AJHE2601 - CME/CMLE - Review of flow cytometry findings and associated scoring approaches for identifying myelodysplastic syndrome and the future role of machine learning in improving the diagnostic algorithm-AJHE2601

AJHE2601 - Educational Activity

Pdf Summary

This review summarizes how multiparameter flow cytometry (FC) can support the diagnosis of myelodysplastic syndrome (MDS), particularly in difficult low-grade cases where morphology is subtle and no defining cytogenetic or molecular abnormality is present. Although recurrent FC abnormalities in MDS have been described for decades—aberrant antigen expression, asynchronous maturation patterns, altered cell subset proportions, and changes in light scatter—FC is not yet widely standardized or adopted as a routine diagnostic pillar, largely because findings are variable and often nuanced. The article outlines current WHO/ICC positions that FC cannot independently establish MDS, while highlighting European LeukemiaNet (ELN) guidance and ongoing standardization efforts (preanalytics, processing, and minimum antibody panels).<br /><br />The authors review lineage-specific abnormalities: myeloblast shifts and aberrant markers (eg, altered CD34/CD117 patterns, aberrant CD5/CD7/CD56), granulocytic maturation disturbances (eg, abnormal CD13/CD16 or CD11b/CD16 patterns, hypogranularity reflected by reduced SSC), monocytic abnormalities (eg, reduced SSC and altered CD13/CD14/CD36/CD64 with possible CD56/CD2 expression), and erythroid dysplasia features (notably increased CV of CD36 and CD71, altered CD71 MFI, abnormal CD117 subsets, and emerging emphasis on erythroid SSC). They also note associations between certain FC patterns and genetic lesions (eg, SRSF2-linked maturation patterns; SF3B1-related erythroid findings; CD177 decreases linked to specific mutations).<br /><br />A major focus is comparison of published scoring systems designed to standardize interpretation. Many build on the Ogata score; multiple studies suggest the ELN Integrated Flow Score (iFS), especially when combined with erythroid parameters, performs best overall in comparative cohorts.<br /><br />Finally, the review highlights machine learning as a promising route to integrate subtle multidimensional FC signals, improve accuracy and speed, and uncover new features (eg, erythroid SSC). Approaches using clustering (FlowSOM) plus classifiers and other embedding-based models show high reported performance, but require calibration, sufficient training data, validation, and robust laboratory infrastructure before broad clinical deployment.

Keywords

multiparameter flow cytometry

myelodysplastic syndrome

MDS diagnosis

low-grade MDS

immunophenotyping abnormalities

European LeukemiaNet

ELN Integrated Flow Score

Ogata score

erythroid dysplasia markers

machine learning in flow cytometry