Thyroid Autoimmunity at Onset of Type 1 Diabetes as a Predictor of Thyroid Dysfunction

Between September 1987 and January 1994, 204 patients were consecutively diagnosed with type 1 diabetes at our institution. One patient had developed primary hypothyroidism 3 years before the development of type 1 diabetes, 10 patients were not followed up after a few months from diagnosis, and thyroid autoimmunity at diagnosis was not determined in 17 patients. The remaining 176 patients were included in the study. The mean age at diabetes diagnosis was 24.4 ± 9.4 years, 65% were male subjects, and the percentage of islet cell antibody positivity was 76%. The mean follow-up was 9.6 ± 4.3 years. Thyroid autoimmunity was determined at onset by hemagglutination, measuring thyroperoxidase autoantibodies (TPO-Abs) and tyroglobulin autoantibodies (Tg-Abs), and were considered positive at a dilution of ≥1/100. All patients underwent clinical follow-up, and thyrotropin (TSH) was monitored every 1–2 years using a commercial noncompetitive immunoassay. Thyroid dysfunction was diagnosed in the presence of a serum TSH alteration (normal range 0.3–5 mUI/l) with or without symptoms. Statistical analysis was performed with SPSS version 14.0 for Windows using χ² and Student's t tests. Data are expressed as means ± SE. Statistical significance was established at P < 0.05.

Twenty-five of 176 patients (14.2%) were TPO-Ab positive at type 1 diabetes diagnosis. Eighteen of these 25 patients developed thyroid dysfunction at follow-up (primary hypothyroidism, defined as TSH >5 mUI/l in 17 cases and hyperthyroidism in 1 case). Only 1 of the 151 negative subjects at type 1 diabetes diagnosis developed hypothyroidism; he was a 14-year-old subject at type 1 diabetes onset who later turned TPO-Ab positive and developed hypothyroidism 15 years after diabetes diagnosis. Tg-Abs were less sensitive than TPO-Abs, being negative in 8 of 18 patients who developed hypothyroidism. The mean time to thyroid dysfunction development from type 1 diabetes onset was 8.1 ± 4.2 years. Interestingly, the seven TPO-Ab–positive patients who maintained normal thyroid function after 8.0 ± 5.0 years of follow-up were younger at type 1 diabetes onset than the patients who developed thyroid dysfunction (20.1 ± 6.2 vs. 29.5 ± 11.8 years; P < 0.05). Furthermore, patients who developed thyroid dysfunction were older at type 1 diabetes diagnosis than those who did not (28.7 ± 12 vs. 23.9 ± 8.9 years; P < 0.05). As expected, there were also sex differences, with thyroid dysfunction more frequent in female subjects than in male subjects (11 of 61 [18.3%] and 8 of 114 [7.0], respectively; P < 0.05)

Table 1 shows the prevalence of thyroid dysfunction according to the presence of TPO-Abs at type 1 diabetes onset. Further data analysis showed that the presence of TPO-Abs at type 1 diabetes onset can predict the development of thyroid dysfunction with 95% sensitivity (18 of 19 cases of future thyroid dysfunction were detected), 96% specificity (150 of 157 patients were correctly classified as normal), a 99% negative predictive value (only 1 of 151 patients classified as normal later developed thyroid dysfunction), and 72% positive predictive value (7 of 25 patients remained normal in the follow-up).

The present study confirmed the high prevalence of thyroid autoimmunity and thyroid dysfunction in young adults with type 1 diabetes. In addition, the determination of TPO-Abs at type 1 diabetes onset was highly sensitive in the prediction of the development of thyroid dysfunction.

Although there is general agreement that the high prevalence of thyroid dysfunction in type 1 diabetes subjects justifies screening in all patients (5,6), it is not clear which is the best procedure and how often to perform it. The American Diabetes Association (7) and several authors (8,9) recommend annual screening for thyroid disease in all type 1 diabetes subjects with TSH measurement; this procedure is considered the most sensitive way to identify patients with thyroid dysfunction, as autoantibodies may persist for many years without thyroid dysfunction. However, the American Diabetes Association recommendations note that the presence of thyroid autoantibodies increase the risk for thyroid disease, and Hansen et al. (8) did not find any initial TPO-Ab–negative patients who developed thyroid disease after 3 years of follow-up. Other authors (10,11) recommend screening using TSH and TPO-Abs. In a cohort of 58 type 1 diabetic patients enrolled in the Diabetes Control and Complications Trial and followed for 18 years, Umpierrez et al. (10) observed that TPO-Ab–positive subjects were 17.9 times more likely to develop thyroid dysfunction. These authors recommended annual screening using TSH determination, particularly in patients with positive TPO-Abs. Barker (11) screened type 1 diabetic patients with TPO-Abs and thyroid function at onset and every 1–2 years thereafter and patients with positive TPO-Abs every 6–12 months. Finally, a third group of authors (1,12) recommends TSH determination only in TPO-Ab–positive patients. In our study, the high sensitivity and especially the high negative predictive value of TPO-Abs for the development of thyroid dysfunction over a mean follow-up of 9.6 years support a screening strategy with determination of TPO-Abs in all type 1 diabetic subjects at diagnosis. In agreement with previous studies (8,13), Tg-Abs were less sensitive and did not increase TPO-Ab–positive predictive value. Thereafter, an annual TSH determination would be performed only in subjects with positive autoantibodies. Using this procedure, only one case of hypothyroidism was missed, appearing 15 years after the onset of diabetes in a patient who was a 14-year-old at diabetes onset and who was later TPO-Ab positive.

Considering that the prevalence of TPO-Ab positivity may increase with time (1,3,8), further studies are required to investigate if additional determination of TPO-Abs after diagnosis may be useful, specially in younger patients.

This study was funded by Grants ISCIII-RETIC RD06, PI052099 and PI051540 from the Fondo de Investigación Sanitaria, Carlos III Health Institute, Spain.