Data-Informed, Refined Immunophenotypic Classification of Adult T-Lineage Acute Lymphoblastic Leukemia

Introduction T-lineage acute lymphoblastic leukemia (T-ALL) is a subtype comprising around 20 - 25% of all ALL cases, which represents specific clinical challenges. Current classification systems are heterogeneous with varying reproducibility. We developed a data-informed classification system of T-ALL. Methods To assess T-ALL immunophenotypes, we analyzed flow cytometry data from T-ALL patients (pts) treated between 1989-2024 within the framework of the German Multicenter ALL (GMALL) Study Group. From a total of 1696 untreated pts, we selected samples based on availability of a sufficient antibody panel (sCD3, cyCD3, CD1a, CD7, CD2, CD5, CD4, CD8, CD34, TdT, CD10, CD117, HLADR, CD13, CD33). Antigen positivity was defined as ≥20% for cell surface and ≥10% for intracellular antigens. We performed hierarchical clustering and UMAP-analyses to identify phenotypic subgroups. Results were compared to established classification systems (i.e. EGIL, GMALL, WHO). GMALL recognizes thymic (CD1a+), early (CD1a-, either CD2- or CD2+, CD5-, CD4-, CD8-, sCD3-) and mature T-ALL (definitions of thymic and early not met). Identified clusters and categorization along established classification schemes were correlated with available T cell receptor (TR) rearrangement profiles in a subset of pts. Results Data from 1177 pts (841/325/11 male/female/unknown) were available. The median age was 35 years (range 15-85). According to GMALL definitions, 304 pts were classified as early, 532 as thymic, and 341 as mature T-ALL. Comparative reclassification according to EGIL and WHO revealed inconsistencies, particularly with respect to mature, early and pre/pro T-ALL, with 61% (n = 208) of mature T-ALL according to GMALL categorized as pre-T according to EGIL. ETP-ALL (n=167) comprised mature (n=32, 20%) and early (n=135; 80%) according to GMALL and mature (n=5; 3%), pre-T (n=130; 78%) and pro-T (n=32; 19%) according to EGIL. Hierarchical clustering based on antigen expression profiles revealed 5 immunophenotypically distinct groups, which we termed early-T-like (n=336; 28%), ETP-like (n=61; 5%), atypical-thymic (n=119; 10%) typical thymic (n=462; 39%), mature-T-like (n=199; 16%) T-ALL. Within the newly identified subgroups atypical/typical thymic T-ALL, we observed a lower median age compared to other subgroups, whereas sex and date of analysis (before 2000 vs. after 2000) showed no significant associations. TR rearrangement data were available in 121 pts. We found that the data-informed classification system showed significantly lower numbers of TR rearrangements per sample in ETP-like compared to typical-thymic T-ALL (median 2 vs 5, p=0.0003) and 97% of typical-thymic T-ALLs had TR rearrangement profiles occurring later within T-cell developmental trajectories (complete TRB and/or TRD). Likewise, the most mature TR profiles were detected in thymic T-ALL according to GMALL and EGIL (≥97% with complete TRB/TRD). The novel identified subgroup of atypical-thymic T-ALL was characterized by a significantly less frequent expression of more mature T-cell antigens (CD2, CD4, CD8), increased CD10 expression and a trend towards more immature TR profiles compared to typical-thymic T-ALL (81% vs. 97% complete TRB/TRD). This finding is in line with a 24% proportion of early/pre/pro-T-ALL according to GMALL/EGIL within the atypical-thymic group with a median expression of 2% for CD1a, 4% CD2, 3% CD4, 2% CD8 and 85% CD10. To further characterize and capture the heterogeneity within thymic T-ALL, we analyzed expression profiles in a subset of 611 T-ALL pts defined as thymic T-ALL by a ≥10% threshold for CD1a positivity, commonly used to define thymic T-ALL in pediatric trials. We discovered that subgroups of thymic T-ALL could be separated by CD2, CD4, CD8 and CD10 expression. CD1a-low (10-19%) cases had a more immature expression profile with lack of either CD2, CD4 and CD8 compared to CD1a-high (p=0.0026 for CD2, p<0.0001 for CD4 and CD8). Conclusion Our findings identified novel immunophenotypic subtypes within T-ALL, especially among CD1a+ thymic cases. TR rearrangement data aligned with immunophenotypic maturation stages, validating our classification. Remarkably, the most mature rearrangement profiles were present in ‘typical-thymic’ T-ALL. It is essential for comparability of clinical data and categorization of molecular data to develop a standardized classification system for T-ALL.