Introduction

Overstimulation of the thyroid-stimulating hormone (TSH) receptor leads to development of hyperthyroidism, accompanied by thyroid eye disease (TED) as the most common extrathyroid manifestation of Graves’ disease (GD). Graves’ orbitopathy (GO) is a potentially vision-threatening chronic autoimmune disease of the orbit and retroocular tissues.1-3

It is widely observed that individuals previously diagnosed with an autoimmune condition, such as autoimmune thyroid disease (AITD), commonly exhibit a concurrent presence of other autoimmune conditions, a phenomenon commonly referred to as polyautoimmunity.1,4,5 Antithyroid antibodies and disturbances in thyroid function have been commonly observed in individuals diagnosed with rheumatologic autoimmune conditions, including Sjögren syndrome (SS), rheumatoid arthritis (RA), systemic lupus erythematosus, and systemic sclerosis.4,6,7 Other autoimmune diseases (OAIDs) coexisting with AITD also include celiac disease (CD) and type 1 diabetes mellitus (T1DM).8

Zinc transporter 8 (ZnT8) is expressed predominantly in the pancreatic β-cells and has been found to be specifically associated with T1DM.9 Additionally, in a murine model, ZnT8 is also expressed in other secretory cell types, including the cuboidal epithelium of thyroid follicles and the adrenal cortex.10 Also, ZnT8 protein expression was found in thyroid follicular cells and C cells from patients with GD and nontoxic nodular goiter.11 In keeping with that, prevalence of ZnT8 was higher in nondiabetic AITD patients than in healthy controls, but lower than that in T1DM patients with AITD.12,13 Moreover, although antibodies against ZnT8 are regarded as an independent marker of autoimmunity for T1DM diagnosis, positive ZnT8 autoantibody status in patients with autoimmune Addison disease indicates heightened susceptibility to development of additional autoimmune conditions.14

This retrospective study aimed to determine the prevalence of OAIDs in patients with active moderate-to-severe GO. Also, due to a potential relationship with OAID co-occurrence, ZnT8 antibody status was assessed.

Patients and methods

Patients

We conducted a retrospective single-center study involving 185 patients with active moderate-to-severe GO hospitalized from 2014 to 2022 in the Department of Endocrinology, Metabolic Disorders and Internal Medicine at Poznan University of Medical Sciences, Poznań, Poland.

Data were collected from electronic medical records. Demographic and clinical data included age, sex, body mass index (BMI), disease duration, and activity of GO (based on the clinical activity score). We also obtained laboratory test results, including TSH, free triiodothyronine (fT3), free thyroxine (fT4), TSH receptor antibodies (TRAb), thyroid peroxidase antibodies (TPOAb), and thyroglobulin antibodies (TGAb).

OAIDs were recognized in patients only if the diagnosis was confirmed by a proper specialist and recorded in the medical documentation.

Additionally, serum antibodies against ZnT8 were determined in randomly selected patients using Random Integer Set Generator (www.random.org). We requested 1 set with 40 unique random integers, taken from the 1–185 range. ZnT8 antibodies were measured by an enzyme-linked immunosorbent assay kit (RSR Ltd., Cardiff, United Kingdom), which detects autoantibodies specific to arginine, tryptophan, or other nonspecific variants of residue 325. A cutoff value of at least 15 U/ml was considered positive.

Ethics statement

In collecting, analyzing, and reporting data, we adhered to the ethical standards of the Declaration of Helsinki. Due to the retrospective design of the study, a consent of a bioethics committee was not required.15

Statistical analysis

All statistical analyses were performed for 40 patients with GO. Thirty GO patients without autoimmune diseases were matched for sex, age, and BMI to a group of 10 GO patients with autoimmune diseases (3:1 ratio). The patients were matched exactly for sex, for age with a margin of 5 years, and for BMI with a margin of 2.5 kg/m2 using R statistical software (RStudio Inc., Boston, Massachusetts, United States).

Quantitative variables with non-normal distribution were presented as medians and interquartile ranges, and continuous data were compared with the nonparametric Mann–Whitney test. The qualitative variables were compared with the 2-sided Fisher exact test. Also, receiver operating characteristic (ROC) analysis was performed to discriminate GO patients with CADs using levels of selected thyroid function parameters.

The significance level was set at a P value below 0.05 for all analyses. The statistical analysis was performed with Statistica 13.3 software (StatSoft, Kraków, Poland).

Results

Among 185 patients with active moderate-to-severe GO, CADs were reported in 10 individuals (5.4%). These diseases included psoriasis (n = 4), RA (n = 3), myasthenia (n = 1), vitiligo (n = 1), and CD (n = 1).

To compare thyroid function parameters, a group of 30 GO patients without OAIDs matched for sex, age, and BMI was established. The group characteristics are presented in Supplementary material, Table S1.

TSH and fT4 levels were slightly lower in the GO patients but the differences were not significant (P = 0.13 and P = 0.99, respectively). In contrast, fT3 levels increased in the GO patients with CADs, as compared with the control GO patients (P = 0.04).

The GO patients without OAIDs had lower levels of TPOAb than the control individuals with GO (P = 0.04). We observed that TRAb and TGAb levels were lower in the GO patients with OAIDs (P = 0.16 and P = 0.32, respectively).

The ROC analysis confirmed that selected thyroid function parameters could predict the presence of OAIDs in patients with GO. Useful predictors turned out to be TPOAb (area under the curve [AUC; SE], 0.725 [0.086]; P = 0.009) and fT3 (AUC, 0.730 [0.103]; P = 0.03), but not TRAb (AUC, 0.653 [0.085]; P = 0.07).

Figure 1 shows a scatterplot with distribution of both subgroups of GO patients (with and without CADs) with reference to the levels of TRAb, TPOAb, and fT3.

Figure 1. Scatterplot of triiodothyronine (fT3) vs thyroid-stimulating hormone receptor antibody (TRAb) vs thyroglobulin antibody (TPOAb) in patients with Graves’ orbitopathy (GO) with and without other autoimmune diseases

Moreover, 40 randomly selected patients displayed negative ZnT8 antibody status (median level, 8.9 U/ml), irrespective of the presence or absence of OAIDs.

Discussion

In our study, we found a relatively low prevalence of OAIDs in the patients with active moderate-to-severe GO. Only 5.4% of the patients with active GO reported CADs.

This is in contrast with other literature reports. Kelada et al1 found a 13.9% prevalence of OAIDs among TED patients. Among them, RA (3.7%), vitiligo (3%), and psoriasis (3%) were the most common. Moreover, TED patients with OAIDs tended to present significantly higher clinical activity scores than those without OAIDs. In our study, we did not observe such a relationship.

A study by Ferrari et al4 demonstrated that 16.7% of GD patients had a OAID. The most frequently observed ones were vitiligo (2.6%), chronic autoimmune gastritis (2.4%), and RA (1.9%). Furthermore, 1.5% of GD patients presented 3 OAIDs. The patients with GO were more likely to suffer from another autoimmune disease (18.9%) than the non-GO individuals with GD (15.6%). However, the pattern of OAIDs in GD patients was not significantly different.

In subsequent studies, nonthyroid autoimmune diseases were detected in 9.4%16 and 9.67%17 of patients with GD. Similarly, the most common diseases included autoimmune gastritis, vitiligo, RA, CD, T1DM, SS, and multiple sclerosis. However, it is impossible to define a single pattern of the most often OAIDs.

Furthermore, we found that among thyroid function parameters, fT3 levels were significantly higher in the GO patients with OAIDs, as compared with matched GO patients without these conditions. In contrast, TPOAb levels were significantly reduced in the same setting. Based on that observation, we can speculate that elevated TPOAb levels protect against OAIDs. Interestingly, Meneghini et al18 indicated that positive TPOAb was a protective factor for psoriasis. Also, a study by Hansen et al19 reported that psoriasis was associated with increased levels of fT3 but not with TPOAb positivity and clinical or subclinical dysfunction. Another explanation is the fact that treatment modalities used to manage autoimmune diseases, such as immunosuppressive medications or biologic agents, can potentially further suppress immune function. These therapies are designed to reduce autoimmune activity and inflammation, with potential immunomodulatory effects that may extend to the suppression of autoantibody production. Consequently, individuals receiving immunosuppressive therapy for multiple autoimmune diseases may experience a decrease in the production of autoantibodies, contributing to reduced levels of TPOAb.20,21

Moreover, in our study, all selected patients had negative ZnT8 antibody status. Thyroid autoimmunity is common in patients with T1DM, whereas the issue of islet autoimmunity in patients with GD is less studied. Most studies in this area have examined the presence of ZnT8 in patients with both T1DM and T2DM, and only a few studies considered autoimmune diseases, such as GD and Addison disease.8,14,22

A study conducted in Sweden among patients with newly-diagnosed thyrotoxicosis showed that the prevalence of positive ZnT8 autoantibody status was 7.6%.22 Furthermore, Larizza et al23 demonstrated that positive ZnT8 status was relatively low in AITD patients (6.4%), and higher in AITD patients with concomitant CD (12.5%). In turn, in CD patients, ZnT8 was not detected.

The main limitation of our study is its retrospective design, in which biases related to the data collection and their potential incompleteness are to be expected. Not always patients had to report and doctors record all concomitant diseases in the medical documentation. Regarding the presence of ZnT8 antibodies, it may depend on the type and activity of autoimmune disease. In our group, all patients had active GO, but none had concomitant T1DM, and patients were randomly selected to determine their antibody status.

In summary, we observed a relatively low concomitance of OAIDs in patients with active moderate-to-severe GO. One can speculate that immunosuppressive therapy implemented during the course of OAIDs may have a protective effect and indirectly reduce the risk of development or progression of ocular symptoms in GD. The data regarding the impact of biologic drugs used in rheumatologic diseases on thyroid autoimmunity are divergent and undoubtedly require further research.24 However, it should be noted that some of these drugs, such as rituximab or tocilizumab, are used in patients with GO.25 Finally, in patients developing nonspecific symptoms, a screening for these diseases is indicated. Also, monitoring antibody levels may be helpful in assessing the dynamics of autoimmune processes. Therefore, further prospective studies on large populations are necessary.