Background. Diffuse large B-cell lymphoma (DLBCL) displays a high degree of molecular heterogeneity and may be classified into molecular clusters harboring specific lesions and therapeutic vulnerabilities. To date, DLBCL molecular clusters have been identified on the tissue biopsy and do not consider mutations deriving from different anatomical compartments. Currently, DLBCL prone to become long-term survivors after R-CHOP or, alternatively, destined to relapse early after treatment initiation cannot be unequivocally identified a priori. The quantification of circulating tumor DNA (ctDNA) on the liquid biopsy at the time of diagnosis allows to stratify the outcome of DLBCL patients but its integration with molecular clustering has so far not been evaluated.

Purpose. The aims of this study were: i) to evaluate the potential contribution of liquid biopsy to the identification of DLBCL molecular clusters; and ii) to evaluate whether the assignment to a specific molecular cluster may improve patients’ stratification in combination with ctDNA load.

Methods. A multicenter cohort of newly diagnosed DLBCL patients provided with ctDNA from the plasma and genomic DNA (gDNA) from the lymph node (LN) biopsy represented the basis of this study. gDNA extracted from granulocytes was analyzed for comparative purposes. The LyV4.0 CAPP-Seq assay comprising a panel of 59 genes relevant to B-cell malignancies was used. The LymphGen tool was utilized for cluster analysis.

Results. The cohort included 77 newly diagnosed DLBCL patients treated prospectively with R-CHOP. The median age of the studied cohort was 67 years, 68.8% presented with stage III-IV and 22.1% had a high-risk IPI. After a median follow-up of 33.9 months, the 40-month progression-free survival (PFS) and overall survival (OS) were 66.5% and 80.2%, respectively. Mutational analysis identified at least one somatic non-synonymous mutation in 92.2% of patients (71/77) in the LN biopsy and in 87.0% (67/77) in the ctDNA. The most frequently mutated genes in both compartments and in more than 20% of patients were KMT2Dfollowed by PIM1 and HIST1H1E. To assign patients to molecular clusters, we applied the LymphGen tool to both ctDNA and LN. This tool allowed the cluster assignment in 40.3% (27/67) DLBCL on ctDNA and in 46.5% (33/71) on the LN biopsy. On ctDNA, 9 patients were classified as MCD, 5 as ST2, 7 as EZB, 5 as BN2 and 1 as molecular composite (BN2/ST2). By evaluating the clinical impact of molecular clusters, patients assigned to the ST2 and BN2 clusters in both ctDNA (p=0.032) and in LN biopsy (p=0.007), displayed a very good PFS compared to other patients. By recursive partitioning for PFS, patients with a baseline ctDNA load <2Log10hGE (N=17) had a 40-month PFS of 94.1% compared to 54.6% for patients (N=49) with higher ctDNA levels (p=0.003) and were not further stratified. Conversely, patients with ctDNA levels >2Log10hGE were further stratified by the assignment to the ST2 and/or BN2 clusters. Patients assigned to cluster ST2 or BN2 (N=10), despite having ctDNA levels >2Log10hGE, had a 40-month PFS of 100% compared to 44.4% for patients (N=39) classified as MCD or EZB or not classified (p=0.004). Therefore, by combining ctDNA levels and molecular clusters identified on the liquid biopsy, we identified 2 different groups of patients with a significantly different PFS and OS (both p<0.001). Patients with a <2Log10hGE and/or assigned to cluster BN2 and ST2 presented a 40-month PFS and OS of 96.2% and 100%, respectively. Conversely, patients with a ctDNA level >2Log10hGE and not assigned to cluster ST2 and BN2 presented a 40-month PFS and OS of 44.4% and 61.5%, respectively (Figure 1). To evaluate the ctDNA dynamics according to a specific molecular cluster, we analyzed the ctDNA of 48 patients after one cycle of R-CHOP. Ten were classified as ST2 or BN2 at the time of diagnosis of which, after 1 cycle of R-CHOP, eight (80%) reached an early molecular response (EMR). Interestingly, the 2 ST2/BN2 patients who did not achieve an EMR have remained in CR and have not progressed at the last follow-up.

Conclusions. Molecular clustering on the liquid biopsy reflects in most cases the prognostic impact of clusters identified in the tissue biopsy. If validated, the combination of ctDNA levels <2 Log10hGE and ST2/BN2 molecular clusters on the liquid biopsy represents a powerful biomarker to identify patients who can benefit most from R-CHOP.

Rossi:Janssen: Consultancy, Honoraria, Other: Travel Support, Research Funding; Gilead: Other: honoraria, advisory board fees , Research Funding; AbbVie: Consultancy, Honoraria, Other: Travel Support, Research Funding; AstraZeneca: Consultancy, Honoraria, Other: Travel Support, Research Funding; MSD: Other: advisory board fees ; BMS: Consultancy, Honoraria, Other: Travel Support; BeiGene: Consultancy, Honoraria, Other: Travel Support, Research Funding. Gaidano:Janssen: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Incyte: Membership on an entity's Board of Directors or advisory committees; Beigene: Membership on an entity's Board of Directors or advisory committees; Astra-Zeneca: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees; Abbvie: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau; Roche: Honoraria, Membership on an entity's Board of Directors or advisory committees.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution