Autoimmune thyroid disease 

Autoimmune thyroiditis

In focal thyroiditis, the thyroid is usually normal in size and contains foci of lymphocytes which are predominantly T cells, although lymphoid follicles can also occur. Thyroid cells adjacent to these foci are usually atrophic and deficient in colloid, but away from the foci, thyroid follicular architecture is normal (5). Focal thyroiditis may also be prominent adjacent to a papillary carcinoma or other neoplasm. By contrast, the whole thyroid is usually involved in Hashimoto’s thyroiditis. The lymphocytic infiltrate is more extensive, diffuse, and composed mainly of T cells, with prominent germinal centres containing B cells scattered through the gland (Fig. 3.2.6.1). Macrophages, dendritic cells, and sometimes giant cells may be prominent. The thyroid follicles go through variable degrees of destruction, depending largely on chronicity, and in the process undergo hyperplasia and oxyphil metaplasia, giving rise to so-called Hürthle or Askanazy cells. These cells are generally absent in juvenile autoimmune thyroiditis.

• Click to view larger
• Download figure as PowerPoint slide

Fig. 3.2.6.1

Histological features of (a) normal thyroid, (b) atrophic thyroiditis, and (c) Hashimoto’s thyroiditis (original magnification ×100; photomicrographs courtesy of Dr K. Suvarna).

However, the relative proportion of lymphocytic infiltrate, thyroid follicular cell change, and fibrosis varies greatly, in keeping with the suggestion made previously that there is a broad spectrum of changes which may ultimately result in atrophic thyroiditis. In this condition, the thyroid is small, has extensive fibrosis mixed with a scattered lymphocytic infiltrate, and there is a marked reduction in thyroid follicular cells. Attempts have been made to subdivide these histological entities further, including a mixed variant of chronic thyroiditis, but the clinical value of this is limited. The pathology in postpartum and silent thyroiditis generally resembles mild to moderate Hashimoto’s thyroiditis, although without the oxyphil metaplasia. Germinal centres are usually absent.

Graves’ disease

It is now unusual to see the full histological picture of Graves’ disease as patients are almost all treated with antithyroid drugs which diminish the lymphocytic infiltrate (5). Even after such treatment, however, there is often a diffuse or focal lymphocytic thyroiditis, predominantly of T cells, sometimes with germinal centre formation. As an aside, lymphoid hyperplasia may also involve the lymph nodes, thymus, and spleen in Graves’ disease, once again being reversed by antithyroid drugs. The thyroid follicles are both hypertrophied and hyperplastic, with scalloping and reduction in colloid (Fig. 3.2.6.2). The epithelial cells are columnar and extend as papillae into the lumen. These changes are also attenuated by antithyroid drugs, so that after prolonged treatment, the colloid reaccumulates, the papillae regress, and the epithelium becomes cuboidal.

• Click to view larger
• Download figure as PowerPoint slide

Fig. 3.2.6.2

Histological features of Graves’ disease (original magnification ×100; photomicrograph courtesy of Dr K. Suvarna).

Genetic factors

These are dealt with extensively in Chapter 3.2.1. However, a brief discussion is given here, in relation to genetic effects on the autoimmune process. It is obvious clinically that thyroid diseases cluster in families more often than expected by chance, although the association of Graves’ disease and autoimmune hypothyroidism in such families, and their coassociation with autoimmune polyglandular syndrome type 2, indicates that at least some of the susceptibility is determined by genes that control a generalized tendency to organ-specific autoimmunity. One such determinant in white people is (p. 418) the HLA-DR3 specificity, which is associated with all of the major autoimmune endocrinopathies (6, 7). As HLA-DR3-positive healthy individuals differ from those who are DR3 negative in a number of immunological measurements, such as immune complex clearance, circulating T-lymphocyte subsets, immune responses to particulate antigens, and production of the cytokine tumour necrosis factor, this association may simply reflect a heightened nonspecific immune responsiveness. Thus, if a DR3-positive individual develops thyroid autoimmunity for any reason, this will be more likely to progress to florid disease.

Another reason why the highly polymorphic alleles of HLA class II genes (also called major histocompatibility complex or MHC class II genes) are associated with autoimmunity is that their products are expressed by antigen-presenting cells and are crucial in initiating any immune response (Fig. 3.2.6.4). Autoimmune disease may arise because a certain class II allele is able to bind and present a crucial fragment of an autoantigen, called an epitope, to a CD4⁺ T cell. Alternatively, the effect of class II alleles in determining immune responsiveness may be exerted in the thymus during development, at which stage future autoreactive T cells may be deleted (negative selection) or allowed to develop (positive selection). Finally, some class II molecules may determine selection of regulatory T cells, and deficiencies in these cells have been postulated as a cause of autoimmunity. It still remains unclear whether other genes in linkage disequilibrium with HLA-D-region (p. 419) genes confer additional susceptibility and it is possible that class I genes have a role independent of those in the class II region.

• Click to view larger
• Download figure as PowerPoint slide

Fig. 3.2.6.4

Key steps in antigen presentation and T-cell activation. The dotted line represents an inhibitory pathway. (From Weetman AP. Recent progress in autoimmune thyroid disease: an overview for the clinician. Thyroid Today, 1996; 19(2): 1–9, with permission).

The existence of non-HLA susceptibility genes is shown by the higher frequency of thyroid autoimmunity in monozygotic twins than in HLA-identical siblings, which in turn is higher than non-HLA identical siblings. The critical role of CTLA4 in costimulation of T-cell responses is discussed below and this has made it an excellent candidate to test as a susceptibility gene. It is now clear that polymorphisms in this gene have a significant role in autoimmune thyroid disease as well as several other autoimmune disorders that are associated with thyroid disease clinically. More recently it has become clear that polymorphisms in other T-cell regulatory genes, including PTPN22 and interleukin-2 receptor/CD25, can similarly increase susceptibility to autoimmune thyroid disease and related disorders. Overall it is now clear that many genes exerting small effects contribute to these diseases and their influence varies between individuals, which in turn may explain the diverse clinical presentations of thyroid autoimmunity.

Environmental factors

The lack of complete concordance for Graves’ disease in monozygotic twins and the clinically obvious lack of a family history in many patients with autoimmune thyroid disease, point to a major role for environmental factors in determining susceptibility, as does temporal variability in incidence (8). Furthermore, at least part of the family clustering of disease could be the result of shared exposure to environmental triggers. Some of the best evidence for the involvement of the environmental factors shown in Fig. 3.2.6.3 comes from animal models of experimental autoimmune thyroiditis (Table 3.2.6.2) which resemble Hashimoto’s thyroiditis (9). Excess dietary iodide exacerbates the severity of the lymphocytic thyroiditis in rats with experimental autoimmune thyroiditis, and (p. 420) leads to enhanced production of thyroglobulin antibodies; similar observations have been made in the Obese strain chicken which develops spontaneous autoimmune hypothyroidism. Indeed, rigorous depletion of iodide, starting at the stage of egg formation, virtually abolishes autoimmune thyroiditis in this strain of chicken. This observation neatly demonstrates the interaction possible between genetic and environmental factors. Excess iodide may have several effects, including an action directly on the immune system, the formation of an important part of a major T-cell epitope on the iodinated thyroid antigen thyroglobulin, and the generation of toxic metabolites within the thyroid which damage thyroid cells. There is epidemiological evidence to support a similar effect of excess iodide on human autoimmune thyroiditis (10), which is of some concern given the recent tendency for iodide intake to increase.

Infections could precipitate an autoimmune response by target-cell damage, leading to release of autoantigens, by altering target-cell expression of autoantigen or immunoregulatory molecules, such as HLA, or by molecular mimicry, in which an immune response against microorganism antigens that resemble host autoantigens triggers an autoimmune response. Despite the appeal of the notion and the success of animal models, there is surprisingly little evidence linking infection to human autoimmune disease. Autoimmune hypothyroidism occurs with increased frequency after congenital rubella infection, and epidemiological as well as serological studies have suggested a role for Yersinia infection in Graves’ disease. On the other hand, studies showing a lower frequency of thyroid and other types of autoimmunity in areas with a poor standard of hygiene suggest that in some settings infections may enhance immune responses in a way that avoids the emergence of autoimmunity, perhaps through skewing of the cytokine secretion of T helper cells, discussed below, (11). Despite many attempts, no convincing role for retroviruses in autoimmune thyroid disease is proven. Taking the opposite view, subacute thyroiditis is caused by a wide variety of viruses, and gives rise to thyroid destruction, yet rarely (if ever) triggers autoimmune thyroid disease. Only low and infrequent levels of thyroid antibodies occur in the course of infection and then disappear, although subacute thyroiditis may lead to permanent hypothyroidism in individuals who have coincidental subclinical autoimmune thyroiditis.

Stress now appears to be an important precipitant of Graves’ disease, based on analysis of preceding life events in newly diagnosed patients (12). A note of caution is necessary since retrospective recall and the influence of the evolving disease itself (both on recall and preceding interpersonal relationships) may have biased results, but reports are generally consistent and suggest an effect on susceptibility of roughly the same magnitude as HLA. The mechanism is presumed to be via the neuroendocrine effects of stress, in particular those mediated via the hypothalamus–pituitary–adrenal cortex axis. Interestingly, the Obese strain chicken, which develops spontaneous autoimmune thyroiditis, has an abnormal corticosteroid secretion profile that may be one of the genetic determinants of this disease (9).

As the therapeutic armamentarium expands, an increasing number of iatrogenic factors precipitate autoimmune thyroid disease. Mantle irradiation for lymphoma and other conditions is associated with an increased frequency of Graves’ disease and autoimmune thyroiditis, and rare cases of Graves’ disease have been reported following radio-iodine treatment of nodular thyroid disease. While these examples could be the result of thyroid injury, leading to autoantigen release, the lack of a parallel response in the wake of virally induced thyroid damage suggests additional mechanisms, such as a differentially suppressive effect of radiation on critical immunoregulatory T cells. An increased prevalence of thyroid autoantibodies has also been reported in children exposed to fallout form the Chernobyl nuclear reactor explosion (13). Lithium treatment is also associated with an increased prevalence of thyroid autoantibodies, hypothyroidism, and probably Graves’ disease.

Therapeutic doses of cytokines precipitate autoimmune hypothyroidism, but rarely Graves’ disease. The major culprit is α‎-interferon, probably because it is the most extensively used (14), but granulocyte-macrophage colony-stimulating factor, interleukin 2 (IL-2) and IL-4 have also been implicated. How these effects relate to the role of the same cytokines, at far lower endogenous concentrations, in untreated patients is unknown. However, an association exists between attacks of allergic rhinitis and the time of relapse of Graves’ disease, which may well depend on the nonspecific enhancing effects of cytokines released during the allergic response.

Environmental pollutants and toxins are theoretically important factors but remain underinvestigated. Administration of anthracene derivatives to genetically predisposed rats can precipitate experimental autoimmune thyroiditis. The potential of pollutants to operate in this way in humans is illustrated by the association between cigarette smoking and thyroid-associated ophthalmopathy (Chapter 3.3.10) as well as, to a lesser extent, Graves’ disease, whereas smoking appears to decrease the risk of Hashimoto’s thyroiditis (15).

Endogenous factors

The most impressive effect is imposed by pregnancy, which can lead to postpartum thyroiditis in around 5% of ostensibly healthy women (Chapter 3.4.6). However, the frequency of Graves’ disease is also increased in the 2 years postpartum and transient autoimmune hypothyroidism is frequently encountered after an episode of permanent hypothyroidism, indicating that pregnancy can produce a longer-lasting bias of the autoimmune response. Hyperprolactinaemia has only been inconsistently associated with an increased frequency of autoimmune thyroiditis, but clear evidence for a role of sex hormones has come from work on experimental autoimmune thyroiditis. Female animals given testosterone have a reduced frequency of thyroiditis, while castrated males, or those given oestrogen, have an increased frequency, which approaches that of females (9). These effects explain in large part the much higher rates of autoimmune thyroid disease in women, although it remains to be seen whether any other effects are encoded on the sex chromosomes to explain this dichotomy. Fetal microchimerism or skewed inactivation of the X chromosome are alternative possibilities (16). Low birthweight has only inconsistently been associated with an increased prevalence of thyroid autoimmunity in later life; presumably any such increase in risk depends on altered hormonal status determined in utero.

Thyroglobulin

Thyroglobulin is a homodimeric 660-kDa glycosylated iodoprotein which is secreted by thyroid follicular cells and stored in the luminal colloid; thyroglobulin also circulates. There are around 100 tyrosine molecules in each molecule of thyroglobulin and around 25 are normally iodinated, but this varies greatly depending on iodine uptake and thyroid activity. The iodination reaction depends on thyroid peroxidase and occurs at the apical border of the thyroid cells. Four thyroglobulin domains, termed A to D, have been identified from analysis of internal homology, and contain between them four to eight hormonogenic sites, two of which, at residues 5 and 2746, correspond to sites of preferential T₄ and triiodothyronine (T₃₎ synthesis, respectively. When stimulated by TSH or thyroid-stimulating antibodies, thyroglobulin is endocytosed and hydrolysed in lysosomes to release T₃ and T₄.

Although iodination of thyroglobulin plays a major role in the antigenicity of the molecule in animal models of autoimmune thyroiditis, the place of iodination in human autoimmune thyroid disease is less clear, with continuing uncertainty over whether the hormonogenic sites are part of T- or B-cell epitopes. As the immune response diversifies with time, an increasing number of epitopes are recognized, especially by sera with high levels of thyroglobulin antibodies, but patients with autoimmune thyroid disease show greater restriction of epitope recognition by autoantibodies than those who have autoantibodies but remain clinically euthyroid (17). These epitopes are largely conformational, although certain Hashimoto sera recognize linear determinants; all thyroglobulin antibodies cloned from patients so far recognize native but not denatured thyroglobulin. The immunopathogenic nondominant nature of thyroid autoantibody epitopes suggests that the disease may arise from unmasking of cryptic epitopes, which leads to a loss of tolerance (18).

The antibody response to thyroglobulin is relatively restricted, with a predominance of IgG₁ and IgG₄ subclasses and over-representation of certain immunoglobulin variable (V) genes. However, thyroglobulin antibodies, even of the IgG₁ subclass, do not fix complement due to the wide spacing of epitopes, which prevents cross-linking. The potential role of these antibodies in pathogenesis is considered below. Less is known about T-cell epitopes on thyroglobulin, information about which could lead to important insights regarding molecular mimicry with other self-determinants or microbial antigens.

Thyroid peroxidase

Thyroid peroxidase is a glycosylated haemoprotein which exists in two alternatively spliced forms of 100–105 kDa. The predominant form, TPO1, is responsible for tyrosine iodination and coupling to form thyroid hormones and is predominantly located at the apical border of the thyroid cell, anchored by a transmembrane segment near the C-terminus, with the catalytic domain facing the follicular lumen. TPO2 has no enzymatic activity and is restricted to the endoplasmic reticulum: its role in autoimmunity is unknown.

Initial studies of B-cell epitopes on thyroid peroxidase found two sequences, C2 (amino acids 590–622) and C21 (amino acids 710–722), which are linear epitopes recognized by the majority of Hashimoto sera and a smaller proportion of Graves’ sera (19). It is likely that these and other linear determinants identified subsequently are only the target of antibodies late in disease when degradation of thyroid peroxidase allows spreading of the immune response. In the initial stages, however, conformational epitopes are probably involved in antibody binding, and these have been identified by human and mouse monoclonal antibodies. There are two large overlapping domains, A and B, which are the target of more than 80% of thyroid peroxidase antibodies in Graves’ disease and Hashimoto’s thyroiditis and, in the absence of thyroid peroxidase crystals, modelling has allowed prediction of the structure of these. Furthermore the immunoglobulin V gene usage of thyroid peroxidase antibodies is remarkably restricted, with domain B-binding antibodies using a particular light-chain sequence (Vκ‎ 012), irrespective of heavy chain, although heavy-chain V gene usage is also relatively restricted. Relative binding of thyroid peroxidase antibodies to the individual domains varies little over time, indicating a genetic component to the control of thyroid peroxidase antibody formation. Thyroid peroxidase antibodies in general show the same type of IgG subclass restriction as those against thyroglobulin but are able to fix complement.

T-cell epitopes are multiple and individual patients respond to different combinations of epitopes without any apparent correlation with disease type or chronicity (9). As the T-cell response is likely to have had many months to diversify or ‘spread’ by the time of diagnosis, this observation is not surprising, but it does emphasize how difficult identification of any dominant epitope (which might cross-react with a microbial epitope) will be.

Thyroid-stimulating hormone receptor

The TSH receptor is a typical G-protein-coupled receptor, with an extracellular domain of 398 amino acids, a transmembrane region of 266 amino acids organized in seven loops, and an intracellular domain of 93 amino acids (20). There are two subunits, A (55 kDa) and B (40 kDa), which correspond to the extracellular and transmembrane domains and are joined by disulphide bonds. The A subunit can be shed from the cell surface, which may have immunological consequences by allowing greater access of the autoantigen to the immune system. Polymorphisms in the gene encoding the TSH receptor have been associated with Graves’ disease but not with autoimmune hypothyroidism (21), and elucidating the basis for this may be illuminating in understanding the differential expression of autoimmune thyroid diseases.

Although clearly highly expressed in the thyroid, where the receptor is fundamental for cell activation, there is now considerable evidence that the TSH receptor is expressed in fat, particularly preadipocytes, where it may make a contribution to thyroid-associated ophthalmopathy (see Chapter 3.3.10). The main physiological regulator of the TSH receptor is obviously TSH, which causes a rise in intracellular cAMP and, at high concentrations, activation of other signalling pathways, such as phospholipase C. These actions are mimicked by thyroid-stimulating antibodies in Graves’ disease, with the possibility that activation of different signalling pathways leads to disease heterogeneity, including goitrogenesis. (p. 422) The interaction of TSH-receptor antibodies with the receptor is even more complex, with additional antibodies blocking the effect of TSH (leading to hypothyroidism) and others appearing to bind without effects on function (neutral antibodies). The terminology of these antibodies has been obscure, and Table 3.2.6.3 gives an overview.

As would be predicted from the heterogeneous nature of TSH-receptor antibodies, multiple B-cell epitopes have been identified. In summary, the majority are conformational and comprise discontinuous sequences. Both stimulating and blocking antibodies bind to sites on the receptor which overlap with, but are distinct from, the TSH binding site (21, 22). The greatest separation between the binding of these three entities occurs at the N-terminal region of the receptor. Much is still to be determined, including whether the receptor dimerizes and how this could affect activation, and whether receptor desensitization might explain the poor correlation between circulating TSH-receptor stimulating antibody levels and the degree of abnormal thyroid function in patients. TSH-receptor stimulating antibodies show restriction immunoglobulin of heavy and light chain usage, implying oligoclonality of the B cell repertoire.

TSH-receptor T-cell epitopes have been identified and, as with thyroid peroxidase, there is considerable heterogeneity both within and between patients in the regions recognized, with no clear dominant epitope. Certain TSH-receptor sequences are recognized by 10–20% of healthy individuals, but it is not known whether these represent potentially pathogenic T cells kept in check by regulatory mechanisms or low-affinity nonspecific interactions of unlikely relevance to the initiation of Graves’ disease.

Animal models

T cells play a vital role in the pathogenesis of experimental autoimmune thyroiditis. Disease is easily transferable with T cells, whereas attempts to transfer disease using serum or antibodies produce only weak or inconsistent effects at best. Full-blown disease requires the transfer of both CD4⁺ and CD8⁺ cells from an animal with experimental autoimmune thyroiditis to a naïve recipient (disease being established in the donor by immunization with thyroglobulin in adjuvant). However, a subpopulation within the CD4⁺ cells also has an important regulatory function, being capable of preventing the action of thyroglobulin-specific, disease-inducing T cells (9). In essence, these findings are consistent with a model in which autoreactive T cells are largely, but not completely, deleted or rendered anergic in the thymus during development. These T cells are normally kept in check, either because they are controlled by a regulatory T-cell subset or through clonal ignorance in which the T cells fail to react to antigen in the absence of an appropriate costimulatory signal (Fig. 3.2.6.5). Animal strains particularly prone to experimental autoimmune thyroiditis have genetic defects either in positive/negative selection of T cells (which make it more likely that the adult animal has sufficient autoreactive T cells to develop disease) or in the regulatory T cell subsets, and these defects interact with environmental factors to result in disease (Table 3.2.6.4).

• Click to view larger
• Download figure as PowerPoint slide

Fig. 3.2.6.5

Alternative outcomes of major histocompatibility complex (MHC) class II molecule expression by thyroid cells, depending on the provision of co-stimulatory signals from antigen-presenting cells (APCs).

An appropriate balance of T helper cells (Th1 and Th2) (Table 3.2.6.5) is needed for full expression of disease, and the reciprocal inhibition between these two subsets (Fig. 3.2.6.4) may be one of the most important regulatory pathways controlling the activity of autoreactive T cells. For instance, blocking IL-2 receptor activation or removing γ‎-interferon leads to a granulomatous rather than lymphocytic thyroiditis, and the production of (p. 423) high levels of thyroglobulin antibodies, due to preferential Th2 activation (23). Typical experimental autoimmune thyroiditis is most likely Th1 dependent through the action of thyroid-specific cytotoxic T cells.

Further support for this T-cell-dependent mode of pathogenesis comes from the induction of experimental autoimmune thyroiditis by modulation of the T-cell repertoire alone, without the need to immunize animals with thyroid antigen (Table 3.2.6.2). Certain strains of rat or mice develop experimental autoimmune thyroiditis after thymectomy, sometimes coupled with sublethal irradiation, when performed at a critical stage of postnatal development (9), and T-cell depletion/reconstitution or ciclosporin A can have similar effects. Disease is reversed by a subset of CD4⁺ T cells from untreated donors. One major regulatory CD4⁺ T-cell population can be identified because it expresses CD25 and Foxp3. Depletion of this T-cell subset causes severe thyroiditis in certain mouse strains and this subset also appears to be reduced when thymectomy is performed (24). From these studies it is clear that thyroid-reactive T cells are present early after birth and that preferential removal of a critical regulatory subset of CD4⁺ T cells can induce organ-specific autoimmune disease. Transgenic mice have been used to confirm that tolerance, imposed in the thymus or periphery, is a major step in the production of thyroid reactivity; in contrast, B cells were not tolerized in animals overexpressing a membrane-bound antigen specifically on thyroid cells, presumably because the antigen is sequestered from B but not T cells (25). These B cells are harmless (or ‘ignorant’) unless specific T cells are available in a nontolerized state, in which case help in the form of B-cell stimulation might be provided, leading to thyroid antibody formation. The frequency of thyroid antibodies (and focal thyroiditis) in the healthy population may be due to the existence of such untolerized B cells, which can be partially activated if T cell tolerance is disrupted or bypassed, e.g. by the provision of B-cell-stimulatory cytokines by nonthyroid-specific T cells.

Human studies

The methods used to examine thyroid-reactive T cells in humans are shown in Table 3.2.6.6 and, despite their limitations, have provided important insights into the pathogenesis of autoimmune thyroid disease. A major problem has been the difficulty of access (p. 424) to thyroid-infiltrating T cells in untreated patients; blood-borne lymphocytes contain only a small proportion of thyroid-specific T cells which happen to be trafficking at the time of sampling, and although Graves’ thyroid tissue is often available for study, such patients have usually received treatment with antithyroid drugs which reduce the severity of the lymphocytic infiltrate, making the remaining T cells unrepresentative. Furthermore, it is obvious that any immune response, initially directed against a single epitope on a single antigen, rapidly diversifies to involve other epitopes and antigens, and this phenomenon of determinant spreading makes any analysis of T-cell reactivity in autoimmune diseases as chronic as those affecting the thyroid very difficult to interpret.

Humoral immunity

The role of thyroglobulin antibodies is uncertain, as they do not fix complement, but these antibodies may be involved in mediating antibody-dependent cell-mediated cytotoxicity. In this, the effector cell is a natural killer cell which binds to the antibody via Fc receptors on the natural killer cell surface. This allows the natural killer cell to destroy a specific target cell, in this case a thyroid cell, as otherwise natural killer cell-mediated destruction is not restricted by recognition of specific antigen. Antibody-dependent cell-mediated cytotoxicity is demonstrable in vitro with both thyroglobulin and thyroid peroxidase antibodies, small numbers of natural killer cells appear in the thyroid infiltrate, and monocytes may also be involved in this destructive pathway (35). However, transplacental transfer of thyroglobulin antibodies is not accompanied by thyroid dysfunction, and similar considerations apply to the frequent presence of thyroglobulin antibodies in euthyroid individuals. Thyroid peroxidase antibodies can fix complement but, for similar reasons, would seem to be of minor importance as primary mediators of thyroid cell destruction. Thyroid peroxidase may well be sequestered from access by autoantibodies until late in the disease process, when cell-mediated injury will permit antibody binding, although there is evidence for some internalization of thyroid peroxidase antibody by thyroid cells, the consequences of which are unknown.

A second reason for the failure of complement-fixing thyroid peroxidase antibodies to destroy thyroid cells is that, in common with all nucleated cells, thyroid cells express complement regulatory proteins which prevent lethal injury by interfering with C3 convertase activity or by impairing terminal complement component formation. The most important of these regulatory proteins functionally is CD59, and its expression is upregulated by IL-1, γ‎-interferon, and tumour necrosis factor, all of which are produced by the lymphocytic infiltrate, thus enhancing the ability of thyroid cells to defend themselves from complement attack (36). There is good evidence that complement is activated in thyroid autoimmunity, with elevated serum levels of terminal complement complexes, and local deposition of such complexes around the thyroid follicles in both Graves’ disease and Hashimoto’s thyroiditis. Unless formed in overwhelming amounts, complement membrane attack complexes do not overcome the thyroid cell’s defences, but none the less, sublethal effects of complement attack are demonstrable in vitro, and include impaired responses to TSH stimulation and the release of cytokines, reactive oxygen metabolites, and prostaglandins, which will contribute to the local inflammatory response (36). Antithyroid drugs block this phlogistic response to complement attack, which may explain the selective immunomodulatory effects of these drugs.

A final mechanism by which antibodies can cause hypothyroidism is through their direct effects on cell function, most clearly illustrated by TSH-receptor antibodies. Although it is nearly certain that all patients with Graves’ disease have TSH-receptor stimulating antibodies, these may be absent in the serum of around 5% of patients when measured using the currently available binding assays (37). As well as assay insensitivity as an explanation, it is possible in these cases that there is exclusively intrathyroidal production of autoantibody which is sufficient to sustain disease. Thyroid peroxidase antibodies can inhibit thyroid peroxidase enzymatic activity operating in vitro, which would contribute to hypothyroidism if also present in vivo, but the importance of this inhibition is questionable.

Cell-mediated immunity

Cytokines released locally by the infiltrating lymphocytes and macrophages may have a number of effects that exacerbate thyroid injury. Some of these effects are related to the metabolic activity of the thyroid cells, such as decreased synthesis of thyroglobulin or thyroid peroxidase, which will ultimately impair thyroid hormone production (Table 3.2.6.7), while others evoke responses by thyroid cells which have direct immunological relevance. One of these has already been discussed, namely the expression of MHC class II molecules induced by γ‎-interferon, but many other effects are being uncovered. Adhesion molecules allow cytotoxic T cells and natural killer cells to bind initially to their targets, and the up-regulation of thyroid cell adhesion molecule expression by cytokines will enhance the susceptibility of thyroid cells to such attack (9). Nitric oxide and reactive oxygen species may play a key role in thyroid injury and their production by thyroid cells is initiated by the intrathyroidal proinflammatory environment which exists in autoimmune thyroiditis (9, 38). Finally, certain cytokines, in particular IL-1, IL-6, IL-8, IL-12, IL-13, and IL-15, are produced by thyroid cells in response to inflammatory cytokines, especially (p. 427) IL-1 (30), and this may set up a mutually reinforcing pathway of cytokine interactions which results in escalation and perpetuation of the autoimmune process (Fig. 3.2.6.7).

• Click to view larger
• Download figure as PowerPoint slide

Fig. 3.2.6.7

Cytokine interactions between the immune system and thyroid cells in autoimmune thyroid disease. (From Weetman AP, Ajjan RA, Watson PF. Cytokines and Graves´ disease. Bailliëre’s Clinical Endocrinology and Metabolism, 1997; 11: 481–97, with permission.)

As well as thyroid cells, vascular endothelial cells in the thyroid are exposed to cytokines which up-regulate expression of selectins and other molecules essential to the egress of inflammatory cells from the blood. Thyroid cells can also produce an array of chemokines, molecules which are able to enhance the recruitment of lymphocytes to the gland in disease. Chemokine synthesis may also be critical in the formation of lymphoid germinal centres in chronically affected thyroid tissue (39). Clearly, these processes of adhesion molecule expression and chemokine synthesis are essential to the recruitment of lymphocytes to the infiltrate, although it is unknown what proportion of these are blood-derived and what proportion result from local expansion.

Specific cytotoxic T cells have long been thought to be key mediators of thyroid cell destruction in autoimmune thyroiditis, but evidence for their existence is surprisingly sparse and best documented in experimental autoimmune thyroiditis (9). As well as releasing cytokines, cytotoxic T cells kill either by insertion of perforin into the target cell membrane, or by interaction of Fas ligand on the T-cell surface with the widely expressed Fas molecule on the target cell. Perforin-expressing T cells are present in the thyroid infiltrate in both Hashimoto’s thyroiditis and Graves’ disease, with slightly differing phenotypes in the two conditions (40). This certainly indicates the potential for perforin-mediated cell destruction, although recent attention has focused on Fas-mediated apoptosis as a major mechanism for thyroid cell death (41). This interest has been sparked by the demonstration of Fas ligand expression by thyroid cells in Hashimoto’s thyroiditis, but not other conditions. Fas ligand expression was enhanced in vitro by IL-1β‎ but not other cytokines, suggesting that, in addition to the classic pathway of apoptosis mediated by T cells, Fas and Fas ligand on thyroid cells could interact and lead to cell suicide. Normally Fas ligand expression is limited to sites of immunological privilege, such as the trophoblast and Sertoli cells, where it is clear that suicide is not an outcome; instead, Fas ligand expression at these sites ensures tolerance by deleting any autoaggressive Fas-expressing lymphocytes specific for these tissues. Thus a major effect of thyroid cell Fas ligand expression in vivo may be the evasion of thyroid cell recognition by T cells.

In summary, thyroid cell dysfunction and destruction result from a wide array of insults (Fig. 3.2.6.8) and, in the initial stage at least, seems dependent on cell-mediated autoimmune processes. It is likely that within the same clinically identified disease there are interindividual differences in the relative contributions from each type of injury. This variation would account for the diversity of pathological processes previously described, and because of this complexity, it is highly improbable that only two types of mechanism predominate, one resulting in atrophic thyroiditis and the other in goitrous thyroiditis.

• Click to view larger
• Download figure as PowerPoint slide

Fig. 3.2.6.8

Main mechanisms involved in thyroid-cell dysfunction in autoimmune hypothyroidism.

Thyroglobulin and thyroid peroxidase antibodies

There are essentially four methods for assaying thyroglobulin and thyroid peroxidase antibodies. The two oldest are haemagglutination and indirect immunofluorescence, which depend on dilution of the test serum to determine the level of antibodies. Although robust and providing reasonably sensitive and specific results, the more modern methods of enzyme-linked immunosorbent assay (ELISA) and radioimmunoassay allow truly quantitative determination of antibodies and, in the case of assays for thyroid peroxidase antibodies, can use antigen of high purity, if necessary for research purposes. Thyroid peroxidase was previously called the microsomal antigen, and assays for these antibodies have relied on positive immunofluorescence staining with an appropriate pattern or, in the case of haemagglutination, have used an excess of (p. 428) thyroglobulin to absorb out thyroglobulin antibody activity when testing crude microsomal extracts of thyroid homogenate. Comparison of assays based on haemagglutination with microsomal antigen and more modern methods with purified or even recombinant thyroid peroxidase has shown a good correlation between the two, although those assays based on thyroid peroxidase are more sensitive. Further improvement in assay standardization has come from the use of reference positive serum samples, such as those obtained from the National Institute of Biological Standards and Control, UK, and increased sensitivity is possible with the use of immunoradiometric assays.

With the most sensitive assays, up to 20% of healthy women have thyroglobulin and/or thyroid peroxidase antibodies, although in the majority the levels are very low. Using more conventional assays, 11% of women and 3% of men were positive in a large community-based survey in the UK (3) and similar results have been reported elsewhere. Antibodies are not entirely stable, appearing or disappearing in 17% and 2% of women, respectively, over a 20-year period (3). Moreover, although there is a rise in the frequency of thyroid antibodies with increasing age, healthy centenarians show a remarkably low frequency (Fig. 3.2.6.9), suggesting that thyroid autoimmunity is associated with senescence-related illnesses (42). Thyroid peroxidase antibodies are found in 80–90% of Graves’ sera and 95–100% of Hashimoto sera, with thyroglobulin antibodies in up to 70% of Graves’ sera and 90–100% of Hashimoto sera, using sensitive assays. Occasional patients with Hashimoto’s thyroiditis are negative for serum thyroid antibodies, although synthesis can be detected locally within the thyroid, presumably at too low a level to be detectable in serum. In most patients, thyroglobulin antibodies are accompanied by thyroid peroxidase antibodies, but thyroid peroxidase antibodies frequently occur in the absence of thyroglobulin antibodies. This has led some centres to abandon routine testing for thyroglobulin antibodies in the diagnosis of autoimmune thyroid disease.

• Click to view larger
• Download figure as PowerPoint slide

Fig. 3.2.6.9

Change in prevalence in antibodies to thyroglobulin (•) and thyroid peroxidase (O) in healthy subjects with age. (From Mariotti S et al. Thyroid and other organ-specific autoantibodies in healthy centenarians. Lancet, 1992; 339: 1506–8, with permission.)

Thyroglobulin and thyroid peroxidase antibodies are found in a variety of other conditions at higher frequency than would be expected by chance (Box 3.2.6.1). As in healthy individuals, the presence of such antibodies is a marker of future thyroid dysfunction, especially if coupled with subclinical hypothyroidism, and all patients with positive thyroid antibodies should be offered annual screening to detect early thyroid failure, while patients with subclinical hypothyroidism should have antibodies measured to stratify their risk (3). Another situation where prospective thyroid antibody testing is particularly worthwhile is in patients starting amiodarone, as those with antibodies are more likely to develop amiodarone-induced hypothyroidism. Antibody testing is also useful in patients with Addison’s disease, as around 25% may develop (p. 429) thyroid dysfunction due to associated autoimmune polyglandular syndrome type 2. Similar considerations apply to pernicious anaemia and other autoimmune disorders which are associated with a high frequency of thyroid autoimmunity (43).

On the other hand, thyroid antibodies can be misleading in goitre as 10–50% of patients with multinodular goitre have thyroid antibodies, although usually only at low or moderate levels. Similarly, 25–50% of patients with papillary or follicular thyroid cancer have thyroglobulin and/or thyroid peroxidase antibodies. Such patients have a somewhat better prognosis than those without thyroid antibodies, and the presence of antibodies is correlated with a coexistent thyroiditis, suggesting that this autoimmune response is beneficial.

Finally, research since 1980 has thrown up some new associations, the clinical importance of which is yet to be fully realized. Thyroid peroxidase antibody positivity is strongly related to postpartum thyroiditis, giving rise to the suggestion that it may be worthwhile screening all pregnant women antepartum, but the positive predictive value of thyroid peroxidase antibodies is quite low (about 40–60%) and some cases have been reported in women who are thyroid peroxidase antibody negative (44). Because of the high frequency of postpartum thyroiditis in type 1 diabetes mellitus, which is approximately 3 times normal, there is a strong case for thyroid peroxidase antibody screening of this group of women antepartum. Women with positive thyroid antibodies, even without clinical thyroid dysfunction, do seem at risk of postpartum depression (albeit mildly) and recurrent first-trimester miscarriage. There is also a threefold increase in the relative risk of depression in perimenopausal women with thyroid peroxidase antibodies (45). Whether the mental disorder is the cause or result of the autoimmune response is not yet known, but presumably some interaction between the neurological, endocrine, and immune systems is involved, akin to the adverse effects of stress described previously. An unexplained association also exists between thyroid autoantibody positivity and breast cancer, with an improved prognosis in those women who have positive thyroid peroxidase (46). Most recently, an association between the presence of thyroid antibodies and miscarriage has been identified (47), which in turn has led to the concept that thyroxine treatment should be considered even in euthyroid individuals with thyroid antibodies and a history of such an adverse outcome in pregnancy.

Thyroid-stimulating hormone-receptor antibodies

The terminology of TSH-receptor antibodies (Table 3.2.6.3) has evolved from the methods used for their measurement. In essence there are two current methods: (1) the binding assay, which measures the capacity of immunoglobulins to inhibit the binding of radiolabelled TSH to purified or recombinant TSH receptor and (2) bioassays which measure the stimulatory or inhibitory effects of immunoglobulins on some aspect of thyroid cell function (48). Generally, cAMP production is used as the endpoint in bioassays, but there has been an irreversible move away from using primary cultures of animal or human thyroid cells in these assays, with their attendant problems of supply and standardization, to using either cell lines, such as rat FRTL5 cells, or Chinese hamster ovary cells transfected with TSH receptor, such as JPO9 cells. With the most sensitive bioassays for TSH-receptor stimulating antibodies almost all patients with Graves’ disease are positive, but these antibodies are rarely found in its absence, and then would be associated with a greatly increased risk of future hyperthyroidism. This is shown most clearly by the finding that 30–50% of euthyroid patients with thyroid-associated ophthalmopathy have TSH-receptor antibodies, and this proportion increases if the most sensitive assays are used.

As would be predicted, there is only a weak or absent correlation between levels of TSH-receptor antibodies measured in the binding and stimulatory bioassays. Up to 95% of patients with Graves’ disease are positive using modern binding assays, as are 10–20% of patients with autoimmune hypothyroidism. In the latter, the binding activity is mostly due to TSH-receptor blocking antibodies, but neutral antibodies with binding but not biological activity could theoretically also be detected. Antibodies against the TSH receptor are present at much lower concentrations than thyroid peroxidase antibodies and this makes the development of robust and simple solid-phase assays very difficult, compounded by problems in expressing the TSH receptor in its native form.

TSH-receptor antibody testing is not recommended for routine use in the diagnosis of Graves’ disease when the diagnosis is clinically obvious, such as when there is coincident ophthalmopathy, or when such information will not influence management, for instance if the decision has already been made to proceed with radio-iodine treatment (49). Measurement of thyroid peroxidase antibodies, coupled (p. 430) with clinical examination and, if necessary, a thyroid scan to confirm a diffuse goitre, are also reasonable alternatives to TSH-receptor antibody testing for diagnostic purposes, if the latter are not readily available. When it is necessary to test for these antibodies, second-generation binding assays give information which is essentially comparable to bioassays, providing the results are interpreted in the light of the patient’s clinical and biochemical status. Prediction of outcome after antithyroid drugs has been another suggested use for these assays, but although there is no doubt that the presence of detectable TSH-receptor antibodies after treatment is associated with a higher rate of relapse, the sensitivity and specificity of this measurement is too poor to be used in clinical prognosis. The one situation where TSH-receptor antibody measurement is definitely indicated is during pregnancy in Graves’ disease; a high level of maternal antibodies at the beginning of the third trimester is a strong predictor of neonatal thyrotoxicosis (50).