How I Diagnose Primary Mediastinal (Thymic) Large B-Cell Lymphoma

Sarah L. Ondrejka, DO; German Ott, MD


Am J Clin Pathol. 2021;156(4):497-512. 

In This Article

Diagnostic Challenges and Questions for the Experts

Diagnosis of PMBL requires the incorporation of clinical, morphologic, and immunophenotypic and molecular information. Nonetheless, the lack of an accepted "common definition" for this entity and the variability in cytologic features can make interpretation challenging.[5] It has been suggested that cases with similarities to PMBL may arise outside the mediastinum, and GEP data have supported this notion, but these tests are not part of the routine diagnostic workup in most pathology practices.[53] The mediastinal location can pose practical problems in obtaining adequate tissue for diagnosis. Needle core biopsies may not be sufficient, especially if there is histologic variability throughout the tumor or significant fibrosis. Whether true mediastinal large B-cell lymphoma exists at nonmediastinal sites, without a mediastinal mass, is a subject of continued discussion. In situations in which a mediastinal biopsy may be too risky and a peripheral lymph node biopsy is taken, instead, knowledge of a dominant mediastinal mass and careful attention to the histologic features and staining characteristics of PMBL can help infer a diagnosis. It is important to recognize the difference between PMBL and B-cell lymphoma, unclassifiable, with features intermediate between DLBCL and CHL (hereafter referred to as MGZL). An accurate diagnosis of MGZL is challenging. In a consensus study of 68 cases submitted as MGZL by academic centers in the United States, an expert consensus review panel agreed with the diagnosis in only 38% of cases.[55] There remains more uncertainty about how MGZL should be treated because these cases tend to have an inferior outcome and higher stage compared with PMBL.[56] As such, close collaboration between pathologists and oncologists is recommended for completeness of clinical data and accuracy of the final diagnosis.

Question 1: Which Mediastinal Lymphomas Should be Considered in the Differential Diagnosis of PMBL?

The clinical differential diagnosis of an anterior mediastinal mass includes hematolymphoid neoplasms as well as thymoma, germ cell tumors, and carcinomas. These possibilities can often be distinguished on routine H&E sections.

Immunophenotypic information is helpful for excluding T-cell neoplasms occurring in the mediastinum, importantly, T-lymphoblastic lymphoma, which can occur in young men and shares an aggressive presentation, with symptoms attributable to mass effect. Thymic MALT lymphoma is rare and presents in the sixth or seventh decade, with a predominance in women and often an association with autoimmune disease. Typically, routine histologic features will discern thymic MALT lymphoma from PMBL because MALT lymphomas are composed of small to intermediate-sized B cells with centrocyte-like, monocytoid, and plasmacytic differentiation. There is infiltration of the thymic epithelium to form lymphoepithelial lesions Figure 7 but the monocytoid cytologic features might be confused with the cytoplasmic clearing present in PMBL, and poor fixation or sample integrity can introduce artifacts.[57]

Figure 7.

Thymic mucosa-associated lymphoid tissue lymphoma (A, ×20), with cytology (inset), positive for CD20 (B, ×20) and cytokeratin AE1/3 demonstrating altered thymic epithelium (C, ×20). A lymphoepithelial lesion (D, ×100).

The most frequent and important differential diagnostic considerations with PMBL are the nodular sclerosis type of CHL and MGZL Table 2, which involve the mediastinum in approximately 78% and 73% of cases, respectively.[58,59] Problems that may arise when considering this distinction are also discussed in question 2. The interrelationship of Hodgkin and non-Hodgkin lymphomas is well recognized,[60] and there are reports of B-cell lymphoma and CHL occurring as composite lymphomas as well as sequentially.[18,61] Given accumulating evidence of the relatedness between these entities and the fact that some mediastinal lymphomas demonstrate a true blurring of features of B-cell non-Hodgkin lymphoma and CHL, the term mediastinal gray zone lymphoma was conceived.[62] The 2008 WHO classification recognized this for the first time as "B-cell lymphoma, unclassifiable, with features intermediate between diffuse large B-cell lymphoma and classic Hodgkin lymphoma." MGZL cases could be categorized as those that have a morphology that resembles CHL but atypical features, such as increased numbers of large mononuclear cells, lack of the typical mixed inflammatory background, and strong expression of CD20 in the Hodgkin-like cells. Some cases of MGZL had diffuse monomorphous large cells with an atypical phenotype, including weak or absent CD20; positive CD15 staining; and other staining characteristics of CHL, such as weak expression of transcription factors.[62,63] Cases of composite CHL and PMBL, defined as the simultaneous presence of histologically distinctive CHL and PMBL components in the same tumor, have been reported and are not considered equivalent to or diagnostic of MGZL.[64] This finding emphasizes the importance of adequate tissue sampling because a small biopsy could be misleading and result in a diagnosis that does not take into account areas of different histology.

Despite these guidelines, accurate classification of MGZL is difficult even among experts. In a clinicopathologic consensus study, in which 42 of 68 cases submitted as MGZL were reclassified by central review, the majority of reclassified cases were interpreted to be nodular sclerosis CHL. A subset was considered CHL grade 2 with depleted small lymphocytes. In that study, most of the cases submitted as MGZL and reclassified as CHL had strong CD20 expression and relatively weaker CD30 expression, with an inflammatory background and large cells with cytomorphologic characteristics of Reed-Sternberg cells.[55] The largest series of MGZL at the time of this writing is a retrospective Lymphoma Study Association project whereby 86 cases with a CHL-like morphology and large B-cell lymphoma immunophenotype were compared with 53 cases of large B-cell lymphoma–like morphology and CHL immunophenotype. Despite the varied morphologic spectrum, genetic aberrations were similar between the 2 groups.[62] A recent GEP study comparing MGZL with PMBL and CHL found that the clustering of MGZL is distinct but more closely related to CHL than to PMBL, with downregulation of the B-cell program, higher T-cell and macrophage signatures, and higher NFκB activation.[65]

MGZL affects younger individuals (ages 20–40 years), with a predominance in men, and presents with a mediastinal mass but can harbor nodal involvement and may present at a higher stage. Thus, clinical characteristics are distinctive compared with PMBL. The outcome for MGZL is inferior to that of PMBL, even with dose-intensive therapy.[56]

Question 2: How Might a Pathologist Evaluate and Report CHL With CD20 Positivity?

Biologically, PMBL shares many similarities with CHL, including activation of the JAK-STAT and NFĸB pathways and immune evasion.[10,50] The genetic relationship is supported by the transcriptional signature of PMBL, in which more than one-third of genes that are highly expressed in PMBL were shared with CHL. These attributes are reflected in the histologic features and can cause confusion in diagnosis, particularly when CHL demonstrates a syncytial pattern (eg, in the nodular sclerosis variant). The fact that PMBL and CHL both express CD30 can be problematic, but CHL typically shows strong uniform staining rather than the weak and patchy staining described for PMBL. This separation, however, can be even more troubling when the HRS cells assume more of a Hodgkin-like appearance with monolobated nuclei and demonstrate expression of CD20 Figure 8.

Figure 8.

Classic Hodgkin lymphoma (A, ×500) positive for CD30 (B, ×500), inconsistent and weakly positive for CD20 (C, ×500), and positive for CD15 (D, ×500).

PMBL retains expression of a full panel of B-cell markers and transcription factors, including CD19, CD20, CD79a, BOB1, OCT2, PAX5, and PU.1. In contrast, CHL typically expresses weak PAX5 and an incomplete B-cell antigen and transcription program, with variable staining for the other markers and negativity for CD45. CD15 can be helpful when it is positive because it is more often a feature of CHL. The polymorphous inflammatory background composed of histiocytes, plasma cells, neutrophils, and eosinophils is a supportive and characteristic feature of CHL that is typically absent in PMBL. Although both of these lymphoma types exhibit a fibrotic stroma, CHL typically shows broad fibrotic bands of sclerotic collagen separating lymphohistiocytic nodules that contain variable numbers of large HRS cells. This is in contrast to the delicate, interstitial, and compartmentalizing fibrosis that occurs in PMBL. EBV is invariably negative in PMBL,[15] whereas a variable percentage ofnodular sclerosis CHL cases (27%-41%) demonstrates EBV positivity in the HRS cells.[66–70]

It is generally accepted that CD20 positivity in HRS cells, when supported by diminished expression of other B-cell markers and in the presence of the proper inflammatory background, is considered to be an immunophenotypic variant of CHL rather than PMBL.[63] In doubtful cases, expression of CD20 should be compared with that of other B-cell markers, such as CD19 and CD79a, and BOB1 and OCT2 staining should be considered. An adequate tissue biopsy is important for assessing the background histologic features; in some cases, a second biopsy is obtained for contextual interpretation of the transformed cells. Because chemotherapeutic regimens and prognoses differ between PMBL and CHL, it is important to be certain about their distinction when issuing a diagnosis.

Question 3: Which Clinical and Pathologic Data Will Help in the Separation of PMBL From DLBCL NOS?

A combination of clinical, imaging, and pathologic data is necessary to separate DLBCL NOS from PMBL. Some cases of DLBCL may include mediastinal involvement, with lack of a dominant mediastinal mass. There are cases of PMBL in which a dominant mass may be present, but the presence of high-stage disease at the time of diagnosis makes it challenging—if not impossible—to differentiate PMBL from DLBCL NOS. Staging will provide an assessment of disease spread, and an abundance of extrathoracic lymph node or bone marrow involvement favors DLBCL NOS, with a predilection for involvement of mediastinal lymph nodes rather than the thymic area.

Histopathology can be helpful. Features such as fine sclerosis, cellular pleomorphism, and clear cytoplasm are not exclusive to PMBL but fit nicely with the diagnostic interpretation. DLBCL NOS can show a heterogeneous spectrum, from the centroblastic nuclear features typical of germinal center–derived lymphomas to immunoblastic cytology and anaplastic variants, among others. Immunohistochemical expression of MAL, CD200, heterogenous/weak CD30, CD23, c-REL, and TRAF have been shown to distinguish PMBL from DLBCL NOS, as described in the earlier sections of this manuscript.

Interestingly, cases of large B-cell lymphoma with features of PMBL but without detectable mediastinal involvement have been described. In a large series of DLBCL cases, a subset of cases with a GEP signature of PMBL was identified. Pathology review showed that most of these cases were indeed consistent with PMBL, although the classifier also identified an MGZL as well as a subset consistent with DLBCL NOS, with no clinical or radiologic evidence of mediastinal involvement at presentation.[53] The question of how to recognize and interpret these nonmediastinal cases in the course of routine pathology practice without the assistance of GEP remains to be determined.