Cell Free Tumor DNA for DLBCL Genotyping in a Phase II Randomized Trial Comparing Standard RCHOP Versus Brcap As First Line Treatment in Patients with Poor IPI DLBCL

INTRODUCTION

Cell free tumor DNA (cftDNA) based targeted next generation sequencing (NGS) is a novel tool that enables accurate and minimally invasive tumor genotyping.

MATERIAL AND METHODS

Here we have performed cftDNA targeted NGS for the molecular diagnosis of a series of 92 peripheral blood samples obtained at diagnosis in patients included in a phase II randomized clinical trial comparing standard RCHOP versus BRCAP as first line treatment in patients with poor IPI (aaIPI2-3 or aaIPI1 with high b2m, age 18-71, NCT01848132). Centralized histopathological review of diagnostic tissue samples confirmed the diagnosis: 83 DLBCL (68%, 55/81 GCB, 32%, 26/81 non-GCB, according to Hans immunohistochemical algorithm), 4 TCHRBCL, 3 PMBCL, 1 concurrent DLBCL and FL3A, 1 FL 3B. 7 out of 83 cases had DLBCL morphology and DH by FISH (6 MYC/BCL2 and 1 MYC/BCL6, all GCB-type by IHC). A targeted NGS approach using amplicon-based chemistry (Truseq, Illumina) that interrogates exonic regions of 35 relevant genes in DLBCL (1458 amplicons analyzed) was used for library generation from plasma derived DNA. Germline DNA from peripheral blood granulocytes from 89 out of 92 patients was sequenced in parallel and used as control DNA for variant calling. Only somatic variants with a read depth above 100 and Variant Allele Frequency > 5% were considered.

RESULTS

cftDNA concentration (hGE/mL) was associated with LDH levels (U/mL) at diagnosis. Mean cftDNA concentration was lower in patients with low LDH (U Mann whitney p= 0.005). Somatic mutations were identified in 74% (68 out of 92) patient plasma samples. 242 somatic mutations were considered after filtering. Mean of 3.6 mutations per patient. 10 genes were mutated in >10% of the patients, including: FAT2 (19.6%), ARID1A (18%), NOTCH1 (17.4%), NOTCH2 (15.2%), KMT2D (14.13%), SMARCA4 (14%), ATM (13%), EP300 (13%), EZH2 (13%), PLCG2 (13%). Recurrent mutations were found in EZH2 (2 patients). Mutations in BCR/TLR pathway genes were found in less than 10% of the cases (CARD11, 7.6%; MYD88, 4.4%; CD79A, 5.4%; CD79B, 1%; BTK, 4.4%). NOTCH1 mutations were found preferentially in GCB type DLBCL (12/55 GCB vs 1/26 nonGCB, Chi square p value <0.05). The mutational profile of the seven High Grade DLBCL/DH cases was similar to that found in the other GCB-type DLBCL-NOS. PMBCL cases had mutations in ARID1A, NOTCH1, NOTCH2, EP300, PLCG2 and SGK1. Two out 4 TCHRBCL cases had no mutations. One case had a mutation in MEF2B and the other in ARID1A, NOTCH2, SMARCA4, PLCG2, CREBBP, SGK1 and BRAF.

DISCUSSION

cftDNA targeted NGS captures the mutational profile in a significant fraction of DLBCL patients and enables tumor genotyping in the clinical trial setting. In contrast with previously published data mutations in NOTCH pathway, including NOTCH1, NOTCH2 and SGK1 were relatively frequent in our series of poor IPI patients, outnumbering the frequency of mutations in BCR/TLR pathway. NOTCH pathway mutations were also preferentially found in GCB-type DLBCL-NOS, PMBCL and 1 case of TCHRBCL. In summary cftDNA genotyping based on targeted NGS can provide potentially useful information for therapy selection in DLBCL patients.