Graves' disease

The thyroid gland controls the metabolic rate of many organs and tissues. Underactivity and overactivity of thyroid function represent the commonest endocrine problems, have widespread manifestations, and often require long term treatment. Therefore, all practising clinicians have to be aware of thyroid physiology and the consequences of dysthyroidism.

Methods

We have chosen topics in thyroid disease of interest to clinicians in both primary and secondary care: compliance, because it remains a challenge to all clinicians; and subclinical thyroid disease and the effect of amiodarone on thyroid function because they are interesting and evolving topics. We have also addressed some aspects of Graves' disease that have recently generated interest. Our sources included papers from Medline as well as discussions from recent national and international endocrinology meetings.

Compliance with treatment

Management of thyroid disorders usually requires prolonged, and often lifelong, courses of treatment. Hence adequate compliance is needed to achieve and maintain euthyroidism. The sensitive assay for thyroid stimulating hormone has advantages over assays for thyroxine, triiodothyronine, and free thyroxine, and both the free thyroxine index and older versions of the thyroid stimulating hormone radioimmunoassay. The sensitive assay helps to differentiate normal concentrations of thyroid stimulating hormone in euthyroid subjects from low concentrations (for example, hypothyroidism secondary to pituitary insufficiency, subclinical hyperthyroidism). The assay is also independent of changes in concentrations of thyroxine binding globulin, which occur, for example, during pregnancy and hormone replacement therapy.[1]

Poor compliance (box 1) is the most common cause of persistently increased thyroid stimulating hormone concentrations in patients who take excessive doses of thyroxine for their size or who have wide variations in thyroid function test results on the same dose of thyroxine (box 1). In the absence of clear evidence of malabsorption--for example, with bowel bypass surgery or sprue--there is no evidence that malabsorption of thyroxine exists as an isolated entity.[2]

Occasionally patients take multiple daily doses of thyroxine just before their follow up visit. This results in increased free thryoxine concentrations and an increased free thryoxine index but an inappropriately raised thyroid stimulating hormone concentration. As well as increased thyroid stimulating hormone concentrations, poor compliance with thyroxine can result in several challenging presentations.[3]

Improving compliance

Several methods may improve compliance with treatment: (a) thyroid patient groups, by increasing the understanding of thyroid disease[4]; (b) education (particularly of the primary care team) so that the education of patients is improved[5]; and (c) thyroid registers--for example, WAFUR and SAFUR, the Welsh and Scottish automated follow up registers respectively (patients are registered on a computer once they are both clinically and biochemically euthyroid then recalled annually for review and a thyroid function test, the results of which are reviewed centrally thereby limiting follow up only to those with abnormal test results). Funding registers, despite their proved cost effectiveness,[6] remain a problem.

In one study, 48% of patients taking thyroxine had abnormal thyroid stimulating hormone concentrations--27% high and 21% low concentrations. The relation between the prescribed thyroxine dose and the thyroid stimulating hormone concentration suggested that undertreatment and overtreatment were common and that these largely reflected inappropriate dosage rather than poor compliance (high concentrations were found in 47% of patients taking less than 100 [micro]g thyroxine, whereas low concentrations were found in 24% of patients taking more than 100 [micro]g thyroxine).[7] Thyroid registers recall patients with abnormal thyroid function test results who become lost to follow up, thus reducing poor compliance or inappropriate dosage.[8]

The most important way of improving patient compliance is to simplify the treatment regimen--for example, by widening the strength range of thyroxine tablets in hypothyroid patients so that the drug can be taken less frequently.[9] More hyperthyroid patients (83%) were compliant when taking methimazole once daily than when taking propylthiouracil every 8 hours (53%)[10]

It should be noted that patients receiving thyroxine may have total or free thyroxine concentrations above the normal reference range of the laboratory. This should not be taken as indicating a reduction in thyroxine dose, especially if the patient is clinically euthyroid and has normal thyroid stimulating hormone concentrations.

Nodular thyroid disease

Nodular thyroid disease was reviewed recently.[11] We focus on the role of fine needle aspiration and the use of thyroxine in therapy.

Fine needle aspiration

All solitary thyroid nodules should be examined by fine needle aspiration biopsy. Typically, 70% are benign, 4% malignant, and the remainder inconclusive. Surgery is advisable when cytological findings are indeterminate or repeatedly inadequate. If hyperthyroidism was evident, the treatment of choice would be radioiodine therapy after an uptake scan.

In an unselected series of multinodular goitre, the prevalence of clinically important thyroid cancer was less than 1%.[11] Fine needle aspiration biopsy, however, should be offered to all patients presenting with multinodular goitre, especially if cancer is suspected (for example, a recent increase in size or a dominant nodule).

Use of thyroxine in therapy

As with normal thyroid tissue, the growth of solitary thyroid nodules often depends on the concentration of thyroid stimulating hormone, which is why thyroxine is used to suppress the concentrations to help reduce nodule size. Two of four controlled trials" showed no benefit from thyroxine in reducing nodule size. In the third study, 45% of patients taking thyroxine showed decreases in nodule size compared with 26% of patients taking placebo, but the number of patients with more than 50% reduction in nodule size did not differ significantly between the two groups. In the fourth study, 39% of patients taking thyroxine had a 50% reduction in nodule size compared with no patients taking placebo, but the initial nodule volume had to be [is less than or equal to] 10 ml. Therefore, in the absence of data for long term efficacy of thyroxine and its potential for inducing subclinical hyperthyroidism, thyroxine therapy remains controversial.

The only randomised trial of the efficacy of thyroxine in multinodular goitre showed a mean reduction of 25% after 9 months of therapy with a return to baseline after discontinuation.[12] As with solitary nodules, small multinodular goitres seem more responsive to thyroxine therapy than larger ones, and there is no evidence that thyroxine influences the clinical course of the goitre. Thyroid function test results need to be checked before initiation of thyroxine therapy, especially in patients with large multinodular goitres because autonomously functioning nodules are common. If thyroid stimulating hormone concentrations were already suppressed there would be no basis for giving further thyroxine, which would only produce overt hyperthyroidism.

Subclinical thyroid disease

Subclinical thyroid disease is still a controversial topic. Subclinical hypothyroidism was reviewed recently,[13] but subclinical hyperthyroidism is equally challenging.

Subclinical hyperthyroidism

Subclinical hyperthyroidism is defined as low serum thyroid stimulating hormone concentrations (with an immunometric assay) with normal serum thyroxine and triiodothyronine concentrations. In clinical practice, patients with low thyroid stimulating hormone concentrations fall into one of three categories (box 2).

Causes and prevalence

Subclinical hyperthyroidism may be transient or persistent (box 3). In several large scale community studies the prevalence of subclinical hyperthyroidism ranged from 2% to 16%, reflecting a wide range of population characteristics. Prevalence is higher in women, in older age, and in the presence of nodular thyroid disease (20% with multinodular goitre).[15]

Clinical course

Overall, the likelihood of subclinical hyperthyroidism progressing to overt hyperthyroidism is small ([is less than or equal to] 4% per year in autonomous thyroid nodules). Persistent suppression of thyroid stimulating hormone (and progression to overt hyperthyroidism) are most common in those with undetectable thyroid stimulating hormone in a sensitive assay, whereas those with subnormal but not fully suppressed thyroid stimulating hormone concentrations often show a return of their biochemistry to normal.

Clinical effects

Cardiovascular system

During the 10 year follow up of the Framingham cohort, the incidence of atrial fibrillation was related to the extent of suppression of thyroid stimulating hormone (incidences of 8%, 12%, and 21% for thyroid stimulating hormone concentrations of normal, 0.1-0.4 [micro]U/ml, and less than 0.1 [micro]U/ml respectively).[16] Additionally, there are reports that subclinical hyperthyroidism might affect other variables of cardiac function--for example, increased left ventricular systolic function and mass, impaired diastolic function, reduced maximal exercise capacity, reduced ejection fraction during exercise. There is, however, no evidence of increased rates of hospital admission or mortality from ischaemic heart disease,[17] possibly because of beneficial effects on both total cholesterol and low density lipoprotein cholesterol concentrations.[18]

Bone mineral density

Bone mineral density in both exogenous and endogenous subclinical hyperthyroidism was investigated extensively. Two recent meta-analyses evaluated the effect of thyroxine on bone mineral density. The first included 13 studies (750 patients) on long term suppressive thyroxine therapy (5-15 years). Compared with healthy women bone mineral density loss of the distal forearm, femoral neck, and lumbar spine in the premenopausal women was 0.46%, 0.27%, and 0.17% per year respectively; all losses were non-significant. The corresponding values for postmenopausal women were all significant at 1.39%, 0.77%, and 0.92% per year.[19]

The second meta-analysis included all 41 published controlled cross sectional studies (about 1250 patients) of both replacement and suppressive doses of thyroxine.[20] As with the first meta-analysis, suppressive therapy was associated with significant bone loss in postmenopausal women but not premenopausal women. Conversely, replacement thyroxine therapy was associated with loss of bone mineral density of the hip and spine in premenopausal women but not postmenopausal women. Methodological limitations were, however, evident. For example, despite various exclusion criteria to improve the homogeneity of the data, the researchers acknowledged that the controls were usually not matched with the cases for many factors influencing bone mass--for example, body weight, age at menarche and menopause, intake of dietary calcium, smoking, alcohol intake, exercise. Only a large long term prospective placebo controlled trial of thyroxine therapy evaluating bone mineral density (and ideally fracture rate) would provide conclusive evidence.

Treatment

Subclinical hyperthyroidism should be differentiated from other causes of low thyroid stimulating hormone concentrations and must be confirmed to be persistent before any action is taken. In general, observation is the best policy, but therapy may be indicated if the condition develops into frank thyrotoxicosis. Therapy could be considered in elderly patients with atrial fibrillation if there were risk factors for cardiovascular or musculoskeletal disease or if there was a large goitre. If suppression of thyroid stimulating hormone was intentional, concomitant bisphosphonate therapy might be warranted.

Amiodarone and thyroid function

Amiodarone is a potent broad spectrum antiarrhythmic comprising 37% iodide. It has a strong affinity for intralysosomal phospholipids, inhibiting their degradation by phospholipases and leading to phospholipidosis and disturbances of lysosomal function. These inclusion bodies have been found in the lungs, liver, heart, skin, corneal epithelium, and peripheral nerves, which explains the toxic effects in many organs and the proportional relation between toxicity and duration of use and cumulative dosage.[21]

Sequential effects on thyroid homeostasis

The sequential effects of amiodarone on thyroid homeostasis are: (a) amiodarone releases pharmacological quantities of iodide--the standard maintenance dose of 200-600 mg/day releases 75-225 mg organic iodide (normal daily requirement 0.2-0.8 mg); (b) thyroid iodide uptake increases, peaking at 6 weeks. The chronic iodide excess transiently decreases thyroxine production (Wolff-Chaikoff effect), with a consequent increase in thyroid stimulating hormone concentrations; and (c) within 3 months the thyroid gland is free of this inhibitory effect, with normalisation of thyroxine production. In up to 50% of euthyroid patients on long term amiodarone, thyroid function tests may show minimal increases in thyroxine concentration, suppression of triiodothyronine, and sometimes suppression of thyroid stimulating hormone. These changes do not require further management apart from monitoring with thyroid function tests. Box 4 gives a brief review of thyroid dysfunction caused by amiodarone, with emphasis on the practical aspects and recent concepts.[22 23]

Graves' disease

Smoking

Smoking increases the risk both of Graves' disease and of Graves' ophthalmopathy (odds ratio 1.9 and 7.7 in smokers v non-smokers).[24] Orbital fibroblasts cultured under hypoxic conditions (as found with smoking) synthesise more glycosaminoglycans.[24] Typically in Graves' ophthalmopathy, excess glycosaminoglycan deposition with water retention results in muscle swelling. In normal individuals smoking is associated with antibodies to heat shock protein 72, a protein expressed on orbital fibroblasts and present in autoimmune reactions.[25] Smokers with Graves' disease have a lower concentration of soluble interleukin 1 receptor antagonist than do non-smokers, which suggests that the proinflammatory and fibrogenic effects of interleukin I are less inhibited.[26]

Patients with Graves' disease--especially with significant ophthalmopathy--should be advised to stop smoking. More studies on the effect of smoking, and smoking cessation, on the clinical course of Graves' disease are required.

Treatment with radioiodine

The temporal progression of Graves' hyperthyroidism and Graves' ophthalmopathy is independent. Graves' ophthalmopathy appears before Graves' hyperthyroidism in 20% of patients, simultaneously in 40%, and after Graves' hyperthyroidism in 40%. The implication is that almost half of patients with Graves' ophthalmopathy will have it in the aftermath of treatment with radioiodine, leading to the impression that progression of the disease is associated with radioiodine. This is further confounded by the fluctuating course of the disease. Not surprisingly, attempts to address the potential worsening of Graves' ophthalmopathy after controlling Graves' hyperthyroidism, especially with radioiodine, has generated much research and debate.[27]

A randomised controlled trial with 12 months' follow up showed that Graves' ophthalmopathy worsened in more patients treated with radioiodine (15%) than in those treated with methimazole (3%; P (0.001). Treatment with radioiodine and prednisolone prevented the development or progression of the ophthalmopathy (P(0.001).[28] This confirms the results of a previous randomised trial.[29]

On the basis of this finding it was suggested that, unlike antithyroid drugs, radioiodine carries a small but definite risk of development or worsening of ophthalmopathy. Routine steroid use after radioiodine therapy was considered inappropriate as many patients would be exposed to the side effects of high dose steroids to prevent eye changes in only a maximum 15% of patients.[30] It was suggested that the risk factors should be thoroughly assessed for consideration of steroid use. The most important factor seems to be pre-existing active ophthalmopathy. Other predictors include smoking or high serum triiodothyronine concentrations before treatment.

Routine full blood counts with antithyroid drugs

The estimated risk of agranulocytosis with antithyroid drugs (carbimazole and propylthiouracil) is 3 per 10 000 patients per year, mainly in the first 3 months of treatment.[31] Recent recommendations in Drugs and Therapeutics Bulletin were that full blood counts should be monitored fortnightly for the first 3 months of treatment,[32] but this was criticised by many endocrinologists because the incidence of agranulocytosis due to antithyroids is extremely rare in Britain. If the condition does occur, it develops rapidly so that even fortnightly monitoring of full blood counts may miss it. Also, there is no evidence that monitoring would benefit patients, despite the increased cost and complexity of patient management.

The journal countered this criticism with the results of a prospective Japanese study conducted over 12 years in which more than 15 000 patients were treated with methimazole (the active metabolite of carbimazole) or propylthiouracil. During the first 3 months, 55 patients (0.4%) developed agranulocytosis of whom 43 were identified on routine monitoring of full blood counts before the onset of symptoms. All the patients recovered on stopping the treatment--29 without symptoms of infection.[33] The journal also contrasted current recommendations of antithyroid drugs and of sulphasalazine. Both treatments are associated with agranulocytosis, mainly in the first 3 months of treatment. Unlike the recommended monitoring of full blood counts for sulphasalazine, however, the current datasheet for carbimazole only recommends that patients should be warned about the onset of sore throat, mouth ulcers, pyrexia, or other symptoms that might suggest the development of bone marrow suppression. If so, treatment should be stopped, medical advice sought, and a full blood count performed. In response to this debate, the Committee on Safety of Medicines has recently recommended that routine full blood counts are not indicated, but that the above mentioned precautions should be followed.[34]

It is noteworthy that the Japanese study had an unusually high threshold for stopping antithyroid treatment (granulocyte count of 1.5 x [10.sup.9]/l). Unlike agranulocytosis (defined as a granulocytic count 0.25 x [10.sup.9]/l), mild and transient granulocytopenia (1.5 x [10.sup.9]/l) occurs in up to 10% of patients treated with antithyroid drugs. Thyrotoxicosis itself is associated with some granulocytopenia. Therefore, the threshold of the Japanese group may have overestimated the problem. Again, the reported incidence was 0.4%, more than 10 times the reported incidence in Europe, arguing against extrapolating the results to Europe.

Summary points

Simplifying the treatment regimen for thyroid disease is the most important way of improving patient compliance

All solitary thyroid nodules should be examined by fine needle aspiration; the technique may also be helpful in multinodular goitres if carcinoma is suspected

Subclinical hyperthyroidism is defined as suppressed concentrations of thyroid stimulating hormone with normal serum thyroxine and triiodothyronine concentrations

In subclinical hyperthyroidism the incidence of atrial fibrillation increases as thyroid stimulating hormone concentrations decrease, and in postmenopausal women bone mineral density may also be slightly reduced

Smoking increases the risk of both Graves' disease and Graves' ophthalmopathy

Box 1--Causes of increased thyroid stimulating hormone concentrations with adequate thyroxine replacement dose

* Poor compliance

* Malabsorption

* Influence of pharmacological agents:

Reduced absorption

Sucralfate

Cholestyramine

Ferrous sulphate

Aluminium hydroxide

Reduced conversion of thyroxine to triiodothyronine

Amiodarone

Box 2--Differential diagnosis of low concentrations of thyroid stimulating hormone

Hyperthyroidism

Overt

Subclinical

Secondary hypothyroidism

Low thyroxine concentrations with low thyroid stimulating hormone concentrations Clinical and biochemical evidence of pituitary insufficiency

Euthyroidism

* Physiological

Pregnancy (near end of first trimester)--Human chorionic gonadotrophin concentration (with thyroid stimulating activity) reaches its peak, suppressing production of thyroid stimulating hormone for a few weeks. This is magnified in hyperemesis gravidarum, with higher concentrations of human chorionic gonadotrophin and hence higher concentrations of thyroxine and triiodothyronine[14]

Elderly patients--Due to reduced thyroxine clearance, with subsequent suppression of thyroid stimulating hormone

* Non-thyroidal illness

Commonest form includes low concentrations of thyroid stimulating hormone with low concentrations of triiodothyronine (occasionally low thyroxine concentrations) in patients with severe non-thyroidal illness. This might be due to a combination of central suppression of thyroid stimulating hormone (for example, somatostatin or other neurotransmitters) and other factors interfering with peripheral thyroid hormone metabolism and conversion of thyroxine to triiodothyronine (for example, cortisol)

Box 3--Causes of hyperthyroidism

Persistent hyperthyroidism

* Exogenous:

Iatrogenic excessive thyroxine replacement (poor compliance or wrong prescription) Intentional suppression (control of differentiated thyroid cancer or goitre) Surreptitious

* Endogenous:

Graves' disease Multinodular goitre Autonomously functioning nodule

Transient hyperthyroidism

De Quervain's thyroiditis Silent thyroiditis Postpartum thyroiditis Drug induced thyroiditis (for example, amiodarone, interferon alpha)

Box 4--Amiodarone induced thyroid dysfunction

Hyperthyroidism

* Incidence

1.7% in areas with high iodine intake

12% in areas with low iodine intake

Patients with high environmental iodine (for example, in United Kingdom, United States) develop amiodarone induced hypothyroidism more often than amiodarone induced hyperthyroidism--opposite occurs with low iodine (for example, in Italy)

* Pathogenesis

Type I--Iodine induced excessive thyroid hormone synthesis, especially in patients with underlying thyroid disease (for example, diffuse or nodular goitre)

Type II--Amiodarone induced destructive thyroiditis in previously normal thyroid gland with release of thyroid hormones into circulation

* Diagnosis

Increased triiodothyronine (or free triiodothyronine) and free thyroxine concentrations with suppressed thyroid stimulating hormone concentrations

Some authorities argue that suppressed thyroid stimulating hormone and increased free thyroxine concentrations can be compatible with euthyroid state. Hyperthyroidism best confirmed by increases in triiodothyronine or free triiodothyronine concentrations and tissue markers of hyperthyroidism (for example, sex hormone binding globulin)

Fine needle aspiration yielding cytological findings of thyroiditis differentiate type II from type I amiodarone induced hyperthyroidism. Few patients, however, will have goitre, limiting feasibility of technique Profoundly increased concentration of interleukin-6 in type II (marker of thyroid inflammation, increased also in subacute thyroiditis) seems promising as non-invasive tool but not yet available for routine clinical use

* Therapy

Withdraw amiodarone if possible (in view of long half life, withdrawal has no immediate obvious effect)

Type I amiodarone induced hyperthyroidism

Large dose antithyroid drugs is usual first line, although occasionally disappointing. Adding steroids improves outcome through inhibition of 5'-deiodinase activity, blocking conversion of thyroxine to triiodothyronine, in addition to possible direct effect on thyroid

If adding steroids fails, potassium perchlorate (under expert supervision) could be added--it blocks the thyroid iodide trapping mechanism Surgery could offer immediate and effective control (although patients are usually high risk)

Plasmapharesis to remove thyroid hormones offers temporary but expensive option

Type II amiodarone induced hyperthyroidism

High dose steroids

Hypothyroidism

* Incidence

13% in areas with high iodine intake

6.4% in areas with low iodine intake

* Pathogenesis

Failure of thyroid gland to escape from inhibitory Wolff-Chaikoff effect, being compromised by underlying abnormality in thyroid gland--for example, autoimmunity (being female and positive for antithyroid peroxidase antibodies poses sevenfold higher risk)

* Diagnosis

Increased thyroid stimulating hormone concentrations, with low thyroxine and triiodothyronine concentrations

Increased thyroid stimulating hormone concentration not diagnostic of amiodarone induced hypothyroidism in first 3 months of therapy (might be transient)

* Therapy

Continue amiodarone in addition to L-thyroxine therapy

Competing interests: None declared.

[1] Nordyke RA, Gilbert FI Jr. Management of primary hypothyroidism. Compr Ther 1990;16:28-32.

[2] Ain KB, Refetoff S, Fein HG, Weintraub BD. Pseudomalabsorption of levothyroxine. JAMA 1991;266:2118-220.

[3] Sarlis NJ, Brucker-Davis F, Doppman JL, Skarulis MC. MRI-demonstrable regression of pituitary mass in a case of primary hypothyroidism after a week of acute thyroid hormone therapy. J Clin Endocrinol Metab 1997;82:808-11.

[4] Wood LC. Support groups for patients with Graves' disease and other thyroid conditions. End Metab Clin North Am 1998;27:101-7.

[5] Schectman JM, Elinsky EG, Pawlson LG. Effect of education and feedback on thyroid function testing strategies of primary care clinicians. Arch Intern Med 1991;151:2163-6.

[6] Davies PDO, Lazarus JH, Wheeler MH. A computer-assisted thyroid follow-up system. World J Surg 1986; 10:681-6.

[7] Parte JV, Franklyn JA, Cross KW, Jones SR, Sheppard MC. Thyroxine prescription in the community: serum thyroid stimulating hormone level assays as an indicator of undertreatment or overtreatment. Br J Gen Pract 1993;43:107-9.

[8] Tan SY, Gill GV, Hales T Thyroid register audit: a district general hospital experience. Postgrad Med J 1992;68:483-4.

[9] Ain KB, Pucino F, Csako G, Wesley RA, Drass JA, Clark C, et al. Effects of restricting levothyroxine dosage strength availability. Pharmacotherapy 1996;16:1103-10.

[10] Nicholas WC, Fischer RG, Stevenson RA, Bass JD. Single daily dose of methimazole compared to every 8 hours propylthiouracil in the treatment of hyperthyroidism. South Med J 1995;88:973-6.

[11] Hermus Ad R, Huysmans DA. Treatment of benign nodular thyroid disease. N Engl J Med 1998;338:1438-47.

[12] Berghout A, Wiersinga WM, Drexhage HA, Smits NJ, Touber JL. Comparison of placebo with L-thyroxine alone or with carbimazole for treatment of sporadic non-toxic goitre. Lancet 1990;336:193-7.

[13] Weetman AP. Hypothyroidism: screening and subclinical disease. BMJ 1997;314:1175-8.

[14] Goodwin TM, Montoro M, Mestman JH, Pekary AE, Hershrnan JM. The role of chorionic gonadotropin in transient hyperthyroidism of hyperemesis gravidarum. J Clin Endocrinol Metab 1992;75:1333-8.

[15] Marqusee E, Haden ST, Utiger RD. Subclinical thyrotoxicosis. Endocrinol Metab Clin North Am 1998;27:37-49.

[16] Sawin CT, Geller A, Wolf PA, Belanger AJ, Baker E, Bacharach P, et al. Low serum thyrotropin: a risk factor for atrial fibrillation in older persons. N Engl J Med 1994;331:1249.

[17] Leese GP, Jung RT, Gutherie C, Waugh N, Browing MC. Morbidity in patients on L-thyroxine: a comparison of those with a normal TSH to those with a suppressed TSH. Clin Endocrinol (Oxj) 1992;37:500.

[18] Franklyn JA, Daykin J, Betteridge J, Hughes EA, Holder R, Jones SR, et al. Thyroxine replacement therapy and circulating lipid concentrations. Clin Endocrinol (Oxf) 1993;38:453.

[19] Faber J, Galloe M. Changes in bone mass during prolonged subclinical hyperthyroidism due to L-thyroxine treatment: a meta-analysis. Eur J Edocrinol 1994;130:350-6.

[20] Uzzan B, Campos J, Cucherat M, Nony P, Boissel JP, Perret GY. Effects on bone mass of long term treatment with thyroid hormones: a meta-analysis. J Clin Endocrinol Metab 1996;81:4278-89.

[21] Vrobel TR, Miller PF, Mostow ND, Rakita L. A general overview of amiodarone toxicity: its prevention, detection and management. Prog Cardiovasc Dis 1989;31:393-426.

[22] Harjai KJ, Licata AA. Effects of amiodarone on thyroid function. Ann Intern Med 1997; 126:63-73.

[23] Wiersinga WM. Amiodarone and the thyroid. In: Weetman AP, Grossman A, eds. Pharmacotherapeutics of the thyroid gland. Berlin: Springer-Verlag, 1997;225-87.

[24] Metcalfe RA, Weetman AP. Stimulation of extraocular muscle fibroblasts by cytokines and hypoxia: possible role in thyroid-associated ophthalmopathy. Clin Endocrinol 1994;40:67-72.

[25] Prummel MF, van Pareren Y, Bakker O, Wiersinga WM. Anti-heat shock protein (hsp)72 antibodies are present in the patients with Graves' disease (GD) and in smoking control subjects. Clin Exp Immunol 1997;110:292-5.

[26] Hofbauer LC, Muhlberg T, Konig A, Heufelder G, Schworm HD, Heufelder AE. Soluble interleukin-1 receptor antagonist serum levels in smokers and non smokers with Graves' ophthalmopathy undergoing orbital radiotherapy. J Clin Endocrinol Metab 1997;82:2244-57.

[27] DeGroot LJ, Gorman CA, Pinchera A, Bartalena L, Marocci C, Weirsinga WM, et al. Therapeutic controversies: radiation and Graves' ophthalmopathy. J Clin Endocrinol Metab 1995;80:339-49.

[28] Bartalena L, Marocci C, Bogazzi F, Manetti L, Tanda ML, Dell'Unto E, et al. Relationship between therapy for hyperthyroidism and the course of Graves' ophthalmopathy. N Engl J Med 1998;338:73-8.

[29] Tallestedt L, Lundell G, Torring O, Wallin G, Ljunggren JG, Blomgren H, et al. Occurrence of ophthalmopathy after treatment of Graves' hyperthyroidism. N Engl J Med 1992;326:1733-8.

[30] Wiersinga WM. Preventing Graves' ophthalmopathy. N Engl J Med 1998;338:121-2.

[31] International agranulocytosis and aplastic anaemia study. Risk of agranulocytosis and aplastic anaemia in relation to the use of antithyroid drugs. BMJ 1988;297:262-5.

[32] Anon. Drug-induced agranulocytosis. Drug Ther Bull 1997;35:49-52.

[33] Tajiri J, Noguchi S, Murakami T, Murakami N. Antithyroid drug-induced agranulocytosis. The usefulness of routine white blood cell count monitoring. Arch Intern Med 1990;156:621-4.

[34] Anon. Reminder: agranulocytosis with antithyroid drugs. Curr Probl Pharmacovigilance 1999;25:3.

(Accepted 12 May 1999)

Department of Medicine, Prince Charles Hospital, Merthyr Tydfill, Cardiff CF47 9DT

F W F Hanna consultant endocrinologist

Department of Medicine, University Hospital of Wales, Cardiff CF4 4XN

J H Lazarus consultant physician

M F Scanlon professor

Correspondence to: F W F Hanna fahmy@WFhanna. swinternet.co.uk

BMJ 1999;319:894-9

COPYRIGHT 1999 British Medical Association
COPYRIGHT 2000 Gale Group