Blast enumeration in the bone marrow

Clinical Center

Investigators

Abstract

The enumeration of blasts in the bone marrow (BM) is a critical element in the diagnosis, prognosis and therapeutic response evaluation in myelodysplastic and myeloproliferative syndrome neoplasms and acute leukemias. However, few studies have examined the accuracy and precision of blast counting in the BM using standard microscopic procedures. We assessed the true number of blasts in BM biopsy samples by manually counting individual labeled and unlabeled mononuclear cells in CD34-stained sections of BM samples containing CD34-positive blasts. This gold standard was then used to test blast count reproducibility and accuracy among four experienced hematopathologists using both, differential counts on aspirate smears, and visual estimates based on CD34-stained BM sections. There was excellent inter-observer reproducibility. However, the counts in both, smears and biopsies showed high variability and bias when compared to the gold standard. Furthermore, counts on smears and biopsies correlated poorly with each other. Compared to the smears, CD34-stained biopsies showed less variability but a higher positive bias, indicating that counts were overestimated in these preparations. Our results demonstrate that better counting methods than those currently employed are needed if significant diagnostic, prognostic or therapeutic decisions are to be made based on microscopic enumeration of blasts. Counting may be best performed in well preserved marrow tissue within trephine biopsies but efforts should be made to improve how to define blasts, perhaps using a combination of informative labels that could be applied to tissue sections. The counting procedure could also be facilitated by using automated digital imaging technologies, and serious consideration should be given to expressing results as blasts per marrow area rather than as a percentage of marrow cells since non-blast marrow cells may vary in number independently of blasts.