Phenotyping primitive plasma cells.

Abstract

In a recent paper, Rawstron et al (2001) presented a ‘rigorous sequential’ gating strategy in an attempt to resolve perceived discrepancies between various reports of the phenotypic changes which occur during plasma cell differentiation. The authors concluded that discrete stages of plasma cell differentiation could not be distinguished by the level of CD45 expression. It is then suggested that this supports the hypothesis that there is a single stage of neoplastic plasma cell differentiation. We believe that there are several problems with this study which have led to incorrect conclusions. The first issue relates to the accurate identification of plasma cells using immunophenotyping. One obvious omission from the study is the failure to demonstrate either cytoplasmic immunoglobulin or light chain expression. It has been our experience that dual CD38hi and cytoplasmic light chain expression provides the most accurate phenotypic marker of plasma cells (Joshua et al, 1996; Pope et al, 1997). Although both CD138 and CD38hi are commonly used plasma cells markers, each has limitations. CD38 is expressed on several other cell lineages at a lower antigen density and is therefore not suitable for separating plasma cells using antibody-coated magnetic beads. However, flow cytometric electronic gating can overcome this defect and the best discrimination occurs when anti-CD38 with a high intensity fluorochrome such as anti-CD38 PE is used. CD138 cross-reacts with few other cell lineages and is therefore potentially suitable for affinity separation techniques. However, we have previously demonstrated that, in␣a group of 35 patients with myeloma, an average 32% of␣the CD38hi cells were CD138– (Pope et al, 2000). In addition, it has been previously reported that only four out of six commonly used myeloma cell lines demonstrate CD138 expression (Gooding et al, 1999). In a series of flow plots from a single patient (Fig 1) we demonstrate that CD138– cells were present in the CD38hi population (Fig 1A) and that, as 99% of the CD38hi population in this sample were cytoplasmic kappa+ (Fig 1B), it is clear that CD138 does not identify all plasma cells. (A) Scattergram of CD38 versus CD138 expression on bone marrow cells from a patient with myeloma. (B) Cytoplasmic kappa expression on CD38hi cells. (C) CD45 subpopulations of CD38hi cells. (D–F) Cytoplasmic kappa, lambda and Ig expression in CD38hi CD45–, CD38hi CD45+ and CD38hi CD45hi cells. The second issue relates to CD38hi CD45+ cells. We previously suggested that primitive (CD45hi), immature (CD45+) and mature (CD45–) plasma cell populations can be identified (Fig 1C). All these cells express high levels of cytoplasmic Ig and are light chain restricted. The CD38hi CD45+ and CD38hi CD45hi cells were > 99·5% cytoplasmic Ig positive and > 97% kappa positive (Fig 1D, E and F) and did not express CD3. However, as Rawstron et al (2001) discovered, the CD38hi CD45+ population may be contaminated with CD3 cells when the fluorochrome used is anti-CD38 phycoerythrin Cy5 (PECy5) rather than anti-CD38 PE. The complexities of colour compensation without using advanced digital compensation software also contribute to this artefact by reducing fluorescence intensity. Clearly not only the correct antigens but also the most appropriate fluorochromes must be used. Many of the monoclonal antibodies used in the Rawstron study, including anti-CD45, were in-house hybridoma supernatants. CD45 has considerable epitope heterogeneity and comparisons between studies must be made with caution. It is difficult to determine the specificity of the anti-CD45 reagent used in the study of Rawstron et al (2001); however, we use a commercially available pan anti-CD45 (Becton Dickinson). We have noted that other anti-CD45 reagents have different specificity and produce different results. Finally, to claim that there is a single stage of neoplastic plasma cell differentiation based on the results of this study ignores many other important observations. For instance, there is clearly a spectrum of differentiation in the plasma cell morphology of many patients. Those involved in the search for ‘the cell of origin of myeloma’ may argue about a pre- or post-switch germinal centre B cell but none would argue that a precursor cell population does not exist at all. Furthermore, various studies have shown that small subpopulations expressing cytoplasmic µ are evident as is a reservoir of CD19+ precursors. Thus, there is significant evidence to support the concept that the malignant plasma cells of patients with myeloma differentiate from precursor cells. As it is the functional phenotype of the primitive plasma cell and not their number which determines the clinical state of the patient (Joshua et al, 1996; Pope et al, 1997), it is these cells which should be identified and targeted for therapy in the future rather than the bulk of the tumour.