Testing for t(3;8) in MYC/BCL6-rearranged large B-cell lymphoma identifies a high-risk subgroup with inferior survival Free

TO THE EDITOR:

There are conflicting reports on the prognostic significance of MYC and BCL6 rearrangements in large B-cell lymphoma (LBCL).^1-5 In the World Health Organization Classification, fifth edition, these cases are a subtype of diffuse LBCL, not otherwise specified, whereas in the International Consensus Classification, there is a discrete provisional MYC/BCL6 entity.^6,7,MYC and BCL6 rearrangements are usually detected by separate fluorescence in situ hybridization (FISH) break-apart probes. Reciprocal fusions between BCL6 on 3q27 and MYC on 8q24 have been reported in MYC/BCL6 lymphomas and are dubbed pseudo-double hit.^8-10 The effect of these rearrangements is variable, and some can drive high MYC expression dependent on the BCL6 superenhancer, but their prognostic significance is unknown.^11,12 Our regional histopathology service has used a FISH fusion probe to detect t(3;8)(q27;q24) since 2017. We undertook a service evaluation to determine its clinical utility.

FISH was performed on interphase nuclei in fixed tissue sections. We used the Zytolight MYC Break Apart Probe (Zytovision), and if the signal pattern was abnormal, we undertook further testing with Vysis BCL2 and BCL6 break-apart probes (Abbott; supplemental Figure 1A, available on the Blood website). Since 2017, we tested cases with MYC-r or 3′ distal MYC loss, along with concurrent BCL6-r or 3′ proximal BCL6 loss, with a Cytocell MYC/BCL6 fusion probe (Oxford Gene Technology IP Limited; supplemental Figure 1B). All cases of lymphoma tested from 1 March 2015 to 1 June 2023 were identified prospectively, and data on baseline variables, treatment, and survival were collected retrospectively for eligible cases. The inclusion criteria were: age ≥18 years; new diagnosis of large B-cell non-Hodgkin lymphoma, excluding Burkitt and mantle cell, but including high-grade transformation of previously low-grade disease; and successful FISH testing for MYC, BCL2, and BCL6 disruption, demonstrating MYC-rearrangement (MYC-r) and either wild-type (WT) BCL2 and BCL6 or BCL6-rearrangement (BCL6-r) with or without BCL2-rearrangement (BCL2-r). MYC-r BCL2-r BCL6-WT cases were excluded. Overall survival (OS) and progression-free survival (PFS) were compared using Cox proportional hazards. Detailed methods are in the supplemental Data.

Between 2015 and 2023, a total of 2298 cases of lymphoma were tested for MYC-r by FISH, of which 203 met the inclusion criteria, and we collected data for 96 MYC-r BCL6-WT cases (hereafter, “BCL6-WT”), 42 MYC-r BCL2-r BCL6-r cases (hereafter, “BCL2-r BCL6-r”), and 31 MYC-r BCL2-WT BCL6-r (hereafter, “BCL2-WT BCL6-r”; Figure 1A). We did not have diagnosis details for all FISH-tested cases, so we extrapolated from our screening data to estimate that 1997 of the 2298 cases were newly diagnosed LBCL (Figure 1B). With this denominator, BCL6-r cases were 3.9% of LBCL. Results of t(3;8) fusion testing were available for 46 BCL6-r cases, and 22 (48%) were t(3;8)⁺ pseudo-double hit. The presence of t(3;8) was more common in BCL2-r BCL6-r cases (60% vs 33% in BCL2-WT BCL6-r). Most baseline variables were similar between BCL6-WT and BCL6-r, except sex: BCL2-r BCL6-r and BCL2-WT BCL6-r had a higher female proportion (65% and 57% vs 32% in BCL6-WT; supplemental Table 1). At a median follow-up of 26.8 months, OS of BCL2-r BCL6-r and BCL2-WT BCL6-r cases was very similar, and both groups had a hazard ratio (HR) of 1.7 relative to BCL6-WT cases (P = .049 for BCL2-r BCL6-r; P = .06 for BCL2-WT BCL6-r; Figure 1C). Similar trends were seen for PFS (HR, 1.7; P = .003 for BCL2-r BCL6-r; HR, 1.5; P = .14 for BCL2-WT BCL6-r; supplemental Figure 1A).

We compared BCL6-r cases tested for t(3;8) (n = 45) with contemporaneous BCL6-WT cases from 2017 onward (n = 83). Survival was shorter in t(3;8)⁻ cases, in both BCL2-r BCL6-r and BCL2-WT BCL6-r groups (HR, 2.9; P = .017; and HR, 2.4; P = .015, respectively, relative to BCL6-WT; Figure 1D-E). Although BCL2-r identifies the molecular EZB cluster of LBCL,¹³ our results suggest that t(3;8) status has prognostic significance independently of BCL2 genotype. Therefore, we combined the BCL2-r and BCL2-WT cases in univariate and multivariate analyses. The HR for t(3;8)⁻ cases relative to BCL6-WT in univariate analysis was 2.57 (P = .002), and in the multivariable analysis, it was higher (2.66; P = .002), whereas BCL2 status was not significantly associated with survival (P = .36; Figure 1F; Table 1). For PFS, the effects are similar but smaller, with a HR of 2.08 for progression or death associated with t(3;8)⁻ cases (P = .02; supplemental Figure 1B; supplemental Table 2). Comparing baseline variables with BCL6-WT, t(3;8)⁻ cases had the highest lactate dehydrogenase ratio, whereas t(3;8)⁺ cases had the highest proportions of female sex and bulky disease but the lowest Ki67 staining. The number of early events was high in t(3;8)^– cases, which had a lower frequency of curative-intent treatment (78% vs 85% and 86% in BCL6-WT and t(3;8)⁺ respectively). Therefore, we additionally analyzed only cases treated with curative intent, but we still observed inferior OS (HR, 3.1; P = .002) and PFS (HR, 2.4; P = .016) exclusively in the t(3;8)^– group (supplemental Figure 1C).

For high-risk LBCL, some centers use DA EPOCH-R (dose-adjusted etoposide, prednisolone, vincristine, cyclosphosphamide, doxorubicin, and rituximab) or R-CODOX-M/R-IVAC (rituximab, cyclophosphamide, vincristine, doxorubicin, methotrexate, ifosfamide, etoposide, and cytarabine), despite their increased toxicity, cost, and little evidence of superiority.^14-16 We found no survival advantage to R-CODOX-M/R-IVAC over R-CHOP (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisolone) in our cohort of MYC-r LBCL (HR, 1.08; P = .9, in a multivariable model; supplemental Figure 1C-D). However, this service evaluation was not designed to compare regimen efficacy.

Other studies of MYC/BCL6 rearranged LBCL are limited by selection bias. For example, the cohort of Khurana et al⁵ from large referral centers had a low median age of 63 years and 83% PS 0 to 1. Our cohort is representative of all patients with MYC-r LBCL. The median age of our cohort is 71 years, which is comparable with 70 years in the population-based Haematological Malignancies Research Network,¹⁷ and only 69% of our cohort had PS 0 to 1. Our finding that 3.9% of all LBCL carry MYC and BCL6 rearrangements, and only 1.6% without a coincident BCL2 rearrangement, is, to our knowledge, the first using a large denominator representative of real-world LBCL. We found that nearly half of MYC/BCL6 rearranged LBCL were t(3;8)⁺ and are therefore pseudo-double hit. The proportion was higher in MYC-r BCL2-r BCL6-r cases than MYC-r BCL2-WT BCL6-r (60% vs 33%), and these proportions are very similar to a recent report in which 54 cases were tested for t(3;8).¹⁰ Most significantly, we found that although pseudo–double-hit LBCL has a similar prognosis to MYC-r BCL6-WT LBCL, the true double hit t(3;8)⁻ cases have a significantly worse prognosis, independently of age, PS, and stage. This difference is seen in both BCL2-r and BCL2-WT groups, although there is a trend toward inferior survival in the BCL2-r group overall, which may be due to more cases of transformed low-grade lymphoma that are not treatment-naïve. The mechanism behind the inferior survival of t(3;8)⁻ cases may relate to the frequency in this group of immunoglobulin-MYC rearrangements, which confer adverse prognosis.^1,2 In LBCL, half of MYC rearrangements are with immunoglobulin loci,^2,9 and because in our BCL6-r cohort half of the MYC rearrangements are with BCL6, many of the remainder may be with immunoglobulin loci. Zhang et al¹⁰ found in 3 of 5 t(3;8)⁻ cases a IGH-MYC rearrangement.¹⁰ However, because comprehensive testing for IGH/IGK/IGL-MYC rearrangements requires at least 3 separate fusion probes, testing for t(3;8) would be a more parsimonious method to identify this high-risk group if t(3;8) negativity correlates with immunoglobulin-MYC status. Our findings support the inclusion of t(3;8)⁻MYC/BCL6 LBCL as an entity in future lymphoma classifications and provide a rationale to recruit this patient group to clinical trials addressing high-risk disease.

This service evaluation was approved by the University Hospitals Birmingham NHS Foundation Trust.