Surprising discriminating power of intracellular markers detected by flow cytometry in lymphoma diagnosis unveiled by artificial intelligence techniques

Introduction: In a previous study, we demonstrated that Artificial Intelligence (AI), applied to a wide case series of mature B-cells Leukemia and Lymphomas (B-NHL), allows us to define homogeneous groups of neoplasms characterized by the expression of one or more surface markers. Using additional intracellular markers on a series of tissue samples from B-NHL patients, and applying a more articulated system of analysis, we investigated whether we could obtain an AI based classification and estimate which markers are more specific to differentiate Lymphoma groups. Methods: We collected phenotypic analysis of 615 biopsy samples, whose diagnoses, all confirmed by histological analysis, were grouped in 8 major categories of B-NHL. Leveraging the Predictive Power Score, we evaluated the predictive performance of each individual marker against all lymphoma categories. Among all, we identified 10 markers strongly correlated with the diagnosis. We further validated the role of these markers by combining them in a classification tree. Here, each marker is analyzed in combination with all the others leading to a structural relationship tree that separates the entire database in quasi-homogeneous groups of lymphomas. Finally, using the Uniform Manifold Approximation and Projection (UMAP) dimensionality reduction technique, we observed that the 8 lymphoma categories were substantially grouped and separated in clusters. Results: The results obtained demonstrate how the use of surface and intracellular markers allows us to define the major categories of B-NHL with a high degree of accuracy. Ten or less markers seem to be sufficient to achieve an adequate classification capability. Nevertheless, a greater number of markers, combining intracellular with unconventional markers (CD305, CD81), increases the ability of UMAP to separate different entities. Conclusions: It is conceivable that the implementation of AI applied to multiparametric flow cytometry (MFC) could contribute significantly to an optimal diagnostic process in B-NHL, where histopathological examination remains the gold standard. It is still being investigated whether the use of these methods with a large number of markers can also be predictive of categories of neoplasms carrying molecular or genetic alterations, which would be useful for a better classification even for therapeutic purposes of B-NHL.