied to cases deemed high- or low-risk by the proliferation score (p<0.001). In this largest study of the genomic landscape in MCL, we identified a novel prognostic gene expression signature that appears to be driven by microenvironmental features. The immune signature stratifies risk within and across known prognostic groups, including in TP53-aberrant cases. Analyses assessing the impact of the immune signature in pre-treatment vs. relapsed cases will be reported at the meeting.