Damage to the thyroid might release normally sequestered antigens, inducing an immune response. Damage to thyroid cells does indeed occur in viral thyroiditis, such as in association with mumps or in subacute thyroiditis of unknown cause, but autoantibodies appear only transiently at low titer, and progressive lesions of the thyroid do not usually occur (reviewed in 219, 220). External irradiation to the thyroid, including that from nuclear fallout, can also lead to an increase in Graves' disease or thyroid antibody production (221, 222), but it is unclear if this is caused by autoantigen release or an effect on the lymphocytes which are radio-sensitive. Even occupational exposure to ionizing radiation appears to be a risk factor for the development of autoimmune thyroiditis (223). Among current adolescents exposed thirteen to fifteen years earlier to radiaoactive fallout from Chernobyl, the incidence of antithyroid antibodies remains nearly double that found in controls, although there is no evidence of thyroid dysfunction. (Agate L, Mariotti S, Elisei R, Mossa P, Pacini F, Molinaro E, Grasso L, Masserini L, Mokhort T, Vorontsova T, Arynchyn A, Tronko MD, Tsyb A, Feldt-Rasmussen U, Juul A, Pinchera A. Thyroid Autoantibodies and Thyroid Function in Subjects Exposed to Chernobyl Fallout during Childhood: Evidence for a Transient Radiation-Induced Elevation of Serum Thyroid Antibodies without an Increase in Thyroid Autoimmune Disease. J Clin Endocrinol Metab. 2008 Jul;93(7):2729-36)

A powerful argument against the hidden antigen hypothesis is that TG is a normal component of circulating plasma (224). One might turn the first argument around and suggest that thyroiditis results from a lack of exposure to TG at some period, an exposure that is necessary to depress continuously an otherwise inevitable immune response. This suggestion has no clinical or experimental support, and the available evidence indicates that TG is present in the plasma of patients with active immunity. It remains to be seen how sequestered TPO and TSH-R are, but the appearance of T cells capable of proliferating in response to these antigens, in apparently healthy individuals, also argues against any sequestration (54, 225). What is clear is that availability of the thyroid autoantigen is essential to maintain the autoimmune response: complete removal of thyroid antigens following thyroidectomy and remnant ablation with radioiodine leads to disappearance of antibodies to Tg, TPO and TSH-R (226). Although this is not surprising, it does suggest that extrathyroidal sources TSH-R are insufficient normally to maintain an autoimmune response.

A new variant on this theme is that of microchimerism, the persistence of fetal cells in maternal tissues. Studies have found evidence of microchimerism in thyroid tissue from patients with and without AITD (227). Could such sequestered fetal material make the thyroid prone to an alloautoimmune response, and be responsible for the exacerbation of AITD seen in the postpartum period? If so, this phenomenon would help to explain the high frequency of AITD in women. Recent animal work supports such a possibility, as fetal cells of immune origin could be shown to accumulate in the thyroid glands of mice with EAT during pregnancy and the early postpartum period (228). In contrast a recent survey of over 3700 women has found no correlation between thyroid autoantibodies and parity, casting some doubt on this as a mechanism at least for the appearance of thyroid autoimmunity at the B cell level (229).

After pregnancy, major changes in lymphocyte-endothelial interactions can be induced by estradiol and progesterone, and direct effects on the cytokines produced by T cells can also be demonstrated. It seems likely that sex steroids play a role in determining the autoimmune response. Another hypothetical reason for the unequal sex ratio is that skewed X chromosome inactivation, which has recently been demonstrated in scleroderma and in autoimmune thyroid disease, could contribute through the failure of some autoantigens expressed on one X chromosome to be expressed at a critical point in the disease pathway (230).

An abnormal antigen might also serve to produce an immune reaction. The protein abnormality could be either congenital or acquired by an injury such as a virus infection. To date there is no evidence which indicates that TG, TPO, or other proteins of the thyroid of a patient with autoimmunity are abnormal. Minor allelic differences apparently do occur but attempts to associate thyroid disease with polymorphisms of the TPO and TSH-R genes have been unsuccessful.

The theory that immune reactivity to an environmental antigen could lead to antibodies that cross-react with thyroid antigens has been bolstered by studies which show a relationship between Graves' disease and antibodies to the common enteropathogen Yersinia enterocolitica. An increased incidence of antibodies to Yersinia is found by some, but not all authors, in patients who have Graves' disease (219), and there are saturable binding sites for TSH on Yersinia proteins (231). After infection by Yersinia, human sera contain Igs that bind to TEC cytoplasm (216), and IgGs which appear to compete with TSH for binding to thyroid membrane TSH receptors (232). The antigens involved may in fact include proteins encoded by plasmids present in the Yersinia, rather than intrinsic Yersinia proteins, but that does not alter the general concept (233). Arguing strongly against a role for Yersinia is the fact that there is no unique pattern of serological immunoreactivity to Yersinia antigens in patients with AITD (234), and most patients with this infection do not develop Graves’ disease.

Heat shock proteins (HSPs) are produced by prokaryotic and eukaryotic cells in response to heat and other forms of cellular injury. Because their structure is highly conserved in all cells studied, the potential cross-reactivity between HSPs from pathogens and human is great. A number of pathogenetic organisms present HSPs as antigens in animal and human models. It is therefore possible that immunization against HSPs from a pathogen may lead to cross-reactivity with autologous HSPs released from damaged tissues. The HLA region contains genes for the major HSP70, and polymorphisms in the HSP70 gene are associated with Graves' disease (235). This finding could provide another link between HLA and susceptibility to infection and autoimmunity. HSPs are expressed at a high level in thyroid cells from patients with Graves' disease, and in fibroblasts from patients with exophthalmos (236). These changes may be secondary to cytokine stimulation, but could be involved in a secondary immune response.

Antibodies to TG sometimes also recognize TPO (237). The exact reason for this is uncertain, but may -- or may not -- relate to short stretches of shared peptide sequence and hence shared B cell epitope. In theory an initial response to one antigen might proceed by reacting to the other antigen, and thereby spread and augment the autoimmune process. In the context of T cell autoreactivity there is much greater scope for molecular mimicry whereby a response to an exogenous epitope leads to a cross-reactive response to an endogenous autoantigenic epitope. Simple sequence homology is insufficient to predict this, as shown elegantly by the cross-reactivity of two TPO epitopes showing a similar surface but not amino acid sequence (238). This makes the prediction and study of molecular mimicry much more difficult than is generally appreciated (239). For these reasons, it may be naïve to believe that the putative orbital antigen responsible for ophthalmopathy will be the identical protein (eg. TSH-R) to that expressed in the thyroid.

Virus infection has for years been speculated to be an etiological factor in most autoimmune diseases, by causing cell destruction and liberating antigens, by forming altered antigens or causing molecular mimicry, by inducing DR expression, or by inducing CD8 ⁺T cell responses to viral antigens expressed on the cell surface. Antithyroid antibodies are elevated transiently after subacute thyroiditis, which is thought to be a virus-associated syndrome, but no clear evidence of virus-induced autoimmune thyroiditis in humans has been presented. In this regard it is of great interest that persistent, apparently benign virus infection of the thyroid can be induced in mice (240), and that infection of neonatal mice with Reo virus induces a polyendocrine autoimmunity (Fig. 7-11). These agents could work by liberating thyroid antigens. Virus infection might also augment autoimmunity by causing non-specific secretion of IL-2, or by inducing MHC class II expression on TEC. Despite many attempts to implicate retroviruses in AITD, results to date remain inconclusive (219, 241), although a recent study has detected elevated levels of reverse transcriptase in well conducted experiments with Graves’ thyroid tissue, reviving this concept (242). Human T lymphotrophic virus-1 has been repeatedly associated with various autoimmune disorders, including Hashimoto’s thyroiditis; presumably the virus alters immunoregulatory pathways allowing autoimmunity to emerge (243).

Figure 11. Autoantibodies to thyroid in sera of reovirus-infected mice detected by indirect immunofluorescence. (a) Frozen section of normal mouse thyroid incubated with sera obtained from mice 21 days after infection, showing staining of colloid characteristic of antithyroglobulin antibody (original magnification, X200). (b) Section of normal mouse thyroid (fixed in Bouin's solution) incubated with sera obtained from mice 21 days after infection, showing staining of thyroid acinar cells (original magnification, X 200). Reproduced with permission from I. Okayasu and S. Hatokeyama, Clin. Immunol. Immunopath., 31:334, 1984.

Apart from the evidence that some TSI may have an oligoclonal origin (88, 244), there is no evidence to support a clonal B cell abnormality in AITD. V gene usage by TSI will need to be analysed to determine whether Graves’ disease has a unique pathogenesis determined by germ-line immunoglobulin genes. Thyroid-reactive T cells are present in healthy animals and man, as noted above, and therefore a defect at the clonal T cell level is less likely as a primary event in etiology than previously thought. A few autoreactive T cells can be expected to escape tolerance normally, particularly if the autoantigen in question is not available to delete T cells in thymus during fetal development. Stochastic events later in life affecting such undeleted T cells could readily explain the lack of complete concordance for AITD in genetically identical twins (245), and this lack of such concordance argues against an inherited pathogenic TCR as a primary event in AITD.

A role of heredity in AITD is clearly demonstrated by family studies (246, 247). The role of heredity in AITD is clear, since there is an increased frequency of AITD among family members, first degree relatives, and twins of patients with the illness (248). Indeed a recent careful analysis of concordance in Danish twins with Graves’ disease came up with the estimate that 79% of the liability for this disorder was attributable to genetic factors (249). A recent study from the USA has found similar concordance rates for Graves’ disease as in this Danish study, while among the unaffected monozygotic twins of the patients with Graves’ disease, 17% had chronic thyroiditis, while 10% had pernicious anemia or other autoimmune disorders (250). In an investigation of the relatives of a group of propositi with high circulating antibody levels and clinical thyroid disease, approximately half of the siblings and parents (first-order relatives) were found to have significant titers of thyroid antibodies, many being without clinical thyroid disease (251) but the transmission of thyroid autoantibodies is a more complex trait than the dominant inheritance originally thought (252, 253).

Together, such observations suggest that these diseases have a common genetic defect, although other genes are likely to be disease-specific in their effects, as reviewed extensively elsewhere (254). The most important susceptibility factor so far recognized is the inheritance of certain MHC class II genes. Inheritance of HLA-DR3 causes a 2 to 6-fold increased risk for the occurrence of Graves' disease or autoimmune thyroiditis in Caucasians, and inheritance of HLA-DR4 and DR5 has been found in some studies to increase the incidence of goitrous hypothyroidism (255). In post-partum painless thyroiditis an association is found particularly with HLA-DR5 (256). Recent studies have identified HLA-DQA1*0501, which is often linked to DR3, as having an even more pronounced predisposing effect in Caucasians with Graves’ disease (257). HLA-DRB1*07 may be protective (258). One intriguing observation which needs further work is the observation that familial clustering of juvenile thyroid autoimmunity has a higher risk when the children’s fathers, not mothers, are HLA-DR3-positive, especially if the fathers also have TPO antibodies (259). This may imply an X-linked gene interaction with the HLA haplotype.

The HLA linkages found in Caucasians are not found in American Blacks, and different HLA associations are found in other ethnic groups such as Koreans, Chinese, and Japanese (255). This tends to suggest that such HLA associations do not depend on critical binding between selected class II molecules and epitopes from thyroid antigens (so-called determinant selection). Instead, the frequent association of HLA-DR3 with many autoimmune diseases may reflect a non-specific enhanced immunoresponsiveness encoded by DR3-linked haplotypes. It is noteworthy also that the relative risks conferred by HLA alleles is rather modest, borne out by the relatively low concordance for Graves’ disease in HLA-identical siblings of patients with Graves’ disease (259). This suggests the operation of other genetic susceptibility loci, also emphasised by the weak lod scores for linkage with the HLA region in family studies of AITD (260, 261).

The nature of these other loci is unclear and their identification is likely to require an extensive analysis involving several hundred families in studies using modern molecular techniques coupled to either genome screening or the transmission disequilibrium test. Association studies have been the method of choice to date, investigating various candidate genes, but with mixed success. Inconclusive results have been reported for associations of AITD with TCR polymorphisms, immunoglobulin allotype and TSH-R polymorphisms. The most consistent non-HLA association is between polymorphisms in the CTLA-4 gene and both Graves’ disease and Hashimoto’s thyroiditis (262, 263). Despite claims to the contrary, there appears to be no additional risk conferred by CTLA-4 (or HLA) polymorphisms in Graves’ patients with clinical evidence of ophthalmopathy (264), but these CTLA-4 polymorphisms may partially determine outcome after antithyroid drug (265, 265a). Given the most important role of the interaction between CTLA-4 on T cells and the B7 family of molecules on APCs, it is possible that this association represents a genetic effect on immunoregulation, although, as with HLA-DR3, this is not specific for thyroid autoimmunity; the same polymorphism is also associated with type I diabetes mellitus and several other autoimmune disorders. Recent fine mapping of the CTLA-4 region has confirmed that it is indeed this gene, rather than those in linkage disequilibrium, which is responsible for the associations, and the polymorphisms may exert their effects by causing variation in levels of soluble CTLA-4, which in turn may after T cell activation, especially in Treg cells (266).

Another recent possible candidate is the association of a polymorphism of the vitamin D receptor with Graves’ disease, an association which has some biological plausibility as vitamin D has immunological effects (267). However a very large survey comprising 768 patients with Graves’ disease from the UK, compared to 864 controls, found no evidence of an association (268). Polymorphisms in genes encoding molecules involved the NFkB inhibitor pathway modulating B cell function (FCRL3 amd MAP3K7IP2) may also be involved in susceptibility to Graves’ disease (269, 269a). The most likely candidate for a third genetic susceptibility locus in Graves’ disease, besides HLA and CTLA-4, is polymorphism in the lymphoid tyrosine phosphatase gene, which has been associated with functional changes in T cell receptor signaling. A recent study of 549 patients and 429 controls found that a codon 620 tryptophan allele conferred an odds ratio of 1.88 (270), although it should be noted that similar effects have been seen in many other autoimmune diseases. This result has recently been confirmed (271a) and a fourth likely locus is the IL-2 receptor alpha (CD25) gene region, another locus associated with other autoimmune diseases like type diabetes (271b). Finally there seems to be conclusive proof from both linkage disequilibrium and association studies, that polymorphisms in the TSH-R confer susceptibility to Graves’ disease but not autoimmune hypothyroidism (271). This is one of the few susceptibility factors that segregates with one rather than both types of thyroid autoimmunity.

A different approach has been genome scanning although huge effort is required to undertake such studies. Based largely on this approach, other loci which may be important have been identified on chromosomes 14q31, 20q11 and Xq21 (261, 273), but larger studies are now required to have sufficient statistical power to confirm these exciting results. Indeed in the largest genome scan to date in autoimmunity, involving 1119 relative pairs, there was no replication of these findings (273). The importance of a gene on the X chromosome is supported by the increased frequency of AITD in women with Turners syndrome, especially those with an isochromosome-X karyotype (274). However, the Xq21 linkage with AITD was not supported by a second study, although linkage to 9 locus at Xp11 was found (275). Clearly, most work should focus initially on the X chromosome.

As an aside, it should be noted that low birth weight, a known risk factor for several chronic disorders, has not associated with clinically overt thyroid disease or with the production of thyroid autoantibodies in one study (276) but others have come to an opposite conclusion, with prematurity irrespective of birth size being another risk factor (276a, 276b).

The co-existence of AITD and other diseases possibly of autoimmune cause has often been reported, and suggests some intrinsic abnormality in immune regulation. An extensive review of these associations has recently been published (277). A striking association is with pernicious anemia. Perhaps 45% of patients with autoimmune thyroiditis have circulating antigastric antibodies (278), and the reverse association is almost as strong (279). Up to 14% of patients with pernicious anemia have primary myxedema, and pernicious anemia is increased in prevalence in patients with hypothyroidism (280). The association of Sjogren's syndrome and thyroiditis is not uncommon and both systemic lupus erythematosus (SLE) and rheumatoid arthritis are also significantly associated with AITD (281, 282). A high frequency of antibodies to nucleus, smooth muscle, and single-stranded DNA (26-36%) is found in AITD (283).

Autoimmune Addison's disease and/or type I diabetes mellitus and AITD occasionally co-exist and this forms the autoimmune polyglandular syndrome (APS) type 2 (284). This is an autosomal dominant disorder with incomplete penetrance and is often associated with other disorders, such as vitiligo, celiac disease, myasthenia gravis, premature ovarian failure and chronic active hepatitis (285, 286). AITD is an infrequent feature of the much rarer APS type I (287) and there is no association between mutations in the AIRE gene, which causes APS type I, and sporadic AITD (288).

Together these data provide convincing proof of an association of other autoimmune phenomena with AITD. Most typically, this immunity is organ specific, but in one subset of patients, antithyroid immunity develops in association with the non-organic-specific collagen diseases. A syndrome, or running together, of course, does not prove a causal association. Nevertheless, the plethora of associations and their familial occurrence indicates that a defect in the immune system may be more likely than primary defects in each organ. This in turn suggests a shared immunoregulatory defect, which is at least partly genetically determined, as these diseases often share similar genetic associations, including HLA, CTLA-4, PTPN22 and CD25 gene polymorphisms.

Possible abnormalities in immunoregulation have been addressed in hundreds of studies. The basic hypothesis of this work is that a deficiency of functional T suppressor cells -- either antigen-specific or nonspecific -- may allow uncontrolled T and B cell immune responses to thyroid (or other) antigens. As noted above, this concept is a major theme in experimentally induced or naturally occurring thyroiditis in animal models. Immune regulation is extremely complex and still only partially understood. The studies described below have been reported over more than two decades, during which time our understanding of lymphocyte function, terminology, and methods of analysis have continually changed. Most of these studies define immunoregulatory responses in relation to in vitro assays done in unique conditions, or a group of cells bearing a specific surface antigen (e.g. CD4, CD8, etc.). As we have previously noted, T cell antigen expression and function can vary depending on stimulating event, culture conditions, etc. Further, whether a unique group of “suppressor cells" actually exists is uncertain. Thus we present these observations as reported (by us and others), and in the terms used by the authors.

Sridama and DeGroot found decreased suppressor cells, defined as CD8 ⁺peripheral blood T cells in patients with Graves' disease (289, 290). These results have been challenged, and some investigators have reported depression of CD4 ⁺cells in AITD (291). However, overall, there is now agreement that, in thyrotoxic patients with Graves' disease, a decrease in CD8 ⁺T cell number (292, 293) is characteristically present, and that a similar abnormality exists in the thyroid. CD8 ⁺cells return gradually toward normal during therapy, and are usually but not always normal at the end of therapy (292) (Fig. 7-12). The phenomenon is present but less evident in Hashimoto's thyroiditis patients. It has been attributed by some workers to increased thyroid hormone levels (294), although this issue is clouded, since there are reports disproving the idea that hyperthyroidism per se induces suppressor cell abnormalities in humans, and reduced suppressor T cells (Ts) are found present long after thyrotoxicosis is cured (295). Our interpretation is that the abnormality is not due specifically to excess T ₄in blood, but is a manifestation of ongoing active autoimmunity, for reasons which are unclear. Reduced nonspecific "suppressor" T cell function may be in part an inherited abnormality, and is probably also a manifestation of the augmented immune reactivity ongoing in toxic Graves' disease patients. It may be largely a secondary phenomenon, but one which augments and continues the immunological disease. The mechanism causing reduced Ts number and function is unclear. Ts could be reduced during an active immune response by binding immune complexes on their surface, which could inactivate the lymphocytes or cause them to be removed from circulation. Ts could also be removed by cytotoxic antibodies. These findings need to be related to recent developments in understanding Treg function, studies which have yet to be undertaken.

Figure 12. Influence of a 6 month course of carbimazole on peripheral blood T cell subsets of 29 patients with hyperthyroid Graves' disease (Mean SD). OKT ₄= CD4 ⁺, OKT ₃= CD3+, OKT8 = CD8 ⁺, ** = p < .001 vs. zero time value. (From Reference 265)

T cells from patients with Graves' disease were unable to suppress Ig synthesis when mixed with B cells, in comparison to T cells from normal individuals (290). Okita et al. (296) suggested that this is due to a low number of histamine H2 receptor-positive CD8 ⁺cells in Graves' disease patients. Another possibly related phenomenon is the decreased pokeweed mitogen (PWM) responsivity of PBMC found in Graves' disease patients during illness (297) and also when cured of disease (298).

An alteration in helper T (Th)/Ts ratio may also predispose to the occurrence of postpartum transient thyrotoxicosis. We have shown a decrease in CD4 ⁺cells in normal pregnancy (299), possibly representing one of the factors causing the diminished immunoreactivity typically found in pregnancy. A rebound increase in CD4 ⁺cells, which occurs during the first two or three months following delivery, may lead to a recrudescence of immuno-reactivity, including antibody levels and the occurrence of postpartum transient thyroiditis in some women (300).

Thyrotoxic Graves' disease patients and those with active Hashimoto's thyroiditis have a high proportion of DR+ T cells in their peripheral circulation (292, 301), which indicates the presence of activated T cells. It is unlikely that these cells (> 20% of circulating T cells) are all responsive related to thyroid antigens, so they must include DR+ T cells with TCRs for many other antigens. There is also a marked increase in circulating soluble IL-2 receptors in thyrotoxic Graves' disease, but this appears to be typical of thyrotoxicosis per se, and not specifically Graves' disease (302). Nevertheless, there is no evidence for a generalized ongoing immune hyper-responsiveness in thyrotoxic patients. Perhaps these T cells (for many different specificities) are stimulated by IL-2, but in the absence of the required "second signal" provided by antigen exposure, do not induce B cell proliferation or cytotoxic responses.

Diminished, non-specific suppressor cell function is also observed in many autoimmune diseases including lupus, and multiple sclerosis and the results in AITD are equally non-specific. Functional assays attempting to show a deficiency in antigen-specific suppressor cells have been reported by several groups (303 - 308) in particular using the MIF assay or measuring effects on antibody synthesis in vitro. The pathophysiological relevance of some of these systems is questionable, results have not always been reproducible and there continues to be controversy over the nature of antigen-specific suppressor activity (309-311). The most likely explanation for many “suppressor” phenomena is the reciprocal inhibition of Th1 and Th2 cells by their cytokine products, and powerful evidence shows how important this regulatory mechanism is in exacerbating or inhibiting autoimmune disease, at least in animal models. In addition, work is now needed to review the potential role of CD4 ⁺CD25 ⁺Treg cells, discussed earlier, in AITD. One recent study has found that despite increased numbers of CD4 ⁺T cells bearing the T regulatory cell markers CD25, Foxp3, GITR and CD69, in both thyroid and PBMC of patients with AITD, there is a non-specific defect in regulatory function in vitro, which in turn must explain somehow why the increased number of reguilatory T cells are so patently ineffective (312). Analysis in the earliest phases of disease may of course yield different results and unlocking how T regulatory cells can be activated seems an obvious but at present unrealizable therapeutic strategy.

Many studies have examined T cell subsets in thyroid tissue of patients with active AITD (255). For example, Margolick et al (313) found increased CD8 ⁺cytotoxic/suppressor cells and also increased CD4 ⁺T helper cells, and a normal Th/Ts ratio. Canonica et al (314) found increased proportions of cytotoxic/suppressor T cells in thyroids of Hashimoto's thyroiditis patients. Infiltrating cytotoxic/suppressor cells in Hashimoto's thyroiditis were found usually to be activated and to express DR antigen, whereas this response was not so obvious in Graves' disease (315). Canonica et al (314) reported an increased proportion of activated T helper/inducer cells in both Graves' disease and Hashimoto's thyroiditis, and increased cells thought to represent cytotoxic T cells in Hashimoto's thyroiditis. Chemokine expression within the thyroid is likely to be an important determinant of this infiltration (316).

Increased CD8 ⁺CD11B- cells, presumed to be cytotoxic cells, were found in Graves' disease thyroids (in comparison to PBMC of Graves' disease or normal subjects), whereas "dull" CD8 ⁺CD11B+ natural killer cells were diminished (317). Other studies have suggested a reduction in NK cells in Graves' disease and an increase in Hashimoto's thyroiditis. Tezuka et al found decreased NK cells in Graves' disease thyroid tissue, no differences in the NK activity of PBMC between Graves' and normal patients, and that the NK cells in Graves' disease did not kill autologous thyroid epithelial cells (318). We have already indicated other reports of normal NK and ADCC in Hashimoto's PBMC, and of increased ADCC in Hashimoto's thyroiditis. Most studies that have looked at Graves' disease tissues also indicate an increased proportion of B cells compared to peripheral blood subsets.

Cell cloning has also been used to examine thyroid and peripheral blood lymphocyte subsets. Bagnasco et al (319) found a predominance of cytolytic clones, releasing IFN-g, in Hashimoto's thyroiditis but not in Graves' disease. Del Prete et al (320) found a high proportion of cytolytic cells with the CD8 ⁺phenotype in clones from thyroid tissue, and felt these results may relate to autoimmune destruction of TEC but the non-specific methods used to derive such cytotoxic T cells raises questions about any pathophysiological relevance. There is no clear predominance of Th1 or Th2 cytokines in the thyroid of patients with Graves’ disease or Hashimoto’s thyroiditis (165, 321), although Th1 clones seem to predominate in the retrobulbar tissues in ophthalmopathy (322). It might simplistically be thought that Graves’ disease represents a Th2 response, but the fact that some patients end up with hypothyroidism itself indicates the likely presence of a Th1 response too. This is supported by evidence from an animal model of Graves’ disease: immune deviation away from a Th1 response, in g-IFN knockout mice, did not enhance the response to TSH-R cDNA vaccination (323).

A general summary of these data is difficult. The results probably at least indicate there are increased B cells, increased DR+ T cells, increased CD4 ⁺DR+ T helper cells, decreased CD8 ⁺DR+ T suppressor/cytotoxic cells, and possibly lower NK cells in Graves' disease AITD tissue and in blood than among normal subjects' PBMCs. The intrathyroidal T cells are a mix of Th1 and Th2. Such studies have been performed primarily on patients with well developed and often treated disease, and do not bear directly on early stages of the disease, nor on whether the changes represent primary or secondary phenomena.

To date there has been no certain indication that a non-specific or specific suppressor cell defect exists in patients who are genetically predisposed to have AITD, or in most patients who have recovered from the illness. Rather, the data suggest, at least for the better studied changes in antigen nonspecific T cell subsets, that the changes may be a secondary, although possibly still important part of disease pathology for augmentation and continuation, rather than initiation.

Whereas anti-idiotypic antibodies are thought to play a physiological role in immunoregulation, there is little evidence for participation in, or abnormality of, this function in AITD. Immunoglobulins from some patients with Graves' disease bind TSH (324). This suggests that anti-idiotypes to TSH antibodies are present and might theoretically function as thyroid stimulating immunoglobulins; or conversely that anti-idiotypes to thyroid stimulating antibody exist and can bind TSH. Either possibility remains to be confirmed. TG antibodies can be induced in experimental autoimmune thyroiditis by idiotype-anti-idiotype manipulation (182) and Sikorska (325) demonstrated the presence of antibodies in sera of AITD patients which inhibit binding of TG to monoclonal anti-TG antibodies, and interpreted these as anti-idiotypes. We have looked very carefully for anti-TG anti-idiotypes in patients with autoimmune thyroid disease and failed to find them (326 - 327). On the other hand, weak anti-idiotypes of the IgM class have been found which bind to TPO antibodies and are present in pooled normal immunoglobulins as well as certain patient sera (327). Although one could postulate that a failure to produce anti-idiotype antibodies could be a feature of AITD, a more likely hypothesis is that anti-idiotypic antibodies are simply rarely produced at a detectable level. Since anti-idiotype antibodies raised in animals will suppress in vitro anti-TG antibody production, the theory that lack of anti-idiotype control is causal in AITD remains attractive, but data to support it are scant.

De novo expression of HLA-DR on thyroid epithelial cells, from patients with Graves' disease, was first reported by Hanafusa et al (328) and was proposed as the cause of autoimmunity by Bottazzo et al. (329) who suggested that de novo expression of MHC class II molecules on these cells, which are normally negative, allows them to function as APCs. Lymphocyte-produced IFN-g augments the expression of HLA-DR (also DP and DQ) on thyroid epithelial cells, and that TNF-a further increases the induction caused by IFN-g (330, 331). HLA-DR+ TECs definitely can stimulate T cells (332, 333) but this is critically dependent on the requirements of the T cell for a costimulatory signal, as Graves’ TECs do not express B7-1 or B7-2 (334, 335). In contrast B7.1 expression on Hashimoto TEC has been recorded, but how this is differentially regulated, compared to Graves’ disease, is unknown (336). We have shown that TECs can present antigen to T cell clones which no longer require costimulation through B7, yet not only fail to stimulate B7-dependent T cells but also induce anergy in these cells by at least two mechanisms, one of which is Fas-dependent (337, 338). Perhaps the most conclusive proof that class II expression by thyroid cells cannot induce thyroiditis comes from the creation of transgenic mice expressing such molecules on TECs – such animals have no thyroiditis and have normal thyroid functionm (338a). For reasons which remain unclear, thyroid follicular and papillary cancers may express B7.1 and B7.2, and B7.2 expression is associated with an unfavourable prognosis (339).

HLA-DR is also expressed on TECs in multinodular goiter and in many benign and malignant thyroid tumors, and this does not appear to induce thyroid autoimmunity (340). Aberrant DR expression has not been shown to develop before autoimmunity. Normal animal thyroids not expressing class II molecules can become the focus of induced thyroiditis, and then express class II molecules (341). Furthermore, HLA-DR expression on Graves' disease thyroid tissue is lost when tissue is transplanted to nude mice (342). Thus a consensus position is that class II expression could be important, but is a secondary phenomenon in AITD, dependent on the T cell-derived cytokine, g-IFN, and only allowing TECs to become APCs for T cells which have already received B7-dependent costimulation elsewhere. This could clearly exacerbate AITD once initiated, but teleologically the role of class II expression seems to be as a peripheral tolerance mechanism, allowing the induction of anergy in potentially autoreactive but still naive (ie. B7-dependent) T cells (Fig. 7-13). The recent description of hyperinducibility of HLA class II expression by TECs from Graves’ disease suggests that such patients may be genetically predisposed to display a more vigorous local class II response and this would increase the likelihood of disease progression (343). The genes controlling this response are therefore worthwhile candidates for future studies of genetic susceptibility.

Figure 13. Alternative outcomes of MHC class II expression by thyroid follicular cells. Reproduced from Weetman (1997) New Aspects of Thyroid Autoimmunity. Hormone Research 48 (Suppl 4), 51-54 with permission.

Environmental factors include viral and other infections, discussed above. Strong evidence for an important role for environmental factors is provided by the incomplete concordance seen in the monozygotic twins or other siblings of individuals with AITD. Also, there may be abrupt changes in disease incidence, such as the recent rise in Graves’ disease in children in Hong Kong, an effect that can only be environmental (344). Such studies also show that environmental factors may change rapidly, making their ascertainment difficult and challenging. IL-2 administration for treatment of cancer leads to the production of antithyroid antibodies, and hypothyroidism (and possibly a better tumor response) (345). IFN-a administration and other cytokines (118), as well as highly active antiretroviral therapy for AIDS (346), have a similar effect, although interferon-b1b treatment has no significant adverse effect on AITD (347). Recently however long-term follow up studies have shown that around a quarter of multiple sclerosis patients treated with this latter cytokine may develop autoimmune thyroid disease within the the first year of treatment (348). It remains unclear how relevant any lessons from these observations are for AITD pathogenesis, as of course the doses of cytokines and drugs used therapeutically are vast. However, it has been reported that thyrotoxicosis tends to recur following attacks of allergic rhinitis (349). Possibly this is due to a rise in endogenous cytokines and the recent association of raised IgE levels with newly diagnosed Graves’ disease indicates that this may be mediated by preferential Th2 activation (350). Cigarette smoking is associated with Graves' disease, and with ophthalmopathy (reviewed in 351). The mechanism is uncertain and is doubtless more complex than a local irritative effect. Environmental tobacco smoke induces allergic sensitization in mice, associated with increased production of Th2 cytokines, but a reduction in Th1 cytokines, by the respiratory tract (352). It is therefore possible that modulation of cytokines contributes to the worsening of ophthalmopathy with smoking. On the other hand smoking exposure was associated with a lower prevalence of thyroid autoantibodies in a large population survey of over 15000 US citizens (353).

The role of dietary iodine is clearly established in animal models of AITD and circumstantial evidence exists for a similar role in man (196, 255, 354, 355). The response is complex and recently it has been shown that iodide may exacerbate thyroiditis in NOD mice but not affect the production of TSH-R antibodies in the same strain (356). Such findings are intriguing as they raise the possibility that the thyroiditis which accompanies Graves’ disease may not be due to the immune response to the TSH-R. Dietary selenium may also be important as high dose selenium supplementation reduced TPO antibody levels in women with AITD (356a). Stress is also likely to be important in the etiology of Graves’ disease, although studies to date have had to rely on retrospective measures of this (reviewed in 357). Moreover stress does not appear to be asociated with the development of TPO antibodies in euthyroid women (358). Presumably stress acts on the immune system via pertubations in the neuroendocrine network, including alterations in glucocorticoids, but the complex interaction between the nervous, endocrine and immune systems includes the actions of neurotransmitters, CRF, leptin and melanocyte stimulating hormone as well and so unravelling the pathways whereby stress may alter the course of autoimmunity is difficult in the extreme (359). Indirect support for such a mechanism, mediated through norepinephrine, comes from experiments showing dramatic enhancement of delayed-type hypersensitivity by acute stress, the result of sympathetic nervous system activation on the migration of dendritic cells and subsequent enhanced T cell stimulation (360). Given the diversity of these environmental factors, presumably operating on different genetic backgrounds, it will be difficult (if not impossible with current tools) to establish the relative importance of each in AITD.

"Normal" people express antithyroid immunity, as previously described, and this must be important in understanding the overall mechanism of AITD. Many people with low levels of antibodies but without clinical disease can be shown to have lymphocyte infiltrates in the thyroid at autopsy. B cells from normal individuals can be induced to secrete anti-TG antibody in vitro. These observations clearly show that incomplete deletion of clonal self-reactive T cells is indeed the normal (and indeed perhaps necessary) circumstance, and provide strong support for the idea that disordered control of this low level immunity may be important in the etiology of AITD (115).

Antithyroid drugs are used in Graves' disease to decrease production of thyroid hormone, and also lead to diminution in TSI and other antibody levels. Clinical studies show that antithyroid drug administration also leads to a diminution in antibody production in thyroxine replaced Hashimoto's thyroiditis patients (361), proving that their effect is not simply due to control of hyperthyroidism in Graves' disease. Surprisingly (362), administration of KClO ₄to patients with Graves' disease leads to diminished serum antibodies, suggesting that the effect of treatment is not specific for thionamide drugs, but could be mimicked by this compound. Antithyroid drugs inhibit macrophage function, interfering with oxygen metabolite production (362).

Following antithyroid drug treatment of active Graves' disease, there is a prompt short-term increase of DR+CD8 ⁺T cells in the bloodstream as described above. Antithyroid drugs inhibit the production of cytokines, reactive oxygen metabolites and prostaglandin E ₂by TECs and the reduction in these inflammatory mediators may explain the site-specificity of the immunomodulation produced by antithyroid drugs (156). Another pathway for an immunomodulatory action of these drugs is via the upregulation of Fas ligand expression, which may then attenuate the autoimmune response of Fas-expressing T cells (363). Only approximately 50% of patients enter remission after treatment with antithyroid drugs, a fact which must be accommodated in any hypothesis concerning an immunomodulatory action of these agents. Those patients with Graves’ disease who have the highest IgE and IL-13 levels in the circulation are the most likely to relapse (364). In turn, this suggests that antithyroid drugs only effect remission in individuals who do not have a strong Th2 response; those with the strongest such responses seem unlikely to be affected by the relatively weak action of such drugs.

Thus one is led to the uncomfortable position that AITD is probably not caused by a single factor, but rather due to many factors which interact. We have divided the roles of these potential disease activity factors into a series of stages, emphasizing the predisposing events, antigen driven responses, and then the secondary and nonspecific amplification which ensues.

Figure 14. Theoretical Sequence of Development of AITD.

Stage 1 -- In the basal state, Stage 1, immune reactivity to autologous antigen occurs as a normal process. This probably exists at a physiologically insignificant level, since not all T or B cells reacting with TSH-R, TPO or TG are clonally deleted, and Ag is normally present in the circulation. If assays become sensitive enough, we probably will find some level of antibodies to TSH-R, TPO and TG present in some or all healthy persons, increasing in prevalence and concentration with age, and especially in women, since "femaleness" somehow augments antithyroid immunity manyfold. Patients who have inherited certain susceptibility genes, such as those encoding HLA-DR3 or DR5 specificities, will be especially prone to develop AITD because their T and B cell repertoire includes cells recognizing self-antigen, or their immunocytes are especially good at collecting, presenting, and responding to antigen, or are unable to effectively clear immune complexes from their circulation.

Stage 2 -- Possibly viral infection, or other causes of cell damage, or cross-reacting antibodies present after Yersinia (or other) infection, leads to release of increased amounts of (or possibly modified) thyroid antigens which, in genetically prone individuals, leads to an increased but still a low level immune response. Nonspecific production of TNF-a and IFN-g, in response to any infection or immune response, may augment MHC class expression on TECs, allow these cells to function as APCs, and increase production of the already established, normally occurring low levels of antibodies. The process may be affected by stress, although the mechanism remains quite uncertain. The process may go on over years, and wax and wane, as it has been shown that thyroiditis can be clinically apparent and then disappear. Factors involved in temporary or permanent suppression of the autoimmune response may include diminished thyroidal release of antigen, B cell anti-idiotype feedback, or the normal auto-regulatory induction of T cells with a suppressor function, including those engendered by the mutual regulation of Th1 and Th2 subsets. In some individuals, thyroid cells may be less able to express DR, or may secrete TGF-b and suppress immune responses. Glucocorticoid administration and other immunosuppressives can also temporarily prevent the expression of nascent autoimmunity.

Stage 3 -- If suppressive factors do not control the developing immune response, the disease progresses to a new intensity, now driven by specific antigens, inducing cell hyperfunction (TSI), or hypofunction (TSH blocking or NIS antibodies), or cell destruction. Direct T cell cytolysis and apoptosis, ADCC, and K or NK cell attack presumably play an important role at this stage, and now the disease becomes clinically evident.

Stage 4 -- As the disease develops, a variety of secondary factors come into play, and augment antithyroid immunoreactivity. Any stimulus which causes increased DR expression on thyroid cells, such as T cell release of IFN-g, combined with increased TSH stimulation, may allow TECs to function as APCs. Although perhaps poor in this function, they are large in number and localized in one area. The TECs may also participate in the autoimmune process by several other pathways, including the expression of adhesion molecules, Fas, Fas ligand, CD40 and complement regulatory proteins, and the production of a number of inflammatory mediators such as cytokines, reactive oxygen metabolites, nitric oxide and prostaglandins. These events are, like class II expression, dependent on cytokines and other signals generated by the intrathyroidal lymphocytic infiltrate. Some patients may inherit diminished antigen-specific suppressor cell function. Development of hyperthyroidism, or more likely the ongoing immune reaction itself, may lead to nonspecific suppressor cell dysfunction, further augmenting immunoreactivity. Antigen "non-specific", and antigen specific suppressor T cells, may be reduced by binding immune complexes.

Stage 5 -- T cell derived cytokines may non-specifically induce bystander antigen specific T and B cells to be activated and produce antibody. Autoreactive cells will now accumulate in thyroid tissue because of the many strongly DR+ positive lymphocytes and TECs, and augment the developing response by lymphokine secretion or cytolysis, in a manner independent of thyroid antigens. At this stage in the disease, non-specific autoreactive immune processes may dominate a disease process which no longer depends upon antigen for its continuation.

Stage 6 -- As the concentration of activated T and B cells builds in thyroid tissue, and autoreactive and antigen nonspecific T cells become progressively involved, cell destruction may lead to release of new antigens. Cross-reacting epitopes, and nonspecific stimulation of T cells in genetically prone individuals, may cause the addition of new immunologic syndromes (exophthalmos, pretibial myxedema, atrophic gastritis) typical of older patients with more long standing and florid disease.