Article Text
Abstract
Objectives Development of non-Hodgkin’s lymphoma (NHL) is the major adverse outcome of Sjögren’s syndrome (SS) affecting both morbidity and mortality. Preliminary evidence suggested that, although not deregulated compared with sicca controls, miR200b-5p levels are decreased in the minor salivary glands (MSGs) of SS patients with NHL. The aim of the current study was to evaluate the MSG expression of miR200b-5p in SS-associated NHLs and its potential predictive value for the identification of patients with SS susceptible to develop NHL.
Methods miR200b-5p expression was investigated in MSG tissues of patients with SS who were at: (A) low risk and did not develop NHL during follow-up (n=27; median follow-up time on biopsy performance, range: 8.9 years, 1.33–14 years), (B) high-risk and diagnosed with NHL during follow-up (prelymphoma; n=17; median follow-up to until lymphoma diagnosis, range: 3.67 years, 0.42–8.5 years) and (C) had NHL (n=35), as well as non-SS sialadenitis controls (sarcoidosis and hepatitis C virus (HCV) infection, four each). The differential miR200b-5p expression, correlations with disease features and its discriminative/predictive value, was evaluated by appropriate statistical approaches.
Results The MSG levels of miR200b-5p were significantly downregulated in patients with SS who will develop or have NHL and strongly discriminated (p<0.0001) them from those without lymphoma or non-SS sialadenitis. Furthermore, they were reduced long before clinical onset of lymphoma, did not significantly change on transition to lymphoma and, importantly, were proved strong independent predictors of patients who will develop NHL (p<0.0001).
Conclusions These findings support that miR200b-5p levels in MSGs represent a novel predictive and possibly pathogenetic mechanism-related factor for the development of SS-associated NHL, since its expression is impaired years before lymphoma clinical onset.
- sjøgren’s syndrome
- autoimmune diseases
- autoimmunity
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Introduction
Primary Sjögren’s syndrome (SS) is an autoimmune disease with a diverse clinical picture ranging from benign, mild exocrinopathy to severe, systemic, disorder with high prevalence (5%–10%) of B cell non-Hodgkin’s lymphoma (NHL).1 2 NHL is the major adverse outcome of the disease, affecting both morbidity and mortality.3–5 Several clinical, laboratory and histological features, including high EULAR SS disease activity index (ESSDAI) score, salivary gland enlargement (SGE), purpura, vasculitis, leucopenia, cryoglobulinaemia, hypocomplementaemia, rheumatoid factor, formation of germinal centres (GCs) in the histopathological lesion and infiltration by certain cell types, such as macrophages, have been associated with the development of lymphoma in SS.3 4 6–11
The miR200 micro-RNA (miRNA) family, consisting of miR200a, miR200b, miR429, miR141 and miR200c miRNAs, possesses a central role in oncogenesis, tumour metastasis and drug resistance. The miR200b-3p and miR200b-5p miRNAs are considered powerful regulators of epithelial-to-mesenchymal transition (EMT) and as such they have been implicated in the oncogenesis of solid tumours.12–17 Recently, in the context of investigating the expression of several miRNAs that are predicted to target the Ro/SSA and La/SSB autoantigens, we examined the expression of miR200b-3p and miR200b-5p in the minor salivary glands (MSGs) of SS patients and sicca controls.18 Although their expression in the MSGs of patients with SS was not deregulated compared with sicca controls, miR200b-5p levels were significantly reduced in four SS patients with mucosa-associated lymphoid tissue (MALT) lymphomas compared with those without.18
Prompted by this finding, we sought to: (A) validate the hypothesis that miR200b-5p levels are decreased in the MSGs of patients with SS-associated NHL by studying its expression in an adequate population of SS patients with or without NHL, as well as in sialadenitis controls; (B) evaluate its predictive value by investigating its expression in MSG samples from both low-risk patients with SS who did not develop lymphoma during follow-up and high-risk patients who developed SS-associated NHL in the future; (C) test its independent predictive utility over that of previously identified adverse predictive factors for the development of SS-associated NHL.
Materials and methods
Patients
MSG samples obtained from 79 patients with primary SS and 8 non-SS sialadenitis associated with sarcoidosis and HCV infection (four each) were studied after informed consent. Patients with SS were diagnosed according to the American–European classification criteria.19 The patients with SS included 27 low risk who did not develop lymphoma during follow-up (without lymphoma (SSwo); median follow-up time on biopsy performance, range: 8.9 years, 1.33–14 years), 17 high risk who were diagnosed with SS-associated NHL during follow-up (prelymphoma (SSpL); median follow-up time to NHL diagnosis, range: 3.67 years, 0.42–8.5 years) and 35 with SS-associated lymphoma at the time of biopsy (lymphoma (SSL)). The low-risk group consisted of SS patients with low probability to develop lymphoma,11 20 including 17 without risk factors for lymphoma development, 5 with low serum C4 levels, four with SGE and one with both low serum C4 levels and SGE (median follow-up time of the patients expressing adverse predictive factors: 8.47 years, range: 4.91–12.91 years). The prelymphoma SS group included 14 MALTs, 2 nodal marginal zone lymphomas (NMZLs) and 1 diffuse large B cell lymphoma (DLCBL). The SS-associated NHLs consisted of 28 MALTs, 2 NMZLs, 2 DLCBLs, 1 B cell bronchial associated lymphoid tissue (BALT), 1 lymphoplasmacytic (LP) and 1 small lymphocytic (SLL) lymphoma. In 14 cases (11 MALTs, 2 NMZLs and 1 DLCBL), the SSpL and SSL samples were paired sequential specimens from the same patient (before and on lymphoma onset). All prelymphoma and lymphoma SS samples that had available MSG specimens from a total of 84 SS patients with NHL who were followed up during 1993–2016 in the Department of Pathophysiology, School of Medicine, National and Kapodistrian University of Athens, Greece, were studied.
The medical records were retrospectively evaluated for various clinical, laboratory and histological parameters of SS and lymphoma that are described in table 1. The characteristics of the patients are summarised in table 2, whereas lymphoma features in table 3 and online supplementary table S1. Three SS patients had received corticosteroids, four hydroxychloroquine, one cyclophosphamide and five B cell depletion therapy (anti-CD20) prior to biopsy performance.
Supplemental material
Evaluation of miR200b-5p levels by quantitative PCR (qPCR)
MSG tissues were obtained during diagnostic biopsy, fresh-frozen in RNAlater stabilisation reagent (Qiagen, Venlo, The Netherlands) according to manufacturer’s instructions and stored at −80°C until RNA isolation. Total RNA including small RNA molecules, such as miRNAs, was isolated from an MSG lobe using the mirVana PARIS kit (Ambion, Applied Biosystems, Carlsbad, California, USA) according to the manufacturer’s instructions. The time of storage has not been found to significantly affect the quantity or quality of the isolated RNA (data not shown). Subsequently, 0.25 μg RNA were reverse-transcribed by TaqMan miRNA reverse transcription kit and miRNA-specific primers (TaqMan-MicroRNA Assays, Applied Biosystems), followed by miRNA-specific amplification by qPCR using primers from TaqMan-MicroRNA Assays and TaqMan Universal PCR Master Mix, no AmpErase UNG (Applied Biosystems). RNU48 small nucleolar RNA (snoRNA) was selected for normalisation after NormFinder analysis.21 Three snoRNAs, namely U6, RNU44 and RNU48, have been examined in preliminary experiments, including samples from sicca controls, low-risk SS patients without lymphoma and patients with SS-NHL (six in each group) for identifying the most appropriate reference. In these experiments, a common reverse transcription using Megaplex RT Primers, Human Pool B V.3.0 (Applied Biosystems) was performed in order to exclude variations due to distinct RT reactions. NormFinder analysis21 revealed that RNU48 had the higher stability value (0.006) compared with RNU44 (stability value: 0.008) or U6 miRNA (stability value: 0.016). The relative quantification of miRNA expression was performed by the 2−ΔΔCT method using non-malignant parotid gland tissue from a patient subjected to parotidectomy as calibrator. The relative occurrence of various cell types, including epithelial cells, B cells, T cells and macrophages, in MSG samples was evaluated by qPCR for keratin-8 (KRT8), CD19, CD3D and CD68, respectively, using TaqMan expression assays. Samples were processed randomly and without grouping. All samples were run in duplicates. Additionally, in a proportion of patients (six from each SS subgroup and four sialadenitis controls; randomly selected) the infiltration by CD20+ B cells, CD3+ T cells and CD68+ macrophages (number of infiltrating cells per mm2 tissue area), as well as its correlation with miR200b-5p levels were analysed immunohistochemically using ImageJ software, as previously.7 22
Statistical analyses
The differential expression of miR200b-5p levels among the subgroups of patients with SS and sialadenitis controls was evaluated by the non-parametric Tukey’s multiple comparison test. Significant differences in miR200b-5p levels between patients expressing or not various clinical, histological and serological markers were analysed by the non-parametric Mann-Whitney U test, whereas associations with patient features and overtime changes by Spearman’s rank correlation and Wilcoxon’s matched pairs tests, respectively. Holm-Bonferroni sequential correction was applied for correcting for multiple comparisons.23
Analyses regarding the diagnostic/discriminative utility of miR200b-5p were performed in two levels: (A) all patients with SS who did not have lymphoma at the time of biopsy (SSwo and SSpL) were compared with lymphoma patients (SSL) and (B) comparisons among the three SS subgroups. The 14 patients that were common in the prelymphoma and lymphoma SS group were excluded from lymphoma group, thus remaining 21 SSL patients in the analyses. The diagnostic/discriminative ability of miR200b-5p levels was evaluated by receiver operating characteristic (ROC) curve analysis. Categorical variables were compared by the Pearson χ2 or the Fisher’s exact test, when appropriate. HRs are provided with a 95% CI. To identify independent risk factors for NHL development in SS, all variables associated with lymphoma with a p value less than 0.1 in univariate analysis were further evaluated by multivariate logistic or Cox regression analysis using a backward stepwise exclusion method. The predictive ability of miR200b-5p levels was evaluated by Kaplan-Meier lymphoma-free survival curves compared by the log-rank test in the prelymphoma and without lymphoma SS patients, who were split in two groups according to low or not miR200b-5p expression, as this defined by ROC discriminative value.Similar analysis was performed for low-risk and high-risk patients according to previously described models.4 9 11 Descriptive analyses of all data were performed.
GraphPad Prism-5 (GraphPad Software, San Diego, California, USA) and SPSS V.17 software were used. Statistical significance was defined as a p value of less than 0.05 for all comparisons; p values were two tailed. Only the statistically significant differences are reported. Using two-sided 95% CI, the observed difference of the means was proved to have 96.5% power (OpenEpi, V.3, open source calculator).
Results
The MSG levels of miR200b-5p are reduced in patients with SS who have or will develop NHL and discriminate them from those who will not
The miR200b-5p levels were significantly lower in the MSGs of high-risk patients with SS who were diagnosed with NHL during follow-up (SSpL; mean relative expression±SD: 0.31±0.33) and lymphoma SS patients (SSL; 0.21±0.25) compared with the low risk that did not develop lymphoma during follow-up (SSwo; 0.72±0.37; p≤0.01 and p≤0.0001 for prelymphoma and lymphoma, respectively) or non-SS sialadenitis controls (0.95±0.84, p≤0.01 and p≤0.001, respectively) (figure 1A). Low miR200b-5p levels were also detected in patients with SSL who had not lymphoma in MSGs. Interestingly, low miR200b-5p levels (0.17) were also detected in an HBV-patient that had MALT lymphoma (patient excluded from the analysis). Additionally, miR200b-5p levels were not found to significantly change on transition to lymphoma, as indicated by the analysis of 14 sequential paired samples from SS patients before and on lymphoma diagnosis (figure 1B). The significantly lower expression of miR200b-5p in the MSGs of SS patients with NHL long before lymphoma diagnosis indicates that it is possibly implicated in the pathogenesis of SS-associated lymphoma. miR200b-5p levels correlate with clinical, laboratory and histologic factors of adverse outcome and/or NHL development, as well as NHL prognosis.
The low levels of miR200b-5p were associated with several clinical, laboratory and histological features indicative of adverse outcome and lymphoma development that are summarised in table 4. Hence, miR200b-5p levels were negatively correlated with ESSDAI, whereas they were positively correlated with serum C4 levels (table 4). Significantly lower miR200b-5p levels were detected in SS patients with SGE, purpura, peripheral neuropathy, cryoglobulinaemia, hypergammaglobulinaemia and rheumatoid factor compared with those without (table 4). Despite the lower age of SSpL and SSL patients compared with SSwo (table 2), miR200b-5p levels were not correlated with patient age (p=0.1), whereas they were negatively correlated with biopsy focus score (table 4); however, this did not affect their higher expression in SSwo, compared with SSpL and SSL patients (supplementary table S3 and supplementary figure S2). Furthermore, it was negatively correlated with CD3D and CD68 mRNA expression in MSGs (r=−0.5613, p<0.0001 and r=−0.5048, p<0.0001, respectively), which are indicative of T-lymphocyte and macrophage infiltration, respectively, but not CD19 (B cell; r=−0.3072, p=0.068) or KRT8 (epithelial cell; r=0.1536, p=0.23) expression. Subsequent immunohistochemical evaluation of the number of infiltrating CD3+ T cells, CD20+ B cells and CD68+ macrophages per tissue area (mm2) in randomly selected samples from each group of patients with SS (six patients/group), as well as sialadenitis controls (n=4) revealed that the levels of miR200b-5p in MSGs were negatively correlated with all these types of infiltrating cell populations (table 4).
The levels of miR200b-5p in MSGs were not found to associate with the type, stage or number of involved sites of NHL, as well as event-free or overall-free survival, whereas there was a trend to associate with high-intermediate/high international prognostic index (IPI: 3–4) (mean±SD: 0.24±0.28 vs 0.10±0.06 in patients with low/low-intermediate IPI (0–2), p=0.07). miR200b-5p levels in MSGs discriminate patients with SS who have versus those who do not have NHL, as well as those who will develop NHL versus those who will not.
ROC curve analysis revealed that miR200b-5p strongly discriminates prelymphoma and lymphoma SS patients from those without lymphoma. Thus, the lymphoma SS patients were discriminated from SS patients without lymphoma at the time of biopsy (SSwo and SSpL) with AUC and cut-off values 0.840 (p<0.0001) and 0.3164 (sensitivity=0.952, specificity=0.750), respectively. More importantly, prelymphoma and lymphoma SS patients were discriminated from those that did not develop lymphoma during follow-up (SSwo) with AUC values 0.863 and 0.986 (p<0.0001), respectively, and cut-off values 0.4156 (sensitivity=0.765, specificity=0.926) and 0.3164 (sensitivity=0.952, specificity=1), respectively (figure 1C–E).
Kaplan-Meier analysis of patients split into two groups according to miR200b-5p expression levels of 0.4156 (as defined by the specificity–sensitivity analysis) revealed that patients with miR200b-5p levels ≤0.4156 had a 4.8-fold (HR: 4.81, 95% CI 3.15 to 6.47, p<0.0001) higher risk to develop lymphoma compared with patients with miR200b-5p levels >0.4156 (figure 1F).
miR200b-5p levels in MSGs predict SS who will develop NHL versus those who will not, independently from other known predicting factors
The known outcome of the NHL development in the patients with SS who did not have lymphoma at the time of MSG biopsy, involving the low-risk ones who did not develop NHL during follow-up and those who evolved to an SS-associated NHL, enabled the evaluation of the utility of miR200b-5p levels in the prediction of lymphoma development. Cox regression analysis was employed to evaluate the value of miR200b-5p in the prediction of lymphoma development along with other disease parameters that were associated with lymphoma development in univariate analysis. These included high ESSDAI (defined as score ≥5, p<0.0001), SGE (p=0.002), purpura (p=0.0001), vasculitis (p=0.024), anaemia (p=0.058), splenomegaly (p=0.024), lymphadenopathy (p=0.005), C4-hypocomplementaemia (p=0.005), RF (p=0.001), GCs (p=0.007), cryoglobulinaemia (p=0.001) and hypergammaglobulinaemia (p<0.0001).
Multivariate analysis confirmed that miR200b-5p was independently associated with development of lymphoma (HR per 1-unit change: 0.10, 95% CI 0.01 to 0.87, p=0.012) along with high ESSDAI (p=0.024), SGE (p=0.012), purpura (p=0.057), vasculitis (p=0.043), splenomegaly (p=0.047), cryoglobulinaemia (p=0.032) and hypergammaglobulinaemia (p=0.055).
Possible role of miR200b-5p levels in monitoring of therapeutic response
Preliminary observations from nine patients with SSL before and after treatment suggest that MSG levels of miR200b-5p may apply in the prediction of therapeutic response. Thus, miR200b-5p levels remained stable or decreased (0.22 to 0.07) in refractory to treatment MALT (n=5) and DLCBL (n=1) lymphomas, respectively, reduced in two MALT-SSLs who relapsed (0.54 and 0.31 to 0.2 and 0.09, respectively) and increased in a MALT-SSL who reached complete remission (0.07 to 0.47).
Discussion
This study supports that the MSG expression levels of miR200b-5p constitute a novel, strong, predictive biomarker for NHL development in SS, since its expression is impaired years before lymphoma clinical onset. Indeed, reduced MSG expression of miR200b-5p characterised patients with SS who have or will develop NHL. Low levels of miR200b-5p were correlated with disease parameters, previously associated with adverse outcome and NHL development. Importantly, miR200b-5p levels strongly discriminated the three groups of patients with SS, namely low-risk, prelymphoma and lymphoma, with high sensitivity and specificity and independently predicted lymphoma development. Furthermore, miR-200b-5p levels in MSGs were downregulated long before the clinical onset of lymphoma, supporting its potential as a predictive biomarker. In addition, our preliminary observations imply that miR200b-5p may also be significant for the therapeutic monitoring of patients with NHL, signifying the need for appropriate prospective studies.
The mechanisms underlying the reduced miR200b-5p expression in MSGs and its role in SS-related lymphomagenesis have not been delineated. The reduced levels of miR200b-5p long before lymphoma development suggest that it is implicated in SS lymphomagenesis, although the pathways that regulate remain unknown and are currently under investigation. Despite the well-established role of the reduced miR200b-3p expression in EMT and associated oncogenesis, tumour metastasis and invasion,12–17 little are known for miR200b-5p, possibly because it represents the star strand, which is generally considered to degrade during miRNA biogenesis.24 Recently, it has been reported that miR200b-5p controls the non-canonical EMT in synergy with miR200b-3p by targeting PRKCA and PIP4K2A molecules in the RHOGDI pathway.25 Interestingly, miR200b miRNAs have been almost exclusively associated with solid tumours. A recent study suggests that the elevated miR200b expression and subsequent inhibition of zinc finger E-box-binding homeobox 1 transcription factor and increased BCL6 protein expression is associated with the better prognosis of the Helicobacter pylori-positive gastric diffuse large B cell lymphomas compared with H. pylori-negative ones.26 To this end, it would be interesting to investigate whether miR200b-5p levels are downregulated in other lymphomas that are not associated with SS.
Despite the clinical progress, the mechanisms underlying SS-related lymphomagenesis have not been delineated. It is considered as a multistep antigen-driven process taking place in the inflammatory MSG lesions that arises from the chronic continuous and/or inappropriate B cell stimulation, which increases the risk of chromosomal translocations, activation of proto-oncogenes and inactivation of tumour-suppression genes resulting in malignant transformation.27 28 Most likely, the reduced miR200b-5p expression in the MSGs of patients with SS that will develop or have lymphoma does not involve B cells or other types of infiltrating lymphocytes, since we have been previously unable to detect its expression in the peripheral blood B cells.18 29 Interestingly, miR200b-5p levels were negatively correlated with biopsy focus score and infiltration by T cells, B cells and macrophages, suggesting that the reduction of miR200b-5p in prelymphoma and lymphoma SS patients may result from ‘dilution effect’ due to increased infiltration by inflammatory cells. However, this reduced expression could be coincidental and reflective of reduced expression in epithelial cells, which are a possible source of miR200b-5p, in severe lesions. This is supported by the higher expression of miR200b-5p in low-risk patients compared with those with prelymphoma and lymphoma and similar MSG infiltration and the lack of correlation with KRT8 expression, which is possibly reflective of downregulated epithelial expression of miR200-5p in MSGs of patients with SS who will develop or have lymphoma and not alteration in epithelial cell number. In contrast to peripheral blood mononuclear cells, cultured salivary gland epithelial cells express miR200b-5p.18 29 Epithelial cells are key regulators of SS autoimmune responses, whereas preliminary data suggest that they can drive B cell activation and differentiation.30 Thus, it would be tempting to hypothesise that the reduction of miR200b-5p in MSG epithelia promotes the efficient interaction with B cells, which eventually leads in lymphomagenesis. In this context, the detection of low miR200b-5p levels in NHLs without MSG involvement is not paradoxical. The cellular types expressing miR200b-5p in the MSGs of patients with SS, the effect of its downregulation in their phenotype and the mechanisms underlying its reduction is of high importance for the discovery of novel therapeutic targets and is currently investigated in our lab.
In summary, the expression levels of miR200b-5p in MSG tissues constitute a novel, strong, disease mechanism-related, predictive biomarker for NHL development in SS. Their predictive value will be further validated during the HarmonicSS project (European Union Grant-731944) that includes 21 European patient SS cohorts.
Acknowledgments
The authors gratefully acknowledge Haralampos M Moutsopoulos, MD, FACP, FRCP(hc), Master ACR, for his contribution in patient selection and manuscript preparation and editing.
References
Footnotes
Handling editor Josef S Smolen
Contributors EKK designed the study, performed the experiments, analysed the data and wrote the paper. AP evaluated the clinical data of patients with non-Hodgkin’s lymphoma (NHL) patients, ADP performed statistical analyses, MV evaluated the clinical data of patients with NHL and contributed in data analysis and manuscript preparation and AGT designed the study and wrote the paper. AGT and EKK had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Funding The research has been financed by grants of the Research Grant from the Greek Rheumatology Society and Professional Association of Rheumatologists and the European-funded multicentric protocol ’HARMONIzation and integrative analysis of regional, national and international Cohorts on primary Sjögren’s Syndrome (pSS) towards improved stratification, treatment and health policy making' (HARMONICSS; H2020-SC1-2016; grant agreement no: 731944).
Competing interests AGT has received research grants from Novartis, Pfizer, UCB, AbbVie and GSK pharmaceutical companies, through the National and Kapodistrian University of Athens, outside the submitted work.
Patient consent Not required.
Ethics approval Ethics Committee of School of Medicine, National and Kapodistrian University of Athens, Greece (protocol no: 1516023881).
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no additional unpublished data related to the study.