A single-cell co-expression score of MYC and BCL2 identifies high-risk BCR-addicted DLBCL benefiting from polatuzumab vedotin in the POLARIX trial

Abstract Introduction: Double Expressor Lymphoma (DEL) refers to diffuse large B-cell lymphoma (DLBCL) with overexpression of the oncogenes MYC and BCL2 and carries a poor prognosis. However, the clinical utility of DEL is limited by variability in definitions for thresholds of positivity of these markers and a potential protective role of BCL6. We previously showed using multiplex imaging that the DEL phenomenon is explained at single cell resolution by a sub-population of cells co-expressing MYC and BCL2 in the absence of BCL6 (M+2+6−) (Hoppe et al., Cancer Discov 2023). Importantly, the precise extent of these M+2+6− cells for a given case can be derived from single marker information from routine IHC or bulk gene expression data through a simple probabilistic formula +2+6−% = ((%)/100 × (2%)/100 × ((100 − 6%))/100) × 100% using single marker values of MYC, BCL2, and BCL6, which we describe as the scDEL (single-cell DEL) score. In single-cell RNA sequencing data (Ye et al. Cell Rep 2022; n = 17), M+2+6− cells were strongly enriched for B-cell receptor (BCR) signalling, suggesting this subpopulation of cells driving relapse may depend on CD79B for survival, therefore use of polatuzumab vedotin in first line treatment may overcome the negative prognostic impact of DLBCL with a high scDEL score. Methods We retrospectively analysed bulk RNA expression data from biomarker evaluable patients in the Phase III POLARIX trial for first-line treatment in DLBCL (NCT03274492; Tilly et al., N Engl J Med 2022; n = 668; POLA-R-CHP arm = 330; R-CHOP arm = 338) to derive a per-patient scDEL transcriptomic score, previously validated (Hoppe et al, Cancer Discov 2023) in the GOYA trial dataset (NCT01287741; Vitolo et al., J Clin Oncol 2017). Multivariable cox proportional hazards models in the POLARIX population with 3 year follow up outcomes were adjusted for cell-of-origin (COO) subtype and International Prognostic Index (IPI). Analyses were not adjusted for multiple testing. Results In the R-CHOP arm of POLARIX, higher scDEL scores were significantly associated with inferior progression-free survival (PFS) and overall survival (OS). When modelled as a continuous variable, each 1% increase in M+2+6− cells was associated with worse PFS (HR 1.04, p<0.0001) and OS (HR 1.05, p<0.0001). When dichotomised at a pre-specified 15% threshold based on prior studies, patients with high scDEL scores (M+2+6− ≥15%; 94/338 patients) had significantly poorer PFS (HR 2.10; 95% CI, 1.44–3.06) and OS (HR 2.63; 95% CI, 1.54–4.49). This adverse prognostic association remained significant in multivariable analyses adjusting for key clinical covariates, including COO and IPI, among patients receiving R-CHOP. In contrast, in the POLA-R-CHP arm, scDEL scores were not associated with outcome, either as a continuous variable (PFS: HR = 0.99, p = 0.39; OS: HR = 0.99, p = 0.51) or when dichotomised M+2+6− ≥15%; 107/330 patients (PFS: HR 0.76, p = 0.25; OS: HR 0.87, p = 0.65). Treatment-by–M+2+6− cell interactions were significant for both PFS and OS, even when adjusting for common prognostic factors (p<0.001 and 0.01 respectively), further suggesting an association between these clinical endpoints and the scDEL score in R-CHOP patients but not Pola-R-CHP patients. Conclusion These findings validate the scDEL score as a clinically relevant biomarker associated with poor outcomes in DLBCL treated with R-CHOP. This study represents its first evaluation in the setting of frontline polatuzumab vedotin use in DLBCL and suggests that the risk conferred by a high scDEL score may be mitigated by the use of frontline polatuzumab vedotin, potentially through targeting of CD79B and BCR-addicted MYC-BCL2 co-expressing tumour cells. As the scDEL score can also be derived from routine clinical-grade IHC percentages for MYC, BCL2 and BCL6 (Hoppe at al, Cancer Discov 2023), these data suggest a potential biomarker with immediate clinical utility. Prospective validation is warranted, particularly using IHC derived scDEL scores, to guide future biomarker-enriched therapeutic strategies for front-line treatment of DLBCL.