Abstract
Introduction
It was proposed that peripheral blood (PB) monocyte subset analysis evaluated by flow cytometry, hereafter referred to as "monocyte assay", could rapidly and efficiently distinguish chronic myelomonocytic leukemia (CMML) from other causes of monocytosis by highlighting an increase in the classical monocyte (cMo) fraction above 94%. However, the robustness of this assay required a large multicenter validation.
Methods
PB and/or bone marrow (BM) samples from adult patients displaying monocytosis were assessed with the "monocyte assay" by ten ELN iMDS Flow working group centers (6 equipped with BD FACSCanto™ II (BD Biosciences), 3 with Navios™ (Beckman Coulter) and one with BD™ LSRII (BD Biosciences)) with harmonized protocols. The corresponding files were reanalyzed in a blind fashion by a skilled operator and the cMo (CD14 ++CD16 -) percentages obtained by both analyses were compared. Information regarding age, gender, complete blood count, marrow cytomorphology, cytogenetics and molecular analysis was collected. Confirmed diagnoses were collected when available as well as follow-up for CMML patients.
Results
The comparison between cMo percentages from 267 PB files provided by the 10 centers and the centralized cMo percentages showed a good global significant correlation (r=0.88; p<0.0001; FigA) with no bias (FigB). Confirmed diagnoses were available for 212 files, namely 101 CMML according to the WHO criteria, 99 reactive monocytosis, and 12 MPN with monocytosis. A phenotype in favor of CMML, either classical with accumulation of cMo ≥94% or a bulbous aspect (FigC), was observed respectively in 81 and 14 patients. Hence, a total of 95 out of the 101 CMML patients translated into a sensitivity of 94% (FigD). Assessment of C reactive protein counts were available in seven of the 14 patients with the characteristic bulbous profile and correlated with an inflammatory state, showing a median of 93.0 [7.0-157.4] mg/L. Conversely, a phenotype not in favor of CMML (FigC) was observed in 83 of the 99 patients with reactive monocytosis and in 10 of 12 patients with MPN with monocytosis, leading to a 84% specificity (FigD). We established a Receiver Operator Curve (ROC) and again obtained a 94% cut-off value of cMo with an area under the ROC curve (AUC) of 0.865 (FigE).
The second aim of this multicenter study was to assess the feasibility of the monocyte assay on 117 BM samples provided by 7 out of the 10 ELN centers, 43 of which being paired to PB samples. The comparison between cMo percentages provided by the 7 centers and the centralized cMo percentages showed a lower global significant correlation compared to PB samples (r=0.74; p<0.0001; FigF) with a slight underestimation of cMo percentage by the participating centers (FigG). The comparison between PB and BM samples cMo% obtained by centralized reanalysis showed an excellent global correlation (r=0.93; p<0.0001; FigH) with a higher percentage in the marrow (FigI). Seventy-nine files were associated to a confirmed diagnosis, as expected mostly CMML (n=69), only seven reactive monocytosis and three MPN with monocytosis. Thus, we determined a sensitivity of the "monocyte assay" on BM samples of 87% (a phenotype in favor of CMML being observed in 60 out of the 69 CMML with 6 bulbous aspect profiles) and a specificity of 80% (a phenotype not in favor of CMML being observed in 5 of the 7 patients with reactive monocytosis and in 3 of the 3 patients with MPN with monocytosis).
Conclusions
This ELN multicenter study demonstrates the robustness of the monocyte assay with only limited variability of cMo percentages, validates the 94% cutoff value, confirms its high sensitivity and specificity in PB and finally, also confirms the possibility of its use in BM samples.
Figure 1 Figure 1.
Disclosures
Kern: MLL Munich Leukemia Laboratory: Other: Part ownership.