TNFRSF14 and EZH2 Mutations, Chr2p Gain and Copy Number Changes Targeting Genes Whose Proteins Interact with the Microenvironment In Transformed Follicular Lymphoma

The accurate prediction and optimal management of transformed follicular lymphoma (t-FL) remains a significant clinical challenge. Investigation of the prognostic significance of genetic factors and the immune microenvironment in FL provides insight into the molecular pathogenesis of transformation and should aid in the development of robust molecular predictors of disease.

To address this issue, molecular profiling was performed on 91 samples including paired FL/t-FL samples from each of 31 patients (with additional germline (n=19) and relapse (n=10) samples where available). Median age at FL diagnosis was 46 years (range 22–71) with a median time to transformation of 4 years (1-16). Median follow-up of surviving patients was 10 years (2-19).

The two most frequently mutated genes detected in FL/t-FL were TNFRSF14 (chr1p36.32) and EZH2 (chr7q36.1). Twelve TNFRSF14 mutations were detected in 10 patients (32%) with 3 arising in t-FL and one rendered homozygous on transformation. All were confined to exons 1–6 with 7/12 mutations causing premature termination of the protein. Heterozygous non-synonymous mutations in codon 641 of EZH2 occurred in 9 patients (29%) with 2 arising upon transformation. In mutated cases, the TNFRSF14 and EZH2 loci were subject to recurrent copy number (CN) loss/uniparental disomy (6/12) and gain (6/9) respectively. Genome-wide profiling was performed using the Affymetrix SNP 6.0 array to search for co-operating genetic events. A total of 1120 CN aberrations (CNAs) were identified across the cohort with a median of 5 gains (range 0–13) and 7 losses (0-42) in the earliest available FL (FL1) samples and 8 gains (0-33) and 10 losses (4-32) in the initial t-FL (t-FL1) samples. Recurrent CNAs included loss of chr1p, 6q, 9p, 17p and gain of 1q, 6p, 18 and × in both FL1 and t-FL1.

In 18 cases, FL1 samples had CN changes absent from subsequent relapsed FL/t-FL samples. This is consistent with t-FL evolving from a common progenitor cell (CPC). Divergent changes seen in both FL and t-FL at relapse after t-FL1 further augment this hypothesis. CNAs common among different samples identify early events in FL/t-FL development. Recurrent regions of loss and gain affecting ≥20% of cases ranged in size from 4 kb to 60 Mb. Of these, gain 2p16.1-p15 (including REL and BCL11A) in FL1, predicted worse clinical outcome (survival from FL diagnosis; logrank p = 0.004, hazard ratio = 7). No cases with gain 2p16.1-p15, gain 8q24.13-q24.3, del 9p21.3 or gain 18q21.2-q21.33 were alive. Novel CNAs included gain 16p12.1 (containing IL4-R, IL21-R, NSMCE1) in 5 t-FL1 cases, recurrent deletion of ETS1 (n=4) on chr11q24.3 and further micro-deletions targeting CREBBP (chr16p13.3) and ALK (chr2p23.2).

The high mutational frequency of TNFRSF14 together with recurrent gain of IL4-R suggests a link between intra-cellular events and B-T cell interactions in the microenvironment. Enrichment of EZH2 mutations, the prognostic impact of gain REL/BCL11a and recurrent CN changes in CPC cases complete a core set of genetic changes that are potential molecular markers in t-FL.

No relevant conflicts of interest to declare.