This study suggests that SMZL can be divided into two different types based on VH gene mutational status, one with unmutated VH genes (all had >98% homology to germline VH gene segment) and the other with mutated VH genes (all had <97% homology to germline). Although some of the unmutated cases had occasional single nucleotide differences from germline, VH genes with 98% or greater homology to published germline counterparts are typically considered unmutated because VH gene polymorphisms, which are very common, can account for that degree of difference. 23 In normal B cells, VH gene hypermutation is restricted to the germinal center stage of development and plays a role in the affinity maturation process. 22 For the most part, VH gene mutation patterns in malignant B cells appear to parallel normal B-cell development in that neoplasms of naïve pregerminal center B cells have unmutated VH genes whereas neoplasms of germinal center and postgerminal center B cells display mutated VH genes. 7,22 Our findings are therefore consistent with the proposal that the unmutated type of SMZL originates from naïve pregerminal center B cells and the mutated type of SMZL originates from postgerminal center memory B cells.
It is interesting that normal splenic marginal zone B cells also appear to be heterogeneous consisting of both memory and naïve B cells. 5 Analysis of VH genes expressed by normal splenic marginal zone B cells revealed that many are mutated consistent with these representing memory B cells. 24,25 However, normal splenic marginal zone B cells without VH gene mutations or only occasional mutations were also found in these studies suggesting that some marginal zone B cells represent naïve B cells. Additional support for a population of naïve splenic marginal zone B cells comes from evidence suggesting the splenic marginal zone is enriched for B cells that respond to thymus-independent type II antigens. 5 Our proposal that SMZL originates from both naïve and memory B cells is, therefore, in line with the apparent heterogeneity of normal splenic marginal zone B cells.
Our finding significant numbers of SMZLs with unmutated VH genes differs from most earlier studies. 8-10 However, these earlier studies only examined a small number of cases and occasional SMZLs with VH genes that would be considered unmutated were identified. In contrast to these earlier studies, the recent report by Camacho and colleagues 11 found a majority of their SMZL cases studied (four of five) to have unmutated VH genes. It may be important that the SMZLs in this study were restricted to those that had progressed to large cell lymphomas and may therefore represent a subset with a more aggressive clinical course. The possibility that SMZLs with unmutated VH genes may have a more aggressive clinical course than SMZLs with mutated VH genes parallels chronic lymphocytic leukemia (CLL), in which CLL with unmutated VH genes are more aggressive clinically than CLL cases with mutated VH genes. 23,26 Our proposal that VH gene mutational status may be used to differentiate two types of SMZLs, therefore, is not without precedent and similar to what has already been proposed for CLL.
The World Health Organization lymphoma classification as well as other recent reports suggest that most cases of SMZL should be positive for both IgD and IgM. 1,2 Therefore, finding five of eight SMZL cases in our study lacking IgD seems to be atypical and could raise doubts regarding our diagnoses even though all of the other characteristic histological and immunophenotypic features were present. One potential factor contributing to finding fewer than the expected number of IgD-positive cases may be related to a low sensitivity of our immunoperoxidase detection technique even though it clearly detects normal mantle zones. Indeed, this possibility is supported by our finding one SMZL case that appeared negative by immunoperoxidase to be IgD-positive by flow cytometry. Similar to our findings, however, other studies of SMZL have also reported a significant proportion of cases to be negative for IgD. 8,10,18,27 It is possible, therefore, that there are more IgD-negative SMLZ cases than now generally appreciated. This idea is also more consistent with frozen tissue studies of normal splenic marginal zone B cells that demonstrate that approximately half are negative for IgD whereas the remaining are weakly positive and a few are strongly IgD-positive. 5,6
Recent studies of normal B cells have indicated that high levels of surface IgD and IgM expression are more typically found on naïve B cells with unmutated VH genes. 28-30 This is consistent with our SMZL findings and suggests that high IgD expression may also mark some naïve B cells in the splenic marginal zone. However, co-expression of IgD and IgM is not a specific marker for naïve B cells because a small percentage of peripheral blood memory B cells with mutated VH genes and CD27 expression have this phenotype. 30 Moreover, CD27 and CD148 appear to identify memory B cells in the spleen some of which appear to express low intensity IgD with IgM. 31 In addition, co-expression of IgM and IgD may be less restricted in B-cell neoplasms as suggested by follicular lymphomas often being positive for both IgD and IgM. 32,33 Therefore, the correlation we have observed between IgD expression and unmutated VH genes may have to be modified somewhat after additional cases and quantitative levels of IgD are studied.
Finding evidence of ongoing mutation in most of the SMZL cases with mutated VH genes raises the possibility that these neoplasms originate in germinal center B cells rather than postgerminal center memory B cells. 12,22 Furthermore, it is well recognized that follicular lymphomas involving the spleen can sometimes morphologically resemble SMZL. 34 However, none of the four mutated SMZL cases resembled follicular lymphoma, none expressed CD10, and none of the three tested cases harbored a (14;18) translocation. It is also possible the initial transforming event occurs in a germinal center B cell that could theoretically result in sustained activation of the VH gene mutator even if the final transformation event occurs in a postgerminal center B cell. An alternative explanation for the observed low-level ongoing mutation is that SMZL cells may sometimes re-enter the permissive germinal center, 35 which has been described for normal marginal zone B cells. 36 Therefore, finding low-level on-going VH gene mutation does not necessary imply an origin from germinal center B cells.
The IgD-positive SMZL cases with unmutated VH genes raise the possibility of a mantle cell origin for this subset because normal mantle cells and mantle cell lymphomas typically demonstrate strong IgD positivity and have unmutated VH genes. 37 However, none of these cases had the expected CD5 or cyclin D1 positivity indicative of mantle cell lymphoma. The possibility that these represent a previously undescribed lymphoma derived from mantle zone cells that differentiate to cells resembling marginal zone cells could still be considered. 38 However, because it appears naïve B cells with unmutated VH genes are also present in normal splenic marginal zones, we favor a marginal cell origin for the unmutated SMZL cases.
This study also suggests that SMZL shows biased or nonrandom use of VH gene segments with overrepresentation of the VH1 and VH4 families (seven of eight cases) and underrepresentation of the largest VH3 family (one of eight cases) that contains ~50% of the functional VH gene segments. 15,20 Although the number of cases is small, finding repeat use of VH4-34 (two cases), V1-69 (two cases), and V1-2 (two cases) further suggests there may be preferential use of specific gene segments from the VH1 and VH4 families. Additional support for this possibility comes from the study of Dunn-Walters and colleagues, 8 in which two of four SMZL cases analyzed appeared to express V4-34, and the study of Miranda and colleagues, 9 in which one of four cases used V4-34 and two of four used a V1-2 gene segment. Although none of the SMZL cases studied by Tierens and colleagues 10 used V4-34 or V1-2, three of four cases used other VH gene segments from the VH1 and VH4 families. Preferential use of specific VH gene segments or families suggests that selection by a limited number of antigens may be playing a role in the development of SMZL. 13 This is also supported by analyses of VH gene mutations in the mutated SMZL cases, which suggested there was selection for expression of functional immunoglobulin molecules.
Although two types of SMZL can be distinguished by VH gene mutational status, differences in the use of specific VH segments between the two types were not apparent because both types had representative cases using V1-69, VH4-34, and V1-2. Furthermore, differences in histological features were also not apparent as a similar number of cases showing plasmacytic features or having well-defined dimorphic white pulp nodules were also represented in both groups. Finding similar histological features and patterns of VH gene use among the two SMZL types is consistent with the proposal that that both originate from splenic marginal zone B cells.
In conclusion this study suggests that SMZLs can originate from pregerminal center naïve B cells that have unmutated VH genes as well as from postgerminal center memory B cells that have mutated VH genes. Because the normal splenic marginal zone appears to contain both memory B cells and naïve B cells, 5 we propose that the mutated and unmutated types of SMZL originate from different types of splenic marginal zone B cells. Even though differences between the proposed two types of SMZL were not identified other than IgD expression, additional studies looking at more parameters are needed to adequately address this issue. In particular, it will be important to obtain longer term clinical data to see if there may be differences in SMZL patient survival similar to what has been described for the recently proposed two types of CLL that are also distinguished by VH gene mutational status. 23,26
After this study was submitted for publication, an article examining VH genes using 35 cases of SMZL was published. 39 Similar to our findings, approximately half of these SMZL cases had unmutated VH genes, and there seemed to be preferential use of VH1 family gene segments, especially V1-2, among both unmutated and mutated types. Survival data from this article indicated that SMZL with unmutated VH genes may indeed have a more aggressive clinical course as suggested above.