Date Published: January 25, 2019
Publisher: Public Library of Science
Author(s): Francesco Carubbi, Alessia Alunno, Paola Cipriani, Gino Coletti, Barbara Bigerna, Mirko Manetti, Paola Di Benedetto, Onelia Bistoni, Gianluca Cipolloni, Vasiliki Liakouli, Piero Ruscitti, Elena Bartoloni, Roberto Giacomelli, Roberto Gerli, Silke Appel.
A standardization of minor salivary gland (MSG) histopathology in primary Sjögren’s syndrome (pSS) has been recently proposed. Although there is strong agreement that germinal center (GC)-like structures should be routinely identified, due to their prognostic value, a consensus regarding the best protocol is still lacking. Aim of this study was to compare the performance of different histological techniques and operators to identify GC-like structures in pSS MSGs. MSG biopsies from 50 pSS patients were studied.
Three blinded operators (one pathologist and two rheumatologists with different years of experience in pSS MSG assessment) assessed 50 MSGs of which one slide was stained with haematoxylin and eosin (H&E) and consecutive slides were processed to investigate CD3/CD20, CD21 and Bcl-6 expression.
By assessing 225 foci, the best agreement was between H&E-stained sections evaluated by the rheumatologist with more years of experience in pSS MSG assessment and CD3/CD20 segregation. In the foci with CD21 positivity, the agreement further increased. Bcl-6- foci could display a GC, detected with other staining, but not vice versa.
GC assessment on H&E-stained sections should be performed with caution, being operator-dependent. The combination of H&E with CD3/CD20 and CD21 staining should be recommended as it is reliable, feasible, able to overcome the bias of operator experience and easily transferrable into routine practice.
Primary Sjögren’s syndrome (pSS) is a chronic inflammatory disease mainly affecting exocrine glands . Minor salivary gland (MSG) biopsy is widely used in the classification criteria of pSS [2,3], and the histopathological hallmark of the disease is the focal lymphocytic sialadenitis (FLS), with sensitivity and specificity >80% . Aggregates of at least 50 infiltrating lymphocytes (namely foci) are often organized into tertiary ectopic lymphoid structures (ELS), showing segregated T- and B-cell zones, high endothelial venules and networks of CD21+ follicular dendritic cells (fDCs). As recently reviewed, some of these structures display functional features of classical germinal centers (GCs) such as the expression of the enzyme activation-induced cytidine deaminase (AID) and in situ B cell affinity maturation and clonal selection . The formation of ELS is not peculiar of pSS as they can be observed in other rheumatic diseases or organ specific autoimmune conditions, solid tumors, chronic infection and graft rejection . In recent years, several studies explored the presence and features of GC-like structures in MSGs and parotid gland of patients with pSS. From most of these studies, it emerged that the detection of GC-like structures in the lymphoid infiltrates of the salivary glands (SG) is clinically relevant, since the presence of these structures seems to be associated with more severe disease [6–8], and higher risk of lymphoma development [9, 10]. However, previous studies aimed at assessing GC-like structures in pSS reported a highly variable prevalence ranging from 18% to 67% based on different methods [6–8]. In particular, the peculiar structure of fully formed GCs, namely a well-distinguished light and dark zone segregation, detectable with haematoxylin and eosin (H&E) staining, is sufficient to allow pathologists to detect these structures in secondary lymphoid organs. However, when GCs ectopically develop in non-lymphoid tissues such as MSGs, this detection is more challenging since the classical dark/light zone may lack and only a lighter are within the follicle can be present. Therefore, additional GC-specific immunostainings, such as B-cell lymphoma (Bcl)-6, CD21 and AID, have been suggested to assess ELS in MSGs . Recently, the EULAR Sjögren’s syndrome study group (eSSential) provided standardized consensus guidance for the use of MSG histopathology in classification and clinical trials . Although there is strong agreement that the presence of GC-like structures should be reported in routine clinical practice, the consensus regarding which technique should be employed is lacking . The lack of consensus, in association with the difficulty to identify these structures and the variability of pSS samples used in the literature (i.e. parotid glands versus MSG samples), may explain the different prevalence of GC-like structures in pSS biopsies , and different associations found from both a pathological and clinical point of views [7, 9, 10, 13, 14]. The harmonization of GC-like structure assessment and, subsequently, the development of recommendations to be used in clinical practice are compelling as proven by the ongoing intense debate in the literature [11, 15, 16]. On this basis, the purpose of our study was to compare, for the first time, the performance of different histological techniques and operators with variable histopathology expertise in the assessment of GC-like structures in MSGs of pSS patients.
A total of 225 foci from 50 MSGs were scored. The focus score (FS) ranged from 1 to 12 (namely confluent foci) with a mean value of 2.02±2.2 (standard deviation) and a median value of 1.45. The number of positive MSG samples, according to the presence of at least one focus scored as possible GC-like structure by the operators on H&E or by different histological techniques (alone or in combination), is shown in Table 1.
To our knowledge, this is the first study assessing the agreement and disagreement as well as the contribution of operators’ experience for the proposed procedures to identify GC-like structures, in a cohort of pSS patients. In fact, standardization of histopathological patterns in pSS MSGs, including the presence of GC-like structures, is still a major unmet need as confirmed by the high variability of GC-like structures prevalence, detected by each of the currently available techniques [11, 15, 16]. Overall, the prevalence of GC-like structures in MSG specimens (namely the presence of at least one focus positive for at least one operator or histological technique) ranged between 26 and 52% according to the different operator and to the fact that different staining protocols identify different cell populations. When assessing each focus separately, the best agreement was observed between Op#2 and B/T cell segregation. In this regard, our study shows that the H&E scoring by a rheumatologist with over 5 years of experience in pSS MSG histopathology, ensured the identification of almost 80% of foci with B/T-cell segregation. When considering the co-existence of B/T cell segregation and CD21 positivity in the same focus, this agreement reached 86%. CD21 is expressed by fDCs, but the presence of these cells does not necessarily imply that GCs are present. In fact, lymphoid infiltrates in SG tissue of patients with pSS may contain fDC networks in the absence of GCs , with consequent overestimation of the number of GCs in SG tissue using only CD21 staining. On the basis of these observations, and according to Fisher et al. who recommended to perform H&E, CD3/CD20 and CD21 staining, our paper reinforces that H&E staining should not be performed alone, in order to prevent operator-dependent bias . Furthermore, the presence of fDC network together with B/T-cell segregation would be expected in all GC-like structures, and this combined approach would avoid the risk of overestimating GC-like structures if relying on CD21 alone . H&E is considered sufficient to allow accurate detection of a fully formed GC by experienced pathologists, in particular in normal secondary lymphoid organs or lymphomatous tissues, with a well-distinguished light and dark zone segregation. Conversely, in MSGs this detection is more challenging and GCs are often only appreciable as a lighter area within the follicular structure, without the classical dark/light zone segregation. In our study, we defined GCs in H&E-stained sections as lighter areas within the lymphoid infiltrate composed of both lymphoid cells (centrocytes, centroblasts) and cells of non-lymphoid nature (macrophages and fDCs), as previously described . Interestingly, we noticed that the pathologist tended to underestimate the number of possible GC-like structures compared to other operators, probably due to the lack of classical dark/light zone segregation. Conversely, the rheumatologist with over 5 years of experience may tend to overestimate. This may also explain the fact that in our study the pathologist is the operator with the highest agreement with Bcl-6 positivity. Bcl-6 is a transcription factor consistently expressed at high levels by GC B-cells. Although it is routinely used worldwide in pathology laboratories for the diagnosis of lymphoma, the applicability of staining protocols for such nuclear factor has not been extensively validated in other tissues . In our study, the agreement between Bcl-6+ staining, B/T-cell segregation and CD21 positivity was poor. Bcl-6 was positive in less than 40% of foci showing B/T cell segregation and CD21 positivity. To note, however, the foci showing this triple positivity were more frequently scored as GC-like structures by the pathologist compared to those where B/T-cells were segregated, CD21 staining was positive, but Bcl-6 was not detectable. Furthermore, isolated Bcl-6+ staining was detected in the 2% of foci where B- and T-cells were not segregated and CD21 staining was negative. Interestingly, these foci were scored as no GC-like structures by the three operators. Although Delli et al proposed Bcl-6 as an unequivocal marker of GC , our data pointed out that the absence of Bcl-6 does not necessarily rule out the presence of the GC. These observations, together with the well-recognized technical difficulties that may occur with nuclear staining protocol for Bcl-6, make this marker less likely to be considered for routine assessment of GC-like structures in MSGs in clinical practice. It should be noted, however, that our study is focused only on MSGs, being the sample routinely collected in the majority of Rheumatology centres. Hence, an additional value of Bcl-6 staining in parotid gland cannot be excluded. Finally, it is of interest the finding that activation-induced cytidine deaminase (AID) seems to be expressed if the GCs are functional . However, accurate staining for this marker is technically challenging and, therefore, its use appears to be confined, at the moment, only to research fields rather than routine clinical practice . Another aspect deserving consideration pertains to the possible morphological differences outlined at different cutting levels. We acknowledge that studies conducting for research purpose always include samples from different cutting levels. However, this is not always the usual procedure in real life setting for diagnostic purposes. To overcome this discrepancy, the definition of cutting levels as a standardised requirement for a satisfactory assessment of MSGs would facilitate the implementation of this aspect in daily practice.