Clinical manifestations including the EULAR Sjögren's Syndrome Disease Activity Index in anti-Ro52/SS-A antibody-seropositive patients with Sjögren's syndrome CURRENT STATUS: POSTED

Background: Relationship between anti-Ro52/SS-A antibody (anti-Ro52) and clinical manifestation of Sjögren's syndrome (SS) was sporadically described regarding involvement of internal organs. Objectives: To determine the clinical factors relevant to anti-Ro52 with SS. Methods: We conducted a retrospective study of subjects with suspected SS (n=149), patients with rheumatoid arthritis (RA) (n=62), and healthy subjects (n=50). We analyzed components of the American-European Consensus Group (AECG) criteria, Raynaud's phenomenon (RP), anti-centromere antibody (ACA), serum IgG, rheumatoid factor, and the EULAR Sjögren's Syndrome Disease Activity Index (ESSDAI). Results: Among the 149 suspected SS subjects, 115 subjects were classified as having SS. Anti-Ro52 was observed in 70 SS patients (60.9%), of whom 31 patients had markedly elevated anti-Ro52 (>500 U/ml). In the SS group, three patients with anti-Ro60−/anti-Ro52+ and 17 patients without anti-Ro were observed. Compared to ACA, the relevance of positive anti-Ro52 toward positive anti-Ro60 was significantly higher (p<0.05). We observed relevance between the anti-Ro52 concentration and anti-Ro60 that was significantly affected by xerophthalmia, xerostomia, ACA seropositivity, RP, serum IgG level, and RF. The anti-Ro52 concentration well discriminated six clinical factors (ROC AUC >0.75) for ACA seropositivity, ESSDAI score ≥1, and RF or moderately high serum IgG, focus score ≥1, and anti-La/SS-B antibody seropositivity (ROC AUC >0.7). A linear relationship between the ESSDAI score and anti-Ro52 was observed. Conclusion: Significant relevance between anti-Ro52 and anti-Ro60 as well as significant items including the ESSDAI regarding the anti-Ro52 concentration was revealed. Regarding relevance toward anti-Ro60, anti-Ro52 had a higher association than ACA.


Introduction
Sjögren's syndrome (SS) is a systemic autoimmune disease that has unique clinical manifestations including xerostomia, xerophthalmia, and extraglandular clinical manifestations such as interstitial pneumonia and tubulointerstitial nephritis [1,2]. SS also has characteristic autoantibodies including We enrolled 149 subjects suspected of having SS who visited Nagasaki University Hospital between 2009 and 2018 (Fig. 1). The classification of primary (p) SS was determined based on the 2002 AECG classification criteria [5]. Among the 149 subjects with suspected SS, 115 subjects were classified as having SS based on the AECG criteria (Fig. 1). The 115 patients with SS were classified as 87 primary SS and 28 secondary SS. There were 28 patients with secondary SS; these cases were complicated with RA in 13 patients, with SLE (n=6), mixed connective tissue disease (n=5), systemic sclerosis (SSc; n=2), polymyositis (PM; n=1) and anti-phospholipid syndrome (n=1). One PM patient had neither anti-Ro52 nor anti-Jo-1 antibody. The degree of mononuclear cell infiltration in the subjects' labial salivary glands (LSGs) was evaluated based on the Chisholm & Mason grading or a focus score (FS) determined by Greenspan [12,13]. We retrospectively evaluated the disease activity of SS by using the ESSDAI [14,15], which is composed of 12 items used to determine the systemic involvement of SS. The present FS calculation was subjected to the standardization method recommended by the EULAR SS Study Group [16].
As a disease control group, we examined the cases of 62 patients with rheumatoid arthritis (RA) that had been classified based on the 2010 ACR/EULAR classification criteria or the 1987 American Rheumatism Association criteria for RA [17,18]. For the exclusion of latent SS, we excluded RA patients with sicca symptoms including xerostomia and/or xerophthalmia and anti-Ro60 that was determined by the Mesacup SS-A/Ro test and SS-B/La Test. Blood samples from 50 healthy subjects who showed participation intention were used as normal controls. Healthy subjects were tested only for anti-Ro52 by EliA Anti-Ro52.
For the calculation of the focus score (FS), images of salivary glands were captured by a microscope (BZ-X-710, Keyence, Osaka, Japan), and whole areas of salivary glands were measured by the hybrid cell count system, then FS was manually calculated as the number of foci per 4mm 2 according to the method described by Fisher et al [16]. This case-control study involving patients' medical records and sera was performed with the disclosure of information according to the approval of the Clinical Studies Ethics Committee of Nagasaki University Hospital (approval no. 18121007). Handling blood samples from healthy subject was approved by this Committee because this opt-out disclosure operation is based on ethical guidelines for medical and health research involving human subject determined in Japan.

Statistical analysis
For comparison of age and sex among the patients with SS, the patients with RA and healthy subjects, Wilcoxon rank-sum test and Fisher exact test were used, respectively. We included the following items in the analyses as clinical factors of interest: AECG criteria components including xerostomia, xerophthalmia, salivary secretion by Saxon test, lacrimal secretion by Schirmer's test, the focus score (FS), labial salivary gland biopsy (LSGB) grading (0-2/3/4), anti-Ro/SS-A antibody, anti-La/SS-B antibody, and relevant items including RP, ACA, serum IgG, rheumatoid factor (RF), and the ESSDAI score.
The concentration of anti-Ro52 and anti-Ro60 was transformed to logarithm with base 10. The association of clinical factors with relevance between anti-Ro52 and anti-Ro60 was inferred as the 95% confidence interval (95%CI) of the regression coefficient of a term of interaction between anti-Ro52 and the clinical factor in a linear regression model, which regressed anti-Ro60 onto the interaction term with covariates of anti-Ro52 and the clinical factor. The model did not include other covariates to control for confounders. The regression coefficients were estimated by MM-estimator with bisquare psy-function [19,20] (for all of the linear regression analyses). We used the multiple imputations by chained equation (MICE) method [21] to estimate the regression coefficients. The patterns of missingness are shown in Supplemental Figure S1 (labeled "NA" on the x-axis) and Supplemental Figure S2 (the right column of each plot). The contour of two-dimensional probability density was drawn based on a kernel density estimate with the Gaussian kernel with bandwidth selected by the "solve-the equation" estimator [22,23]. The linear relationship between the ESSDAI score and the levels of antibodies against the two respective subtypes of Ro/SS-A antigen were analyzed by MM-estimator with bisquare psy-function.
We compared the odds of anti-Ro60 positivity between the positivity of ACA and that of anti-Ro52 by using the ratio of the two odds ratios (the ratio of ORs). The null hypothesis that the ratio is 1 was tested via a permutation test of anti-Ro60 positivity. We assessed the associations between markedly elevated anti-Ro52 (>500 U/ml) and each of the clinical manifestation by determining the OR, and the statistical significance of each association was tested by Fisher's exact test. We evaluated the associations between respective clinical factors and the concentration of antibodies against the two respective subtypes of Ro/SS-A antigen by determining the maximal information coefficients [24]. The null hypothesis of independence between respective clinical factors and the concentration was tested by a permutation test. The odds of positivity in respective characteristics given the concentration titer of anti-Ro52 were evaluated with the area under curve (AUC) of a receiver operator characteristic (ROC) curve [25]. The confidence intervals of sensitivity and specificity were obtained from 2,000 bootstrap samplings. Hypothesis testing was conducted with the significance level of 0.05 without adjustment for multiple comparisons. All statistical analyses were conducted under the R environment

The seroprevalence of anti-Ro52 in the SS patients
The 115 patients with SS were classified as 87 primary SS and 28 secondary SS. Among these 115 patients with SS, 70 patients (60.9%) were anti-Ro52 positive ( Fig. 2A). The p value for group differences of age between the patients with SS vs healthy subjects and the patients with SS vs the patients with RA was <0.001 and 0.018, respectively. In addition, the p value for group differences of sex between the patients with SS vs healthy subjects and the patients with SS vs the patients with RA was <0.001 and 0.025, respectively. Therefore, no significant differences in the distribution of age and sex were observed among these 3 groups. There were two cases (3.2%) of positive anti-Ro52 among the 62 patients with RA, and there was no positive case among the 50 normal control subjects patients with anti-Ro60 (82.6%). There were 34 non-SS subjects who were not classified as SS by AECG criteria. In the non-SS subjects, 7 subjects were positive for anti-Ro52 ( Fig. 2A).
The background characteristics of the patients with SS according to the presence or absence of anti-Ro52 are summarized (Table 1). Five items were significantly high in the SS patients with anti-Ro52.
We also examined the subjects' treatment for sicca symptoms and the use of immunosuppressants.
There was no case of congenital heart block (CHB) in the patients' medical records. According to their status as being with or without anti-Ro60, the SS patients with anti-Ro52 were divided 2 groups. Anti-Ro60+ SS patients included anti-Ro60+anti-Ro52+ patients (n=67), and the anti-Ro60+anti-Ro52− patients (n=28). In contrast, anti-Ro60-SS patients included the anti-Ro60-anti-Ro52+ patients (n=3), and the anti-Ro60-anti-Ro52− patients (n=17) (Fig. 2B). We also observed markedly high anti-Ro52 (≥500 U/ml) in 31 (44.3%) of the 70 anti-Ro52-positive patients with SS (Fig. 2C).  Clinical characteristics associated with relevance between anti-Ro60/Ro52 As described above, our present findings replicate the association between the positivity of anti-Ro52 and that of anti-Ro60 that had been reported previously [29,30]. We also determined the clinical characteristics with respect to the increment of anti-Ro60 concentration toward increment of the anti- xerophthalmia (−) and xerophthalmia (+). It is possible that the significance of the regression coefficients of the interaction terms is attributable to this clustered unbalancedness. As we noted above, it was also clear that the relationships between the levels of anti-Ro52 and anti-Ro60 were not linear, but we used a linear regression in order to grasp the characteristics of the global relationship.

The association between clinical characteristics and anti-Ro60/Ro52
We compared the concentration of anti-Ro52 and anti-Ro60 between the subgroups of subjects classified by components in the AECG criteria and by other items including RP, ACA, RF, and serum IgG (Suppl. Fig. S1). The comparison revealed that three characteristics (RP, ACA, and IgG) with which subgroups were created were significantly different according to the titer for both antibodies (p<0.05 for each characteristic). Regarding the anti-Ro52 concentration, the differences were significant in three additional characteristics: xerophthalmia, anti-La/SS-B antibody, and RF. For the anti-Ro60 concentration, there was no additional characteristic with which the association was significant aside from RP, ACA, and IgG.
We determined the maximal information coefficients to identify the clinical characteristics with significant dependency between anti-Ro52 or anti-Ro60 or both, and the results demonstrated that the dependency between anti-La/SS-B and both anti-Ro52 and anti-Ro60 was significant (p<0.05 for each antibody) (Suppl. Fig. S2). We also investigated which clinical characteristics are associated with the anti-Ro52 concentration by determining the ROC curve (Fig. 4). The characteristics in which positivity was highly discriminated by the concentration of anti-Ro52 were as follows: ACA, ESSDAI ≥1, and RF (AUC >0.75, respectively) (Fig. 4). Moderately discriminated characteristics were serum IgG, FS ≥1, and anti-La/SS-B antibody (AUC >0.70, respectively) (Fig. 4).
We analyzed the linear relationships between the ESSDAI score and levels of antibodies against the two respective subtypes of Ro/SSA antigens. The range of our subjects' ESSDAI scores was 0 to 17. In  Fig. S6).

Discussion
The results of our retrospective analyses demonstrated that anti-Ro52 was more frequently detected in the sera from patients with SS compared to the sera from RA patients and normal subjects. There were 7 patients with anti-Ro52 among 34 non-SS subjects. Among these 7 patients, 5 subjects were positive for anti-Ro60 and 2 subjects were positive for ACA without salivary gland biopsy. The reasons why these 7 subjects were not classified as SS by AECG criteria might be coexistence of anti-Ro antibodies and lack of information for pathological findings. In terms of relevance toward positive anti-Ro60, the odds of anti-Ro52 positivity were significantly high compared to anti-Ro52 and ACA. In a simple comparison without a test of the relevance between anti-Ro52 and anti-60, we observed that the group of SS patients who were anti-Ro52-positive had significantly high prevalences of anti-Ro60, anti-La/SS-B antibody, and RF, high serum IgG levels, and high ESSDAI scores. The extent of the increase in the level of anti-Ro52 for the extent in the level of anti-Ro60 was significantly higher in the subjects with high serum IgG and high RF. Contrary to these two characteristics, the extent was significantly lower in the subjects with the following items compared to those without them: xerophthalmia, xerostomia, ACA, and the presence of RP. Our results also identified six clinical parameters (including the ESSDAI score) that were significantly associated with the anti-Ro52 level.
Although Ro52 was initially reported as a component of SS-A particle, it is structurally distinct from Ro60. Ro52 has four structural domains: RING, B-box, coiled coil (CC), and B30.2/PRYSPRY regions [31]. Because motifs that have three regions including RING, B-box, and CC are describes as tripartite motif proteins (TRIMs), Ro52 is also described as TRIM21 [4,32]. With regard to the function of Ro52, the ubiquitination of intranuclear target molecules as well as E3 ligase activity are known [33]. Type 1 interferon is known to induce the translocation of Ro52 to the nucleus [34]. A relationship between the above molecular mechanisms and the clinical manifestations identified in the present study (including high serum IgG level and ESSDAI score) has not yet been determined, but the ubiquitination of Ro52 antigen or a change in cellular distribution might be associated with anti-Ro52-positive patients with SS. Since it is possible that the change of ubiquitination by anti-Ro52 may increase inflammation in each organ in patients with SS, the serum IgG level, RF, FS, and ESSDAI score (which showed an AUC >0.7 with respect to anti-Ro52) might be explained by the effect of anti-Ro52 on ubiquitination. In addition, this biochemical action of anti-Ro52 and the wide distribution of Ro52 antigen in various organs [35,36] might be related to the relevance between ESSDAI among the items with an AUC >0.7 and the change of anti-Ro52 concentration.
Clinically, anti-Ro52 is found in SS as well as other autoimmune diseases [37,38]. Although the specific biochemical reasons are not known, a high frequency of the co-expression of anti-Ro52 with anti-Jo-1 antibody was reported in patients with inflammatory myopathy, although there was no difference regarding Ro52 epitope recognition in the presence or absence of anti-Jo-1 antibody [39].
In addition, organ specificity of anti-Ro52 was repeatedly reported in patients with interstitial lung diseases [40] or CHB [41]. In contrast, there are some controversial data from patients with SLE or SSc [42], in which isolated anti-Ro52 had no significance to estimate the clinical activity of these autoimmune diseases; rather, those data indicated that comorbid autoantibodies had higher diagnostic value than anti-Ro52, as noted in a review [43].
Regarding the relevance between the anti-Ro concentration and clinical manifestations in SS, there are few reports (and no reports of a large study). In a 2006 study of pregnancies presenting a risk of CHB, decreased levels of anti-Ro52 IgG1 and IgG4 were observed [44]. It was recently reported that the concentration of anti-Ro, anti-Ro52, and anti-Ro60 were independent factors for fetal CHB [41].
The anti-Ro concentration was observed to fluctuate in SS patients with skin vasculitis, suggesting that a fluctuating concentration titer of anti-Ro was associated with the disease activity of SS, although that study reported only two skin vasculitis cases among 15 patients with SS [45]. In our present analyses, the anti-Ro52 concentration showed moderate (AUC >0.70) or high (AUC >0.75) discriminability for six clinical factors. The significant linear relationship between the ESSDAI scores and anti-Ro52 detected by the linear regression model in Figure 5 indicates that the anti-Ro52 level was relevant to clinical parameters including the disease activity in SS.
A user's guide to the ESSDAI [14] was released to precisely assess the clinical activity of SS. Ramos-Casals et al. examined a cohort of 921 Spanish patients with SS, and they reported that the most frequently involved organs were the joints, skin, and peripheral nerves [46]. Although a direct correlation between anti-Ro52 and ESSDAI score was not noted in that study, the reason why the details of our patients' ESSDAI scores (Suppl. Fig. S3) differed from the findings reported by Ramos-Casals et al. may be because our ESSDAI results revealed a low frequency of articular involvement and no peripheral nerve symptoms. In this report, the total ESSDAI score was lower than that in the SS Group of the Autoimmune Disease Study Group (GEAS) registry [46] in Spanish centers. In the registry, frequency of glandular and articular domains was 34% and 56% that was much higher than frequency in Suppl. Fig. S3. Since we estimate that low frequency of specific domains might influence on low ESSDAI score in Japanese patients with SS, ethnic differences should be taken into account since components of the ESSDAI might not be the same among different geographic regions. Our recent data supported that the frequency of specific domains in ESSDAI was certainly low according to ethnicity [47]. We observed non-ESSDAI cardiovascular items including 14.8% (17/115) of Raynaud's phenomenon that was reported recently [48] and 1 case of pulmonary arterial hypertension. In addition, 3 cases of autonomic dystonia were observed in 115 patients with SS.
Because presence of these clinical manifestations might influence on the frequency of items in ESSDAI, we should carefully observe extra glandular manifestations considering regional characteristics.
Regarding ACA, Earnshaw et al. identified centromere proteins (CENPs) including CENP-A, B, and C [49], which was followed by the molecular cloning of CENP-B [50]. Although CENP-A and C were shown to be involved in the assembly of the kinetochore [51], the function of CENP-B remains unknown.
Because CENP-B can bind the specific CENP-B box [52], it was predicted that CENP-B regulates the dynamic state of heterochromatin in centromeres. In a Spanish cohort, anti-Ro52 was found in 35.6% of the patients with SSc, and the patients with anti-Ro52 had a high prevalence of ACA (61.9%) [53].
Regarding the coexistence of ACA and anti-Ro52 in SSc, the Canadian Scleroderma Research Group (CSRG) observed that CENP-B was present in approx. 43% of 194 patients with anti-Ro52-positive SSc, although a biochemical relationship between anti-Ro52 and ACA was not demonstrated [54]. In contrast, we observed significantly low odds of anti-Ro52 seropositivity in our ACA+ SS patients. This paradoxical result suggests a potential difference in the coexistence of anti-Ro52 in ACA+ between SS and SSc. Concerning the differential diagnosis between ACA+ SS and ACA+ SSc, the potential future development of SSc during the course of ACA+SS is an important issue. Given the descriptions that anti-Ro52+SSc had a high prevalence of ACA, the low prevalence of anti-Ro52 in ACA+ SS patients can be helpful to distinguish these two disorders.
Our study has some limitations. We did not use the 2016 ACR/EULAR criteria because the ocular staining test, which is one of the main items in 2016 ACR/EULAR criteria, was not performed in many of our patients. In addition, the use of the Saxon test (one of the stimulated salivary secretion tests       The relevance of anti-Ro52/60 antibodies according to ESSDAI score. The line with the intercept and slope obtained from the linear regression (gray line: 95%CI) and box-whisker plots (boxes: interquartile ranges; whiskers extend from the upper/lower hinges to the highest/lowest value no further than 1.5 times the ranges of the 1st and 3rd quartile points).
The y-axis data were scaled on logarithmic with base of 10 in panels A and B and not scaled in panels C and D. ESSDAI: EULAR Sjögren's Syndrome Disease Activity Index.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.