Multicentric Validation of the "Monocyte Assay" for Chronic Myelomonocytic Leukemia Diagnosis By Flow Cytometry

Introduction:Accumulation of classical monocytes (MO1:CD14⁺⁺CD16^-) ≥94% of total peripheral blood monocytes analyzed by flow cytometry (FCM) was demonstrated to be a specific and sensitive tool that distinguishes chronic myelomonocytic leukemia (CMML) from reactive monocytosis (Selimoglu-Buet D et al, Blood 2015). The 94% threshold was further validated in an independent study (Talati C et al, Blood 2017) and is increasingly used as a diagnostic marker of CMML. Since this "monocyte assay" has been largely adopted by French diagnosis laboratories, we evaluated its performance in a multicentric evaluation.

Methods: A short survey was sent by personalized emails between July and September 2016 to 48 French laboratories performing routinely FCM diagnosis of myeloid neoplasms. Afterwards, a few selected centers were asked to send their raw data files as well as the corresponding MO1 percentage evaluation in order to perform a centralized blind analysis using Kaluza^® Software (Beckman Coulter^®). Biological and clinical data were also provided when available.

Results: We obtained a comprehensive reply to the survey with a 100% rate of response. Thirty of 48 centers (63%) use routinely the "monocyte assay": apart from the three centers that participated to the initial study and keep using it, 27 centers spontaneously implemented it, most often following the initial publication and/or a meeting presentation. Eighteen out of 27 (67%) centers use at least the published exclusion panel for non-monocyte mature blood cells (i.e. CD2 and CD56 for T lymphocytes and NK cells, CD24 for B lymphocytes and granulocytes). Ten centers use at least one exclusion marker while two use no one. Nine centers perform 1 to 2 tests per month, fifteen 3 to 6, three 6 to 10 and three more than 10 tests. All the surveyed cytometrists consider the "monocyte assay" useful for CMML diagnosis. Among the 18 centers that do not use the "monocyte assay", 8 intend to develop it soon, the 10 remaining centers not excluding a possible implementation.

In a second step, five centers (named A to E) were selected according to their frequency of use and their antibody panel, excluding the three centers which were part of the initial study. They provided 329 FCM raw data files with the corresponding MO1 percentage. All these files were reanalyzed in a blind fashion. This centralized analysis compared to the MO1 percentage determined by the centers showed an excellent correlation (r=0.93; p<0.0001, Figure 1A) without any major bias (Center A: r = 0.93; B: r=0.97; C: r=0.89; D: r=0.93, E: r=0.93; p<0.0001). As the cut-off value of 94% MO1 is the most relevant parameter in the "monocyte assay", we compared the patients with MO1 percentage ≥94% by centralized analysis (i.e. suspected of being diagnosed as CMML) and found an excellent concordance between both analyses (matching for 110 analyses/ 115 (95.7%), Figure 1B). Conversely, of the 214 patients with a fraction of MO1 <94% by the centralized analysis, 201 (93.5%) displayed MO1 percentage <94% in the original center. In most of the 18 divergent cases, the difficulty in drawing MO1 gate was related to an insufficient number of classical monocytes analyzed. Hence, we decided to exclude all the files showing MO1 event number below 10,000 (Figure 1C), thereby improving the correlation (r=0.94) and the concordance for the 245 remaining files (total matching files: 95%). Among these latter files, we collected clinical data for 84 patients for whom an overt diagnosis was made, namely 46 CMML, 24 reactive monocytosis and 14 non-CMML malignancies. MO1 accumulation ≥94% determined by centralized analysis was observed in 43 of the 46 CMML (Figure 1D), indicating a sensitivity of 93.5% in accordance with previous results. A specificity of 86.8% was calculated as 5 of the 38 "non-CMML" patients displayed MO1 <94% determined by centralized analysis. Interestingly, using these data we established a new Receiver Operator Curve (ROC) and obtained yet again a 94% cut-off value.

Conclusion: This study demonstrates a strong adhesion of the French laboratories to the "monocyte assay" for CMML diagnosis set-up in 2015, with 30 user centers so far. We also demonstrate that this FCM test requires the acquisition of a minimum of 10,000 events. Finally, we observed a low variability of the MO1 percentage quantification between centers and validated the published 94% cut-off value with a 93.5% sensitivity and a 86.8% specificity.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal