Thyroid-Stimulating Antibodies Predict Hyperthyroidisma

OVER 50% of patients with hyperthyroid Graves’ disease relapse after a 12-month course of antithyroid drugs, the actual percentage varying among populations and with their iodine intake. The measurement of TSH receptor antibodies (TSHR-Abs) by competitive radioreceptor assay or by thyroid cell bioassay (1) in patients with Graves’ disease can be a useful predictor of relapse and remission (2). However, controversy has surrounded the clinical utility of TSHR-Ab measurements ever since my colleagues and I published the first report of their use in the Lancet in 1977 (2). In that study, we analyzed the usefulness of TSHR-Abs to predict relapse and remission after 6 months of antithyroid drug treatment and found that highly positive titers of the autoantibody were indeed useful predictors of recurrence (Fig. 1). Similar data had been published previously using indirect measurements of TSHR-Abs—for example, the pioneering work of Alexander, who had showed clearly that the failure of T3 to suppress radioiodine uptake increased the chances of recurrence because a non-TSH thyroid-stimulator of radioiodine uptake was present (3). Similar studies continue to confirm such observations (4) of the failure to suppress T4 levels with T3 therapy. The assay for TSHR-Ab that we had used at that time, the original simple thyroid membrane radioreceptor assay of Smith and Hall (5), was new, but the conclusions were similar to the data of the time. The presence of an abnormal thyroid stimulator indicated a higher chance of recurrence than in patients in whom such a stimulator could not be detected. While a number of the publications that followed our report confirmed this observation (6–14), others were less enthusiastic (15–21). The original observation certainly had a number of problems. The most serious was that the follow-up time for detecting recurrence was only 6 months. But I had a fellowship to move to at the NIH, and the study had to end! Furthermore, sensitive TSH assays were not available to help us in the management of the patients. Nevertheless, I believe time has confirmed the original general conclusions. The presence of significant titers of biologically active TSHR-Abs accurately predicts recurrence in up to 90% of such patients. The original data showing prediction of recurrent Graves’ disease after 6 months treatment with antithyroid drugs (combined carbimazole and thyroxine). Data were expressed as a TSHR-Ab Index, where a low index indicated high titers of TSHR-Abs. On the left are shown those patients who remained in remission, and on the right are those patients who had recurrence defined as increased T3 and T4 levels. (Data redrawn from ref. 2). As discussed above and elsewhere, there have been many publications examining the clinical usefulness of TSHR-Ab measurement. I would like to focus here on a recent meta-analysis published in JCEM that has not, to my knowledge, been discussed at length in the literature (22). The conclusion from this attempted meta-analysis of 18 different reports was that “the available methods for TSHR-Ab do not allow sufficiently high prediction of relapse or remission after antithyroid drugs for the individual patient.” I find this conclusion inconsistent with the data presented by the authors and other commentators (23). The meta-analysis included only certain of the available studies according to a number of clear criteria: TSHR-Ab measurements were available at the end of antithyroid drug treatment. The follow up period after stopping therapy was at least 1 yr. Individual rather than grouped data were available. No thyroid hormone therapy was given after stopping antithyroid drugs. Each of these criteria was appropriate and strictly adhered to in the studies selected. However, no criteria at all were applied to the assay of TSHR-Abs except to divide the data into studies with radioreceptor assays and studies with bioassays. The overall assumption of the meta-analysis, therefore, was that all the different assays used in the 18 different studies were of equal sensitivity, equal precision and, most importantly, equal specificity. However, many of the studies used their own“ in house” methods for the measurement of TSHR-Abs, sometimes for the first time. We lack, therefore, essential data on the assay performances, which would make a meta-analysis possible to interpret. For example, in one study (24) only 3 out of 184 patients had detectable TSHR-Abs at the end of antithyroid drug treatment. In contrast, in another study 142 out of 391 patients had TSHR-Abs detected (18). In addition, the actual titer of antibody was not taken into consideration in any of the studies analyzed. Borderline positive results were given the same weight as very high titers. Furthermore, what did the different authors mean by remission? Was the TSH normalized in these patients? Relapse and remission were also not defined in this meta-analysis of relapse and remission. The body of literature on this subject is, therefore, difficult to interpret without careful scrutiny of each study, and inclusion/exclusion criteria would need to be so strict that there would be insufficient data to perform a meta-analysis. Yet we have to conclude that some reports have clearly shown that more than 90% of higher titer TSHR-Ab positive patients had a recurrence of Graves’ disease after withdrawal of antithyroid drugs (13, 25), while this was not seen in other reports (18). How can we explain these disparities? I would like to begin with the only logical conclusion there can be. The presence of TSH receptor antibodies with stimulating activity in the serum of a patient with recent hyperthyroid Graves’ disease must influence thyroid function, unless there is some interference in the transduction of such a signal. There are a number of possible reasons for thyrotoxic recurrence after antithyroid drugs not being as predictable as my logic suggests: 1. Antibody quantity and affinity. In theory it is possible that the level of TSHR-Abs could be so insignificant or their affinity so low that their influence on the TSH receptor would be negligible. However, under such circumstances it is unlikely that the TSHR-Abs would be detectable using currently available assays, which remain relatively insensitive to TSH equivalents of less than 10 mU/L. Furthermore, even a slightly increased TSH level is known to be capable of increasing thyroid hormone levels, as seen in patients with thyroid hormone resistance. Hence, the very presence of detectable TSHR-Abs with stimulating activity should lead to hyperthyroidism unless the assay systems employed are providing false positive results or unless there is a target defect (i.e. thyroid dysfunction). 2. Antibody quality. Not all TSHR-Abs are stimulating antibodies. Blocking TSHR-Abs have been well characterized in many laboratories, and some patients with Graves’ disease have been shown to have both types of TSHR-Abs present in their serum (26). Of course, in order for hyperthyroidism to develop there must be a lower concentration or affinity of the blocking variety (27–30). This has led to discussions on the factors responsible for the balance between stimulating and blocking antibodies and the differing methods for measuring TSHR-Abs, discussions that have occupied clinical thyroidologists for many years. The technique used for the measurement of biologically active TSHR-Abs is indeed critical. Thyroid cell assays have been replaced with more sensitive TSH receptor transfected Chinese Hamster Ovary (CHO) cell stimulation assays (31) and remain the only in vitro way to distinguish stimulating from blocking activity. Many argue that it is never possible to infer the biological activity of TSHR-Abs from competitive binding assays, and in some circumstances this is certainly true. However, if the patient is already thyrotoxic or has recently been thyrotoxic, it does not take a great brain to deduce that the patient’s TSHR-Abs are predominantly of the thyroid-stimulating variety. The real question, therefore, is whether TSHR-Abs change from stimulating to blocking during the course of antithyroid drug treatment. The literature to date is relatively unhelpful in this regard, but this appears to be an unusual event (32, 33). Hence, the measurement of TSHR-Abs by an expensive bioassay is not necessary for everyday clinical practice in patients with recent hyperthyroid Graves’ disease. 3. Problems with thyroid function. There are two major potential disturbances in the thyroid reserve that may prevent an appropriate thyroid hormone response to TSHR-Abs of the stimulating variety—iodine deficiency and autoimmune thyroiditis (34). Iodine deficiency remains the major cause of global thyroid disease, and moderate iodine deficiency continues to affect thyroid function even in many parts of Western Europe such as Germany, where 15% of the population may have iodine deficiency goiters (35). It is not surprising, therefore, that low titers of TSHR-Abs would have difficulty inducing hyperthyroidism in certain parts of the world because of the lack of intrathyroidal iodine. However, in countries such as the USA, Japan, and Korea, where there are high iodine diets, this restriction should not apply. This certainly complicates the meta-analysis from Feldt-Rasmussen (22), where many of the studies were from countries with mild iodine deficiency. Similarly, the onset of significant autoimmune thyroiditis would reduce the thyroid reserve considerably and would require enhanced thyroid stimulation to raise thyroid hormone levels above normal. Hence, the presence of TSHR-Abs would not always predict the hyperthyroid state. No data on the onset of hypothyroidism was provided for the 55 patients in remission with detectable TSHR-Abs in large multicenter studies (18) or in any of the other published prediction studies, but the geographic location of the patients would suggest that iodine deficiency was more important that thyroiditis. Furthermore, it appears to take many years for thyroiditis to induce thyroid failure in some Graves’ patients (36). We have all seen examples where this can happen much more suddenly during treatment, but it is unusual. 4. Assay sensitivity. I believe the major problem today, with easily available and relatively precise and specific assays, is not false positive TSHR-Ab results but the large group of patients who develop hyperthyroidism in the absence of detectable TSHR-Abs. Clearly, in these patients a lack of thyroid reserve is not an issue. Accepting the evidence that the presence of TSHR-Abs often predicts recurrence in the presence of iodine sufficiency, does their absence predict remission? Many studies of this problem have shown that a negative result, as so often in medicine, is of little help to the clinician. Of course the loss of previously detectable TSHR-Abs indicates that the immune system may have become more tolerant, but such patients may still have a high recurrence rate after discontinuing antithyroid drugs. Is this due to problems with detecting low levels of TSHR-Ab? Following the earlier discussion, there is much evidence to suggest that only small increases in TSH (and, therefore, TSHR-Ab) lead to thyroid hyperfunction. However, as also mentioned earlier, the bioassays and radioreceptor assays used for the measurement of TSHR-Abs are insensitive and/or imprecise to TSH concentrations of less than 10 mU/L. Low levels of receptor antibody may not be detected. In addition, thyroid autoantibodies in the circulation are merely a reflection of their intrathyroidal production. For example, patients with Hashimoto’s disease and negative anti-Tg and anti-TPO still have intrathyroidal B cells capable of secreting such antibodies (37). No doubt the same applies to patients with Graves’ disease and negative TSHR-Abs. Hence, assay sensitivity is likely to be the major factor in the analysis of recurrent hyperthyroidism. Studies based on in-house assays of dubious validity from inexperienced investigators are likely to accentuate this intrinsic problem. Of course TSHR-Ab negative patients may be patients who do not have Graves’ disease. The diagnostic criteria used in published studies should also be closely scrutinized. One of the most common arguments against the clinical use of TSHR-Abs has been that the test is not cost-effective. In the United States, laboratory costs are not reflected in the cost of the test performed, but are based more on what the market will pay. In essence, the radioreceptor assay for TSHR-Abs need only cost $10–20, but in practice the patient is more likely to pay $70–80. Under laboratory capitation contracts, this problem is eliminated. Furthermore, both costs are low compared with testing methods in other diseases and specialties. Assuming that all Graves’ patients are tested for TSHR-Abs at diagnosis, then only those patients with higher titers (>20% binding inhibition) need be tested again at the end of their antithyroid drug regimen. Those patients who remain positive should continue on antithyroid drugs for another 6–12 months before retesting, as shown in the algorithm (Fig. 2). With a 90% prediction success in positive patients the cost-effectiveness of this approach is self evident. However, if only 5% of patients are TSHR-Abs positive at the end of a course of antithyroid drug treatment, then this approach may be impractical. This is not the clinical experience in most centers. Physicians using this approach with antithyroid drug regimens of 18 months or more may find many of their patients to be TSHR-Ab negative and the test unhelpful as many such patients will remit after such long-term treatment. However, in my experience, a majority of patients can be withdrawn from treatment much earlier, and those in whom such discontinuation should be avoided can be identified by using the TSHR-Ab testing algorithm (Fig. 2). Nothing can be more cost effective than treating patients for as short a time as possible while avoiding unnecessary thyrotoxic recurrences. A suggested clinical pathway for the treatment of Graves’ disease with antithyroid drugs using the TSHR-Ab test as an aid to decision making. In areas of iodine deficiency or where drugs are used for very prolonged periods, the pathway may need to be adjusted. The validity of predicting neonatal Graves’ disease using the measurement of biologically active TSHR-Abs in the mother has been well documented over many years (38, 39). The rarity of this problem demonstrates the success of modern medical care with the early treatment of Graves’ disease. This is, however, the one clinical situation where the bioactivity of the TSHR-Abs may need to be known to predict the neonatal thyroid function (40). Luckily, pregnancy is a time when thyroid autoantibodies generally decrease in titer, due to the secretion of trophoblast factors, which are immunosuppressive, and bioactivity assessment is not necessary in many patients. However, in those Graves’ patients with higher titers of TSHR-Ab by radioassay in the third trimester, particularly those patients who have not been recently hyperthyroid, a subsequent bioassay will allow the prediction of neonatal hyperthyroidism or blocking antibody-induced hypothyroidism (27, 41). This approach applies most importantly to women previously treated for Graves’ disease and who are euthyroid or on thyroxine replacement but have persisting titers of TSHR-Abs. Many physicians would pay dearly to obtain a marker for the disease that interests them most. Those of us interested in caring for patients with Graves’ disease have such a marker. Knowing the titer of TSHR-Abs in patients is useful in the prediction of thyrotoxic recurrence after antithyroid drugs. A careful review of the literature shows strong evidence for the observation that their accurately measured presence predicts hyperthyroidism in more than 90% of cases in iodine-sufficient areas. The assay of TSHR-Abs, therefore, remains a most useful addition to the clinical armamentarium and a low-cost help in treatment planning. Assays to distinguish blocking from stimulating antibodies remain expensive but are only clinically important in patients with pregnancy (40). The major hurdle remains one of increasing the sensitivity of the available assays for TSHR-Ab so that their usefulness can be applied successfully to an even greater proportion of patients with Graves’ disease. a Supported in part by: the National Institutes of Health Grants DK52464, DK35764, and DK45011 from NIDDK. Please note that Dr. Davies is a Board member of the Kronus company, San Clemente, California, which sells assays for the measurement of thyroid autoantibodies including TSH receptor autoantibodies. The views expressed are those of the author only.