Evaluation of high dimensional reduction and clustering in the phenotypic discrimination of CD5+ B-cell chronic lymphoproliferative diseases

BAEZ, NS; ARROYO, DS; MANZONE-RODRÍGUEZ, C; WANG, JM; RODRIGUEZ, CM; IRIBARREN, P

Buenos Aires

Congreso; Reunión Anual SAI; 2020

Introduction: Mature (peripheral) B-cell malignancies represent themalignant counterpart of normal mature B-cells that have differentiatedinto naive B cells or their progeny. Integration of a complexset of immunophenotypic, morphological, clinical, and cytogeneticinformation is essential for the subclassification of B-cell chroniclymphoproliferative diseases (B-CLPD).Phenotyping is essential for the diagnostic classification of manyB-CLPD cases, but the current immunophenotyping strategies alsoface several difficulties. In view of these issues, new methods tofacilitate the identification of abnormal B-cell populations in routineclinical flow cytometric data would be desirable.Methods: We used both 12 and 8 colour staining panels and weapplied high dimensional reduction (viSNE) and clustering tools todiscriminate between CD5+ B-CLPD cases. Samples from alreadydiagnosed CLL and MCL were analyzed (n=6) and healthy patients?cells were used as controls.Results: High dimensional reduction (viSNE) revealed at least 6individual clusters that corresponded to each CD5+ B-CLPD sample.In addition, the two dimensions spatial distribution of the populationsshowed segregation by disease type (n=6, p<0.05). Conversely,healthy control samples were separated in two clusters, butall the samples showed overlapping (p=NS). Clustering suggestedthe heterogeneity in markers expression in each disease sample.For instance, we detected clusters with higher expression of Ig-kand CD20, that corresponded to MCL samples, p<0.05 and p<0.01,respectively. Deeper analysis of these samples is currently underinvestigation.Conclusions: These preliminary results suggest that combinationof high dimensional reduction and clustering might be an additionaltool that can be used, at least, to distinguish between CD5+ B-CLPD.Further research is required to confirm these results and to evaluatethe power of these tools in the classification of atypical forms of theB-CLPD.