GENE EXPRESSION PROFILING USING a RTMLPA ASSAY ALLOWS FOR AN ACCURATE CLASSIFICATION OF PERIPHERAL T-CELL LYMPHOMA AND HIGHLIGHTS NOVEL SUBGROUPS WITHIN PTCLS-NOS

International audienceMore than 20 Peripheral T‐cell lymphoma (PTCL) entities are recognized in the WHO classification. Their prognosis is usually very poor and their diagnosis is often challenging for pathologists. Up to 30% of cases thus remain not classifiable (PTCL Not Otherwise Specified, NOS) and there is an important need for alternative diagnostic strategies. Here, we developed a parsimonious GEP assay applicable to a routine diagnostic workflow to differentiate the main PTCL entities and characterize the heterogeneity of PTCL‐NOS.A Reverse Transcriptase‐Multiplex Ligation dependant Probe Amplification (RT‐MLPA) assay was designed to evaluate the expression of 20 markers. It simultaneously addresses the expression of 18 genes routinely tested by immunohistochemistry (IHC) or selected from GEP studies. It also assesses the EBV infection status (EBER1) and the presence of RHOAG17 V and IDH2R172K/T mutations.Unsupervised hierarchical clustering of RT‐MLPA data from 102 control cases validated the capacity of our assay to identify the main PTCL entities. All Angioimmunoblastic T‐cell lymphomas (AITL; n = 29), Anaplastic large T‐cell lymphomas (ALCL; n = 23) ALK+, NK/T‐cell lymphomas (NKTCL; n = 16), Hepatosplenic T‐cell lymphomas (HSTL; n = 6) and Adult T‐cell Leukemia/Lymphomas (ATLL; n = 12) were correctly identified. AITLs classified according to the expression of Tfh markers (CXCL13, CXCR5, ICOS, BCL6) and RHOA mutations (n = 18); NKTCLs according to EBER1, GZMB and Th1 markers (TBX21, IFNγ); HSTLs to CD56, GATA3, TBX21 and BCL6; ALCL ALK+ according to CD30, ALK and cytotoxic markers (PRF, GZMB); ATLLs to ICOSand Th2 markers (GATA3, CCR4). Interestingly, ALCL ALK‐ cases (n = 16) divided into two CD30+ subgroups: one associated with expressions of cytotoxic markers which clustered with ALCL ALK+ cases (n = 10), and a second which did not expressed PRF and GZMB but the two GATA3 and CCR4 Th2 markers (n = 6). We next developed a support vector machine based predictor combined with a centroid categorization. Applied to a series of 125 PTCL‐NOS, this algorithm reclassified 36 Tfh (AITL‐like), 6 CD30/Th2, 6 ALCL ALK‐ like, 3 HSTL‐like and 5 NKTCL‐like PTCLs. After exclusion of these cases, unsupervised clustering analysis identified 17 cytotoxic/Th1 (GZMB, PRF,TBX21, IFNγ) cases, 14 Th2 (GATA3, CCR4) cases and 14 TH2/Tfh (GATA3, CCR4, CXCR5, ICOS) cases. Finally, 24 cases (10.5% of the cohort) did not show any recognizable signature.This study demonstrates the applicability of a robust RT‐MLPA classifier for the classification of PTCLs. Its simplicity and its applicability on FFPE samples makes it an attractive alternative to high throughout GEP approaches. In combination with conventional pathological evaluation and IHC, it may participate to improve the classification of PTCLs and the management of these aggressive tumors.Keywords: gene expression profile (GEP); peripheral T‐cell lymphomas (PTCL)