Graves’ orbitopathy as a rare disease in Europe: a European Group on Graves’ Orbitopathy (EUGOGO) position statement

The data provided by one of the major publications on the incidence of GO [20] did not include age-specific intervals. However, the authors stated that “age distribution differed in GO and Graves’ hyperthyroidism, with GO being less than 2% of Graves’ hyperthyroidism at age 20–40 year and 8% at age 40–60 year.” We therefore estimated age and sex-specific incidences derived from [20] and used them to calculate prevalence (Table 2).

This is the only study which directly assessed the incidence of GO in a large European population prospectively and included new cases between 1992 and 2011, based on approximately 8.9 million person-years of observation [20]. The incidence of moderate-to-severe GO was 0.161/10,000/year with a median age at onset of 50 years [20]. Mild cases of GO were not included due to difficulties in defining mild GO in large-scale epidemiological surveys. Distinguishing mild GO from normality and from the transient ocular effects of thyrotoxicosis of any cause, can be a challenge, especially if multiple observers are involved with variable degrees of expertise. In recent cross-sectional studies from secondary or tertiary centres, conducted by highly trained observers and with well-defined criteria for GO, about 65% of all cases of GO were found to have mild GO, and about 2% sight-threatening, disease [11, 16]. We assumed that the cases of mild GO which were excluded by Laurberg et al. [20], accounted for 2/3 (65%) of all cases of GO. Hence, the adjusted figure for incidence of all grades of severity of GO, based on the above studies [11, 16, 20], is 0.483/10,000/year (incidence of moderate-to-severe disease 0.161 multiplied by 3, in view of the fact that it represents 1/3 of all cases = 0.483). The data published by Laurberg et al. [20] are unclear as to whether sight-threatening GO was included within the moderate-to-severe category. We therefore used the frequency of sight-threatening GO of 2% cited by other publications [11, 16] to calculate the prevalence of sight-threatening GO from data by [20]. Using the approach described in “methods” the prevalence of GO is estimated to be 8.97/10,000 population. It can be further broken down to mild (5.83/10,000), moderate-to-severe (2.96/10,000) and sight-threatening GO (0.18/10,000) (Table 3a).

Based on data for incidence of Graves’ hyperthyroidism, the calculated incidence of GO in Europe has been reported to be 10–30/10,000/year [26‐28]. These studies relate to epidemiological data collected 10–20 years ago and the estimates may not be valid today since the incidence of GO appears to be declining [11, 14, 15] although this may be partly due changing referral patterns. In a recent study, Abraham-Nordling et al. [18] reported the incidence of Graves’ hyperthyroidism derived from a Swedish population of about 3.5 million. The incidence of Graves’ hyperthyroidism was 2.10/10,000/year. Mild (“non-infiltrative”) GO was present in 15.2% of patients with Graves’ hyperthyroidism and moderate-to-severe/sight-threatening (infiltrative) in 4.9% of cases. Thus, the incidence of all cases of GO was 0.42/10,000/year (incidence of mild GO 0.32/10,000/year and of moderate-to-severe/sight-threatening 0.10/10,000/year) [18]. Using the same methodology for calculating prevalence as before, the prevalence of GO is shown in Table 3a. The study by Zaletel et al. [19] yielded remarkably similar results for the incidence of Graves’ hyperthyroidism as that by Abraham-Nordling et al. [18] (2.08/10,000/year), and identical figures for the prevalence of GO (Table 3a).

The figures for overall prevalence of GO derived from [18] and [20] vary almost by a factor of 2. The reason is unclear but it may represent a true variation in different populations, or may be accounted for by the different methodologies in identifying cases. Laurberg et al. [20] presumed that all patients with moderate-to-severe GO were seen in a single tertiary centre. Hence, the data have included some patients with GO seen elsewhere and patients who were not referred. The other study [18] identified patients with Graves’ hyperthyroidism through a registry and it may have captured more cases. One of the weaknesses of our calculations is that the estimates of prevalence of GO are derived from a small (n = 3) number of epidemiological studies. However, this is all that is available in the recent literature and it is unlikely that more extensive data will be forthcoming for a while. The consistency between the figures from the three studies quoted above is sufficient in our view to justify estimates of prevalence based on these data.

A number of distinct clinical variants of GO exist, however these are poorly studied and deserve attention. Euthyroid/hypothyroid GO is a variant comprising of patients who are euthyroid, or have overt or subclinical hypothyroidism, before or within 6–12 months after the onset of GO and constitutes 0.2–11% of all cases of GO [29‐33]. Patients with GO associated with dermopathy have characteristic skin changes and comprise 1.5% of all cases of GO [7]. Patients with GO associated with acropachy display typical nail and subperiosteal changes and make up 0.3% of all cases of GO [7]. Asymmetrical GO, usually defined as a difference of ≥2 mm in proptosis between the two eyes, is reported to occur in 10–50% of patients with GO [30, 34‐36]. Unilateral GO, defined as one of more features of GO in one eye without any in the contralateral eye, is reported to occur with a frequency of 5–15% of all cases of GO [35‐38]. Previously reported data on the frequency of euthyroid and asymmetrical/unilateral GO are likely to be overestimates as awareness of IgG4-related ophthalmic disease has been low [39, 40], but is impossible to quantify. The estimated prevalence of these variants (inclusive of all grades of severity) based on the previously estimated overall prevalence of GO of approximately 10/10,000 population, and the reported frequencies of the variants is shown in Table 3b.

One of our assumptions in calculating the prevalence of GO was that the duration of GO is life-long. In a minority of patients this is not the case. In a study published 20 years ago by Bartley et al. [41], 40% of patients responding by questionnaire reported that their eyes were normal 10 years after the diagnosis of GO, but this was not possible to confirm objectively. A study published in 2002 by Terwee and colleagues [42] reported on outcomes in a cohort of patients after 11.7 years of follow-up, but the data did not report on the frequency of complete regression of the features of GO. In a more recent study a significant proportion of patients had transient GO [11]. However, this was a selected group of patients with Graves’ hyperthyroidism who had no GO at the time of recruitment and were treated with anti-thyroid drugs, from a single institution. It is therefore impossible to correct for transient GO with an acceptable degree of confidence without taking a risk to over-state the case for variants of GO being rare. However, it is important to highlight that the data in Table 3 are likely to be overestimates of the prevalence of GO.

Designation of rare disease in Europe requires that several conditions be satisfied: prevalence of less than 5 per 10,000 population [43], distinct phenotype, distinct aetiology and mechanism of pathogenesis [44] and seriously debilitating or life-threatening impact on patients [45]. Euthyroid GO achieved rare disease status on the basis of the above criteria, and the consideration that the euthyroid status of such patients reflects mechanistically unique processes that are distinct from Graves’ hyperthyroidism. We speculate that the other clinical variants described above may also have underlying unique and distinct pathogenetic processes from Graves’ hyperthyroidism and each other. Future research focusing on deep phenotyping and molecular characteristics of variants of GO is likely to be productive in understanding the pathogenesis of GO.