Background & Aims: Accurate plasma cell enumeration is required to diagnose and monitor plasma cell neoplasms. Although generally performed on aspirate marrow, the number of plasma cells is also determined on bone marrow biopsy specimens. This usually includes immunostaining for CD138 and/or MUM1, and manual estimation of the plasma cell burden rather than accurate counting. We have assessed the accuracy of an automated digital enumeration platform to calculate the percent CD138 and MUM1-expressing plasma cells in trephine specimens of multiple myeloma.
Methods: CD138 and MUM1 expression in plasma cells was evaluated in bone marrow trephine biopsy specimens of multiple myeloma (n=91) patients using immunohistochemistry. Manual estimation and digital quantification of the percent plasma cells across the entire trephine section were performed. Intraclass correlation analyses (ICC) with associated Spearman’s correlations were evaluated between observers, antigens (i.e. CD138 versus MUM1) and enumeration methods (i.e. digital versus manual).
Results: Digital counts (range=0.05–93.5%) could be achieved for all specimens with varying cellularity (mean=100,084 cells per specimen, range=16,484–1,118,868 cells). Concordance between digital and manual counts for CD138 and MUM1 were 0.63 and 0.89 respectively. This was highest with diffuse MUM1 infiltrates and poorest when the plasma cells were present in foci. Relative to digital evaluations, CD138 manual counts overestimated plasma cell burden by a mean of 26.4%, but only by 9.7% with MUM1. Most of these cases showed microaggregates or large nodules of plasma cell activity and inappropriate manual overestimation of their number.
Conclusions: Automated digital enumeration of CD138 or MUM1-positive plasma cells in multiple myeloma is highly accurate, even to levels as low as 0.05%. Our digital methodology highlighted the inherent inaccuracies and overestimation of plasma burden by visual determination. The differences were most significant for plasma cells immunophenotyped by CD138 and those in clusters. Digital counting avoids the subjective over-interpretation of plasma cells number due to their size and underestimation of the total number of cells.