The early thymic precursor (ETP) immunophenotype identified by flow cytometry has defined a high-risk subset of T Lymphoblastic Leukemia (T-ALL) by relying on the following criteria: CD5 negative/1 log lower than mature T cells, <5% of CD1a and CD8 and > 25% of one or more stem cell or myeloid antigens (CD34, CD117, HLA DR, CD13, CD33) on T-lymphoblasts. Subsequently, a subset of T-ALL cases were recognized to fulfill all but the CD5 criterion for ETP and termed Near ETP, having levels of residual disease and outcomes similar but not identical to ETP. Subsequent sequencing studies of T-ALL have identified an expanded subset of cases termed ETP-like, having primitive characteristics and encompassing all immunophenotypically defined ETP, many Near ETP, and few Non-ETP cases. However, the genomic recognition of ETP-like T-ALL relies on RNAseq, a technique that is not currently in routine clinical use. The current study defines revised immunophenotypic criteria to allow identification of ETP ALL that matches the genomic classification with high sensitivity and high negative predictive value, an approach suitable for rapid, routine screening by flow cytometry at the time of diagnosis.

A subset of samples (N=99) from Children's Oncology Group (COG) trial AALL0434 were identified having both comprehensive genomic sequencing and immunophenotypically defined ETP status. In addition, these samples had been evaluated for expression of 9 novel antigens identified as differentially expressed between T-ALL and normal mature T/NK cells using a high- throughput flow cytometric screening assay containing antibodies against 242 unique antigenic specificities (LyoPlates; Becton-Dickinson). This cohort included 7 ETP, 20 Near ETP and 73 Not ETP cases. The molecular subgroups were ETP-like (22), TAL1 DP-like (21), TAL1 αβ-like (17), TLX3 (14), NKX2-1 (4), MLLT10 (4), TLX1 (3), KMT2A (3), TME-enriched (3), HOXA9 TCR (2) LMO2 γδ-like (2), NUP98 (1), NKX2-5 (1), SPI1 (1), and NUP214 (1). Comparison between immunophenotype and molecular profile showed all 7 ETP cases to be ETP-like (100%), as were 10/20 (50%) Near ETP and 5/72 (7%) Not ETP.

The current immunophenotypic ETP criteria showed poor sensitivity for the identification of ETP-like (31%), high specificity (100%) and high positive predictive value (PPV; 100%). Systematic assessment of components of the current ETP definition revealed that the CD5 criterion was principally responsible for the poor sensitivity and its removal resulted in marked improvement in sensitivity (77%) at the expense of reduced specificity (90%) and PPV (68%). In fact, use of any of the remaining three criteria (CD1a, stem cell, CD8) alone resulted in much higher sensitivity (90-95%) but with reduced specificity (60-65%) and poor PPV (40-44%). The best of these was CD1a having high sensitivity (95%) and negative predictive value (NPV; 98%) with reduced specificity (60%) and poor PPV (40%). Elimination of any single criterion in the absence of CD5 minimally improved sensitivity (81-86%) with reduced specificity (79-86%) and PPV (53-63%). Consequently, while removal of CD5 is necessary, the remaining criteria alone or in combination are insufficient and additional markers are needed to improve specificity and PPV.

Examination of the 9 novel antigens identified by high throughput screening revealed a pattern of decreased CD165 and CD184 expression that in combination segregated most ETP-like cases. This combination alone showed high sensitivity (100%) and NPV (100%) with moderate specificity (75%) and reduced PPV (54%). CD165/CD184 with inclusion of CD1a was the best of any combination showing high sensitivity (96%) and NPV (99%), good specificity (90%) and modest PPV (72%). False positives were NKX2-5 (1), TAL1 αβ-like (1), TLX3 (3), TME-enriched (1), NUP98 (1) and LMO2- γδ-like (1) (can be recognized by its high sCD3 and high CD45). The single false negative was due to marginal CD1a expression of 8.7%.

We conclude that our revised immunophenotypic criteria using decreased expression of CD165 and CD184 in combination with absent CD1a provides an easy, practical and accurate method to screen for primitive T-ALL (genomically defined ETP-like) without the use of sequencing. Our approach has high sensitivity and NPV while retaining high specificity and acceptable PPV setting the stage for evaluation of therapies against primitive T-ALL.

This investigator-initiated trial was supported by Sandoz Inc.

This content is only available as a PDF.
2025
Sign in via your Institution