This is the first Brazilian study to investigate the morphological and immunohistochemical aspects of Nodular lymphocyte predominant Hodgkin lymphoma/B-cell lymphoma (NLPHL/NLPBL) and its typical growth patterns and variants. Our main finding is the applicability of these patterns in Brazilian cases, with a high reproducible categorization of typical versus variants even by general pathologists (Cohen’s kappa = 0.894). To the best of our knowledge, this is also the first report to assess inter-rater reliability for separating typical versus variant growth patterns of NLPHL/NLPBL.
We include 50 cases from 45 patients seen in a single cancer center between 2014 and 2022. Our cohort of NLPHL/NLPBL is composed predominantly by men (64.0%) in the fourth decade of life (mean age 38 years), with a wide age distribution (between 6 and 82 years), compatible with the epidemiological profile described in international studies[10, 12]. The authors did not identify any Latin American or Brazilian studies indexed for comparison in PubMed/MEDLINE or LILACS/bvs databases.
The distribution of typical and variant primary patterns is similar to the original study by Fan et al. (Table 1), demonstrating the predominance of typical primary patterns (27/47, 57,4%) over variant patterns[8]. We also noted a higher frequency of the "E" pattern among variant patterns (9/20, 45,0%), in addition to the rarity of the "F" pattern (not identified in our study) [8]. However, the German Hodgkin Study Group (GHSG) included 413 cases and identified a considerably higher proportion of typical patterns (78% vs. 57.4%). The "D" pattern was the most frequent among the variants, followed by the "C" pattern. Pattern “E” was only seen in 11 cases[9].
A high reproducibility in discerning typical and variant patterns is an important factor for the applicability of such groups in routine diagnosis. The Cohen's kappa coefficient of 0.894 among general pathologists for reporting patterns "A" or "B" versus "C", "D", "E", or "F" reflects the ability of different observers to reach the same conclusion with regards to the presence of neoplastic cells in nodular or diffuse arrangements and a microenvironment predominantly of B-cells (in typical cases) or T-cells (in most variant cases). It may nevertheless be possible to reliably convey the adverse prognostic significance of identifying any variant pattern. Also, looking at only CD20 from low power is rapid and usually enough to classify most patterns, with an FDC marker helping most in cases where nodularity by CD20 is not prominent. The ability of different pathologists to quickly identify and report this additional parameter might facilitate its adoption in clinical judgment by a greater number of hematologists.
There are three cases with major disagreement in our cohort (3/48, 6,25%) when considering primary pattern assignment. In two of them, there was disagreement concerning the extent of the reactive T-cell area in which the neoplastic cell component needs to be embedded to characterize a diffuse area ("E" pattern) or just an extensive internodular area ("C" pattern). Case 33 is one of them and illustrates the additional difficulty in making this decision in core needle biopsies (Fig. 1). Also, the vague delineation of nodular areas with reactive B cells in this case led to the interpretation of a secondary "A" pattern by the second observer. Upon reevaluation of the case, such areas were considered more compatible with an exclusive "C" pattern.
Case 37 shows the architectural complexity that can be better appreciated in excisional biopsies, with the coexistence of several patterns. The presence of "A" pattern areas was initially noted by both observers, but with disagreement regarding their classification as primary or secondary pattern. Another disagreement was seen regarding the presence of a "C" or "E" pattern. The reevaluation of the case considered the presence of both. Neoplastic cells in interfollicular areas were evident, but there was an extensive focus composed neoplastic cells in a diffuse distribution and with a background rich in reactive T cells, considered as sufficient for the diagnosis of pattern "E" by one of the expert hematopathologists. It is emphasized that there is no objective measure for this distinction, which is still based on the experience and final interpretation of the pathologist or a group of pathologists.
The major divergence in case 42 occurred, again, in a core needle biopsy. One of the observers considered the extent of the changes insufficient to define the presence of a variant pattern. However, the well-defined foci of neoplastic cells forming nodules with virtually no reactive B cells would characterize at least a "D" pattern as classified by the second observed and confirmed during reassessment. Interestingly, in this case, CD21 and CD23 were negative, demonstrating an absence of follicular dendritic cell meshwork. This is a rare finding, which deviates from the original description of the "D" pattern and was evidenced later in the report of 4 cases[8, 13]. The four cases in the literature showed more aggressive behavior, consistent with the clinical evolution of a variant growth pattern and are likely part of the continuum between NLPHL/NLPBL and THRLBCL[13]. Given the prognostic significance of reporting the presence of variant patterns, it was considered important to state this finding clearly, even in small samples. Therefore, in cases of incisional biopsies, it is preferable to corroborate the pattern with a second observer. Any doubts should be emphasized by requesting excisional lymph node samples.
The reliable interobserver identification of growth patterns aids the proper choice of differential diagnoses for each morphological picture. As pointed out by Tousseyn et al. the "A" and "B" patterns bring as main differential diagnoses non-neoplastic lesions–such as progressive transformation of germinal centers and neoplastic lesions with a wide spectrum of clinical aggressiveness, from lymphocyte-rich classical Hodgkin's Lymphoma (CLHL) to nodal TFH cell lymphoma (nTFHL), follicular type[10]. On the other hand, the "D-F" patterns bring as diagnostic considerations THRLBCL and nTFHL, including the angioimmunoblastic type and the type not otherwise specified (NOS)[10].
Differential diagnosis with lymphocyte-rich CHL begins with a detailed morphological evaluation. LP cells are morphologically distinct from the neoplastic cells of CHL, showing large, multilobated nuclei with pale chromatin. Some cases may demonstrate prominent nucleoli, but it is less evident and not as strongly eosinophilic as that of a Hodgkin or Reed-Sternberg cell. In our study, 16 cases (16/47, 34.0%) demonstrated binucleated neoplastic cells, simulating Reed-Sternberg cells, but all cases still showed irregular and multilobed nuclear membranes. The pathologist's attention to morphological details may increase the degree of suspicion of a NLPHL/NLPBL and the selection of appropriate immunohistochemical markers for its characterization.
The immunophenotype of LP cells is well established in the literature. Positivity for several B-cell markers contrasts with CHL, which loses the normal B-cell expression program and shows only weak or, in rare cases, negative PAX5[14, 15]. Our findings were consistent with the English language literature, with frequent positivity for CD20 (50/50, 100%), PAX5 (37/40, 92.5%), CD79a (8/11, 72.7%), OCT2 (6/10, 60%) and BOB1 (4/4, 100%)[16]. However, many cases from our institution were not investigated with a broad panel of B-cell markers. Most were evaluated by the combination of positivity for CD20 and negativity for CD30 and CD15 to differentiate them from CHL. The second most used B-cell marker for distinction from CHL, after CD20, was CD79a (11/50, 22%). This marker was positive in almost all cases in which it was employed and allowed proper characterization of NLPHL/NLPBL. Nevertheless, the current recommendation is that the panel should include additional markers, especially OCT2, in addition to CD79a[10]. OCT2 is the most reliable marker for this characterization, being positive even in cases with negative CD20[10]. CD79a is also frequently preserved, although it may show attenuation of its expression like what is seen with PAX5 [17]. Given the variability of B-cell markers that may be lost in rare cases, it is critical to perform a broad characterization of the B cell immunophenotype. To the best of our knowledge, differences in these markers between typical and variant patterns have not been reported, and our study also did not identify any obvious trend.
Furthermore, the immunophenotype of LP cells is compatible with the molecular characterization of a clonal germinal center cell with active and aberrant somatic hypermutation. All our cases showed strong positivity for BCL6 in the LP cells, including in variant morphological patterns (Fig. 6). This feature aids in the differential diagnosis with CHL and with nTFHL, both of which demonstrate weak/absent positivity in neoplastic cells in CHL, or positivity in immunoblasts induced by neoplastic T-cells in nTFHL[18, 19].
Another important aspect of the evaluation of the neoplastic cells is the presence of criteria for transformation into aggressive/large B-cell lymphomas (LBCL). According to WHO-HAEM5, transformation should be considered when there are confluent sheets of atypical B-cells (in which case one should consider Diffuse large B-cell lymphoma) or scattered atypical cells without, however, characterizing any of the previously described variant patterns, often associating with loss of nodular architecture, FDC meshwork, and increased numbers of reactive T-lymphocytes and histiocytes (thus considering transformation to THRLBCL)[6]. With regards to the definition of a “diffuse pattern” of neoplastic cells, there is no well-defined cut-off point as to the extent from which clusters of neoplastic cells become large enough to characterize "confluence of a sheet of atypical B-cells"[20, 21]. This may hamper the reproducibility of such an important distinction to make, as major therapeutic and prognostic decisions differ when transformation to LBCL is diagnosed[11]. Considering the initial evaluation by a general pathologist and not an expert hematopathologist, we consider important to attempt standardization by arbitrarily defining “clusters of neoplastic cells” as the presence of 5–10 cells with no reactive lymphocytes between them. Thus, we identified 2 cases that met this criterion, both demonstrating variant patterns. In both, the clusters are not confluent, and it is the authors' consensus that there is no transformation to LBCL. Because of the scarce description in the literature of borderline cases that show a few clusters, but confluent sheets of neoplastic cells cannot be confidently diagnosed, the prognostic significance and reproducibility of this finding alone in cases of NLPHL/NLPBL with variant patterns remains uncertain.
Morphology of the tumor microenvironment should also be taken into consideration when characterizing NLPHL/NLPBL. We noted epithelioid histiocytes in more than half of the cases, predominantly as isolated cells (16/28, 57.1%). Granuloma formation around neoplastic nodules is well described, and we observed 5 cases with this finding[10]. Two cases also demonstrated intense sclerosis delimiting nodules, which may pose an additional challenge in the differential diagnosis with CHL, nodular sclerosis[8, 22]. Noting these morphological features is important so that the differential diagnosis of NLPHL/NLPBL is not disfavored.
Characterizing the immunophenotype of the cells that compose the lymphoma microenvironment is essential for the diagnosis of NLPHL/NLPBL and its patterns. We highlight B- and T-cell markers for pattern characterization, TFH or T CD8+ markers for differential diagnosis, and FDC markers for both purposes.
Notably, we did not identify any cases with a variant "F" pattern. The absence of this pattern may represent its rarity that requires larger cohorts to be detected, or the difficulty in identifying it by the pathologists involved in this study. Part of this challenge may be related to the amount of background reactive B lymphocytes required to differentiate a "F" pattern from an "E" pattern. In pattern "E", there is a marked predominance of background reactive T-lymphocytes, but a minimal amount of background reactive B-lymphocytes is allowed, similarly to THRLBCL (less than 5% of the microenvironment, empirical cut-off point established by WHO-HAEM5)[6]. The "F" pattern, on the other hand, is described as rich in diffusely arranged reactive B-lymphocytes, without evident nodular or serpiginous areas that could correspond to the typical "A" or "B" patterns, respectively[8]. However, cases with few background reactive B-lymphocytes representing more than 5% of the tumor microenvironment but not delineating nodules remain with uncertain categorization.
The presence of PD1+ T-cell rosettes was observed in almost all cases that had PD1 immunohistochemistry available (25/27, 92.6%). The expression of PD1 in T-cell rosettes is a sensitive marker for NLPHL/NLPBL and is a desirable diagnostic criterion by the WHO-HAEM5, aiding mainly in the differential diagnosis with reactive conditions[6, 23]. In the study by Visser et al. whose aim was to characterize the tumor microenvironment of NLPHL/NLPBL in viable cells obtained from fresh material and analyzed by flow cytometry, positivity for PD1 in CD4+ T cells was seen in approximately 70% of NLPHL/NLPBL cases. In contrast, only 30% of reactive cases had PD1 positivity in such cells[24]. However, this sensitivity is reduced in cases with variant patterns, predominantly in the "E" pattern, in which partial loss of rosettes is described[25].
In variant patterns, the utility of PD1 is also reduced in the differential diagnosis with "de novo" THRLBCL, which may show areas with PD1+ T-cell rosettes. The initial description of using this marker to separate NLPHL/NLPBL from THRLBCL was not substantiated in the study by Churchill et al. who identified PD1+ rosettes in up to 40% of THRLBCL and CHL cases[23, 26, 27]. In our cases, rosettes are consistently present in the variant patterns. However, we did not compare NLPHL/NLPBL cases with possible differential diagnoses.
CD57 is another marker used to characterize T-cell rosettes. We found only 1 case in which CD57 highlights most of the cells in rosettes and noted a weak and heterogeneous labeling pattern in 10 cases (Fig. 7). Its sensitivity is lower than that of PD1 to identify TFH lymphocytes from the tumor microenvironment that interact directly with neoplastic LP cells, both in cases with typical and variant patterns[26].
Negativity for CD10 is a critical factor to support the distinction mainly with nTFHL[18, 28]. Although not very sensitive, CD10 expression on T lymphocytes is highly specific for the identification of nTFHL in the appropriate clinicopathological context, and its expression outside the germinal center should increase the level of suspicion for this group of entities[29–31]. Similarly, reactive TFH cells in the tumor microenvironment of NLPHL/NLPBL should not express this marker, aiding in the differential diagnosis with nTFHL. None of our cases demonstrated consistent CD10 expression (0/28, 0%).