Multiparametric integration improves diagnostic categorisation of mature, non-CLL B-cell leukemias including atypical CLL - results from the prospective enable-NGS study

Abstract

Abstract Introduction Enhanced diagnostic categorisation improves patient access to information, novel therapies and clinical trials. We previously reported the use of a capture based next-generation sequencing (NGS) panel to improve diagnosis of unclassifiable leukemic indolent B-cell Non-Hodgkin's Lymphomas (LI BNHL) in a prospective multicentre study. We report updated findings including 4-year clinical follow up. Methods 119 patients with LI B-NHL deemed unclassifiable on standard peripheral blood (PB) diagnostics were recruited prospectively from 14 centres around the UK over 4 years. Classic chronic lymphocytic leukaemia (CLL) with Matutes score 4 or 5, CLL-type monoclonal b-cell lymphocytosis (MBL), mantle cell lymphoma (MCL), hairy cell leukaemia (HCL), high grade lymphoma and non-clonal lymphocytosis on standard PB diagnostics were excluded. Sequencing methods using the EuroClonality-NDC assay and analyses for small variants, copy number variants (CNVs) and translocations were described previously (Cross et al. ASH 2020). EuroClonality-NDC confirmed IGH,IGK and/or IGL clonality in all cases. DNA was analysed by 450/850K Illumina methylation array at Ohio State University to provide a DNA methylation-based classifier, as previously described by Yamaguchi et al. Kaplan-Meier method was used to calculate overall survival. Results At 4-year follow up, combining clinical presentation, imaging, morphology, immunophenotyping, somatic mutations and structural variants, 73/119 (61%) of patients could be assigned a diagnostic category with medium to high confidence in most cases. Categories assigned included marginal zone lymphoma (MZL), atypical CLL (aCLL), MCL, lymphoplasmacytic lymphoma, follicular lymphoma, HCL and non-CLL MBL. Majority of cases were categorised as MZL 32/73 (44%) and aCLL 16/73 (22%). The remaining 46/119 (39%) cases remained difficult to categorise even with peripheral blood NGS. Of the MZL cases, 91% were categorised as splenic MZL with 2 nodal MZL and 1 MZL not otherwise specified. 17/32 (53%) of them presented as incidental lymphocytosis and 4/32 (13%) had B symptoms. 22/32 (69%) had documented clinical or radiological evidence of splenomegaly. CD5, ROR1 and CD43 were negative in most cases and CD200 was positive in the majority. NGS detected somatic mutations in 88% of cases. KLF2, TP53 and NOTCH2 were the most common variants. Most common CNVs included 7q-, which was detected in 13/29 (45%) of SMZL cases. Of these 13 cases, 10/13 (77%) had documented splenomegaly. 3 of these cases with 7q- did not have splenomegaly at baseline but developed it during follow up suggesting 7q- could be a predictive marker for development of SMZL. Co-occurrence of trisomy 3 and 18 was found in 3/32 (9%) of cases. t(7;14) CDK6::TRAF associated with SMZL was detected in 1 case. 24/32 (75%) of cases had methylation signatures consistent with SMZL. In the aCLL cohort, 9/16 (56%) presented as incidental lymphocytosis and 3/16 (19%) had B symptoms. 8/16 (50%) and 2/16 (13%) had lymphadenopathy and splenomegaly respectively. Most cases were positive for CD5, ROR1, CD200 and CD81. CD20 expression was moderate to strong in 15/16 (94%) of cases. NGS detected somatic mutations in 71% of cases, with the most common being BIRC3 variant. Most common CNVs include gain of 12 and 13q- in 10/16 (63%) and 4/16 (25%) of cases respectively. t(14;19) IGH::BCL3 involving upstream cluster for BCL3 breakpoint was detected in 2 cases. 13/16 (81%) showed methylation signatures within the HP-CLL or LP-CLL epigenetic clusters. Amongst the whole cohort, 19 cases were lost to follow up and excluded from outcome analysis. 46/100 (46%) of patients needed treatment intervention, including 13/27 (48%) and 10/14(71%) of patients with MZL and aCLL. High grade transformation occurred in 4 patients. 17 patients died with 3 disease-related deaths. 4-year overall survival was 85%, 83% and 92% for the whole cohort, MZL and aCLL respectively. Conclusions Multiparametric integration including NGS data on small variants, copy number changes and structural variants, improves diagnostic categorisation of LI B-NHL on peripheral blood when classic CLL, MCL and HCL have been excluded by immunomorphology and standard FISH testing. This can potentially reduce the need for invasive diagnostic procedures and time to diagnosis while improving access to targeted treatment and clinical trials. A DNA methylation-based classifier is a promising tool for use in these challenging cases.