Splitting from a lump: harmonizing ALCL classification

In this issue of Blood, Feldman et al 1 provide a framework to harmonize multiple anaplastic large cell lymphoma (ALCL) clinicopathologic entities into 2 major subsets dubbed type I and type II ALCL based on gene signatures derived from RNA sequencing analysis of 393 cases.The type I gene expression signature is enriched for STAT3 transcriptional targets, and remarkably, type I and type II ALCL can be distinguished based on immunohistochemical detection of phosphorylated STAT3 (pSTAT3) of 30% of malignant cells in biopsy samples.In contrast, type II ALCL is marked by overexpression of EZH2 and is associated with significantly increased immunohistochemical detection of EZH2 or its biochemical product, trimethylated histone 3 lysine 27, compared with type I ALCL.Because 2 of the clinicopathological subtypes subsumed by type I ALCL, ALK+ ALCL and breast implant-associated ALCL, have favorable clinical outcomes, 1,2 this might suggest that this classification scheme can distinguish clinical outcomes.However, as previously reported, DUSP22-rearranged ALCL, which makes up a large proportion of type II ALCL in this study, shows an absence of JAK-STAT pathway activation and generally shows favorable outcomes. 3 Although other retrospective cohorts of DUSP22 ALCL have reported outcomes less favorable than in this article, 4 these are generally more favorable than those of TP63-rearranged ALCL, which is classified as type II ALCL in this study and is associated with unfavorable outcomes.Ultimately, the authors propose an overarching classification scheme, first of distinguishing cutaneous or breast implant-associated ALCL from systemic ALCL, which is reasonable given clinical management of these subtypes is distinct from one another.Systemic ALCL is then further subclassified to ALK+ ALCL based on positive detection of ALK by immunohistochemistry, then ALK-ALCL is divided based on fluorescence in situ hybridization detection of TP63 and/or DUSP22 rearrangements.This leaves socalled triple-negative (TN) ALCL that this classification scheme breaks into type I TN (TN-1) or type II TN (TN-II) based on immunohistochemical detection of pSTAT3 or not.Thus, after impressive lumping of ALCL subtypes into 2 classes based on gene expression signatures, the authors retain the need for splitting ALCL based on clinical presentation and genetic subtypes based on structural variants.