The DLBCL90 gene‐expression assay identifies double‐hit lymphomas with high sensitivity in patients from two phase II clinical trials with high‐risk diffuse large B‐cell lymphoma

1 Department of Cancer Immunology, Institute for Cancer ResearchOslo University Hospital, Oslo, Norway 2 KG Jebsen Centre for B cell malignancies, University of Oslo, Oslo, Norway 3 Institute of Clinical Medicine, University of Oslo, Oslo, Norway 4 Department of Pathology, Oslo University Hospital, Oslo, Norway 5 Department of Hematology, Aarhus University Hospital, Aarhus, Denmark 6 Department of Oncology, Oslo University Hospital, Oslo, Norway 7 Department of Oncology, Lund University and Skåne University Hospital, Lund, Sweden 8 Helsinki University Hospital Comprehensive Cancer Centre and University of Helsinki, Helsinki, Finland 9 iCANDigital Precision CancerMedicine Flagship, Helsinki, Finland 10 Department of Pathology, Helsinki University Hospital, Helsinki, Finland 11 Centre for Lymphoid Cancer, British Columbia Cancer, Vancouver, Canada

Lymph3Cx primary mediastinal large B-cell lymphoma signature assays [6]. In a population-based cohort of 184 GCB-DLBCL tumors, the DLBCL90 assay classified 23% of the tumors as DHITsig positive (DHITsig-pos), 10% as DHITsig indeterminate (DHITsig-ind), and 66% as DHITsig negative (DHITsig-neg). The DHITsig-pos and DHITsig-ind groups were associated with inferior outcome after R-CHOP therapy, and the combined DHITsig-pos/ind group detected HGBL-DH/TH-BCL2 with a sensitivity of 88% and specificity of 86% [3]. However, the assay has not yet been validated in a prospective clinical trial, and the prognostic value has not been investigated in patients treated with more intensive regimens than R-CHOP.
The first objective of our study was to validate the sensitivity of the DLBCL90 assay, by detection of HGBL-DH/TH-BCL2 in tumors with DLBCL morphology. Although not a perfect measurement of sensitivity, we expect that the broader defined DHITsig-population comprises the majority of HGBL-DH/TH-BCL2. The second objective was to evaluate the prognostic significance of the DHITsig in patients treated with intensified therapy in a clinical trial.
We included patients with high-risk de novo DLBCL from two Nordic phase II clinical trials [7,8] with confirmed DLBCL NOS (WHO classification, 2008) and available FFPE tissue (n = 90). Patients were below 65 years of age and had age-adjusted International Prognostic Index score 2-3 and/or increased risk of CNS recurrence. In both trials, the patients received biweekly R-CHOP with etoposide (R-CHOEP- 14) and systemic CNS prophylaxis with HD-Mtx and HD-Ara-C.
RNA was extracted from representative pre-treatment FFPE tissue and digital gene expression was performed, applying the DLBCL90 assay on the NanoString platform (NanoString Technologies, Seattle, WA). The cutoff for DHITsig positivity was set between the DHITsig-ind and DHITsig-neg group as defined by Ennishi et al [3]. This was based on the optimal cutoff for the DHITsig, defined by the Youden Index, in a receiver operating curve of DHITsig versus HGBL-DH/TH-BCL2 in a large independent cohort of DLBCL [9]. FISH break-apart probes for MYC, BCL2, and BCL6 were used to identify HGBL-DH/TH. Translocation partner for MYC was investigated using FISH fusion probes for MYC and immunoglobulin heavy chain and FISH break-apart probes for kappa and lambda light chains. Double protein expression by immunohistochemistry was scored by expert hematopathologists, and cell-oforigin assignment by immunohistochemistry was determined by Hans algorithm. Details on the patient cohort and experimental procedures are provided in Supporting Information. The data that support the findings in this study are available from the corresponding author upon request.
Successful molecular classification by the DLBCL90 assay was obtained for 89 of 90 cases. Three cases classified as primary mediastinal large B-cell lymphoma were excluded from further analyses. For the remaining 86 cases, the DLBCL90 assay assigned 47 (55%) samples to the GCB-subtype, 26 (30%) to the activated B-cell like subtype, and 13 (15%) to the unclassified group. Patient characteristics were largely representative for the two study cohorts combined (Table S1).
The DHITsig was only detected in the GCB subtype, where 16 (34%) samples were assigned to the DHITsig-pos group (including 11 DHITsig-ind samples) ( Figure 1). The DHITsig-pos group had a higher median age than DHITsig-neg GCBs (61 years vs 54 years, P = .005), otherwise no clinical characteristics differed between the two groups (Table S2). Additionally, two of seven tested were double protein expressers (Figure 1).
The median follow-up of living patients was 65 months. Within the GCB subgroup, there were no significant differences in outcome (OS, PFS) between the DHITsig-pos and DHITsig-neg group. In contrast to previous findings [3], the DHITsig-pos group had excellent outcome, with a 5-year PFS of 81% and 5-year OS of 93% ( Figure 2). Importantly, HGBL-DH/TH determined by FISH also showed similar good outcome (Table S3). This was also observed in the combined total cohort from the two original trials [8].
A possible reason why we did not find a significantly inferior survival for DHITsig-pos and HGBL-DH/TH patients could be the limited sample size. Other reasons may be that the patients were relatively young (median age 55 years) and a selection bias for a more fit double-hit patient population than those presented in population-based cohorts.
Additionally, our cohort included only HGBL-DH/TH with DLBCL morphology, for which current evidence suggests better outcome than HGBL-DH/TH with high-grade morphology [10]. Furthermore, recent studies have shown that the adverse outcome of HGBL-DH/TH is likely to be restricted to cases with an immunoglobulin translocation partner for MYC [10]. In our cohort, only three of the six HGBL-DH/TH-BCL2 had an identified immunoglobulin translocation partner for MYC. Although the numbers are too small to conclude, all three cases achieved complete remission after first-line therapy, and were still in remission at the last follow-up 45, 68, and 84 months after diagnosis. Of importance, the intensive regimen may have contributed to the good outcome. This hypothesis is supported by the favorable results from the total trial cohort [8]. It is also in line with the current understanding that double-hit tumors may benefit from an intensified treatment approach, although not shown in randomized trials [11]. The DLBCL90 assay is a diagnostic tool that has potential for broad clinical application as it can be performed on routinely available FFPE specimens, and requires minimal hands-on time. In contrast to FISH, which detects a limited number of genomic aberrations, the DLBCL90 assay operates on the gene-expression level and the DHITsig may also capture more complex aberrations associated with the double-