Automated Quantitative Assessment of Bone Marrow in Normal Samples: Correlation with the Reference Method.

Abstract Bone marrow (BM) analysis is conventionally performed by microscope examination on films of about 0.3ml of aspirated bone marrow fluid, stained with Romanowsky dyes. Until 1996 the simple automated screening of marrow composition was made very difficult by a number of factors, mainly the lack of the erythroblasts quantitation and the fat interference. From 1996 last generation automated hematology analyzers provide accurate and precise erythroblasts counts; moreover same systems have improved their software reducing the problem of fat interference. We have analyzed data from 100 normal BM samples from patients submitted for diagnostic and/or follow up purposes in our Hematology Day Hospital. BM fluid was harvested from the superior posterior iliac crest. The first 0,3–0.5 ml were used for smears, while the next 1–2 ml of BM, collected into K3-EDTA, were analysed with Coulter LH 750, a fully automated hematology analyzer which provides Complete Blood Count, White Blood Count Differential included Nucleated Red Blood Cells (NRBC) and Reticulocytes count. We used the microscope examination conventionally performed on films stained with Romanowsky dyes as reference method. Quantitative detection BM cellularity was obtained by semi quantitative evaluation based on the evaluation of hematopoietic cells in several marrow particles: physiological differences age related were also taken in account. If hematopoietic cells occupy less than 25% or more than 85% the sample is defined respectively hypocellular or hypercellular (none of our sample was as). Differential cell count was usually performed on two different slides counting 500 cells (1000 when hypercellular, but none of our sample was as). We found a strict correlation between microscope semi-quantitative cellularity evaluation and the instrumental cell count as sum of WBC plus NRBC, the Total Nucleted Cell Count (TNCC). The mean value of the TNCC in normal PM samples was 29,48 x109/L with a range 25,9–54,9 x109/L. These results are in good agreement with normal BM cell count reported in the literature using a cytofluorimetric method, which is 34,5 x109/L (SD28.0). The instrumental mean percentage of BM granulocytes corrected for TNNC was 62% (range: 23,5–93,7) versus a mean microscope percentage of 58,42% (range: 40–72). The automated NRBC BM count corrected for TNCC was 11,38% (range: 2,7 – 39,17) versus a microscopic mean value of 28% (range: 9–45). These results, including the slight NRBC underestimation probably due to partial mature cell lysis, are in line with the data of the literature. This study confirms the feasibility of routine automated cell count using a hematology in normal BM fluid samples. Automated methods will support morphologists quickly providing accurate and precise quantitative information such as TNCC and myeloid/erythroid ratio.