Genomic Profiling of Circulating Tumor DNA Reveals Patterns of Response and Refractoriness in Aggressive B-Cell Lymphoma - a Nordic Lymphoma Group Correlative Study

Introduction

Sequencing of circulating cell-free tumor DNA (ctDNA) has opened a diagnostic avenue for the patients with B-cell neoplasias, including diffuse large B-cell lymphoma (DLBCL). Treatment refractoriness and disease relapse after first-line therapy are associated with dismal survival. We sought to assess the clinical feasibility and prognostic value of ctDNA detection in prospectively collected samples from young high-risk patients treated uniformly in a Nordic phase II study.

Materials and Methods

The patients were treated in a phase II trial with dose-dense chemoimmunotherapy (biweekly R-CHOEP) and early systemic CNS prophylaxis with high-dose methotrexate (NLG-LBC-05; Leppä et al., 15-ICML; ClinicalTrials.gov NCT01325194). Longitudinal plasma samples from all trial centers were gathered and the plasma purified for cell-free DNA (cfDNA) in Helsinki, Finland (Figure 1A).

We designed a 225 kb custom panel targeting most commonly mutated genomic drivers, regions of aberrant somatic hypermutation and immunoglobulin genes. Next-generation sequencing was performed on longitudinal cfDNA samples and primary tumor samples together with matched germline controls from the first 32 patients in the study. Sequencing was completed on Hiseq2500 platform and depth after duplicate removal with unique molecular identifiers was on average 1,926x (ctDNA), 3,073x (primary tumors) and 2,171x (germline controls). Multi-tumor variant calling was performed by Mutect2.

Results

cfDNA was successfully extracted from 96 patients (Figure 1B). Among the patients with completed sequencing and bioinformatics pipeline, we identified a median of 87 traceable somatic mutations in 97% of the patients. In plasma samples, ctDNA was detectable in 93% of the patients prior to treatment. Compellingly, we observed that the levels of total cfDNA and ctDNA were higher at baseline in the patients who relapsed.

We evaluated the concordance between pretreatment ctDNA and primary tumor biopsy mutations in patients with available material. The percentage of shared somatic mutations ranged between 14 - 100%, suggesting substantial spatial heterogeneity in some cases. While the mutations in known lymphoma genes were mostly shared between the baseline ctDNA and the diagnostic tumor biopsy, some lymphomas displayed vast discordance mostly in the immunoglobulin loci, suggesting active somatic hypermutation processes upon clonal divergence in these lymphomas (Figure 1C).

In all longitudinally analyzed plasma samples, the levels of ctDNA dropped after the first three treatment courses. At the end of therapy, the levels of ctDNA were undetectable in most of the patients with durable remissions (Figure 1D #1). However, in the patients with progressive disease, the ctDNA levels increased at the end of the therapy (Figure 1D #2). In the patients with late relapse (>12 months progression free survival), we observed the appearance of a subset of the mutations that were identified at diagnosis, suggesting clonal selection (Figure 1D #3).

Conclusion

Our findings elucidate the biology of ctDNA in B-cell lymphomas, and the ctDNA kinetics in homogeneously treated patients with primary high-risk DLBCL.

Figure 1. A) Collection of cfDNA samples in the CHIC trial. B) Swimmers plot of the patients with extracted cfDNA in the liquid biopsy cohort. C) Concordance between primary tumor and ctDNA mutations at diagnosis. CHIC_100: high concordance. CHIC_88: low concordance. D) Kinetics of ctDNA in patients with different diseases courses.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal