Introduction

Epstein–Barr virus (EBV)-positive mucocutaneous ulcer (EBVMCU) was first described as a distinct clinicopathological entity in 2010, when Dojcinov et al. reported 26 patients with ulcerative lesions confined to the oropharynx, skin, and gastrointestinal tract [1]. These patients were immunosuppressed, with either age-related immunosenescence or by iatrogenic immunosuppression. Subsequently, EBVMCU has been reported in patients with primary immunodeficiencies [2,3,4], solid organ or bone marrow transplant recipients [5,6,7,8] and human immunodeficiency virus/acquired immune deficiency syndrome (HIV/AIDS) [9]. In general, patients with EBVMCU exhibit good prognosis with spontaneous regression or complete remission following reduced immunosuppression [10]. EBVMCU was later described by the World Health Organization as a new disease concept, which was distinct from lymphomas, and recognized as a specific type of immunodeficiency-associated lymphoproliferative disorders (LPD) [11].

Histologically, EBVMCU is characterized by EBV-positive, polymorphous, atypical proliferating B cells that can resemble Hodgkin and Reed–Sternberg (HRS) cells. These atypical cells accompany dense polymorphic inflammatory cell infiltration, such as plasma cells, eosinophils, and histiocytes. The EBV-positive cells demonstrate B-cell immunophenotypes, i.e., CD20 expression. Thus, it is difficult to distinguish EBVMCU from lymphomas [1, 10, 11]. To clarify clinical, pathological and molecular characteristics between EBVMCU and diffuse large B-cell lymphoma (DLBCL), we examined clinicopathological features and gene rearrangements for immunoglobulin heavy chain (IGH) and T-cell receptor (TCR).

Materials and methods

Patient samples

We analyzed the clinicopathological features of 34 patients with EBVMCU. All cases were retrieved from the surgical pathology consultation files of the Department of Pathology at Okayama University between 2009 and 2019. Of the 34 EBVMCU patients, 33 were biopsy specimens and one was a resected specimen. We excluded patients with a history of lymphoma; other lesions identified by radiation diagnosis i.e., positron emission tomography and computed tomography; as well as samples with inappropriate material for analysis via polymerase chain reaction (PCR), such as microtissues or necrotic tissues. We also selected 24 patients with EBV-positive DLBCL and 25 patients with EBV-negative DLBCL as control groups, all of which were not on active therapy by immunosuppressants. All EBVMCU, EBV-positive DLBCL and EBV-negative DLBCL cases were reviewed independently by three pathologists (TI, TY, and YS) to confirm the diagnosis and immunophenotype.

This study was approved by the Institutional Review Board of Okayama University.

Histological examination and in situ hybridization

Specimens were fixed in 10% formaldehyde and embedded in paraffin. Serial, 3-μm thick sections were cut from paraffin-embedded tissue blocks for staining procedures. Sections were stained with hematoxylin and eosin or were immunohistochemically stained with antibodies specific for CD3 (clone: LN10, 1:200; Novocastra Laboratories, Ltd, Newcastle upon Tyne, UK), CD5 (clone: 4C7, 1:100; Novocastra Laboratories, Ltd.), CD10 (clone: 56C6, 1:100; Novocastra Laboratories, Ltd.), CD15 (clone: Carb-3, 1:50; DAKO, Glostrup, Denmark), CD20 (clone: L26, 1:100; DAKO), CD30 (clone: Ber-H2, 1:40; DAKO), CD79a (clone: JCB117, 1:50; DAKO), and Ki-67 (clone: MIB-1, 1:2500; DAKO). Staining was performed using automated Bond Max Stainer (Leica Biosystems, Wetzlar, Germany) according to manufacturer’s instructions. Immunoglobulin kappa or lambda light chains (Igκ or Igλ) were detected by in situ hybridization with Kappa and Lamda probes (PB0645, PB0669, respectively; Leica Biosystems) using automated Bond Max Stainer (Leica Biosystems). EBV was detected by in situ hybridization for EBV-encoded small RNA (EBER, fluorescein-conjugated oligonucleotide probe: PB0589; Leica Biosystems) using automated Bond Max Stainer (Leica Biosystems).

PCR assays of the IGH locus and TCR gene rearrangements

Tissue sections were scraped from the lesion and placed in AmpliTaq Gold Buffer (Applied Biosystems, Inc., Foster City, CA, USA). DNA was extracted by incubating at 94 °C for 45 min in an automated GeneAmp PCR System 9700 thermocycler (Applied Biosystems, Inc.). DNA was quantified using NanoDrop ND1000 spectrophotometer (Thermo Fisher Scientific, Inc., Waltham, MA, USA). All gene rearrangement analyses were performed by a method that has been previously described [12]. All primers were purchased from Sigma-Aldrich (Sigma-Aldrich Japan, Tokyo, Japan). PCR products were analyzed using ABI PRISM 310 Genetic Analyzer with GeneScan Analysis and GeneMapper software (Applied Biosystems, Inc.) [12]. For IGH rearrangements, DNA was amplified using Framework Region II and III primers. JH consensus primer was fluorescently labeled (6-carboxyfluorescein) [12]. For EBVMCU cases, additional IGK rearrangement assays were performed to investigate rearrangements involving Vk loci. IGH, IGK, IGL, and TCR gene rearrangements were analyzed and evaluated using BIOMED-2 protocol [12]. The results of fragment analysis were interpreted as monoclonal, oligoclonal, or polyclonal. If the exponential amplification of PCR is a single high peak, cell samples are monoclonal. In the same way, two high peaks indicate oligoclonal samples and multiple peaks indicate polyclonal samples. If peaks were not visible, the samples were deemed to have an undetectable expression.

Statistical analysis

Statistical differences between EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL were determined using the Mann–Whitney test and the Chi-squared (χ2) test. P < 0.05 was considered statistically significant. All statistical analyses were performed using SPSS for Windows software version 14.0 (SPSS Inc., Chicago, IL, USA).

Results

Clinical features

The clinical features of 34 EBVMCU patients have been summarized in Table 1. The median age of the 34 cases was 73 years (range: 54–91 years). The male-to-female ratio was 0.89:1. A total of 30 patients were treated with MTX for rheumatoid arthritis (RA) or polymyalgia rheumatica. One patient was treated for polycythemia using hydroxycarbamide. Further, one patient was treated with tacrolimus (TAC) as an immunosuppressant in addition to MTX. Another three patients were not on active immunosuppression therapy. The median soluble interleukin 2 receptor (sIL-2R) value was 652 U/mL (range: 263–2786 U/mL, n = 22). The median lactate dehydrogenase (LDH) value was 212 U/mL (range: 151–397 U/mL, n = 28) and the median serum albumin (SA) value was 3.8 g/dL (range: 2.9–4.6 g/dL, n = 25). All patients had mucosal or cutaneous ulcers with no apparent mass lesions [gingiva (n = 13, 38%), tonsil (n = 7, 21%), pharynx (n = 3, 9%), tongue (n = 2, 6%), oral cavity (n = 3, 9%), buccal mucosa (n = 1, 3%), nasal cavity (n = 2, 6%), and skin (n = 3, 9%)] (Fig. 1). Multiple ulcers were identified in the oral mucosa or bilateral tonsils in three cases (Case Nos. 4, 13, 15). Patients did not have lesions in other regions.

Table 1 Clinical and pathological findings of patients with Epstein–Barr virus-positive mucocutaneous ulcer.
Fig. 1: Macroscopic findings of Epstein–Barr virus-positive mucocutaneous ulcer in buccal mucosa (Case No. 31).
figure 1

The ulcer appeared while the patient was undergoing methotrexate treatment (a). After reducing methotrexate, the lesion spontaneously disappeared (b).

After the diagnosis, MTX or hydroxycarbamide was discontinued in 29 patients. In 28 patients (97%), lesion remission occurred and treatment was discontinued. No additional chemotherapy was found to be necessary. Only one patient did not present with resolved lesions. This patient then received chemotherapy (R-THP-COP; combined rituximab, pirarubicin, cyclophosphamide, vincristine, and prednisolone) and had complete remission (Case No. 8). Since the disease concept of EBVMCU was not established at the time of diagnosis, one patient was treated with chemotherapy without reducing MTX (Case No. 4). For the same reason, another one patient with no history of immunosuppression was treated with chemotherapy (R-CHOP; combined rituximab, cyclophosphamide, vincristine, adriamycin, and prednisolone) and was in complete remission (Case No. 7).

Among the 24 patients with EBV-positive DLBCL, the median age was 77 years (range: 33–94 years) and the male-to-female ratio was 3:1. One patient had a nodular skin lesion with no ulcer and another 23 patients had no mucocutaneous lesions. The median sIL-2R value was 4585 U/mL (range: 460–18600 U/mL, n = 22), the median LDH value was 265 U/mL (range: 146–903 U/mL, n = 22) and the median SA value was 3.3 g/dL (range: 2.0–4.9 g/dL, n = 17). Of the 17 patients who received chemotherapy, four exhibited complete remission after treatment and eight patients died of the primary disease.

Among the 25 patients with EBV-negative DLBCL, the median age was 73 years (range: 33–92 years) and the male-to-female ratio was 1.5:1. The median sIL-2R value was 2392 U/mL (range: 199–18832 U/mL, n = 25), the median LDH value was 277 U/mL (range 142–1576 U/mL, n = 25), and the median SA value was 3.9 g/dL (range: 2.8–4.4 g/dL, n = 20). Of the 15 patients who received chemotherapy, nine exhibited complete remission after chemotherapy and three died of the primary disease. We have shown the clinical features of EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL in Table 2 and Fig. 2. The median follow-up durations were 20.4, 30.9, and 30.3 months for patients with EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL, respectively.

Table 2 Comparison of laboratory findings in EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL.
Fig. 2: Comparison of sIL-2R and LDH in EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL.
figure 2

a Differences and distribution of sIL-2R in EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL. b Differences and distribution of LDH in EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL. Box plot explanation: upper horizontal line of box, 75th percentile; lower horizontal line of box, 25th percentile; horizontal bar within box, median; upper horizontal bar outside box, 95th percentile; lower horizontal bar outside box, 5th percentile.

Histological and immunohistochemical features

EBVMCU presented with localized mucosal or cutaneous ulcers. The disease is characterized by the presence of atypical lymphoid cells of various sizes and is accompanied by dense polymorphic infiltration with variable inflammatory cells such as plasma cells, histiocytes, and granulocytes. The large cells often resemble HRS cells. Angioinvasion was observed in 14 of the 34 EBVMCU patients (Fig. 3). Atypical lymphoid cells did not infiltrate into the epithelium. Accordingly, we classified EBVMCU into the following morphological types:

Fig. 3: Pathologic findings of angioinvasion in EBVMCU.
figure 3

Various sized atypical lymphoid cells infiltrating the wall of blood vessels. These atypical lymphoid cells are positive for EBER. (HE, EBER, ×400).

Polymorphous

Polymorphous refers to cases with various small to large atypical EBER-positive lymphoid cells. Some cases include few HRS-like cells. Atypical lymphoid cells are found in various densities such as dense and scattered. This type is often associated with necrosis and angioinvasion. Of the 34 EBVMCU patients, 20 cases (59%) were classified as polymorphous (Fig. 4).

Fig. 4: Pathologic findings of polymorphous EBVMCU.
figure 4

a A gingival ulcer of a 71-year-old female undergoing methotrexate treatment (Case No. 28). Atypical polymorphous lymphoid cells with few HRS-like cells are seen with granulocytes (×400). Atypical lymphoid cells infiltrated and destroyed the blood vessel wall. The atypical cells infiltrating the vessel wall are positive for CD20 and EBER. The background consists primarily of numerous CD3-positive cells without atypia (b: HE, CD20, CD3, EBER, ×200).

Large cell-rich

The atypical lymphoid cells primarily consist of large and monomorphic, with dense proliferation observed similar to those in monomorphic proliferation DLBCL. Seven of the 34 EBVMCU patients (21%) were classified as large cell-rich (Fig. 5).

Fig. 5: Pathologic findings of large cell-rich EBVMCU.
figure 5

A gingival ulcer of an 85-year-old female resembling DLBCL (Case No. 27). The lesion shows a monomorphic and dense proliferation of atypical large lymphoid cells. In situ hybridization shows atypical cells positive for EBER. (HE, EBER, ×400).

Classic Hodgkin lymphoma (CHL)-like

This type consists of many HRS cells and various sized atypical lymphoid cells. The aggregation of CD30-positive HRS cells is highlighted similar to those in CHL, however EBER-positive small to medium sized atypical lymphoid cells are also seen. Some cases also contain epithelioid granulomas or eosinophil infiltration. Of the 34 EBVMCU patients, four cases (12%) were classified as CHL-like (Fig. 6). Interestingly, two cases of these cases exhibited spontaneous regression. This clinical outcome differed from the CHL type of other iatrogenic immunodeficiency-associated LPD [13].

Fig. 6: Pathologic findings of CHL-like EBVMCU.
figure 6

A tonsillar ulcer of an 83-year-old male undergoing methotrexate treatment (Case No. 26). a Lymphoid cells infiltration with epithelioid granuloma observe under the ulcer (×100). b This lesion includes small to large sized atypical lymphoid cells and many HRS cells with epithelioid granuloma (×200). c Hodgkin cell. d Reed–Sternberg cell with epithelioid cells. The HRS cells and other polymorphous atypical lymphoid cells are positive for EBER. The HRS cells are also positive for CD30 (C, D, CD30, ×400, EBER, ×200).

Mucosa-associated lymphoid tissue (MALT) lymphoma-like

The atypical lymphoid cells show small to medium sized, centrocytic-like features, and/or plasmacytic features, which proliferate in the expanded interfollicular zone. One case showed mature plasmacytic differentiation with Russell bodies (Fig. 7). All cases exhibited light chain restriction and had no lymphoepithelial lesions. Of the 34 EBVMCU patients, three (9%) were classified as MALT lymphoma-like, two of which exhibited spontaneous regression.

Fig. 7: Pathologic findings of MALT lymphoma-like EBVMCU.
figure 7

A lingual ulcer in a 66-year-old female undergoing methotrexate treatment (Case No. 14). a This lesion shows ulceration and dense lymphoid cell proliferation (×40). b The atypical lymphoid cells show centrocytic-like feature with plasmacytic differentiation proliferating in the expanded interfollicular zone (×100). c Atypical plasmacytic cells with Russell bodies are seen (HE). The atypical lymphoid cells are positive for EBER. (EBER, ×400).

Immunohistochemistry

Positive CD3, CD5, CD10, CD20, CD30, and EBER staining was observed in 0% (0/34), 0% (0/12), 10% (1/10), 94% (32/34), 92% (11/12), and 100% (34/34) of the atypical lymphoid cells in samples, respectively. Two cases were negative for CD20 and positive for CD79a.

Atypical B cells exhibited diffuse or scattered infiltration primarily in subepithelial tissue. The background consisted mainly of numerous CD3-positive T cells (Fig. 4). Similarly, in EBV-positive DLBCL, B cells exhibited a wide range of infiltration patterns, from diffuse to scattered. CD3-positive T cells did not clearly form a rosette around HRS-like cells.

All three cases of MALT lymphoma-like EBVMCU showed Igκ-monotype by in situ hybridization.

Molecular features

Table 3 shows the IGH and TCR rearrangements. In patients with EBVMCU, IGH (FR2) PCR analysis was successful in 23 cases and monoclonality was detected in four patients (17%). IGH (FR3) PCR analysis was successful in 32 cases and monoclonality was detected in 13 patients (41%). In summary, IGH rearrangement was detected in 14 of the 32 patients (44%). TCR rearrangement was detected in nine of the 28 patient cases (32%) that were successfully analyzed by PCR.

Table 3 Comparison of molecular findings in EBVMCU, EBV-positive DLBCL, and EBV-negative DLBCL.

We also investigated light chain rearrangements in EBVMCU. As summarized in Table 3, IGK monoclonality was detected in ten of the 25 patients (40%) that were successfully assayed for IGK by PCR. Of the seven patients with a successful IGL PCR analysis, six (86%) showed an IGL rearrangement.

Table 3 also shows the results of EBV-positive DLBCL and EBV-negative DLBCL cases: EBV-positive DLBCL [IGH: monoclonal/oligoclonal (32%, n = 7), polyclonal (68%, n = 15); TCR: n = 20, monoclonal/oligoclonal (10%, n = 2), and polyclonal (90%, n = 18)]. EBV-negative DLBCL [IGH: monoclonal/oligoclonal (58%, n = 14), polyclonal (42%, n = 10); TCR: n = 20, monoclonal/oligoclonal (15%, n = 3), and polyclonal (85%, n = 17)].

We then compared the IGH rearrangement in EBVMCU, EBV-positive DLBCL and EBV-negative DLBCL, and observed that the IGH rearrangement occurred in 44% (14/32), 32% (7/22, P = 0.377) and 58% (14/24, P = 0.280) of the cases, respectively. In addition, the TCR rearrangement occurred in 32% (9/28), 10% (2/21, P = 0.060) and 15% (3/20, P = 0.176) of the cases, respectively.

Discussion

EBVMCU was first described by Dojcinov et al. in 2010 as a distinct clinicopathological entity occurring in immunosuppressed patients demonstrating either age-related immunosenescence or iatrogenic immunosuppression [1]. Before EBVMCU was defined, several reports described mucosal ulcerations occurring as a side effect of MTX therapy. However, most of these reports did not perform histological evaluation [14]. Soon after, EBVMCU was reported in patients with primary immunodeficiencies [3, 4], solid organ or bone marrow transplant recipients [5,6,7,8] and in patients with HIV/AIDS [9]. EBVMCU was later described as a new disease type by the World Health Organization [11]. Here, it should be noted that other disease concepts, such as MTX-LPD and EBV-positive marginal zone lymphoma (MZL), overlap with EBVMCU [15, 16]. Given that most EBVMCU do not require chemotherapy, we believe that prioritizing diagnosis of EBVMCU over other immunodeficiency-associated LPD will benefit those patients with overlapping disease concepts.

In this study, we investigated the clinicopathological differences as well as the IGH and TCR gene rearrangements between EBVMCU and DLBCL. Overall, more female patients developed EBVMCU, which may be related to the fact that diseases involved in RA tend to be more common in women [17]. Our data showed that EBVMCU did not increase lymphoma prognosis indicators, such as sIL-2R and LDH [18], compared to that in EBV-positive DLBCL and EBV-negative DLBCL. Since lesions in EBVMCU are local, patients tend to show relatively low sIL-2R or LDH levels. Particularly, the level of sIL-2R exhibited significant differences (Fig. 2). However, it is necessary that the clinical significance of LDH and sIL-2R levels be ultimately entrusted to each clinician. Of note, many EBVMCU patients had autoimmune diseases that caused chronic inflammation. Therefore, hemoglobin and SA values were not found to differ significantly between EBVMCU and EBV-positive DLBCL, EBVMCU and EBV-negative DLBCL cases.

In this study, one patient with age-related EBVMCU received chemotherapy. The lesions in 24 patients with iatrogenic EBVMCU achieved complete or partial remission after discontinuation of the immunosuppression therapy. One patient suffered relapse after hydroxycarbamide readministration. These results indicate that EBVMCU relapse is associated with repeated immunosuppression. Given that patients with EBVMCU generally reach complete remission, we suggest that further long-term observations may be required. Moreover, even if the ulcers improved, EBVMCU often destroys normal tissues, making clinical complete remission difficult to determine.

Herein, we classified EBVMCU into four morphological types as described above. It is necessary to be aware of this histological variety in EBVMCU to accurately distinguish it from lymphomas.

In the best of our knowledge, MALT lymphoma-like EBVMCU has not been previously reported. However, some reports have proposed EBV-positive MZL occurring in immunosuppressed patients [15, 16]. These reports found that most patients had clinically indolent disease with response to reduced immunosuppression. In this regard we consider the disease concept of EBV-positive MZL to overlap with MALT lymphoma-like EBVMCU. However, clinical findings differ between MALT lymphoma-like EBVMCU, and EBV-positive MZL. For instance, in previous reports, EBV-positive MZL was described as being extraoral with non-ulcerative tumoral lesions, while MALT lymphoma-like EBVMCU is characterized by oral and ulcerative lesions. Considering the clinical findings, MALT lymphoma-like EBVMCU should be distinguished from EBV-positive MZL.

Few previous reports have investigated the clonality in EBVMCU. Dojcinov et al. first reported that 38% of the cases exhibited IGH rearrangements and 31% exhibited TCR rearrangements, however, there were no control groups included in this study [1]. In 2011, relatively low frequency of the clonal IGH rearrangements were described in patients with age-related EBVMCU compared to those in EBV-positive DLBCL and EBV-negative DLBCL [19].

In the current study, the clonal IGH rearrangements in EBVMCU and EBV-positive DLBCL patients tended to occur less frequently than in patients with EBV-negative DLBCL (44, 32, 58%). However, there were no significant differences between the three groups. Thus, IGH rearrangements are not useful for distinguishing between EBVMCU and DLBCL.

Previous reports have shown that most EBV-positive cases exhibit polyclonal multiplication [13, 20]. EBV-positive DLBCL showed a relatively low frequency of clonal IGH rearrangements, although this may be due to possible false-negative PCR results, which can be attributed to the presence of EBV-positive blastocytes. Since EBV-positive cases tend to have a low IGH monoclonality, the IGH rearrangement was also found to be of low frequency in EBV-positive DLBCL [13, 20].

A previous report also demonstrated that serum CD8-positive T cells were elevated after MTX reduction [21]. Meanwhile, another report showed that B-cell posttransplant LPD were associated with clonal expansion of CD8-positive T cells [22]. Thus, we considered that EBVMCU also may be associated with T-cell clonal expansion as a result of reduced immune surveillance. Another report showed the presence of TCR beta chain variable restriction in more than half of the elderly patients [23]. In our study, clonal TCR rearrangements in EBVMCU tended to occur more frequently than in EBV-positive DLBCL and EBV-negative DLBCL cases (32, 10, 15%). Our data, therefore, supports that T-cell clonal expansion in EBVMCU is associated with immunosuppression, as with other immunodeficiency-associated LPD [21, 22]. In addition, T-cell clonality does not indicate EBVMCU or the possibility of T-cell lymphoma.

While T-cell clonality was observed, the HRS-like cells may not have an immune evasion mechanism. For example, previous reports have shown the presence of PD-L1 in most EBV-positive DLBCL cases [24,25,26]; however, PD-L1 was absent in all EBVMCU cases [27]. These results suggest that there may be no immune evasion mechanism as seen in EBV-positive DLBCL.

In summary, EBVMCU in the Japanese population represents a wide pathological spectrum, similar to that exhibited by neoplastic lesions, that includes polymorphic LPD, sometimes, DLBCL, CHL, and MALT lymphoma. Furthermore, the IGH rearrangement in EBVMCU has been confirmed, though at a lower frequency than that observed in EBV-negative DLBCL. Regardless of the histological feuture and the IGH rearrangement, EBVMCU shows good prognosis.

In our study, EBVMCU was histologically and genetically difficult to distinguish from lymphoma, suggesting that the collection of clinical information, particularly that related to medical history, lesion location, and laboratory data may lead to more accurate diagnosis.