Understanding Hashimoto's thyroiditis: why the immune system attacks the thyroid

Hollowell and colleagues measured TSH, thyroxine and both thyroid antibodies in more than 17,000 people aged twelve and over as part of the large US health survey NHANES III. The sample was drawn so that it reflects the geographic and ethnic distribution of the US population.

TPO antibodies were found in 11.3 percent, Tg antibodies in 10.4 percent. Both were more common in women and increased with age. An important secondary finding: TPO antibodies were clearly linked with over- or underactive thyroid function, whereas isolated Tg antibodies were not.

For you this means two things. Antibodies alone are not yet a disease, they are widespread. And the TPO value is the more informative of the two.

The Whickham cohort in northern England is one of the few studies that has followed an entire population sample over twenty years. Vanderpump and colleagues were able to reconstruct the course for over 97 percent of the original 2,779 participants.

Women with positive thyroid antibodies had roughly an eightfold higher risk of developing an underactive thyroid by the end of the observation period, with a confidence interval of 5 to 15. When elevated TSH values were added, the odds ratio rose to 38 in women and to 173 in men. The authors also showed that TSH values above 2 mU/l already raised the probability.

This is the reason antibodies are measured at all. They do not tell you how you are doing today. They say something about the direction of the coming years.

Liu and colleagues determined the proportions of Th17 and regulatory T cells in the blood of people with Hashimoto. The ratio of Th17 to Treg was increased and correlated positively with the level of thyroglobulin antibodies.

In addition they found an elevated level of the messenger GITRL, which in turn was linked to the proportion of Th17 cells. Their interpretation: GITRL may contribute to disturbing the balance between the two cell types.

An Iranian research group independently measured the corresponding transcription factors in 40 women with Hashimoto and 40 healthy controls. FOXP3, the marker of the brake, was significantly reduced, while the activators for Th1 and Th2 were increased.

Weetman summarises the state of knowledge like this: HLA-DR alleles as well as the genes for CTLA-4, PTPN22, FCRL3 and probably the interleukin-2 receptor jointly lower the baseline threshold for autoimmunity. Whether a thyroid disease develops from this is co-determined by other factors, among them the genes for thyroglobulin and the TSH receptor.

His decisive sentence for practice: every affected person brings their own bundle of genetic and environmental factors, which only partly overlaps with that of other patients carrying the same diagnosis.

Hasham and Tomer add the epigenetic level. Their thesis is that risk genes only become effective in interplay with environmental stimuli via epigenetic changes. This is mechanistically plausible and in part supported, but not yet clarified in a way that would allow a prediction for individual people.

Cayres and colleagues in Brazil examined stool samples from 40 people with Hashimoto and 53 control subjects, along with cytokines and zonulin in the blood and dietary habits.

They found shifts in bacterial composition, as well as differences in the consumption of vegetables, fruit, animal protein, dairy products and carbohydrates. Systemic cytokines did not differ, while zonulin concentrations in the blood were increased in those affected. From this they concluded a more permeable gut barrier.

What this study shows: an association. What it does not show: a direction. Whether the altered flora helps set off autoimmunity, or whether the disease and its accompanying circumstances change the flora, cannot be answered from a cross-sectional design.

Wang and colleagues investigated which proteins of the gut bacterium Yersinia enterocolitica are recognised by antibodies against the TSH receptor. Using mass spectrometry and structural modelling they identified the outer membrane protein ompF.

This bacterial protein shares cross-immunogenicity with a particular region of the human TSH receptor. The recognised section lies between amino acids 190 and 197.

Important for the framing: this work was carried out in Graves' disease, not in Hashimoto, and it is mechanistic, not clinical. An independent Italian group had already found matches between the TSH receptor and proteins of Yersinia as well as Borrelia burgdorferi by sequence comparison in 2004, with identities between 23 and 50 percent.

A Norwegian-Italian group around Weider examined thyroid tissue from 35 people with autoimmune thyroid disease and 18 control subjects for ten different viruses, using cell culture plus gene search plus protein detection.

In 40 of 53 samples, that is 75 percent, a replication-competent virus was found. The most common were enteroviruses at 51 percent, human herpesvirus 6 at 30 percent and parvovirus B19 at 22 percent. Epstein-Barr virus and cytomegalovirus were found in only a few cases.

And now the decisive sentence: the distribution of viruses did not differ statistically between those affected and the controls. The authors themselves write that it remains open whether these are triggers, cofactors or mere co-inhabitants.

An Amsterdam group followed 790 women with healthy thyroids, all of them with first- or second-degree relatives with autoimmune thyroid disease. Over five years, life events, daily hassles and mood were recorded annually.

In the women who newly developed TPO antibodies over the course, stress levels were no higher than in the women who did not develop any. Nor were differences found in those who developed manifest over- or underactive thyroid function.

The same group had previously looked at 759 women with healthy thyroids in a cross-sectional design. 24 percent had TPO antibodies. The number of daily hassles, the number of life events and mood scores were practically identical between antibody-positive and antibody-negative women.

Fiore, Tonacchera and Vitti summarised the evidence on iodine prophylaxis and thyroid diseases. After the programmes were introduced, an increased frequency of thyroid autoimmunity and hypothyroidism was observed.

The probable mechanism: heavily iodinated thyroglobulin is more visible to the immune system than weakly iodinated thyroglobulin. The Rayman review puts it as highly iodinated thyroglobulin being more immunogenic.

And now the sentence that is often missing in guidebooks: the same authors emphasise explicitly that the benefit of a balanced iodine status clearly outweighs the risks. This is not about avoiding iodine, it is about dose and pace.

Chen and colleagues measured urinary iodine excretion and thyroid antibodies in 4,635 pregnant women in Shanghai. The median iodine value was 138 micrograms per litre, and a good half of the women were undersupplied.

Women with an iodine excretion below 100 micrograms per litre had an increased risk of positive thyroid antibodies, with an odds ratio of 1.32. The analysis using flexible splines showed a non-linear relationship between iodine intake and autoimmunity.

So the picture is not "the less iodine the better", but rather a curve with a favourable middle range. In everyday language that means: iodised table salt and sea fish are not the enemy. High-dose iodine supplements and seaweed products in large quantities belong in medical hands in Hashimoto.

A Swiss group around Huwiler systematically evaluated all randomised studies on selenium in Hashimoto in the journal Thyroid in 2024. Out of 687 screened datasets, 35 studies remained.

In people without hormone replacement the TSH value fell slightly, with a standardised mean difference of minus 0.21. TPO antibodies fell more clearly, with minus 0.96 across 29 cohorts and 2,358 participants. Side effects did not occur more often than under placebo. The authors rate the overall certainty of the evidence as moderate.

Heterogeneity between the studies was very high, however. And the Cochrane review by van Zuuren, which applied stricter inclusion criteria, concluded on the basis of four studies that the evidence is not yet sufficient for a clear recommendation. Both belong in the same account.

Taheriniya and colleagues evaluated 42 observational studies. In Hashimoto the vitamin D level was on average 6.05 nanograms per decilitre lower than in controls, with a confidence interval from minus 8.35 to minus 3.75.

An earlier meta-analysis by Wang and colleagues across 20 case-control studies arrived at a similar picture. People with autoimmune thyroid disease more often had a vitamin D deficiency, with an odds ratio of 2.99.

What follows from this and what does not: a good vitamin D supply makes sense for many reasons and is easy to check. That raising the level influences the course of Hashimoto is not shown by this. Observational data fundamentally cannot deliver that.

Margaret Rayman of the University of Surrey summarised the evidence on iodine, iron and selenium in thyroid diseases. Her point about iron is biochemically precise: thyroid peroxidase is a haem enzyme and only becomes active after binding haem.

She also describes that people with autoimmune thyroid disease have an above-average frequency of iron deficiency, because autoimmune gastritis and coeliac disease as accompanying conditions reduce absorption or cause losses.

The finding I find most interesting: in about two thirds of women with persistent hypothyroid complaints despite an appropriately adjusted hormone dose, symptoms improved after ferritin was raised above 100 micrograms per litre. That is an observation from a review article, not a proof. I find it important enough to know the ferritin, and not only the haemoglobin, when exhaustion persists.

Lee and Pearce summarised the state of knowledge on thyroid disorders in pregnancy and the puerperium for Nature Reviews Endocrinology. Postpartum thyroiditis can occur up to a year after delivery and has to be distinguished from other forms of thyroid dysfunction, because the treatment looks different.

The immunological background is easy to follow. In pregnancy the mother's immune system is dampened so that the child is not rejected. After birth it ramps up again. In some women it overshoots the previous level.

Caturegli and colleagues explicitly classify silent thyroiditis, both sporadic and postpartum, as a variant within the Hashimoto spectrum.

Belin and colleagues determined smoking status objectively in NHANES III via the breakdown product cotinine in the blood, not via self-report. 15,592 people were evaluated.

Smokers had thyroid antibodies less often than non-smokers, 11 percent versus 18 percent, and 13 percent after adjustment for age, sex, ethnic background and iodine status. Elevated TSH values were also found less often, 2.6 versus 5.5 percent.

The decisive second half of the finding: smoke exposure went along with a threefold probability of a low-normal TSH, that is a shift towards overactive thyroid function. In review articles on Graves' disease, smoking counts among the established environmental factors.

Taken together this means: smoking is not protection against autoimmunity. It shifts the risk within the spectrum, away from the Hashimoto side and towards the Graves' side. For the eye involvement in Graves' disease, smoking is a particularly well known risk factor.