Objectives The aims of this study were (1) to assess the effect of rituximab (RTX; anti-CD20) treatment in patients with primary Sjögren's syndrome (pSS) based on sequential parotid biopsies obtained in a placebo-controlled, randomised clinical trial, and (2) to assess the prognostic value of the histological characteristics of parotid gland tissue with regard to responsiveness to RTX treatment.
Methods In a double-blinded, placebo-controlled trial, sequential parotid gland biopsies were taken from 20 RTX-treated and 10 placebo-treated patients with pSS, at baseline and 12 weeks after treatment. The relative amount of lymphocytic infiltrate (stained for CD45), absolute number of T cells and B cells per mm2 parenchyma (stained for CD3 and CD20, respectively), focus score, number of germinal centres and of lymphoepithelial lesions per mm2 in parotid gland parenchyma were assessed. Histopathological data were compared between clinical responders (decrease in European League Against Rheumatism Sjögren's Syndrome Disease Activity Index (ESSDAI) score of ≥3 at 12 weeks compared with baseline) and non-responders (change in ESSDAI<3) to RTX treatment.
Results In RTX-treated patients, a significant reduction in the number of CD20+ B cells/mm2 parenchyma was observed, while no such reduction was observed in placebo-treated patients. The number of CD3+ T cells/mm2 in parenchyma did not change in either group. Furthermore, the number and the severity of lymphoepithelial lesions/mm2 and number of germinal centres/mm2 was significantly reduced in RTX-treated patients, but did not change in placebo-treated patients. When comparing the pretreatment characteristics of clinical responders with non-responders, the median number of CD20+ B cells/mm2 parenchyma at baseline was significantly higher in responders (1871 vs 353 cells/mm2, p<0.05). Other histopathological baseline characteristics were not predictive for response to RTX treatment.
Conclusions RTX treatment in pSS leads to a major reduction of lymphocytic infiltration and to fewer B cells, germinal centres and lymphoepithelial lesions in parotid gland parenchyma. A high pretreatment number of CD20+ B cells/mm2 parotid gland parenchyma predicts better responsiveness of patients with pSS to RTX treatment. Pretreatment parotid gland histopathological characteristics could therefore contribute to a more personalised treatment approach to pSS.
- Sjøgren's Syndrome
- B cells
- Disease Activity
- Outcomes research
Statistics from Altmetric.com
Primary Sjögren's syndrome (pSS) is a common rheumatic disease, with a prevalence of 60.8 (95% CI 43.7 to 77.9) cases per 100 000 inhabitants in the total population.1 pSS commonly affects salivary and lacrimal glands, resulting in a sensation of dry mouth (xerostomia) and dry eyes (keratoconjunctivitis sicca). Although the exact pathogenic mechanism has not been fully elucidated, in patients with pSS the minor and major salivary glands are characteristically infiltrated by mononuclear lymphoid cells, which form periductal foci. The classic glandular lesion is composed of a lymphoid infiltrate of T and B lymphocytes, whose distribution may vary according to lesion severity.2 A central role is attributed to B cells, which tend to be hyperactive.3 Patients with pSS have an increased risk of developing lymhoproliferative diseases, which is about 4% during the first 5 years, 10% at 15 years and 18% after 20 years post diagnosis.4 Consequently, about 7.5% of patients with pSS develop malignant B cell lymphoma. In 48–75% of these cases, this is the mucosa-associated lymphoid tissue type of lymphoma.5–7 Most commonly, these lymphomas arise in the parotid glands. The assumed role of hyperactive B cells in the pathogenesis of pSS is supported by the observed beneficial objective and subjective clinical effects of B cell depletion by rituximab (RTX), a chimerical monoclonal antibody that binds to the B cell surface antigen CD20.8–16 A significant response was observed in most trials, except from one large randomised clinical trial, the TEARS study.17 Post hoc application of the Sjögren's syndrome response index, also showed significant response rate difference between RTX and placebo in TEARS.18 Because there are some concerns about the efficacy of RTX, the TRACTISS study is aiming to provide evidence whether RTX improves the clinical outcomes.19 The final results of the TRACTISS study, including a subanalysis on responders and non-responders, are eagerly awaited.
In a previous open-label phase II study, based on sequential parotid biopsies of five patients with pSS, we showed that RTX treatment might result in restoration of secretory tissue at a glandular level in responding patients.20 In that study we observed a reduction of the lymphocytic infiltration with partial to complete loss of germinal centres (GCs) and redifferentiation of lymphoepithelial lesions (LELs) to regular striated ducts. However, major limitations of the study by Pijpe et al20 were the small number of patients and lack of a placebo group. Therefore, the aims of the current study were (1) to assess the effect of RTX treatment in patients with pSS based on sequential parotid biopsies obtained in a placebo-controlled, randomised clinical trial, and (2) to assess the prognostic value of the histological characteristics of parotid gland tissue with regard to responsiveness to RTX treatment.
Materials and methods
Thirty patients with pSS were treated in a randomised double-blinded placebo-controlled trial on day 1 and day 15 with either 1000 mg intravenous RTX (Roche, Woerden, the Netherlands; n=20) or intravenous placebo (n=10) at the University Medical Center Groningen, the Netherlands, as described before.14 All patients fulfilled the American European Consensus Group Criteria criteria for pSS.21 To minimise side effects (infusion reactions, serum sickness), all patients, including the placebo-treated patients, were premedicated with methylprednisolone (100 mg intravenous), acetaminophen (1000 mg oral) and clemastine (2 mg intravenous), and received 60 mg oral prednisone on day 1 and day 2, 30 mg on day 3 and day 4, and 15 mg on day 5 after each infusion.
An incisional biopsy was taken under local anaesthesia from the same parotid gland before and 12 weeks after therapy.22 The European League Against Rheumatism Sjögren's Syndrome Disease Activity Index (ESSDAI)23 ,24 was assessed at the same time points by two experienced rheumatologists, who were blinded to the treatment group, in order to evaluate possible systemic complications.16 With regard to response to RTX treatment, patients were categorised into two groups: clinical responders, if the decrease in ESSDAI was ≥3 points at 12 weeks after treatment compared with baseline, and clinical non-responders, if the change of ESSDAI was <3 points. The cut-off of 3 points was chosen because this difference indicates a clinically relevant effect.25
Parotid gland tissue biopsies were fixed in 4% formaldehyde, embedded in paraffin, cut at a thickness of 3 μm and stained with H&E. The focus score (defined as ≥50 lymphocytes per 4 mm2 glandular tissue) and the number of GCs/mm2 parotid gland parenchyma were determined. GCs were defined as a clearly visible lighter area in a lymphocytic infiltrate containing cells that are usually present in classical GCs, like follicular dendritic cells, centrocytes or centroblasts and macrophages. LELs are expressed in LELs/mm2 of parotid gland parenchyma, excluding intraparenchymal connective and fat tissues. LELs were evaluated in immunohistochemically stained tissue sections, stained for CD20 as B lymphocytes predominate in the LEL. A LEL was defined as a cross section of a striated duct with infiltration of CD20+ B cells within the basement membrane combined with hyperplasia of the epithelium. To evaluate regeneration of the ducts, LELs were subcategorised into three stages (figure 1A20): stage 1: LEL affecting less than 50% of the epithelium of the striated duct (partial LEL); stage 2: LEL affecting between 50% and 100% of the epithelium of the striated duct (developed LEL); stage 3: LEL with fully circumferentially affected epithelium without lumen (occluded LEL). For histopathological evaluation, biopsies were independently scored by two investigators (RPP and SI) in a blinded setting. In case of discrepancy, a definitive score was established by consensus.
Immunostaining was used for the analysis of lymphocytic infiltrate and was performed as follows. Parotid glands were fixed in 4% buffered formaldehyde, embedded in paraffin wax and sectioned into 4-μm-thick serial sections. Sections were stained after deparaffinisation, pretreatment with Ultra CC1 (Ventana Medical Systems, USA), antigen retrieval and endogenous peroxidase blocking using the Benchmark machine. Sections were immunohistochemically stained with anti-CD45 (dilution 1:25, Dako, Heverlee, Belgium, clone 2B11+PD7/26), anti-CD79a (dilution 1: 100, Dako, Heverlee, Belgium, clone JCB117), anti-CD20 (dilution 1:200, Dako, Heverlee, Belgium, clone L-26) and anti-CD3 (dilution 1:20, Monosan, Uden, the Netherlands, clone PS-1) antibodies. The sections were then treated with peroxidase-labelled secondary antibody and visualised with the chromogen DAB (3,3′ diaminobenzidine) solution.
The relative amount of CD45-positive lymphocytic infiltrates was assessed in relation to the total amount of tissue parenchyma by morphometry with use of ImageJ software (v1.46). Using HistoQuest software, V.3.5.3.0171, two markers were created, the DAB master marker (CD20) and the haematoxylin non-master marker (nucleus). For the master marker, multiple reference shade was set on 8 with a background threshold range of 5–255. Ring mask and identified cell mask were used. By using a colour picker, the shade was chosen directly from a positively stained CD20 cell. One whole section was analysed excluding intraparenchymal connective and fat tissues leaving multiple regions of interest (ROIs). For the assessment of CD20+ cells, scattergrams were created for each ROI, allowing the visualisation of corresponding positive cells in the source ROI, using the real-time back gating feature. To correct false events, a specific gate according to cell size and intensity of CD20 staining was defined and applied to all analysed samples. CD20+ cells were quantified according to the selected marker and gate. By using the real-time back gating feature, automatically counted CD20+ cells were visualised and controlled. The CD20+ cell count (number of cells/mm2) for each analysed ROI was obtained. The same procedure was followed for CD3+ cell count.
Analysis was carried out with IBM SPSS Statistics 20 (SPSS, Chicago, Illinois, USA). Mann-Whitney U test was used to compare differences between the RTX and placebo groups or between clinical responders and non-responders. Wilcoxon signed-rank test was used to compare differences over time within groups. Spearman's correlation coefficient was used to analyse the relationship between histopathology and ESSDAI. Correlations (ρ)<0.3 were interpreted as a poor association, 0.3–0.6 as moderate, 0.6–0.8 as good and >0.8 as excellent.15 p Values<0.05 were considered as statistically significant. Power analysis was performed with Statistical Power Calculator (DDS Research, Washington DC, USA).
From the total group of 30 patients, 5 patients had to be excluded from histopathological analysis, due to serum sickness (n=1, RTX-group) or insufficient biopsy material (n=3, RTX-group); 1 patient dropped out of the study (placebo group). Thus, complete evaluation of parotid gland biopsies taken from 16 RTX-treated patients and 9 placebo-treated patients could be performed.
Lymphocytic infiltrate in parotid glands
No differences at baseline between the RTX-treated group and the placebo-treated group were found regarding the focus score, relative area of CD45 staining, numbers of CD20+ B cells and CD3+ T cells, and proportion of biopsies containing GCs (data not shown).
After treatment, the focus score did not change significantly in either the RTX-treated group or the placebo-treated group (table 1). However, CD45 staining demonstrated a significant decrease of the relative area of infiltrates at 12 weeks after RTX treatment. In the placebo group no change was observed between baseline values and 12-week post-treatment values (figure 2 and table 1).
By counting the number of CD20+ cells/mm2 of parenchyma, a significant decrease was observed in the number of B cells (1172 cells/mm2 vs 355 cells/mm2, p=0.001) in the glandular tissue at 12 weeks after RTX treatment compared with baseline (figure 3 and table 1). In the placebo-treated group the number of CD20+/mm2 of parenchyma of the parotid glands at week 12 was not statistically different from the number of CD20+ cells at baseline (table 1).
The number of CD3+ cells/mm2 of parenchyma remained unaffected after 12 weeks in the placebo and RTX-treated groups (table 1).
GCs were present at baseline in 67% and 68% of the parotid glands of the placebo-treated and RTX-treated patients, respectively. RTX treatment resulted in a significant decrease in the total number of GC/mm2 (figure 4 and table 1). Twelve out of 16 parotid glands (75%) were completely devoid of GCs 12 weeks after treatment with RTX. In the placebo group, no significant difference was observed in the number of GCs/mm2 between baseline levels and 12 weeks after treatment.
At baseline, no differences were observed in the presence of LEL in the parotid gland parenchyma between the group of RTX-treated patients and the placebo-treated patients (data not shown). In the RTX-treated group, a significant decrease in the total number of LELs/mm2 was observed after 12 weeks of treatment (figure 1B and table 1). In 6 out of 16 patients (38%), LELs were completely absent after RTX treatment. In the placebo group, no significant change was observed in the amount of LELs/mm2 after 12 weeks (figure 1B and table 1). Besides the number of LELs/mm2, the severity of the lesions also appeared to decrease; all stages of LEL seemed to transform to a less severe stage. Detailed data regarding the presence of all stages of LELs/mm2 in placebo-treated and RTX-treated patients at week 0 and week 12 is presented in figure 1C.
Histopathology and ESSDAI
Of the 16 patients that were treated with RTX, 11 patients (69%) improved by ≥3 ESSDAI points and were therefore considered to be clinical responders.25 The other five were considered to be non-responders. The online supplementary table shows the number (%) of patients having any degree of activity per ESSDAI domain (score at least 1) before and after RTX therapy, stratified for responders and non-responders.
The baseline (pretreatment) histopathological parameters (CD20+ cells/mm2, CD3+ cells/mm2, CD45+ relative infiltrate, GCs/mm2, LELs/mm2 and focus score) as well as CD19-positive B cell subsets determined by flow cytometry in the peripheral blood (ie, CD38Low CD27−, CD38High CD27−, CD27 Low CD38−, CD27High CD38−, CD38Low CD27Low, CD38High CD27High, CD27Low CD38High, CD27High CD38Low, CD38− CD27−) of responders and non-responders to RTX treatment were subjected to additional statistical analysis. In responders, the baseline number of CD20+ cells/mm2 was significantly higher in comparison to non-responders (1871 (Q1–Q3=801–4310) cells/mm2 vs 353 (Q1–Q3=35–2102) cells/mm2; figure 5). At an α level of 5% it was calculated that the number of responders and non-responders would give us a power of 94.2% to assume that their baseline number of CD20+ cells/mm2 could serve as a potentially prognostic factor with regards to responsiveness to RTX treatment. The other baseline histological characteristics, as well as baseline B cell subsets determined by flow cytometry in the peripheral blood, did not differ significantly between responders and non-responders. Of note, there was no correlation between the absolute numbers of CD20+ B cells/mm2 of parenchyma and ESSDAI or between B cell subsets in peripheral blood and ESSDAI.
Furthermore, in RTX-treated patients the change in ESSDAI correlated with the change in the number of CD20+ cells/mm2 of parenchyma (ρ=0.706 and p<0.05). No other statistically and clinically significant correlations were found for the changes between baseline and 12 weeks.
We demonstrated that RTX treatment significantly reduced the overall lymphocytic infiltrate with a major loss of the B cell component and number of GCs/mm2 of parotid gland parenchyma in patients with pSS. In addition, a major reduction of the quantity and severity of LELs was apparent, reflecting significant restoration of the striated ducts.
RTX treatment results in a considerable decrease in the number of B cells in the parotid gland tissue. Although this is reflected by a decrease in the amount of infiltrate, as measured by staining for CD45, this is not manifested by a decrease in the focus score. This apparent discrepancy can be explained by the fact that the foci also contain high numbers of T cells, which may outnumber the number of B cells,2 and which are not affected in significant numbers by RTX treatment. The focus score is therefore not an appropriate criterion to measure the local effect of RTX on the periductal lymphocytic infiltration. Although RTX treatment results in the almost complete absence of B cells in the peripheral blood of patients with pSS,26 this is thus not accompanied by a complete loss of B cells in parotid salivary gland tissue. These results are in line with other studies in pSS20 ,27 ,28 and rheumatoid arthritis29–31 showing a certain degree of persistence of B cells in the local tissue after RTX treatment. In contrast, Devauchelle-Pensec et al10 reported a total depletion in B cells in labial salivary glands of patients with pSS after RTX treatment. However, in that study only a very low number of the patients with pSS (6 out of 15) showed significant numbers of B cells in the periductal infiltrates at baseline. This is remarkable, since B cells usually make up to 20–60% of the lymphocytes in the infiltrates of the labial glands of patients with pSS, depending upon the grade of the lesion.2
In this study at baseline the included patients had high systemic activity, as indicated by the relatively high ESSDAI scores, and high numbers of GCs/mm2.32 We observed a strong reduction of GCs in the parotid tissue after RTX treatment; in several patients we even observed a complete absence of GCs. This is striking, since not all B cells are depleted in the parotid glands, and GC B cells may be relatively more resistant to anti-CD20 therapy compared with other B cells, as shown by Gong et al33 in a murine model for human CD20 expression. A possible explanation for the strong depletion of GCs in the parotid tissue might be that RTX treatment also results in a significant reduction of follicular T helper cells (TFH), as indicated by analysis of peripheral blood samples (Verstappen et al 2015, manuscript in preparation). TFH cells are essential for the development of GCs at local sites. These cells are present in the salivary gland tissue of patients with pSS,34 where they may drive GC formation and generation of plasma cells. It is therefore possible that the relative absence of TFH in the salivary gland tissue after RTX treatment contributes to the loss of GC activity in these patients with pSS.
LELs develop in striated ducts in patients with pSS, particularly in the parotid glands. The epitheliotropic autoimmune inflammation of the intraepithelial lymphocytes results in the reaction of the epithelium and induction of these lesions.35 RTX treatment results in a significant reduction of the number CD20+ B cells in the periductal infiltrates, and in a recovery of the LELs, as revealed by a considerable reduction of the severity of the lesions at all stages (figure 1C). Such a restoration/redifferentiation of LEL was also observed in the small RTX treatment study (five patients) described by Pijpe et al.20 Apparently, RTX treatment also results in depletion of B cells within the basement membrane of striated ducts. To explain this, we have hypothesised that the trigger for LEL formation is diminished, and as a result less epithelial reaction takes place leading to reduced proliferation and finally anatomical restoration of the striated ducts. The trigger for LEL formation is unknown, but B cell derived cytokines may possibly be responsible for this. This notion is in line with the finding of Pollard et al,36 who showed in the same cohort of RTX-treated patients that the serum levels of proinflammatory cytokines (eg, interleukin 6) decreased significantly.
Patients with pSS have different genetic backgrounds, demographic features and prognoses, and exhibit a wide variety of clinical manifestations, involving a number of pathophysiological pathways.37 Personalised treatment, that is, providing ‘the right patient with the right drug at the right dose at the right time’,38 will therefore be the key to treating pSS. An important finding in our study was that clinical responders to RTX treatment had a higher number of CD20+ B cells/mm2 of parenchyma parotid gland tissue at pretreatment (baseline) compared with non-responders. Furthermore, we also observed a correlation between the change in the number of CD20+ cells/mm2 of parenchyma and the change in ESSDAI. When higher numbers of B cells are present in the parotid gland parenchyma, it is therefore possible that RTX treatment may result in depletion of more absolute numbers of B cells responsible for the disease activity (measured by ESSDAI) than when lower numbers of B cells are present in the tissue. The baseline number of B cells/mm2 of parenchyma of parotid gland may therefore determine patients’ response to treatment with RTX and may be considered as a biomarker for a more personalised treatment approach to patients with pSS. The nature of these disease-associated B cells that are reduced after RTX treatment needs to be elucidated. These cells are probably not antibody producing cells, since antibody producing cells persist in the parotid salivary glands after RTX treatment.28 Alternatively, these cells may represent cytokine producing B cells.36
Although the change in number of B cells in the infiltrates of the salivary glands correlated to the change in ESSDAI after RTX treatment, the absolute number of B cells at baseline did not correlate to ESSDAI. Furthermore, the focus score of the salivary glands did not correlate to ESSDAI. However, Risselada et al39 showed a significant correlation at baseline between the focus score and the cumulative ESSDAI in labial salivary glands of 174 patients with pSS (ρ=0.166–0.284, p≤0.04). This discrepancy could be ascribed to the fact that the study by Risselada et al was retrospective, where ESSDAI was assessed at any time point during disease (not necessarily at diagnosis and biopsy), that 21% of patients used immunomodulating medication at the time of the biopsy and correlations were considered to be significant even if ρ was as low as 0.166–0.284. Moreover, the size of the focus (as we assessed by CD45 staining) is probably more relevant than the absolute number of present foci in the salivary gland tissue.
In conclusion, we demonstrated that in parotid gland tissue of patients with pSS:
RTX treatment leads to major reduction of B cells, and a significant reduction in the number of GCs and LELs. This reduction in the LELs may be the consequence of a major decrease of local B cell infiltration and may result in structural regeneration of the glands, especially the striated ducts.
The baseline number of CD20+ B cells/mm2 of parenchyma may serve as a prognostic biomarker to predict response to RTX treatment. As a result, baseline histopathological characteristics of a parotid biopsy may strongly contribute to a more personalised treatment approach to patients with pSS.
The authors thank Mr Klaas Sjolemma, UMCG Microscopy and Imaging Center (UMIC), University Medical Center Groningen, for his assistance during the immunohistochemical analysis of specimens with HistoQuest software and Dr Suzanne Arends, Department of Rheumatology and Clinical Immunology, University Medical Center Groningen, for critical reading of the manuscript.
Handling editor Tore K Kvien
KD and EAH contributed equally.
Twitter Follow Konstantina Delli at @KonstantinDelli
Contributors KD, EAH, FGMK, AV, HB, FKLS: conception and design, analysis and interpretation of data, drafting the article or revising it critically for important intellectual content. RPP, SI: analysis and interpretation of data, revising the article critically for important intellectual content. BvdV: revising the article critically for important intellectual content.
Funding Supported by unconditional grants of Roche (Woerden, the Netherlands) and the Jan-Kornelis de Cock foundation (Groningen, the Netherlands).
Competing interests None declared.
Patient consent Obtained.
Ethics approval Medical Ethical Committee of the University Medical Center Groningen (UMCG) (METC approval: 05.229).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.