FLOCK cluster analysis of plasma cell flow cytometry data predicts bone marrow involvement by plasma cell neoplasia.

TitleFLOCK cluster analysis of plasma cell flow cytometry data predicts bone marrow involvement by plasma cell neoplasia.
Publication TypeJournal Article
Year of Publication2016
AuthorsDorfman DM, LaPlante CD, Li B
JournalLeuk Res
Volume48
Pagination40-5
Date Published2016 09
ISSN1873-5835
KeywordsBone Marrow, Cluster Analysis, Flow Cytometry, Histological Techniques, Humans, Immunophenotyping, Multiple Myeloma, Plasma Cells, Plasmacytoma, Predictive Value of Tests, Sensitivity and Specificity
Abstract

We analyzed plasma cell populations in bone marrow samples from 353 patients with possible bone marrow involvement by a plasma cell neoplasm, using FLOCK (FLOw Clustering without K), an unbiased, automated, computational approach to identify cell subsets in multidimensional flow cytometry data. FLOCK identified discrete plasma cell populations in the majority of bone marrow specimens found by standard histologic and immunophenotypic criteria to be involved by a plasma cell neoplasm (202/208 cases; 97%), including 34 cases that were negative by standard flow cytometric analysis that included clonality assessment. FLOCK identified discrete plasma cell populations in only a minority of cases negative for involvement by a plasma cell neoplasm by standard histologic and immunophenotypic criteria (38/145 cases; 26%). Interestingly, 55% of the cases negative by standard analysis, but containing a FLOCK-identified discrete plasma cell population, were positive for monoclonal gammopathy by serum protein electrophoresis and immunofixation. FLOCK-identified and quantitated plasma cell populations accounted for 3.05% of total cells on average in cases positive for involvement by a plasma cell neoplasm by standard histologic and immunophenotypic criteria, and 0.27% of total cells on average in cases negative for involvement by a plasma cell neoplasm by standard histologic and immunophenotypic criteria (p<0.0001; area under the curve by ROC analysis=0.96). The presence of a FLOCK-identified discrete plasma cell population was predictive of the presence of plasma cell neoplasia with a sensitivity of 97%, compared with only 81% for standard flow cytometric analysis, and had specificity of 74%, PPV of 84% and NPV of 95%. FLOCK analysis, which has been shown to provide useful diagnostic information for evaluating patients with suspected systemic mastocytosis, is able to identify neoplastic plasma cell populations analyzed by flow cytometry, and may be helpful in the diagnostic evaluation of bone marrow samples for involvement by plasma cell neoplasia.

DOI10.1016/j.leukres.2016.07.003
Alternate JournalLeuk. Res.
PubMed ID27479652