Dissemination of a Sjögren's syndrome–associated extranodal marginal-zone B cell lymphoma: Circulating lymphoma cells and invariant mutation pattern of nodal Ig heavy- and light-chain variable-region gene rearrangements




Both the genesis and outgrowth of extranodal marginal-zone B cell lymphomas (MZLs) of the mucosa-associated lymphoid tissue (MALT) type are generally thought to represent antigen-driven processes. We undertook this study to analyze lymphoma progression and dissemination outside of the MALT-type lesions.


Histopathologic and Ig heavy- and light-chain variable-region gene (VH/L) analyses were performed in sequential tissue samples from a patient with primary Sjögren's syndrome (SS) with glandular (parotid) manifestations and subsequent nodal dissemination of a low-grade MZL.


This MZL expressed a CD20+,CD27+,sIgM/κ+,IgD−,CD5−,CD10−,Bcl-6−,CD23−,p53−,p21−,MDM2− phenotype and mutated VH1–69/D2–21/JH4α–VκA27/Jκ2 Ig rearrangements. Notably, circulating lymphoma cells from the parotid glands occurred transiently in the patient's blood, as detected by single-cell polymerase chain reaction. In addition, 2 minor B cell clones (clones 2 and 3, with VH3–07/D3–22/JH3b–Vλ3L/Jλ2/3 and VH3–64/D3–03/JH2–VκA19/Jκ2 rearrangements, respectively) were also detected in the parotid glands and blood, and 1 of these (clone 2) was also detected in the lymph nodes. Ig VH/L analyses revealed ongoing (antigen-driven) mutations of the glandular lymphoma rearrangements, but an invariant mutation pattern of their nodal counterparts.


These data indicate coexpansion and transient (re)circulation of the lymphoma clone and 2 additional glandular B cell clones in a primary SS–associated extranodal MZL. Combined histologic and molecular features of the nodal lymphoma subclone reflect a process of “follicular colonization” that eventually froze the mutation machinery after accumulation of additional (antigen-driven) Ig VH/L mutations.

Patients with primary Sjögren's syndrome (SS) are at increased risk of developing B cell lymphomas, most commonly extranodal marginal-zone B cell lymphomas (MZLs) of the mucosa-associated lymphoid tissue (MALT) type (1–4). These lymphomas are thought to originate in the organs targeted by primary SS, frequently in the major salivary glands, but tend to spread to other mucosal or nodal sites (1–4). While the transition from benign chronic lymphoepithelial sialadenitis (LESA) to lymphoma is not well understood, the genesis of extranodal MZL in primary SS is generally considered to represent a multistep, antigen-driven process (1, 2, 5–12).

Recently, the clonal relationship between LESA and an overt MALT-type lymphoma of the lung was identified in a patient with primary SS, indicating the presence of local triggers in both the parotid and lung MALT-type lesions (9). The linkage between autoimmunity, autoantibody-producing B cells, and lymphoma has been emphasized by two examples of parotid gland lymphomas in primary SS expressing monospecific rheumatoid factors (10). Immunoglobulin heavy- and light- chain variable-region gene (VH/L) analyses revealed 1) a biased usage of Ig VH/L gene segments, 2) distinctive features of the heavy-chain third complementarity-determining region (HCDR3), and 3) ongoing mutations with an antigen-selected pattern in these lymphomas (7, 8). Thus, it has been speculated that chronic antigen-mediated B cell stimulation mediated by surface Ig receptors leads to nonmalignant (oligoclonal) expansions, increasing the risk of malignant transformation by a subsequent oncogenic event (1, 2, 5–12). However, in contrast to gastric MZL (12), parotid gland MZL is not yet defined by characteristic molecular genetic abnormalities. To analyze lymphoma progression and dissemination outside of the MALT-type lesions, the current study focused on a sequential analysis of the Ig V repertoire of B cell expansions in a patient with primary SS and MZL, which was primarily diagnosed in her parotid glands but which secondarily infiltrated the peripheral lymph nodes.



A 60-year-old woman with a 4-year history of primary SS was evaluated. She fulfilled the American–European Consensus Group criteria for SS (13) by having ocular and oral sicca symptoms, keratoconjunctivitis sicca verified by rose bengal staining, a positive Schirmer I test result, focal lymphocytic sialadenitis of the minor salivary glands (focus score 2/4 mm2), and antinuclear antibody positivity (titer of 1:10,240, with a fine speckled pattern on HEp-2 cell substrate) including anti-Ro and anti-La positivity. She had rheumatoid factor, polyclonal hypergammaglobulinemia, and recurrent parotid gland swelling and arthralgias, but no further signs of extraglandular manifestations, in particular, no cryoglobulins, diminished C3/C4 levels, or palpable purpura. Other causes of sicca symptoms were excluded.

In the spring of 1996, the patient developed a chronic, nontender enlargement of her left parotid gland, with no weight loss, febrile episodes, sweating, or peripheral lymphadenopathy. Differential blood cell count, peripheral blood smear analysis, and β2-microglobulin level were all normal. However, ultrasound and computed tomography revealed severe nonhomogeneous parenchyma with multiple mass lesions of her left parotid. Following left partial parotidectomy, an MZL was diagnosed histologically. Lymphoma staging showed neither extraglandular nor other glandular manifestation. However, 1 year later, painless swelling of the contralateral parotid gland occurred. Infiltration by the low-grade lymphoma was detected by open exploratory excisional biopsy and histology. Again, lymphoma staging showed no further manifestations, and local radiation was applied. Two years later, following transient major salivary gland swelling, a swollen inguinal lymph node was observed, and manifestation of the lymphoma was verified histologically. Seven months later, a relapse at the same site occurred. Followup by computed tomography showed enlargement of the spleen and liver, and the patient reported a weight loss of 6 kg within a period of 7 months. Therefore, systemic chemotherapy was started. After approval by the local ethics committee and informed consent by the patient were obtained, tissue samples were further processed for phenotypic and molecular B cell analyses (Table 1).

Table 1. Chronology of tissue specimens obtained from a patient with primary Sjögren's syndrome with a low-grade extranodal marginal-zone B cell lymphoma of the parotids, and number of detected independent B cell clones
SampleTissueDate*Duration, monthsMaterial for molecular analysisDetected clones
  • *

    Chronology of therapeutic interventions: August 1996, left partial parotidectomy; September–October 1997, local radiation of the right parotid gland (total dose 40 Gy); May–November 1999, 6 cycles of systemic chemotherapy with mitoxantrone/chlorambucil/prednisone. To screen for lymphoma dissemination in all tissue probes based upon the detected lymphoma immunoglobulin heavy-chain variable-region gene (VH) sequence (clone 1), a polymerase chain reaction (PCR) analysis was also performed using 1 μl of the extracted DNAs, external VH1 family–specific and JH primers in the first round, and VH1 family–specific internal and heavy-chain third complementarity-determining region–specific (5′-AAA-GCC-ACC-ACT-ATC-CGG-TCC-TCT) primers in the second round, with the same PCR conditions used for the Ig VH nested protocol (17). PCR products were visualized on high-resolution agarose gels and sequenced as described in Patient and Methods.

  • Time from diagnosis of lymphoma to biopsy/excision.

  • See Results for descriptions of clones.

1Left parotid glandAugust 19960Paraffin-embedded1
2Blood B cellsAugust 199712Bulk DNANone
3Right parotid glandAugust 199712Paraffin-embedded1, 2, 3
4Blood B cellsJuly 199823Bulk DNA and single-cell sort1, 2, 3
5Lymph node 1September 199825Paraffin-embedded1, 2
6Lymph node 2April 199932Paraffin-embedded1
7Blood B cellsApril 199932Bulk DNANone
8Lower lipMay 199933Bulk DNANone
9Bone marrowJuly 200271Bulk DNA and single-cell sortNone


For light microscopy, parotid gland and lymph node specimens were fixed in neutral buffered formalin and embedded in paraffin. Consecutive sections were routinely stained with hematoxylin and eosin, Giemsa, and periodic acid–Schiff. For immunostaining, 4-μm–thick sections were cut, deparaffinized, and subjected to heat-induced epitope retrieval. Immunostaining was performed as described previously using either sensitive labeled streptavidin–biotin– immunoperoxidase technique (14, 15) or alkaline phosphatase–anti–alkaline phosphatase technique (16). The primary antibodies used included monoclonal antibodies against CD5 (4C7), CD10 (56C6), and CD23 (1B12) (all from Novocastra, Newcastle, UK), CD27 (M-T271; BD PharMingen, San Diego, CA), MDM2 (Ab-1; Oncogene Science, Cambridge, MA), p21 (DCS-60.2; NeoMarkers, Fremont, CA), p27 (57; BD Transduction Laboratories, Lexington, KY), Bcl-10 (151; Abcam, Cambridge, UK), and CD20 (L26), Bcl-2 (124), Ki-67 (MIB-1), p53 (DO-7), and Bcl-6 (594) (all from Dako, Glostrup, Denmark), as well as polyclonal antibodies against IgM, IgD, IgG, IgA, Igκ, and Igλ (all from Dako). Finally, sections were counterstained with hematoxylin.

Analysis of Ig VH/L rearrangements.

DNA was extracted from fresh and/or paraffin-embedded tissue samples from parotid glands, CD19+ peripheral blood B cells, lymph nodes, bone marrow, and minor salivary glands using the QIAamp tissue kit (Qiagen, Hilden, Germany). A polymerase chain reaction (PCR)–based B cell clonality analysis was performed as described previously, using a seminested (third framework region [FR3]–JH) protocol (8, 15). To study somatic mutations and intraclonal variations, Ig VH/L rearrangements were amplified as described previously (17–19) and subsequently cloned using the AdvanTage PCR cloning kit (Clontech, Heidelberg, Germany) following the manufacturer's instructions. Positive clones were randomly selected and subjected to a second round of PCR (17–19). In addition, following a transient episode of parotid gland swelling, CD19+ peripheral blood B cells were sorted into single wells of 96-well plates (Robbins Scientific, Sunnyvale, CA) using a FACStar Plus flow cytometer with an automated single-cell deposit unit (Becton Dickinson, Mountain View, CA), and the rearranged Ig VH/L genes were subsequently amplified using nested PCR protocols as described previously (17–19).

After purification with the QIAquick Gel Extraction Kit (Qiagen), all internal PCR products were directly sequenced using the ABI Prism Dye Termination Cycle Sequencing kit (Perkin Elmer, Palo Alto, CA). Sequence alignments were performed using the V Base sequence directory (MRC Centre for Protein Engineering, Cambridge, UK; http://www.mrc-cpe.cam.ac.uk/DNAPLOT.php) to determine the underlying germline segments. Mutational analyses of Ig VH/L rearrangements were performed as described previously (17–19).

Analysis of p53 mutation status and t(14;18)(q32;q21) translocation.

A genomic sequence of the p53 gene (spanning exons 5–8) was amplified from DNA of the patient's tissue samples as well as from genomic DNA of 2 healthy controls according to the technique described by Tapinos et al (20) using a modified 5′ external primer: 5′-TGA-CTT-TCA-ACT-CTG-TCT-CCT-TCC-TCT. Purified internal PCR products were analyzed by direct DNA sequencing in both directions. Sequence alignments were performed by BLASTN searches against nucleotide databases (National Center for Biotechnology Information, Bethesda, MD; http://www.ncbi.nlm.nih.gov/blast/). Analysis of genomic bcl-2/IgJH rearrangement, specifically t(14;18)(q32;q21) translocation, was performed by a nested PCR protocol as described by Takacs et al (21).


Lymphoma characteristics.

The histologic findings from both the parotid glands and lymph nodes were consistent with a low-grade MZL of the MALT type. This included broad coronas of neoplastic marginal-zone or monocytoid cells (i.e., small, atypical lymphoid cells with slightly irregular nuclei and relatively abundant, pale cytoplasm) that virtually destroyed the glandular structures of the parotids, forming lymphoepithelial lesions (1, 22, 23), and infiltrated the lymphoid follicles of the affected lymph nodes, representing so-called “follicular colonization” (24) (Figure 1). Moreover, the lymphoma cells from parotid glands and lymph nodes shared the same CD20+,CD27+,sIgM/κ+,IgD−,CD5−,CD10−,Bcl-6−,CD23−,p53−,p21−,MDM2− phenotype. In accordance with p53/p21/MDM2 negativity, molecular analysis revealed a wild-type p53 gene sequence in the region that encodes for the DNA binding site of the tumor suppressor protein p53 (exons 5–8) (25–27).

Figure 1.

Histopathologic features of a primary Sjögren's syndrome–associated marginal-zone B cell lymphoma (MZL) in the parotid gland (A, C, E, G, and I) and lymph node (B, D, F, H, and J). A, Parotid gland (hematoxylin and eosin staining) showing a dense infiltrate of the MZL with a germinal center (GC) of a reactive follicle next to a band of dark-appearing mantle cells, and a broad corona of neoplastic light cells invading the epithelium (Ep), forming a so-called lymphoepithelial lesion. B, Lymph node effacement (Giemsa staining) by the neoplastic cells of the MZL with paler-staining cytoplasm showing a predominant distribution within the marginal zone (MZ) in addition to follicular colonization and areas of prominent plasmacellular differentiation (upper inset in F). C, Lymphoepithelial lesion of the parotid gland (higher power view; immunoperoxidase staining with anti-CD20) with broad bands of B cells infiltrating and destroying the salivary duct epithelium. D, Lymph node with expanded meshwork of CD23-expressing follicular dendritic cells (inset shows CD20 positivity). Tumor cells exhibit strong IgM expression both in the parotid gland (E, left) and in the lymph node (F, left) but are negative for IgG (E, right and lower inset in F). The proliferation fraction shown by Ki-67 immunostaining is low (∼5%) in the parotid gland (G, left) and slightly higher in the lymph node (∼15%), with some larger blasts (H, left); p27 is strongly detected in the tumor cells of the MZL in the parotid gland (G, right), while it is down-regulated in a proportion of the tumor cells in the lymph node (H, right; reactive mantle cells serve as a positive intrinsic control). Anti–Bcl-2 protein staining of the primary parotid gland specimen (I, left) and of the secondarily infiltrated lymph node (J, left) shows down-regulation of Bcl-2 protein in nodal lymphoma involvement. Increased expression of p53 is absent from the parotid gland (I, right) and from the lymph node (J, right). (Original magnification × 50 in A and B; × 100 in C, E [left and right], F [left and lower inset], G [right], H [right], I [left and right], and J [left and right]; × 25 in D; × 400 in D [inset] and F [upper inset]; × 200 in G [left] and H [left].)

Although the nodal lymphoma involvement showed a somewhat more aggressive pattern, with interdispersed aggregates of immunoblast-like cells in some areas and a moderately enhanced proliferative fraction, including an enhanced Ki-67 index and diminished p27 expression (27) (Figures 1G and H), it was clearly staged as low-grade MZL by independent pathologists. Notably, Bcl-2 protein was expressed in the tumor cells of the parotid glands but was down-regulated in the nodal lymphoma manifestation (Figures 1I and J). Consistent with extranodal MZL (1), genomic bcl-2/Ig JH rearrangement (i.e., rearrangement of the bcl-2 protooncogene by t[14;18][q32;q21] translocation) was not detected (22). Finally, the lymphoma cells showed weak cytoplasmic expression of Bcl-10.

B cell clonality analysis.

Based upon the detection of identical productive VH–D–JH heavy chain rearrangements with unique HCDR3 sequences (Figure 2), 3 independent clonal B cell expansions were identified in the patient's tissue specimens by an FR3–JH B cell clonality analysis and/or FR1–JH protocols. Remarkably, all 3 B cell clones were also detected outside of the extranodal (glandular) lesions (Tables 1 and 2). However, combined data were consistent with the persistence of an identical dominant B cell clone in all tissue specimens from both parotids, lymph nodes, and a single blood sample, thus representing the malignant lymphoma clone (clone 1). The lymphoma clone used a mutated VH1–69/D2–21/JH4a rearrangement, while the 2 minor clones used mutated VH3–07/D3–22/JH3b (clone 2) and unmutated VH3–64/D3–03/JH2 (clone 3) rearrangements. Analysis of the lymphoma HCDR3 sequence (clone 1) revealed a productive rearrangement of the VH segment to truncated D2-21 and JH4a segments, as well as the insertion of random nucleotides by terminal deoxynucleotidyl transferase activity (Figure 2). By both histopathologic and molecular analyses, there was no evidence of lymphoma involvement in bone marrow, minor salivary glands, or further peripheral blood samples (Table 1).

Figure 2.

Heavy-chain third complementarity-determining regions (HCDR3) of the lymphoma clone (clone 1) and 2 additional minor B cell clones (clones 2 and 3). Positions of HCDR3 are assigned according to Kabat et al (48). Nucleotide consensus sequences are given from sequential tissue samples, and stretches of the closest germline D segments are indicated. Dashes indicate sequence identity. Nucleotide exchanges are marked. N-nucleotides (i.e., nucleotides not encoded by germline segments but putatively introduced by terminal deoxynucleotidyl transferase activity at the VH–D and D–JH joining sites) are underlined. Deduced amino acid sequences of the glandular clonal consensus sequences are shown in boldface (1-letter code).

Table 2. Analysis of clonal Ig VH rearrangements from the patient's tissue specimens*
Tissue, date, cloneNo. of sequencesGermline VH segmentMutations/basepairs (%)R:S ratio of CDRsR:S ratio of FRsIntraclonal variation
  • *

    VH = heavy-chain variable-region gene; R = replacement; S = silent; CDRs = complementarity-determining regions; FRs = framework regions.

  • Mutations of the consensus sequence with respect to the closest germline gene. Uncommon mutations found in more than half of the clonal VH sequences were included in the consensus sequences.

  • All nucleotide exchanges (i.e., additional mutations or lack of consensus mutations) compared with the consensus sequence/no. of clonal sequences analyzed. In FR1, the analysis of somatic mutations was carried out starting from the codon where the sense primer ends. Expected (random) ratios for VH1–69 were CDRs 3.6 and FRs 3; expected (random) ratios for VH3–07 were CDRs 4.9 and FRs 2.8 (according to Chang and Casali [28]).

Left parotid gland, August 1996      
 122VH1–69246/4,862 (5.1)4:1 (4.0)2:5 (0.4)46/22 (2.1)
Right parotid gland, August 1997      
 123VH1–69282/5,083 (5.5)4:01:4 (0.25)78/23 (3.4)
 25VH3–0731/1,282 (2.4)3:01:1 (1.0)6/5 (1.2)
 32VH3–645/520 (1.0)0:00:05/2 (2.5)
Blood, July 1998      
 118VH1–69240/3,978 (6.0)4:1 (4.0)2:6 (0.3)46/18 (2.6)
 22VH3–0717/500 (3.4)3:01:1 (1.0)7/2 (3.5)
 33VH3–645/750 (0.7)0:00:05/3 (1.7)
Lymph node 1, September 1998      
 110VH1–69230/2,210 (10.4)6:4 (1.5)4:9 (0.4)0/10 (0)
 21VH3–075/225 (2.2)3:01:1 (1.0)
Lymph node 2, April 1999      
 119VH1–69437/4,199 (10.4)6:4 (1.5)4:9 (0.4)0/19 (0)

Mutation pattern of the lymphoma Ig VH rearrangements.

Altogether, 92 clonal lymphoma FR1–JH heavy-chain sequences were analyzed from the parotid glands (45/92), blood (18/92), and lymph nodes (29/92). While all Ig VH rearrangements obtained from the parotid glands and peripheral blood shared 8 common mutations compared with the underlying germline sequence VH1–69/DP10, the majority of VH rearrangements from these compartments displayed additional nucleotide exchanges. These nucleotide exchanges appeared to reflect intraclonal variation (Table 2) by ongoing mutations for 2 reasons: first, their frequency (1.9%; 263 mutations/13,923 basepairs [i.e., 4.2 mutations/VH segment]) was significantly higher than the estimated Taq error rate (∼1 × 10−4/basepair [i.e., <0.5 mutations/VH segment]; P < 0.001 by chi-square test); second, their pattern showed a genealogic relationship (Figure 3).

Figure 3.

Genealogy of ongoing mutations in the lymphoma's (clone 1) rearranged heavy-chain variable-region gene segments derived from consecutively obtained tissue specimens (i.e., left parotid gland, right parotid gland, peripheral blood, and lymph nodes). VH1–69/DP10 was identified as the closest germline sequence. Analyzed clonal sequences are shown as shaded circles. Deduced intermediate mutation steps are shown as open circles. Additional mutations (analyzed from codon 25 to codon 94) are given with their codon position beneath the respective circles. Silent or replacement mutations are shown as lower-case letters or underlined capital letters, respectively. Numbers within circles represent numbers of mutations compared with the VH1–69/DP10 germline sequence. Putative relationships are marked by broken lines/arrows with broken lines. Identical clonal sequences that were independently obtained from the patient's right parotid gland and blood are connected with gray lines and shown as solid circles. Asterisk indicates the start point of the genealogy as well as the step dividing the parotid gland clone from its nodal subclone (see Figure 4). All clonally related sequences from the parotid glands and peripheral blood shared 8 common mutations compared with the germline sequence: AGG at position 31, GGc at position 49, CTC at position 53, CGG at position 64, GAC at position 65, ACa at position 68, TCC at position 79, and GAa at position 85.

Figure 4.

Comparison of the mutation patterns of the lymphoma's parotid gland heavy-chain variable-region gene consensus sequence and its nodal counterpart, both aligned with the underlying VH1–69/DP10 germline sequence. The parotid gland lymphoma VH sequence was obtained 4 times from the original left parotid gland lesion and shared the 8 common mutations of all parotid and blood lymphoma sequences (see asterisk in Figure 3). The mutation patterns of the clonal lymphoma VH1–69/DP10 sequences from the lymph nodes were invariant, including 7 of the common mutations and 16 additional mutations, but lacked the position 65 GGC-to-GAC mutation. Eleven mutations that were detected exclusively in lymph node–derived lymphoma VH sequences are shown in boldface. Dashes indicate sequence identity. Positions of framework regions and complementarity-determining regions (CDRs) are assigned according to Kabat et al (48).

Remarkably, the lymphoma VH rearrangements obtained from 2 independent lymph nodes expressed an invariant mutation pattern but were found to be significantly more mutated (10.4%; 667 mutations/6,409 basepairs) than their counterparts from the blood (6.0%; 240 mutations/3,978 basepairs) and parotid glands (5.3%; 528 mutations/9,945 basepairs) (P < 0.0001 for both). However, the lymph node–derived lymphoma VH sequences shared only 7 of the 8 common mutations, but displayed 11 unique mutations (i.e., basepair exchanges compared with the VH1–69[DP10] germline sequence) that were not found in any of the other clonally related lymphoma VH sequences (Figure 4).

Analysis of the distribution of codons with replacement (R) and silent (S) mutations revealed significantly enhanced R:S ratios in the CDRs compared with the FRs of the lymphoma Ig VH rearrangements obtained from the parotid glands, blood, and lymph nodes (P < 0.0001 by Fisher's exact test) (Table 2). The R:S ratios in the CDRs of the lymphoma VH sequences obtained from parotid glands and blood (4.0 for both) were in the range of expected (random) mutations for the underlying VH1–69 germline segment (3.6, according to Chang and Casali [28]), while a markedly diminished R:S ratio (1.5) was found in the CDRs of the lymph node–derived lymphoma VH sequences (Table 2).

Light chain analysis.

Analyses of Vκ and Vλ light-chain rearrangements in sequential tissue samples revealed a productive clonal rearrangement using mutated VκA27/Jκ2 segments in parotid glands, blood B cells, and lymph nodes. This VL rearrangement displayed no insertion of N nucleotides but showed signs of both Vκ (3 basepairs) and Jκ (1 basepair) exonuclease activity, resulting in a CDR3 length of 9 amino acids. By single-cell analysis from CD19+ blood B cells, the clonal VκA27/Jκ2 light-chain rearrangement was found to be paired to the lymphoma Ig VH rearrangement (clone 1).

Altogether, 50 Ig VL rearrangements of the lymphoma clone were analyzed from the parotid glands (28/50), blood (9/50), and lymph nodes (13/50). While the lymphoma VL rearrangements from the parotid glands (mutation frequency 3.1%; 196 mutations/6,412 basepairs) and blood (mutation frequency 3.2%; 66 mutations/2,055 basepairs) exhibited intraclonal heterogeneity by ongoing mutations, their lymph node–derived counterparts displayed an invariant mutation pattern with a significantly higher mutation frequency (5.3%; 158 mutations/2,977 basepairs) (P ≤ 0.0005 for both, by chi-square test), with R:S ratios of 0.5 for the CDR and 1.7 for the FR. The 2 minor B cell expansions (clones 2 and 3) used productive Vλ3L/Jλ2/3 and VκA19/Jκ2 and light-chain rearrangements with no or few mutations, respectively. Notably, the Ig VH/L usage in the polyclonal B cell repertoire of this patient appeared to be very similar to that of patients with primary SS without lymphoma (5, 6, 18, 19, 29, 30).


In the primary SS–associated extranodal MZL reported here, both the distinctive Ig VH/L gene pairing and intraclonal sequence heterogeneity indicative of ongoing (antigen-driven) somatic mutations are consistent with the conclusion that salivary gland lymphomas in patients with primary SS may arise from a selected glandular B cell population (6–10), for example, from rheumatoid factor B cells (8, 10). In particular, the VH1–69 and VκA27 segments encoding the G6 and 17.109 cross-reactive idiotypes, respectively, have been found to be preferentially used in rheumatoid factor B cells, salivary gland B cells of patients with primary SS, and B cell lymphomas (7, 8, 30–32). In this context, a recent study has found positive selection of polyclonal B cells preferentially expressing VκA27/Jκ2 rearrangements in the parotid glands of a patient with primary SS with benign LESA (30). Moreover, the current lymphoma HCDR3 shared remarkably similar features with previously reported VH1–69-encoded MALT-type salivary gland lymphoma (8) and rheumatoid factor (33) Ig heavy chains that differ from those described in blood B cells of healthy normal individuals (34) and chronic lymphocytic leukemia B cells (35).

Of note, circulating lymphoma cells were identified in the patient's peripheral blood by single-cell analysis following a transient episode of parotid gland swelling. Contamination during the amplification process could be excluded for the following reasons. First, although closely related to the glandular lymphoma rearrangements, these subsequently obtained sequences from the peripheral blood B cells showed further evidence of ongoing mutations (Figure 3). Second, paired heavy- and light-chain lymphoma sequences were obtained from identical wells. Finally, in several hundred VH–D–JH rearrangements obtained to date in our laboratory, this particular lymphoma rearrangement has never been obtained again. Therefore, these cells were considered to represent (re)circulating lymphoma B cells from the parotid gland lesions. To the best of our knowledge, this finding has not previously been reported in MALT-type MZLs of the parotid gland, although circulating neoplastic cells have been detected in clinically localized cases of MZLs of various other extranodal sites (36, 37). Thus, the patient's diagnosis was reevaluated to exclude a clinical manifestation of other types of low-grade lymphomas with potential peripheral blood involvement (e.g., nodal or splenic MZL [22]). However, combined clinical, histopathologic, and molecular data confirmed the diagnosis of extranodal parotid gland MZL with secondary nodal involvement (1–3, 38).

In this context, clonally related glandular B cells have been occasionally detected in the peripheral blood of patients with primary SS with benign LESA, and these may represent (re)circulating B cells from sites of chronic lymphoid inflammation (29, 39). Lymphoid proliferation is a key feature of primary SS, and during the course of the disease, different B cell clones can be observed at different points in time (40). In the present study, 2 additional clonal glandular B cell expansions (clones 2 and 3) were also identified in the patient's blood, 1 of which (clone 2) was found in a subsequent lymph node biopsy specimen. Interestingly, clone 2 used a VH3–07/D3–22/JH3b rearrangement with an HCDR3 encoding for 16 amino acids (starting with glycine and aspartate residues [GD single-letter code] partly encoded by N-nucleotides). Again, the features of this Ig VH rearrangement shared remarkable similarities with those preferentially used by rheumatoid factor B cells (33) and MALT-type lymphomas (7, 8), while the VL segments encoding for the light chains of the 2 minor clones also expressed some features of positively selected (30) or rheumatoid factor (41) B cells in patients with primary SS. According to the proposed concept of multistep lymphomagenesis in primary SS (including an autoimmunity–lymphoproliferation–lymphoma sequence [2,6–12]), these 2 additional minor B cell clones (clones 2 and 3) may reflect oligoclonal (nonmalignant) proliferations of (auto)antigen-selected B cells (e.g., rheumatoid factor–expressing B cells), thereby representing a feature of the underlying autoimmune disease.

Analysis of mutations in the Ig VH/L lymphoma rearrangements revealed that the lymph node–derived lymphoma sequences most likely represented the outgrowth of a lymphoma subclone. In detail, this subclone shared some (but not all) common mutations with the lymphoma Ig VH/L rearrangements obtained from the primary parotid gland lesion, but exhibited quite a different pattern of additionally accumulated mutations. Remarkably, the R:S ratios in their CDRs were found to be significantly decreased (i.e., lower than expected by chance), thereby indicating that counterselection against R mutations in the antigen-binding CDRs had occurred. Such negative selection may indicate that at a late stage of somatic evolution, conserved residues in the antigen-binding hypervariable loops (42) were important for sustained lymphoma growth (7, 43). The pattern of additionally accumulated mutations of the lymph node–derived lymphoma sequences is consistent with this conclusion.

Additionally, the nodal lymphoma manifestation exhibited a slightly more aggressive morphology compared with that of the parotid gland. Since a distinct signature of the molecular abnormalities in parotid gland MZL (in contrast to gastric MZLs) has not been defined (12, 22, 44), a number of potential markers of cell cycle deregulation and/or transformation were analyzed to search for potential explanations for the distinctive morphology and Ig V gene mutation pattern of the nodal lymphoma involvement. This analysis revealed a secondary nodal effacement of the MZL that exhibited a moderately enhanced growth fraction (indicated by enhanced Ki-67 index and partial p27 down-regulation [27]) compared with the primary parotid gland lesion, but that clearly remained low grade. Especially, a potential nodal high-grade transformation into a diffuse large B cell lymphoma was excluded (by morphologic and phenotypic features [e.g., Bcl-6 negativity] [22]).

Mutations of the tumor suppressor p53 gene have been reported to be associated with a more aggressive behavior of several low-grade MZLs (25–27). However, both the parotid gland and nodal manifestations of the patient's lymphoma shared the same p53−,p21−,MDM2− phenotype combined with the usage of a wild-type p53 gene (26). Importantly, histopathologic features, combined with the lack of t(14;18)(q32;q21) translocation, also made coexpansion of a follicular lymphoma (45) very unlikely as a possible reason for the nodal subclone. Furthermore, the lack of abnormal nuclear Bcl-10 expression practically excludes a t(1;14)(p22;q32) translocation (12, 46) in the patient's tumor cells from both parotid glands and lymph nodes. Finally, although analysis of further chromosomal abnormalities, especially of trisomy 3, could not be performed retrospectively in the patient's material, findings of a recent study made an influence of this abnormality on the nodal relapse rather unlikely (47).

Instead, the features of the nodal lymphoma subclone may be explained by the phenomenon of “follicular colonization” (24) rather than by a distinct genetic abnormality. In this regard, after reentry of marginal-zone lymphoma B cells into nonneoplastic lymphoid follicle centers, intrafollicular expansion initially acquires additional VH/L mutations but subsequently destroys the follicular microenvironment, with a consequent loss of hypermutation (11, 24). This is consistent with the findings of the current study. We found colonization of nodal lymph follicles by cells that retain their marginal-zone lymphoma cytomorphology and immunophenotype as well as down-regulation of Bcl-2 protein expression (1). Following an intensive accumulation of additional mutations in a distinct pattern of a late germinal center reaction, the Ig VH/L mutation process was frozen. Both the histologic and molecular findings indicate that this process had occurred directly in the affected lymph nodes themselves. Importantly, this hypothesis would implicate antigen contact in an early phase of lymph node involvement (i.e., also outside of the patient's MALT-type lesions). Moreover, one could argue that a loss of mutation activity and a decrease in R:S ratios imply that the tumor has strict requirements for antigen contact (i.e., mutations that interfere with antigen binding might be strictly eliminated and therefore not observed).

In conclusion, these findings strongly indicate a (re)circulation of glandular B cells into the blood of a patient with primary SS with MALT-type lymphoma of the parotid glands, including the malignant lymphoma clone and 2 additional minor clones. Analysis of Ig VH/L rearrangements suggests that these clones may belong to a selected glandular B cell population. However, during subsequent nodal lymphoma involvement, an expanding lymphoma subclone was detected that seemed to be clearly selected by antigen but lacked signs of ongoing mutations during lymphoma progression. Remarkably, the mutation pattern in the Ig VH/L rearrangements of this nodal lymphoma subclone suggests its derivation from an early parotid gland lesion, rather than from the detected (re)circulating blood lymphoma cells.


The authors would like to thank Professor Dr. Harald Stein (Head, Consultation and Reference Center for Hematopathology, Institute of Pathology, Campus Benjamin Franklin, Charité Universitätsmedizin Berlin, Berlin, Germany) for reevaluating the nodal lymphoma pattern. We are grateful to Claudia Heimbächer and Thoralf Kaiser for excellent technical assistance.