Analysis of the IgV_H somatic mutations in splenic marginal zone lymphoma defines a group of unmutated cases with frequent 7q deletion and adverse clinical course

Splenic marginal zone lymphoma (SMZL), a specific type of small B-cell lymphoma, is characterized by a peculiar morphology with micronodular pattern of infiltration, biphasic cytology, and the almost constant presence of marginal zone differentiation.1,2 In addition, the tumor is usually recognized by the presence in peripheral blood of villous lymphocytes and by a characteristic intrasinusoidal pattern of involvement in bone marrow biopsies.3,4 Although in most cases SMZL is an indolent tumor, a small fraction of cases display more aggressive behavior with peripheral lymph node infiltration, histologic progression, and eventually death attributable to the tumor.5 

Different observations coincide in pointing toward the relative molecular heterogeneity of this neoplasia, as observed in the study ofbcl-6 somatic mutation frequency or 7q31-q32 deletions.6-8 Simultaneously, the allegedly marginal zone origin of this lymphoma has increasingly been questioned. This hypothesis was based on the almost universal presence of marginal zone differentiation when splenic sections were examined in this tumor. However, it was brought into question because the examination of other organs failed to discover marginal zone differentiation, thus suggesting that the presence of marginal zone rims was dependent on the localization of the lesion in the spleen. This suggestion has subsequently been proven after showing that other lymphoma types, such as follicular lymphoma, regularly show marginal zone differentiation when infiltrating the spleen.9 

The classification and comprehension of B-cell lymphomas has been aided greatly by the recognition that B cells within the germinal centers are exposed to somatic hypermutation of IgV_H andbcl-6,10-12 which in the case of immunoglobulin genes increases their affinity with the antigens present in the germinal center microenvironment.13-16 Thus, somatic mutations of the genes for the variable region of the B-lymphocyte antigen receptor have proven to be the hallmark of germinal B cells, and sequence analysis of these genes offered a molecular approach to identifying the origin of the tumors.17 Thus, most B-cell non-Hodgkin lymphoma types have been assigned either to hypermutated B cells18-21 or to naive unmutated B cells, such as mantle cell lymphoma.22 In B-cell lymphocytic leukemia (B-CLL) the study of hypermutation in IgV_H genes has also revealed the existence of a strong heterogeneity, an especially relevant finding here being that unmutated cases show a more aggressive clinical course.23-26 

Although some initial data seem to suggest that SMZL could derive from mutated postgerminal center B cells, which would be consistent with their origin in splenic marginal zone B cells (usually showing hypermutated IgV_H sequences), these data were based exclusively on very short series of patients.27-30Here we analyze a large group of samples corresponding to this condition and show that roughly half of the cases are characterized by an unmutated V_H sequence with simultaneous loss of 7q and a more aggressive clinical course. SMZL cases also show a preferential usage of the V_H1-2 gene, thus confirming that involvement of the V_H gene is not random in SMZL cases.

To avoid uncertainty in the selection of SMZL cases, we only analyzed cases that showed a typical histology in their splenectomy specimens, thereby excluding cases selected exclusively on the basis of peripheral blood cell morphology.

Frozen tissue blocks from 35 cases of SMZL were included in this study. These cases were selected from the routine and consultation files of the Pathology Laboratories of the Virgen de la Salud Hospital (Toledo, Spain) and the Spanish National Cancer Center (CNIO, Madrid, Spain), based on the availability of clinical information and DNA for IgV_H studies.

These cases were diagnosed on the basis of splenic morphology according to current criteria.3 31 Small lymphocytic lymphoma, mantle cell lymphoma, follicle center lymphoma, hairy cell leukemia, and lymphoplasmacytoid lymphoma were considered in each case and were excluded according to morphologic, immunophenotypic, and molecular findings. Studies for t(11;14) and t(14;18) were performed and were negative in all cases. The percentage of malignant cells was estimated by histologic examination and by immunohistochemical CD20 staining, and at least 70% of cells were tumoral.

All cases were subjected to routine hematoxylin-eosin and immunohistochemical study in splenic sections. Immunohistochemical analysis was performed on formalin-fixed or B-5 paraffin-embedded tissue. After incubation with the primary antibody, immunodetection was performed with biotinylated antimouse immunoglobulins, followed by peroxidase-labeled streptavidin (LSAB-DAKO, Copenhagen, Denmark) and with diaminobenzidine chromogen as substrate. All immunostaining was performed by using the TechMate 500 (DAKO) automatic immunostaining device. The labeling system was obtained from DAKO. The following antibodies were used: CD20, CD3, CD23, CD43, bcl-2, CD38, bcl-6, cyclin D1, Mib1, IgD, κ, and λ (DAKO), as well as CD5 and CD10 (Novocastra, Newcastle, United Kingdom).

Twenty-two cases were studied for 7q deletion, with cytogenetic and/or loss of heterozygosity (LOH) studies. Cytogenetic studies were performed on lymphoid cells from the spleen in 18 cases according to the methods described by Solé et al.7 In 14 cases, 7q loss was investigated by LOH according to the methods described by Mateo et al.6 These cases have been reported previously.6 7 

Rearranged IgV_H genes were amplified by using a seminested polymerase chain reaction (PCR) method as described previously.14 In the first round of PCR, a mixture of 6 framework region 1 (FR1) V_H family-specific primers and 2 consensus primers for the J_H gene were used. The second round of PCR was performed in 6 separate reactions with one of the 6 V_H FR1 primers and J_H internal primers.

In 4 cases, total RNA was extracted from 5-μm frozen tissue sections, using RNAsy Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Total RNA (1 μg) was used for reverse transcription. Complementary DNA (cDNA) was synthesized by using AMV reverse transcriptase with random primers, following the manufacturer's recommendations.

Briefly, DNA 200 ng or cDNA product at a 1/10 dilution was amplified in a volume of 50 μL with 1× PCR buffer, 200 μM dNTPs, 2.5 mM MgCl₂, 250 nmol/L of each primer, and 1 U Ampli Taq Gold. In the second round of amplification, the same concentrations of reagent were used, except for MgCl₂, which was 1.5 mM. The first-round PCR product (1 μL) was added to the seminested reaction as a template. The PCR conditions of the first round consisted of one cycle at 95°C for 2 minutes, 59°C for 2 minutes, 72°C for 45 seconds followed by 34 cycles at 95°C for 45 seconds, 59°C for 15 seconds, 72°C for 40 seconds, and one final cycle of 72°C for 10 minutes. The second PCR consisted of a total of 25 cycles, using annealing temperatures of 65°C for the V_H3 and V_H4 primers and 61°C for the V_H1, V_H2, V_H5, and V_H6 primers. The denaturing and extension temperatures as well as the cycle conditions were identical to those of the first round of PCR. An aliquot of 10 μL PCR product was visualized in 3% metaphor agarose gel.

Direct sequencing was performed from both strands by using the same primers as in the amplification. The direct sequencing procedure was performed by using an ABI PRISM 310 or 3700 Genetic Analizers (Applied Biosystems, Weiterstadt, Germany), following the manufacturer's procedure. Mutations were identified by comparison with the germline sequence (Ig BLAST and V BASE sequence directory). In all cases of SMZL the sequencing analysis was confirmed by 2 separate experiments. To determine whether the number of replacement and silent amino acid substitutions identified were indicative of antigen selection, the Chang and Casali method was used.32 

A chi square test was used to compare the frequency of patient death because of the tumor between cases with and without IgV_H mutations. Survival analyses were performed by using the Kaplan-Meier life-table method. Differences between survival curves were determined according to the log-rank test. P < .05 was required for significance. The SPSS software package (SPSS) was used for all statistical analyses.

Of the 35 patients included in the study, 17 (48.6%) were women and 18 (51.4%) were men. Their median age was 64 years (range, 50-75). Thirty-two of the cases were diagnosed at clinical stage IV with bone marrow involvement and 2 at clinical stage I. All patients were treated by splenectomy, and 9 of them also underwent chemotherapy. The mean follow-up of the series was 30 months (range, 3-80). During this period, 12 patients died as a consequence of the tumor.

We amplified and sequenced 40 clonal IgV_Hrearrangements in 35 SMZL cases, because 5 cases had 2 different rearrangements. Among the 40 clonal V_H gene sequences, 38 were potentially functional and 2 were rendered nonfunctional by out-of-frame rearrangement (stop codon). In 4 of 5 cases, 2 apparently functional V_H genes were obtained, which probably represent a lack of allelic exclusion, similar to that described in B-CLL.33 

Comparison of clonal V_H genes with the germline gene segments revealed that the 40 V_H genes used by these 35 cases have a striking bias toward the use of V_H1, which was detected in 20 cases, and more specifically V_H1-2, which was detected in 18 cases (Table 1). In comparison with the predicted distribution of the V_Hrepertoire in peripheral blood lymphocytes, in which the use of V_H1 was observed in 19% of CD5⁺ and 13% of CD5⁻ cells, V_H3 is usually observed in 54% to 56% of cells. This finding shows a statistically significant difference (P < .000 01).34 

In this study, 17 (49%) of 35 cases displayed more than 98% sequence homology with the nearest germline V_H gene (unmutated), whereas 18 (51%) of 35 cases showed less than 98% homology with evidence of somatic hypermutation (Tables2 and3).

Most of the mutations were localized in FR3 and complementarity-determining region 2 (CDR2; data not shown). The mutations were represented by single nucleotide substitution, and neither point insertion nor deletion was observed. The frequency of mutations was lower in the cases involving the V_H1 family (8 of 20), but this frequency was not statistically significant.

Interestingly, 2 of 5 cases displaying 2 V_H rearrangements showed a discrepancy in the mutation pattern between 2 rearrangements, one sequence having a high homology with the germinal sequence and the other showing homology of less than 98%. Of the other 3 cases with double clonal rearrangement, 2 cases showed both sequences with 98% or greater homology, and 1 case presented with less than 98% homology. Messenger RNA analysis confirmed expression in 4 rearrangements; one of the rearrangements was out-of-frame (Table4).

Analysis of distribution of replacement and silent mutations were calculated by considering all possible mutations as described by Chang and Casali.32 Eighteen cases from SMZL showed somatic mutation in the IgV_H gene. Evidence for positive selection was shown to be statistically significant in 3 cases (Nos. 253, 371, and 377), whereas the number of replacement mutations in the CDR regions was greater than expected in comparison with the FR regions. In 2 cases (Nos. 202 and 380), evidence for negative selection was observed in which fewer replacement mutations than those expected were seen in the FR regions, indicating pressure to maintain the germinal configuration. In the other 13 cases with somatic mutations, no statistically significant evidence for antigen selection was observed (Table 5).

The frequency of IgV_H mutations was compared with the incidence of 7q loss, as determined by karyotyping and LOH. The frequency of 7q loss was higher in the unmutated cases than it was in mutated cases (P = .038; Table6). No relationship was detected between the frequency of somatic mutation and trisomy 3 or other cytogenetic abnormalities. The expression of CD38 in SMZL cases is shown in Tables 2 and 3. A relationship between the expression of CD38 and IgV_H gene mutation status was not observed in this series.

All of the cases analyzed except 4 showed IgD expression. Curiously, in these cases the IgV_H gene was mutated (Tables 2 and 3). The presence of a relationship between clinical course and the frequency of somatic mutations was also analyzed. The clinical features of patients with and without somatic mutations are show in Tables 2 and 3. Unmutated cases showed an increased probability of death because of the tumor (P = .022; Table 6).

To explore the relationship between tumoral aggressiveness and IgV_H status, we selected a group of cases with morphologic or clinical evidence of tumoral progression and compared the frequency of somatic mutations with the other cases. Of the 17 SMZL cases without somatic mutations, 11 (65%) showed either death attributable to the tumor or large cell transformation, whereas similar clinical aggressiveness was found in only 5 (28%) of 18 cases with somatic mutations. This finding is statistically significant (P = .027, Fisher exact test; Table 6).

The survival curve was drawn up according to the Kaplan-Meier method. The graph showing the survival of cases with mutated and unmutatedIgV_H genes is shown in Figure1. Survival at 60 months was 38% for the unmutated cases and 86% for mutated cases (log-rank test,P = .049).

This study analyzed mutations in V_H genes and the use of the V_H gene repertoire in a large number of SMZL cases. To establish beyond any doubt the diagnosis of these cases, examination of splenectomy specimen slides was required as a prior requisite for the inclusion of a case in the series. The results show that 51% of the cases presented less than 98% homology with the germinal sequence. The frequency of unmutated cases (49% in this series) was underestimated in previous analyses, probably because of the small size of the series previously reported and/or the absence of strict morphologic criteria in the recognition of the entity.27-30 

Thus, previously published descriptions of somaticIgV_H mutations analyzed series with a maximum of 4 to 5 patients, and study of the splenectomy specimen was not formally required for diagnosis. Our analysis also made it possible to confirm the selective use of V_H1-2 genes in a high proportion of SMZL cases, as anticipated by Miranda et al.30 The results obtained here are consistent with previously reported data on the frequency of 5′ noncoding regionbcl-6 gene mutations in SMZL, which already showed the existence of a molecular heterogeneity in this neoplasia.8 

This situation is very similar with what has been observed in B-CLL.25 In both conditions a proportion of cases (roughly 50%) are characterized by unmutated sequences, and strikingly in both types of lymphoma these unmutated cases show a more aggressive clinical course. This finding implies that this molecular heterogeneity is related to the clinical characteristics of the tumors in question. The hypothesis that the presence of unmutated V_H sequences (as observed in B-CLL, SMZL, and mantle cell lymphoma) could be used as a reliable marker of aggressive behavior in small B-cell lymphomas needs to be properly addressed in larger series of patients, but it could justify a role for molecular analysis in the selection of therapy.

The possibility of molecular heterogeneity in this neoplasia is strengthened because 7q deletions are significantly more frequent in unmutated cases. The presence of a selective 7q22-q32 deletion has been claimed to be the most frequent abnormality in this neoplasia,6 7 and it is also associated with a more aggressive clinical course.

The selective use of the V_H1-2 gene in a high proportion of cases also suggests that the origin of this lymphoma could be related to specific subsets of B lymphocytes, primed for its growth by autoantigens or superantigens. Although the clinical reports of a small proportion of SMZL patients have revealed autoimmune phenomena before the open development of the tumor, such as hemolytic autoimmune anemia or autoimmune thrombocytopenia, this situation does not apply to a high enough proportion of these patients to justify the selective use of the V_H1-2 gene. It is not known how this finding of preferential V_H gene involvement relates to tumor genesis. One possibility is that a common antigen could be involved in the preferential usage of individualV_H genes, causing the expansion of certain lymphocyte clones, and that malignant transformation occurs within the expanded clone.

A restricted involvement of IgV_H segments has also been described in B-CLL, in which V_H1-69 is involved in up to 12% of cases,23,25 or some mucosa-associated lymphoid tissue lymphomas, such as salivary gland lymphomas35 (8 of 14 cases), and also in hepatitis C virus-associated lymphoplasmacytic lymphomas.36 In all of these cases a selective involvement of V_H1-69has been observed. This involvement contrasts with the more frequent situation of follicular lymphoma, large B-cell lymphoma, or Burkitt lymphoma in which the expression of the V_H3 orV_H4 gene family is more common, thus reflecting more closely the repertoire observed in normal peripheral blood lymphocytes.19,21,34 37 

The selective involvement of V_H1-2 emphasizes the singularity of this neoplasia, suggesting that this tumor derives from a highly selected B-cell population. This involvement makes it advisable to search for specific antigens pathogenically relevant in the genesis or progression of this tumor. Nevertheless, the Chang and Casalli32 method showed evidence of positive antigenic selection in only 3 of 18 cases, which introduces some caveats about the potential role of antigens in the genesis of this tumoral type.

A relationship between the frequency of somatic mutations andIgV_H has been described in peripheral blood lymphocytes in which lymphocytes expressing V_H1 family genes have a lower frequency of mutation in comparison with the higher frequency observed for V_H3 orV_H4.34 This frequency is reproduced in B-CLL in which the use of V_H1 is accompanied by a frequency of 17%, opposed to a frequency of 90% among lymphocytes with V_H3. Nevertheless, in SMZL this relationship is not detected so clearly; thus, the frequency of mutations for cases expressing V_H1 was 40%, 60% for V_H3, and 50% in lymphomas expressingV_H4, which could be interpreted as supporting the hypothesis that, at least in this entity, heterogeneity in the frequency of somatic mutation is not clearly dependent on the selective involvement of V_H1 genes.

A striking finding in this series is the presence of 5 cases with double IgH rearrangements. Although there are a range of possible interpretations of this observation (bystander B-cell clones, lack of allelic exclusion, or presence of more than one tumoral clone), a similar situation in B-CLL has been linked to lack of allelic exclusion.33 Nevertheless, this situation has not been demonstrated in SMZL and needs to be properly investigated.

These data once again raise doubts about the proposed cell origin of this lymphoma type in the splenic marginal zone, because marginal zone B cells in the spleen have been shown to carry mutated V_Hsequences. Instead, they support the hypothesis that at least a significant fraction of cases could derive from naive cells not exposed to antigen B cells situated in the mantle zone.

We thank Dr A. I. Sáez for her useful help with statistical analysis. Drs T. Flores, T. Alvaro, P. Gonzalvo, L. Bernardó, M. Medina, V. Lescano, A. Santana, and Alvarez from the Pathology Departments at hospitals in Salamanca, Tortosa, Jarrió, Gerona, Osuna, Plasencia, Madrid, and Baracaldo (Spain) for kindly providing cases included in this series.