EBV⁺ diffuse large B-cell lymphoma associated with chronic inflammation expands the spectrum of breast implant–related lymphomas

TO THE EDITOR:

Breast implant–associated anaplastic large cell lymphoma (BI-ALCL) has emerged as a new provisional entity in the revised 2016 World Health Organization classification of lymphoid malignancies.¹  BI-ALCL is a rare T-cell lymphoma arising adjacent to breast implants and composed of large atypical CD30⁺ cells frequently confined to the peri-implant seroma fluid and adjacent capsule, more rarely forming a solid infiltrating mass.^2,3  So far, only exceptional cases of lymphomas other than BI-ALCL have been reported to occur in the vicinity of breast implants, including miscellaneous B-cell lymphomas.^4-7  It remains so far unclear whether these cases are coincidental or could be related to breast implants.

We report 3 cases of Epstein-Barr virus (EBV)⁺ diffuse large B-cell lymphomas (DLBCLs) occurring in contact with breast implants. These cases were also characterized by various degrees of invasion of the periprosthetic capsule but no tumor mass, which make them distinct from classical primary breast DLBCLs.^8,9  To our knowledge, no series of DLBCL adjacent to breast implants has been documented so far. This study was approved by the institutional review board of the Institut Paoli-Calmettes, and all patients gave their informed consent.

In the 3 patients (aged 61 to 72 years), the diagnosis was allowed by excision of the periprosthetic capsule due to esthetical issues in cases 1 and 3 or incidental positron emission tomography (PET) scanner during breast cancer surveillance in case 2 (Table 1). In all cases, the lymphoma tumor was strictly confined to the capsule surrounding breast implants (macrotextured type from Allergan), and the PET computed tomography finding was negative otherwise. The bone marrow biopsy result was also negative. No seroma had been observed prior to capsulectomy in any case, which prevented any fluid aspiration and cytologic analysis. None of the patients had any known immunodeficiency or pharmacologic immunosuppression.

Formalin-fixed and paraffin-embedded capsulectomy samples from the 3 cases were extensively characterized using histological, phenotypical, cytogenetic, and molecular analyses, including targeted next-generation sequencing (tNGS) and array comparative genomic hybridization (aCGH), as described in supplemental Materials and methods (available on the Blood Web site). Clinicopathological and biological features of the 3 cases are detailed in Table 1 and supplemental Tables 1 to 4 and illustrated in Figure 1 and supplemental Figure 1.

The 3 cases shared common pathological features consisting of sheets, clusters, and ribbons of large pleomorphic CD30⁺ EBV-infected B-cells, with EBER expression in virtually all lymphoma cells. The latency profile was type III (LMP1⁺/EBNA-2⁺) in 2 cases, whereas the remaining case was negative for both LMP1 and EBNA2. Postresection plasma EBV levels were positive in the 2 analyzed cases, with a decrease over time (supplemental Table 1). Lymphoma cells were observed on the luminal side of the capsule or suspended in a fibrinoid material with constant thickening and invasion of the capsule. This invasion formed cell aggregates or sheets within the capsule but without tumor beyond it or lymph node involvement. Plasmablastic differentiation was observed in 1 case, whereas cytological features in the 2 remaining cases were mostly immunoblastic, though uneasy to classify, especially within fibrin deposits (Figure 1). The latter pattern was somewhat reminiscent of BI-ALCL.^1,2  Variable amounts of inflammatory cells were present in all cases. They all harbored a major B-cell clone and were of non–germinal center cell of origin. These features could meet the criteria for either DLBCL associated with chronic inflammation (DLBCL-CI), or fibrin-associated DLBCL (FA-DLBCL).¹  DLBCL-CI, which are characterized as EBV-associated DLBCLs occurring in the context of chronic inflammation and developing in cavities or enclosed spaces, sometimes adjacent to a foreign body, have been initially described in the pleural cavity as pyothorax-associated lymphomas.¹  They have been more recently described in bones, joints, or periarticular soft tissues in association with metallic implants.^10,11  Our series suggests that breast implants could be added to the list of foreign bodies that could favor the onset of DLBCL occurring in the context of chronic inflammation.

Although prototypic DLBCL-CI like pyothorax-associated lymphomas were initially reported to present as aggressive mass-forming tumors,¹²  there is accumulating evidence that DLBCL-CI can exhibit variable presentation and clinical course.^1,13,14  Such particular variants include EBV⁺ DLBCLs arising within cardiac myxoma, which are regarded as DLBCL-CI with indolent behavior.¹⁵  Our 3 patients could represent such indolent DLBCL-CI, since the lesions were non–mass forming and strictly localized to the peri-implant capsule.

The clinical presentation of the 3 patients without tumor mass and the localization of lymphoma cells within fibrin deposits are also evocative of FA-DLBCL, another subset of EBV⁺ DLBCL. FA-DLBCL is a rare lymphoma presenting without clinical mass and consisting of aggregates of large EBV⁺ B cells within fibrin layers.^13,14,16  As a matter of fact, the diagnosis was incidental in the 3 patients, who had neither clinical nor biological symptoms.

However, our 3 cases displayed various degrees of capsule infiltration, a feature more suggestive of DLBCL-CI.¹  In addition, case 2 contained a MYC-immunoglobulin H (IGH) translocation, which is in accordance with the known occurrence of MYC alterations in DLBCL-CI, whereas they have not been so far reported in FA-DLBCL.¹⁷  In contrast, aCGH analysis in the 2 remaining cases did not detect any abnormality. Thus, it appears that our series displayed intermediate features between FA-DLBCL and DLBCL-CI, which fits with the emerging concept that FA-DLBCL could be regarded as an incidental form of DLBCL-CI.¹ 

Despite the T-cell origin and EBV negativity of BI-ALCL, which are major differences with BI-DLBCL, it has to be stressed that the differential diagnosis between the 2 entities may be a pitfall. The neoplastic infiltrate of the current 3 cases indeed exhibited both “fibrin-associated” and “infiltrative” patterns, which could mimic BI-ALCL.^2,3  Of note, patient 2 from the current series was initially suspected to present BI-ALCL in view of the clinical presentation together with a preliminary phenotype showing CD20 negativity associated with CD30, EMA, and CD4 positivity. The correct diagnosis could be eventually assessed owing to the demonstration of CD79a positivity and κ light-chain restriction (Table 1). Expression of T-cell markers, like CD4 and CD43 in 2 of our cases, is a well-known phenomenon in some EBV-associated DLBCLs such as DLBCL-CI and plasmablastic lymphomas, which can cause problems in lineage assignment.¹ 

Furthermore, our 3 cases harbored not only major clonal rearrangements of immunoglobulin genes but also minor rearrangements of T-cell receptor genes consisting of multiple small-sized peaks. Since the reactive T-cell infiltrate in each case was minimal (Table 1), this pattern could be due to either pseudoclonality or an oligoclonal/restricted T-cell response. Interestingly, this peculiar clonality profile has been already reported in other EBV-associated DLBCLs, likely reflecting clonal expansion of cytotoxic lymphocytes.^13,18,19 

As expected, tNGS analysis of our series could detect mutations in various genes known to be recurrently mutated in DLBCL, such as CREBBP, GNA13, TET2, and IRF4. Case 2 also carried STAT3 and SOCS1 mutations, which are common in BI-ALCL but also occur in DLBCL.^20-22  Thus, case 2 raises the possibility that the JAK-STAT pathway could favor B-cell lymphomagenesis in the context of BI. Environmental factors in the enclosed space formed by the peri-implant cavity might be also involved, like the IL-6 and IL-10 cytokines, which are suspected to favor DLBCL-CI.^23,24 

It is noteworthy that our patients had all macrotextured implants, which are considered as a risk factor for BI-ALCL and are currently subject to widespread limitation.^2,3,25  Thus, BI-related lymphomas are probably doomed to become extremely rare, though they will likely persist in the next few years. This implies that pathologists must stay aware of the diversity of these tumors.

An indolent clinical course was observed in our series, all patients being in complete remission after an average follow-up of 20 months, although 2 of our 3 patients received no treatment after surgery. Although this apparently indolent behavior has to be confirmed by a longer follow-up, it suggests that a “watch and wait” management of BI-DLBCL could be considered after complete tumor resection when there is no evidence of dissemination. Nonetheless, our small series makes it difficult to draw definitive conclusions, and additional cases are needed to complete the understanding of this rare, previously unrecognized subset of BI-related lymphomas.