Complexity and Challenges of Translating the Dark Zone Signature Into Immunohistochemistry in Diffuse Large B-Cell...

Diffuse large B-cell lymphoma/high-grade B-cell lymphoma with MYC and BCL2 rearrangements (DLBCL/HGBCL-MYC/BCL2) is a distinct disease entity with worse prognosis and is usually diagnosed with identification of MYC and BCL2 rearrangements by fluorescence in situ hybridization analysis. Recent progress in gene expression analysis has identified a dark zone signature (DZsig), which characterizes a gene expression profile (GEP) for DLBCL/HGBCL-MYC/BCL2. Notably, DZsig includes both MYC/BCL2-double-hit and non-double-hit cases with DLBCL and high-grade and Burkitt morphologies; importantly, the latter group is present in numbers that are comparable with or greater than the former group. Although GEP analysis can identify cases exhibiting HGBCL-like biology regardless of double-hit status, GEP analysis will not be globally applied to DLBCL cases because of cost and equipment constraints. Therefore, simpler surrogate approaches for detecting DZsig, such as immunohistochemistry (IHC), are desired. Here, we attempted to develop an IHC approach that can serve as an alternative to GEP or fluorescence in situ hybridization assays using 287 tumors with DLBCL morphology, regardless of double-hit status. Our strategy for detecting DZsig by IHC (IHC-DZ) is based on a 2-step algorithm, which involves applying the Hans classifier and antibodies to MYC, ALOX5, and LMO2. The DZ-IHC algorithm had a sensitivity, specificity, positive predictive value, and negative predictive value of 76.2%, 95.5%, 59.3%, and 98.1%, respectively. The 5-year overall survival rates were 55% for IHC-DZ⁺ germinal center B-cell (GCB)-type tumors, 82% for IHC-DZ^- GCB-type tumors, and 69% for non-GCB-type tumors (P < .01), which is similar to that reported using the GEP assay. To date, this represents the largest cohort evaluated for DZsig selection by IHC; nevertheless, the algorithm is not yet adequate for routine application owing to its suboptimal positive predictive value, and further improvements are needed for clinical implementation.