Thymus variants on imaging of patients with primary Sjögren’s syndrome and polymyositis/dermatomyositis: clinical and immunological significance

Abstract We investigated the presence of radiographic thymus variants using a scoring system and examined their association with clinical and immunological features in primary Sjögren’s syndrome (pSS) and polymyositis/dermatomyositis (PM/DM) patients. Cases of 72 patients with pSS and 47 with PM/DM were randomly selected from all visitors to our department who received chest CT scanning, excluding those with thymoma or thymic cyst, or age <30 years. We quantitatively interpreted and assessed thymus size and attenuation score in axial CT images. Thymic enlargement was identified in 16 (22.2%) pSS and 14 (29.8%) PM/DM patients. A thymus attenuation score ≥ 2 was seen in 11 (15.3%) pSS and 9 (19.1%) PM/DM patients. Thymic enlargement showed a significant association with the titre of rheumatoid factor in PM/DM patients. Thymic enlargement and score showed a significant association with body weight in pSS patients. Radiographic thymus variants are often observed in pSS and PM/DM patients, particularly in cases of PM/DM, and may suggest the role of an abnormal immune response in their pathogenesis.


Introduction
Autoimmune diseases affect 5-8% of the general population. They are due to dysregulation of the immune system, which causes it to attack specific molecules and/or cells in the body [1]. The thymus is the source of T cells, which must be able to react to foreign molecules entering the body. This process is mediated by interactions between T cells and thymic epithelial cells [1]. The development of naïve T lymphocytes in the thymus follows a series of events involving antigen-presenting cells (APCs) [2]. Mechanisms within the thymus ensure that cells with autoreactive specificity are eliminated before they emigrate to the periphery and control the generation of thymic regulatory T cells [3], to finally result in the production of mature functional T cells which are tolerant to self [4]. The thymus therefore plays a critical role in immune system homeostasis.
A previous study evaluated the appearance of the thymus on computed tomography (CT) in 2540 healthy individuals using a four-point score according to ratio of fat and soft tissue, size and morphology [5]. Visual assessment with this four-point thymus score showed the normal spectrum of the thymus and sex differences in fatty degeneration of the thymus by age. The greater the amount of fatty degeneration, the higher the thymus attenuation score. The prevalence of a thymic score ! 2 in that study was 8% (Score 0, 74%; Score 1, 18%; Score 2, 7% and Score 3, 1%).
The thymus is reported to undergo changes in both systemic and organ-specific autoimmune conditions. Myasthenia gravis (MG) is an organ-specific autoimmune condition resulting from the presence of autoantibodies to components of the fascia at the neuromuscular junction. Thymic abnormalities observed in these patients include thymoma or follicular thymic hyperplasia [6]. The presence of antiacetylcholine receptor antibodies in these patients is correlated with the degree of follicular hyperplasia [7]. Some MG patients with thymoma improve following thymectomy, further implicating the thymus in its pathogenesis. In systemic sclerosis, radiographic abnormalities such as thymic hyperplasia, nodularity and incomplete thymus involution are significantly more frequent than in controls, and have been associated with a shorter disease duration and skin involvement [8]. Recently, we reported radiographic thymus variants in rheumatoid arthritis (RA) patients [9]. In that study, the prevalence of thymic enlargement was 19.6%, while the prevalence of a thymus attenuation score !2 was 12.9%. The thymus attenuation score was significantly associated with positivity for RA-related autoantibodies.
Primary Sj€ ogren's syndrome (pSS), a common systemic autoimmune disease, is characterized by lymphocytic infiltration of the exocrine glands [10]. Polymyositis/dermatomyositis (PM/DM) is a rare systemic autoimmune disease characterized by muscle inflammation as well as nonmuscular manifestations [11]. In our previous patient cohort, in which we examined thymic enlargement and thymus attenuation scores among several autoimmune diseases, including RA [9], pSS and PM/DM were the two disease entities with the highest prevalence of thymus variants (Supplementary 1A and, 1B; unpublished data). Although several studies [6][7][8][9] have investigated associations between radiographic thymus variants and serological features in systemic autoimmune diseases, information in patients with other systemic autoimmune disease, especially pSS or PM/DM, is limited. Against this background, we hypothesized that those diseases with the highest prevalence of thymus variants would be the most likely to show an association with immune phenotypes, similar to the previously reported RA [9], and that it might be possible to show that different diseases have different reliance on thymic function.
Here, we investigated the association between radiographic thymus variants and clinical and immunological features in patients with pSS and PM/DM and clarified its significance.

Patients
This large cross-sectional study included 72 pSS patients and 47 PM/DM patients randomly selected from all patients visiting the Division of Rheumatology, Keio University Hospital who were evaluated by chest CT scanning between January 2013 and December 2015. We excluded patients with thymoma or thymic cyst and those aged <30 years [9], following reports that thymus gland involution continues until age 30 years [12].
All procedures in this study were approved by the medical ethics committee of Keio University School of Medicine (approval no.: 20130506). The study was conducted in accordance with the Declaration of Helsinki. All samples were obtained after patients and had given written informed consent.

Radiographic assessment
Thymus size and attenuation score on axial CT images were quantitatively interpreted and assessed by two observers who were unaware of the clinical and serological status of the patients. Figure 1 shows the different thymus appearances and their scores.
Differences in the interpretation between the two observers were resolved by discussion to obtain a consensus. Thymic enlargement was defined as a thickness of >13 mm and graded on a four-point scale (score 0-3) according to previous studies [13,14] as follows: Score 0, complete fatty degeneration; Score 1, predominantly fat density; Score 2, half solid tissue and half fat attenuation, or solid tissue foci >7 mm and Score 3, predominantly solid tissue [9] (Figure 1).

Statistical analysis
Fisher's exact test or Wilcoxon's rank-sum test was used to test the association of serological features with thymic enlargement. The Kruskal-Wallis test or Cochran-Armitage test was used to test the association between serological features and thymus attenuation score. A p-value <.05 was regarded as significant. Baseline variables in thymic enlargement in pSS with p < .05 on univariate analysis were considered in multivariate analysis using logistic regression applied to a nominal logistic regression model. The results with p < .05 were regarded as significant. All statistical analyses were conducted with commercial software (JMP v.12.2.0, SAS Institute Inc., Cary NC, USA).

Radiographic thymus variants in subjects with PSS
The demographics of pSS patients are characterized in Table 1. Mean age at diagnosis was 62.7 ± 12.2 years. Disease duration in the patient population was relatively short (median disease duration: 1.3 years). Thymic enlargement was identified in 16 (22.2%), while a thymus attenuation score ! 2 was seen in 11 (15.3%). The role of radiographic thymus variants in pSS patients was characterized by investigating its association with clinical and laboratory findings (Table 1). Radiographic thymus variants in pSS patients, both thymic enlargement and an thymus attenuation score, showed a significant positive association with body weight (p ¼ .0073 and .037, respectively), and thymic enlargement was significantly positively associated with the rate of oral glucocorticoid usage (p ¼ .036). There was no significant difference between the prevalence of RF-positivity, anti-Ro (SS-A) antibody positivity or anti-La (SS-B) antibody positivity and radiographic thymus variants. On multivariate logistic regression analysis, thymic enlargement was significantly positively associated with body weight (odds ratio [

Radiographic thymus variants in patients with PM/DM
Demographics of PM/DM patients are shown in Table 4. Based on these data, mean age at diagnosis was 56.2 ± 13.7 years. Thymic enlargement was found in 14 (29.8%), with a lower proportion of females (42.9%), while a thymus attenuation score !2 was found in 9 (19.1%). Thymic enlargement was significantly positively associated with titres of serum RF (p ¼ .046). The thymus attenuation score showed a significant positive association with titres of serum IgG (p ¼ .042) and a significant negative association with age (p ¼ .033). There was no significant difference between myositisspecific antibody positivity and radiographic thymus variants.

Discussion
In this study, we showed that the prevalence of thymic enlargement and thymus attenuation score was higher in pSS and PM/DM patients than in the general populations examined in previous reports [5,6,15,16]. Thymic enlargement was significantly associated with titres of serum RF in PM/DM patients. Although the significance of serum RF titre in the pathogenesis of PM/DM remains unclear, high titres may reflect immunological activity, as is seen in RA patients. In addition, thymus attenuation score was significantly associated with titre of serum IgG. In general, high titres of IgG often reflect abnormalities of cytokines such as TNF, IL-1b and IL-23, suggesting an abnormal immune response. Thymic involution occurred according to age. In a previous study of healthy individuals, a four-point scale visual assessment of the thymus negatively correlated with participant age [5]. A total of 74% of participants, whose mean age was 58.9 years,  52.2 ± 7.7 (16) 59.2 ± 13.0 (9) .013 Ã .057 102.3 ± 127.7 (17) 32.9 ± 18.1 (10) 85.6 ± 90.7 (8) .47 demonstrated a completely fatty thymus (Score 0) [5]. On the other hand, in our study, the mean age of PM/DM patients with a thymus attenuation score ! 2 was 48.1 years. Therefore, the statistical analysis may have significantly differed due to incomplete thymic involution with age. In this regard, healthy female controls showed less thymic involution compared with men from 40 to 69 years of age [5].

SS-
A total of 15% of SSc patients had an abnormally enlarged thymus in CT images and 6% had a nodular thymus. These variants were not seen in any control subjects [15]. Further, thymus variants were associated with positivity for anti-Scl70 antibody [15]. In RA patients, incomplete thymic involution-defined as residual thymus tissue measuring >7 mm-was associated with administration of biotherapy, suggesting an association with the severity of disease. The thymus attenuation score was significantly associated with the positivity of RA-related autoantibodies [9]. This thymus variant was also not seen in any control patient [16]. In this study, the presence of myositis-specific antibodies was not significantly associated with thymus variants. This may reflect the   idea that, unlike IgG and RF, myositis-specific antibodies are generally not indicators of immunologic activity. Elevated titres of serum RF and IgG in PM/DM patients with thymus variants were considered indicators that suggested the presence of an abnormal immune response. This may indicate that increased T cell activity within the morphologically abnormal thymus promotes autoantibody production by B cells. Accordingly, the titres of serum RF and IgG in PM/DM patients were considered to support the suggestion that thymus variants indicated the presence of an abnormal immune response. In this study, there was no significant difference between thymus variants and the titres of serum IgG or RF in pSS patients. This is because we did not have sufficient power to detect these differences. If the analysis period was extended and the number of cases increased, IgG and other factors might also appear to be significant indicators of pSS thymus abnormalities. On the other hand, regarding the association between both thymus variants and body weight, one report noted that participants with lower thymus attenuation scores showed significantly higher BMI than those with higher thymic scores [6]. In another report, high BMI was associated with a decline in T-cell production [17]. In contrast, our present results showed the opposite, although we did not investigate thymic T cell populations. Further, our multivariate logistic regression analysis showed a significant positive association between thymus variants and body weight, albeit that the OR was low and clinical significance may accordingly be limited. Further investigation of this association is required to show that it is real and not a Type I error. Several limitations of our study warrant mention. First, it was conducted under a retrospective crosssectional design and included patients under treatment. The effect of treatment on thymus variants cannot therefore be excluded. We previously assessed thymus variants affected by treatment in 27 RA patients during accumulation of imaging data in $500 patients. Thymus changes after treatment were shown to be heterogeneous, including a lower attenuation score, larger size, smaller size or no change. Second, evaluation was limited to the thymus on chest CT scan data and enlargement of secondary lymphoid tissues, such as lymph nodes and spleen, was not evaluated.
Our study showed that radiographic thymus variants can be frequently observed in pSS and PM/ DM patients. In particular, thymic enlargement in PM/DM patients was significantly positively associated with titres of serum RF and the thymus attenuation score showed a significant positive association with titres of serum IgG. Radiographic thymus variants may reflect the presence of an abnormal immune response in the pathogenesis of PM/DM.