Recurrent Patterns of Clonal Evolution in Relapsed-Refractory DLBCL Following Treatment with R-CHOP

Introduction

Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma and is characterized by its genetic and clinical heterogeneity. Patients can develop DLBCL de novo or as a transformation from other lymphoid malignancies, most commonly follicular lymphoma. For patients with relapsed/refractory DLBCL (rrDLBCL), prognosis is extremely poor with 2-year overall survival of 20-40%. While numerous treatments are under investigation to improve patient outcomes, the success of these treatments has been limited as the genetic mechanisms underpinning treatment resistance are largely unknown. Identifying genomic alterations which contribute to relapse may improve salvage therapy for patients with rrDLBCL or allow patients to be stratified prior to frontline treatment.

Methods

To identify genomic alterations which contribute to R-CHOP resistance, we previously collected samples from patients enrolled in four clinical trials exploring candidate salvage therapies for patients with rrDLBCL as well as a retrospective rrDLBCL cohort, totalling 193 cases (133 de-novo DLBCL, 60 transformed). Plasma samples were collected from each patient upon relapse along with diagnostic tissue biopsies where available. A combination of exome sequencing and target-panel sequencing of lymphoma associated genes was performed on circulating tumour DNA and tissue biopsies (if available). Mutations implicated in R-CHOP resistance were identified through two complimentary strategies. First, the mutation frequency of recurrently mutated genes across de novo rrDLBCL samples was compared to a cohort of unrelated diagnostic DLBCL cases (n=1691) to identify genes enriched for mutations. Second, the genomic landscape and tumor clonal structure was compared prior to and following R-CHOP to identify mutations in each patient that underwent clonal expansion following therapy. Anti-CD20 antibody binding affinity of MS4A1 mutants was evaluated using flow cytometry on transfected CHO-S cells.

Results

We have identified five genes enriched for mutations in our rrDLBCL cohort relative to diagnostic DLBCL: KMT2D (Mutated in 49%, Q=0.0385), TP53 (47%, Q=1.07x10^-9), FOXO1 (11%, Q=0.0727), NFKBIE (11%, Q=0.0385), and MS4A1 (8%, Q=0.0522). Consistent with its characterization as a poor prognostic marker, mutations in TP53 were typically present at diagnosis and remained stable following R-CHOP therapy for both de novo and transformed DLBCL (23/27 cases, 85%). Recurrent mutations affecting Arg248 of TP53 (6.8%, Q=0.0413) were also clonally stable and have previously been associated with poor overall survival across several cancer types. The histone methyltransferase KMT2D is dominated by nonsense and frameshift mutations which were stable or underwent clonal expansion following R-CHOP (17/19, 89%). Recurrent missense mutations in MS4A1 targeted the small loop and adjacent transmembrane domains of CD20, including several patients with a Tyr86 mutation. Transfected cells carrying Tyr86Cys or Leu66Arg mutations were not bound by rituximab or other anti-CD20 antibodies including obinituzumab and ofatumumab. Subclonal populations containing MS4A1 mutations underwent clonal expansion (6 cases) or were stable (1 case) following treatment, including one case with multiple MS4A1 mutations in distinct subclonal populations which both underwent clonal expansion. In another unique case, a series of ctDNA samples were available prior to and following R-CHOP and salvage therapy, where we again observed convergent evolution of two mutually exclusive clonal subpopulations containing MS4A1 mutations. The first subpopulation underwent clonal expansion following frontline therapy but was extinguished following salvage therapy, while the other subpopulation underwent clonal expansion following salvage therapy and harboured a transmembrane domain mutation.

Conclusion

Mutations in TP53 and truncating mutations in KMT2D are generally present prior to treatment and will be investigated as biomarkers of treatment failure. Additional mutations are not always present at diagnosis, but their emergence can be detected in ctDNA and relapsed tissue, specifically mutations in MS4A1. As mutations in MS4A1 attenuate rituximab binding and are recurrently associated with clonal expansion, they likely impart a selective advantage and lead to resistance against anti-CD20 antibodies.