A systematic review of drug therapy for Graves hyperthyroidism

Transcription

1 European Journal of Endocrinology (2005) ISSN REVIEW A systematic review of drug therapy for Graves hyperthyroidism Prakash Abraham, Alison Avenell, Christine M Park, Wendy A Watson and John S evan Department of Endocrinology, Health Services Research Unit, Aberdeen Royal Infirmary, Foresterhill, Aberdeen A25 2ZN, UK (Correspondence should be addressed to P Abraham; P.Abraham@arh.grampian.scot.nhs.uk) Abstract We assessed the effects of dose, regimen and duration of anti-thyroid drug therapy for Graves thyrotoxicosis on recurrence of hyperthyroidism, course of ophthalmopathy, adverse effects, health-related quality of life and economic outcomes. We undertook a systematic review and meta-analyses of randomised controlled trials (RCTs). We identified RCTs regardless of language or publication status by searching six databases, and trial registries. Dual, blinded data abstraction and quality assessment were undertaken. Trials included provided therapy for at least 6 months with follow-up at least 1 year after drug cessation. Fixed or random effects meta-analyses were used to combine study data. Twelve trials compared a lock-replace regimen (requiring a higher dose of anti-thyroid drug treatment) with a Titration regimen. Overall, there was no significant difference between the regimens for relapse of hyperthyroidism (relative risk (RR) ¼ 0.93, 95% confidence interval (CI) 0.84 to 1.03). Participants were more likely to withdraw due to adverse events with a lock-replace regimen (RR ¼ 1.89, 95% CI 1.25 to 2.85). Prescribing replacement thyroxine, either with the anti-thyroid drug treatment, or after this was completed, had no significant effect on relapse. Limited evidence suggested months of anti-thyroid drug treatment should be used. The titration regimen appeared as effective as the lock-replace regimen, and was associated with fewer adverse effects. However, relapse rates over 50% and high participant drop-out rates in trials mean that the results should be interpreted with caution, and may suggest that other strategies for the management of Graves disease, such as radioiodine, should be considered more frequently as first-line therapy. There were no data on the course of ophthalmopathy, health-related quality of life and economic outcomes. European Journal of Endocrinology Introduction Hyperthyroidism is common, affecting approximately 2% of women and 0.2% of men (1). The most common cause of hyperthyroidism is Graves disease (1). Methimazole, carbimazole and propylthiouracil are the main drug treatments, blocking thyroid hormone synthesis. They may also help control thyrotoxicosis by immune suppression. Propylthiouracil additionally inhibits the peripheral conversion of thyroxine (T4) to triiodothyronine. Methimazole is the active metabolite of carbimazole, and since the conversion of carbimazole to methimazole is virtually complete, equivalent doses are thought to be comparable. Anti-thyroid drug therapy can be given either by the lock-replace regimen (where a higher dose of antithyroid drug is used with a replacement dose of thyroid hormone) or by the Titration regimen (where the antithyroid drug dose is reduced by titrating treatment against thyroid hormone concentrations). The preferred regimen and duration of therapy remain unresolved with varying duration from 6 to 24 months with either the lock-replace or the Titration regimen. We undertook a substantial update to a Cochrane systematic review (2) assessing the effects of anti-thyroid drug regimen and duration in the treatment of Graves hyperthyroidism. Methods PA was involved in protocol development, searching for trials, quality assessment of trials, data abstraction and data analysis. A A was involved in protocol development, quality assessment of trials, data abstraction and data analysis. W A W was involved in searching for trials, quality assessment of trials and data abstraction. C M P undertook quality assessment of trials and data abstraction. J S provided clinical input, and resolution of differences of opinion. No Ethics approval was required. q 2005 Society of the European Journal of Endocrinology DOI: /eje Online version via

2 490 P Abraham and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Search strategy for identification of studies We identified relevant studies regardless of language or publication status by searching The Cochrane Central Register of Controlled Trials (CENTRAL) (Issue 1, 2004), MEDLINE (1966 to July 2004), EMASE (1980 to July 2004), IOSIS (1985 to July 2004), CINAHL (1982 to July 2004), HEALTHSTAR (1975 to June 2002) and trial registries. We contacted authors of published trials and thyroid researchers and checked the references of retrieved studies and reviews for additional trials. Three trialists provided additional data. Selection We included all published and unpublished, randomised and quasi-randomised controlled trials (RCTs) of patients of any age receiving anti-thyroid drug treatment for Graves hyperthyroidism, where Graves hyperthyroidism had been adequately defined. We pre-specified a minimum duration for drug treatment of 6 months and a minimum duration of follow-up of 1 year from completion of drug therapy to assess the pre-specified outcomes. We sought trials of carbimazole, propylthiouracil, methimazole, lithium or perchlorate. We pre-specified the comparisons of lock-replace vs Titration regimen; short-term (6 months) vs long-term (over 6 months) regimens; high-dose drug therapy (equivalent to 40 mg carbimazole or more) vs low-dose (equivalent to 30 mg carbimazole or less); and continued thyroid hormone replacement with or without continued anti-thyroid medication. The main outcome measures were recurrence of hyperthyroidism, incidence of hypothyroidism and mortality. Additional outcome measures sought were the course of ophthalmopathy (need for corticosteroids, radiotherapy, visual compromise); adverse effects (agranulocytosis, drug rash, hepatitis, vasculitis); symptoms of hyperthyroidism (anxiety, tachycardia, heat intolerance, diarrhoea, oligomenorrhoea); thyroid antibody status; weight change; frequency of outpatient visits and thyroid function tests; health-related quality of life; economic outcomes; compliance; and necessity for surgery or radioiodine. Any differences were resolved by discussion between the reviewers. Quantitative data synthesis Where appropriate, the results of comparable groups of trials were combined for relative risks (RRs) using fixedeffects models, and results are presented with 95% confidence intervals (CIs). Heterogeneity between comparable trials was assessed by the I 2 statistic (3). Random effects models were used where I 2 was 50% or more. Pre-specified subgroup and sensitivity analyses were undertaken. Results Trial flow Twenty-three RCTs were included in the review (Fig. 1). All studies included adults only. Overall 83% of participants were women, and the weighted mean age was 40 years. Validity assessment Quality assessment of RCTs included allocation concealment, whether intention-to-treat analysis was undertaken, comparability of groups at baseline, and blinding of outcome assessors. The summary risk of bias was based on the concealment of allocation. Data abstraction Two reviewers independently abstracted the data and assessed the methodological quality of the studies. Figure 1 Flow of studies through the meta-analysis.

3 EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Drug therapy for Graves hyperthyroidism 491 anti-thyroid therapy (Table 1, Fig. 2) In one study (4, 5) the longer-duration Titration regimen (18 months) had significantly fewer relapses than the 6 month group (RR ¼ 0.63, 95% CI 0.41 to 0.99). One study (6) found there was no significant difference between 12 and 6 months (RR ¼ 0.86, 95% CI 0.52 to 1.43) using the lock-replace regimen. Two other studies used longer durations of Titration regimen, comparing 12 months with 24 months (7) and 18 months with 42 months of therapy (8). Combining these showed no significant difference between the longer and shorter durations of treatment (RR ¼ 0.88, 95% CI 0.67 to 1.16). lock-replace vs Titration regimen (Table 2, Figs 3 and 4) The anti-thyroid drug used was carbimazole in eight studies (9 16). The dose ranged between 30 and 60 mg/day in the lock-replace arms of all these studies except for one study (11), where a dose of 100 mg/day was used. Methimazole was used in three studies (17 20) where doses of mg/day were used in the lock-replace arms of the trials. One study (21) used either propylthiouracil or methimazole but the doses were not reported. The duration of therapy was 6 months in two studies (11, 19), 18 months in four studies (9, 10, 15, 20) and 12 months in the remaining trials (duration of therapy in one study (21) is unknown). Including all trials, irrespectively of length of followup, there was no significant difference in the relapse rates between the lock-replace (322 of 636; 51%) and Titration (332 of 614; 54%) regimens (RR ¼ 0.93, 95% CI 0.84 to 1.03) when the losses to follow-up were not considered. However, there were large losses to follow-up, with 234 and 223 in the lock-replace and Titration groups respectively. Excluding one study (17, 18) with a loss to follow-up of 44% had no major effect on the overall results (RR ¼ 0.91, 95% CI 0.81 to 1.03). When analysis was carried out making the extreme assumption of relapsed hyperthyroidism in participants lost to follow-up there was a significant difference between the two groups, with relapse rates of 63% (549 of 870) in the lock-replace group and 68% (566 of 837) in the Titration group (RR ¼ 0.93, 95% CI 0.87 to 0.99). Therefore the true relapse rates with either of these anti-thyroid drug regimen would be between 51 and 68%. Data regarding side-effects and number of participants withdrawn from therapy due to side-effects were available in seven studies (11, 13, 14, 16 19). The number of participants reporting rashes was significantly higher in the lock-replace trial arms Table 1 Trials comparing long vs shorter duration of anti-thyroid therapy. Methodological rating for concealment of allocation Relapse rates (% (numbers)) Patients randomised/assessed at close of study follow-up therapy Trial Location Intervention France CZ 60 mg 1:18 months 2 years 1:57/46 1:37 (17/46) 2:6 months 2:57/48 2:58 (28/48) Allannic et al. (1990,1991) (4, 5) UK CZ 60, then 40 mg þt4 1:12 months 1 year 1:NS/51 1:35 (18/51) C 2:6 months 2:NS/59 2:41 (20/49) Weetman et al. (1994) (6) Spain CZ 30 mg 1:24 months 5 years 1:27/24 1:83 (20/24) 2:6 months 2:29/28 2:86 (24/28) Garcia-Mayor et al. (1992) (7) France CZ mg 1:42 months 2 years 1:82/62 1:29 (18/62) 2:6 months 2:92/72 2:36 (26/72) Maugendre et al. (1999) (8) Abbreviations: CZ: carbimazole; T4: levothyroxine; NS: not stated. Methodological rating for concealment of allocation: A, method did not allow disclosure of assignment;, small, but possible chance of disclosure of assignment, or states random but no description given; C, quasi-randomised (alternate allocation to groups).

4 492 P Abraham and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Figure 2 Long vs shorter duration of anti-thyroid therapy participants relapsing. (Allannic et al. (1990,1991) (4, 5), Weetman et al. (1994) (6), Garcia-Mayor et al. (1992) (7), Maugendre et al. (1999) (8)). (10%, 63 of 616) as compared with the Titration arms (5%, 31 of 622) (RR ¼ 2.04, 95% CI 1.36 to 3.06). The lock-replace regimens also had more patients with agranulocytosis compared with the titration group (nine vs three). There was one further report of agranulocytosis in one study (20) but the treatment group was not mentioned. The number of participants withdrawing due to side-effects was also significantly higher on the lock-replace regimen (RR ¼ 1.89, 95% CI 1.25 to 2.85). The study (11) using the highest dose of 100 mg carbimazole had 7 of 17 patients in the lock-replace regimen withdrawing due to side-effects including two cases of agranulocytosis and five with rashes, compared with one case of agranulocytosis from participants on the Titration regimen. Initial and continued low-dose anti-thyroid drug therapy followed by additional T4 compared with no additional T4 (Table 3, Fig. 5) Combining the data from three studies (22 24) adding in T4 led to fewer relapses (RR ¼ 0.65, 95% CI 0.14 to 3.00, random effects model). There was marked heterogeneity (I 2 ¼ 88%) with the result being largely influenced by one study (22) where the relapses in the T4 group were less than 2% compared with 35% in the placebo group. A sensitivity analysis with the exclusion of this study also yielded no significant difference in the relapse rates. Initial anti-thyroid drug therapy followed by T4 compared with no T4 (Table 3, Fig. 5) Three studies (25 29) gave anti-thyroid drugs for months and then randomised into groups receiving either T4 or no treatment. One study (25, 26) randomised the T4 group further at the end of the first year into groups either stopping or continuing T4 (only the results of the initial randomisation with results at the end of the first year were considered in the analysis). One study (13) had a factorial randomisation such that following a period of anti-thyroid drugs (lock-replace compared with Titration) one group continued T4 for 1 year and the second group had no therapy after anti-thyroid medications. On combining the results of these four studies for T4 there was no statistically significant difference in the relapse rates between the two groups after 12 months of follow-up (RR ¼ 1.09, 95% CI 0.86 to 1.39). Drug choice Among the studies that reported rashes, 7% (19 of 264) participants on carbimazole developed rashes, and 12% (82 of 714) of participants on methimazole.

5 EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Drug therapy for Graves hyperthyroidism 493 Table 2 Trials comparing lock-replace vs Titration regimens. Trial Location Intervention therapy (mean (S.D.)) follow-up (mean (S.D.)) Patients randomised/ assessed at close of study Relapses rates (% (numbers)) Methodological rating for concealment of allocation enker et al. (1998) (17) Edmonds & Tellez (1994) (16) Goni-Iriarte et al. (1995) (12) Grebe et al. (1998) (11) Jorde et al. (1995) (19) Leclere (1994) (10) Lucas et al. (1997) (9) McIver et al. (1996) (15) Nedrebo et al. (2002) (13) Rittmaster et al. (1998) (20) Tuncel et al. (2000) (21) Wilson et al. (1996) (14) European 1: MMI multicentre 40 mg þ T4 (6 countries) 2: MMI 10 mg UK 1: CZ 60 mg þ T4 2: CZ 60 mg Spain 1: CZ 30 mg þ T4 2: CZ 40 mg New Zealand 1: CZ 100 mg 2: CZ 25 mg Norway 1: MMI 60 mg þ T4 2: MMI 30 mg French 1: CZ 60 mg þ multicentre T4 or T3 2: CZ 60 mg Spain 1: CZ 60 mg þ T4 2: CZ 60 mg UK 1: CZ 40 mg þ T4 2: CZ 40 mg Norway 1: CZ multicentre 40 mg þ T4 2: CZ Canada 1: MMI 30 mg þ T4 2: MMI 30 mg Turkey 1: PTU or MMI þ T4 2: PTU or MMI UK 1: CZ 60 mg þ T3 2: CZ 45 mg 12 months 4.3 (1.3) years 1: 258/147 1: 251/ months 2 years 1: 49/34 2: 46/36 14 (5) months 3 years 1: 30/28 2: 46/35 6 months 24 months 1: 17/9 2: 20/16 6 months 24 months 1: 29/19 2: 27/22 18 months 3 years 1: 123/98 2: 123/98 18 months 4.9 (1.6) years 1: 30/30 2: 30/30 18 months 1 year 1: 59/13 2: 52/17 12 months months 1: 110/98 2: 108/91 18 months 27 months 1: 98/98 2: 51/51 NS 9 30 months 1: NS/34 2: NS/39 12 months 1 year 1: 33/28 2: 44/35 1: 54 (85/147) 2: 58 (84/144) 1: 50 (17/34) 2: 67 (24/36) 1: 61 (17/28) 2: 63 (22/35) 1: 66 (6/9) 2: 94 (15/16) 1: 58 (11/19) 2: 77 (17/22) 1: 40 (39/98) 2: 47 (46/98) 1: 67 (20/30) 2: 60 (18/30) 1: 61 (8/13) 2: 47 (8/17) 1: 48 (47/98) 2: 45 (41/91) 1: 58 (57/98) 2: 59 (30/51) 1: 15 (5/34) 2: 18 (7/39) 1: 36 (10/28) 2: 57 (20/35) A A A Abbreviations: CZ: carbimzole; MMI: methimazole; PTU: propylthiouracil; T4: levothyroxine; T3: liothyronine; NS: not stated. Methodological rating for concealment of allocation: A, method did not allow disclosure of assignment;, small, but possible chance of disclosure of assignment, or states random but no description given; C, quasi-randomised (alternate allocation to groups). Five cases of agranulocytosis occurred in participants on methimazole and four cases in participants on carbimazole (two participants were taking very high doses of 100 mg carbimazole/day). Other outcomes None of the trials reported the effects of the different treatment regimens on progress of ophthalmopathy, quality of life, use of healthcare resources, or undertook an economic evaluation. Discussion The optimal medical therapy for Graves hyperthyroidism remains a subject of debate. There are several choices to be made when considering the drug treatment of Graves hyperthyroidism: which drug, dose

6 494 P Abraham and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Figure 3 lock-replace vs Titration regimen participants relapsing. (enker et al. (1998) (17), Edmonds & Tellez (1994) (16), Grebe et al. (1998) (11), Jorde et al. (1995) (19), Leclere (1994) (10), Lucas et al. (1997) (9), McIver et al. (1996) (15), Nedrebo et al. (2002) (13), Rittmaster et al. (1998) (20), Tuncel et al. (2000) (21), Wilson et al. (1996) (14)). and duration of therapy, whether to add T4, and when to discontinue therapy. This review examines the evidence from the RCTs conducted in this field. The results need to be interpreted with some caution as several of the trials had large losses to follow-up. In the lock- Replace vs Titration regimens, there was a total loss to follow-up of 27%. The reasons for the large loss to follow-up are unclear and may include: patients feeling relatively well a few weeks after commencement of therapy and attainment of euthyroidism; relatively younger age group of patients likely to have work and other commitments such that once euthyroidism is attained, they fail to comply with therapy and have a variable lag period before a relapse occurs; other motives leading to poor compliance with therapy including maintenance of weight loss in young to middle aged women; and different health systems in different countries with varying follow-up mechanisms. With regard to the duration of therapy, there is some evidence that months of drug therapy is more effective than shorter durations. This is stronger with use of the Titration regimen (4, 5) than for the lock-replace regimen (6), although the latter study was a quasi-randomised study. Two of the studies (7, 8) using longer durations of therapy did not show any clear evidence of benefit in extending therapy beyond 18 months, although the CIs were wide. Longer-term follow-up data (2 5 years after completion of anti-thyroid drug therapy) showed no evidence to suggest that the lock-replace therapy reduces relapse rates when compared with the standard Titration regimen. None of the studies examined differences in the rates of ophthalmopathy progression or looked at any quality of life indicators. There was in fact a significantly higher rate of drug withdrawal due to side-effects (16 vs 9%), a significantly higher incidence of rashes (10 vs 5%) and more episodes of agranulocytosis (nine vs three) in the lock-replace group. The use of higher doses (carbimazole 100 mg daily) in one study (11) led to unacceptably high rates of side-effects, including a 12% (2 of 17) incidence of agranulocytosis, a potentially life-threatening complication. Seven of the 17 (46%) participants withdrew from the lock-replace group due to the side-effects (11). The increased side-effects of the higher drug doses required for the lock-replace regimen and the lack of evidence in the longer term about its effectiveness over the Titration regimen would prejudice choice in favour of the Titration regimen. Some of the reasons that the lock-replace regimen is preferred include the better maintenance of the euthyroid state and concerns regarding the effects of transient periods of hypothyroidism on quality of life and in Graves ophthalmopathy where hypothyroidism may be detrimental to the progression of the eye disease. Unfortunately none of the RCTs looked at this issue and there is

7 Table 3 Trials comparing additional T4 vs none. Trial Location Intervention Combined ATD and T4 Hashizume et al. (1991) (22) Pfeilschifter & Zeigler (1997) (24) Japan Germany MMI for 6 months then 1: MMI 10 mg þ T4 2: MMI 10 mg þ placebo CZ till euthyroid then 1: CZ 10 mg þ T4 to keep TSH, 0.03 mu/l 2: CZ 10 mg and T4 if TSH. 4 mu/l Raber et al. (2000) (23) Austria MMI for 9 months then 1: MMI 10 mg þ T3 2: No treatment Continued T4 alone Glinoer et al. (2001) (28); Glinoer et al. (2000) (29) Hoermann et al. (2002) (25); Quadbeck et al. (2003) (26) Nedrebo et al. (2002) (13) Mastorakos et al. (2003) (27) elgium multicentre Multicentre Germany, Hungary, Austria Norway multicentre Greece MMI mg (or PTU) with T4 as needed for euthyroidism for 15 months then 1: T4 100 mg 2: Placebo ATD for months, euthyroid off ATD for 1 month then 1: T4 to keep TSH, 0.3 2: No treatment CZ (or CZ 40 mg þ T4) for 12 months then 1: T4 alone 2: No treatment ATD for 18 months, euthyroid off ATD for 1 month then 1: T4 100 m-g 2: Placebo therapy 12 months (then T4 or placebo alone for 3 years) follow-up (mean) Patients randomised/assessed at close of study 3 years 1: 60/60 2: 49/49 12 months 12 months 1: NS/28 2: NS/22 6 months months 1: 31/31 2: 44/44 12 months 12 months 1: NS/42 2: NS/40 12 months 12 months 1: 114/NS 2: 111/NS 12 months months 1: 108/93 2: 110/96 12 months 12 months 1:33/NS 2: 37/NS Relapse rates (% (numbers)) 1: 2 (1/60) 2: 35 (17/49) 1: 61 (17/28) 2: 23 (5/22) 1: 45 (14/31) 2: 52 (23/44) 1: 29 (12/42) 2: 27 (11/40) 1: 16 (18/114) 2: 16 (18/111) 1: 47 (44/93) 2: 46 (44/96) 1: 42 (14/33) 2: 24 (9/37) Methodological rating for concealment of allocation Abbreviations: CZ: carbimazole; MMI: methimazole; PTU: propylthiouracil; T4: levothyroxine; T3: liothyronine; ATD: antithyroid drugs; TSH: thyrotropin-stimulating hormone; NS: not stated. Methodological rating for concealment of allocation: A, method did not allow disclosure of assignment;, small, but possible chance of disclosure of assignment, or states random but no description; C, quasi-randomised (alternate allocation to groups). A EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Drug therapy for Graves hyperthyroidism 495

8 496 P Abraham and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Figure 4 lock-replace vs Titration regimen participants with rashes and withdrawals due to side-effects. (enker et al. (1998) (17), Edmonds & Tellez (1994) (16), Grebe et al. (1998) (11), Jorde et al. (1995) (19), Leclere (1994) (10), Nedrebo et al. (2002) (13), Wilson et al. (1996) (14)). a need for well-designed adequately powered trials which look at the benefits of the lock-replace regimen on the quality of life and progression of ophthalmopathy. There is no clear evidence in favour of giving thyroid hormone supplementation following the initial treatment of Graves thyrotoxicosis with anti-thyroid medication. The discrepant result from one study (22) could be due to higher iodine intake in Japan and/or differences in immunological or genetic factors (30). The trials which have subsequently tried to reproduce this study have all used differing protocols and none has used the original protocol used by Hashizume et al. (22). The relapse rates obtained in our review were 51% for the lock-replace regimen and 54% for the Titration regimen. The relapse rates were 63 and 68% for the lock-replace and Titration regimens respectively when intention-to-treat analysis was carried out making the extreme assumption of relapsed hyperthyroidism in all drop-outs. It can be presumed that the true relapse rate for the anti-thyroid drug regimens is in the range of 51 68%. It could be argued that with such relapse rates, radioiodine therapy should be considered at an earlier stage in the UK, as appears to be the trend among American thyroidologists (31). It is possible that patients may also choose this modality of treatment, creating a stable situation earlier in the course of therapy and saving several years of regular outpatient clinic visits and possible upsets caused by relapses of hyperthyroidism. The possibility of long-term low-dose continuous anti-thyroid use has also been suggested in a recent trial with over 10 years of follow-up for 26 patients on lowdose methimazole (32). Evidence-based recommendations. The Titration regimen is just as effective as the lock-replace regimen and, with a significantly lower incidence of adverse effects, should be considered as the first-line regimen in most situations.. The evidence suggests the optimal duration of anti-thyroid drug therapy for the Titration regimen is months.. There is no benefit from continued T4 replacement after a course of anti-thyroid treatment.. Radioiodine treatment should be discussed at an early stage patients may choose this option when they are made aware of the likelihood of relapse and the need for prolonged outpatient clinic attendance with anti-thyroid drug treatment.

9 EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) 153 Drug therapy for Graves hyperthyroidism 497 Figure 5 Additional T4 vs none participants relapsing. (Hashizume et al. (1991) (22), Pfeilschifter & Zeigler (1997) (24), Raber et al. (2000) (23), Glinoer et al. (2001) (28), Hoermann et al. (2002) (25), Nedrebo et al. (2002) (13), Mastorakos et al. (2003) (27)).. Adequately powered trials which will look at the benefits of lock-replace and Titration regimens on the progression of ophthalmopathy and quality of life are needed. Acknowledgements We are grateful to Prof. Adrian Grant, Ms Sheila Wallace and Ms Cynthia Fraser for their advice; to Dr Amalia Mayo and Ms Aurelie Desbois for undertaking translations; and to Dr R Hoermann, Dr J Leclere and Dr G Nedebro for providing further information about their trials. We thank the Medical Research Council and the Grampian University Hospitals NHS Trust Endowments for funding. The Health Services Research Unit is funded by the Chief Scientist Office of the Scottish Executive Health Department. The study funders had no role in the study design; collection, analysis and interpretation of data; writing of the report; and in the decision to submit the paper for publication. The views expressed are those of the authors. All authors contributed to the writing and revision of this paper. References 1 Franklyn JA. The management of hyperthyroidism. New England Journal of Medicine Abraham P, Avenell A, Watson WA, Park CM & evan JS. Antithyroid drug regimen for treating Graves hyperthyroidism. The Cochrane Database of Systematic Reviews Higgins JPT, Thompson SG, Deeks JJ & Altman DG. Measuring inconsistency in meta-analyses. ritish Medical Journal Allannic H, Fauchet R, Orgiazzi J, Madec AM, Genetet, Lorcy Y, Le Guerrier AM, Delambre C & Derennes V. Antithyroid drugs and Graves disease: a prospective randomized evaluation of the efficacy of treatment duration. Journal of Clinical Endocrinology and Metabolism Allannic H, Lorcy Y, Leguerrier AM, Delambre C, Stetieh H, Madec AM & Orgiazzi J. Antithyrodiens de synthese et maladie de asedow ou le choix d une strategie therapeutique. Presse Medicale Weetman AP, Pickerill AP, Watson P, Chatterjee VK & Edwards OM. Treatment of Graves disease with the block-replace regimen of antithyroid drugs: the effect of treatment duration and immunogenetic susceptibility on relapse. Quarterly Journal of Medicine Garcia-Mayor RVPC, Luna-Cano R, Perez-Mendez LF, Galofre JC & Andrade A. Antithyroid drug and Graves hyperthyroidism. Significance of treatment duration and TRAb determination on lasting remission. Journal of Endocrinological Investigation

10 498 P Abraham and others EUROPEAN JOURNAL OF ENDOCRINOLOGY (2005) Maugendre D, Gatel A, Campion L, Massart C, Guilhem I, Lorcy Y, Lescouarch J, Herry JY & Allannic H. Antithyroid drugs and Graves disease prospective randomized assessment of long-term treatment. Clinical Endocrinology Lucas A, Salinas I, Rius F, Pizarro E, Granada ML, Foz M & Sanmarti A. Medical therapy of Graves disease: does thyroxine prevent recurrence of hyperthyroidism? Journal of Clinical Endocrinology and Metabolism Leclere J. Treatment of asedow disease with synthetic antithyroid drugs. Evaluation of the dose on the efficacy of the long term treatment [French]. Annales d Endocrinologie Grebe SK, Feek CM, Ford HC, Fagerstrom JN, Cordwell DP, Delahunt JW & Toomath RJ. A randomized trial of short-term treatment of Graves disease with high-dose carbimazole plus thyroxine versus low-dose carbimazole. Clinical Endocrinology Goni-Iriarte MJ, Forga Llenas L, Iriarte erioiz A, Anda Apinaniz E, Rodriguez Erdozain R & Menendez Torre E. Recurrence of Graves asedow disease: the influence of treatment schedule [Spanish]. Medicina Clinica Nedrebo G, Holm P, Uhlving S, Sorheim J, Skeie S, Eide GE, Husebye ES, Lien EA & Aanderud S. Graves disease: drug regimens and predictors of outcome. European Journal of Endocrinology Wilson R, uchanan L, Fraser WD, McKillop JH & Thomson JA. Do higher doses of carbimazole improve remission in Graves disease? Quarterly Journal of Medicine McIver, Rae P, eckett G, Wilkinson E, Gold A & Toft A. Lack of effect of thyroxine in patients with Graves hyperthyroidism who are treated with an antithyroid drug. New England Journal of Medicine Edmonds CJ & Tellez M. Treatment of Graves disease by carbimazole: high dose with thyroxine compared to titration dose. European Journal of Endocrinology enker G, Reinwein D, Kahaly G, Tegler L, Alexander WD, Fassbinder J & Hirche H. Is there a methimazole dose effect on remission rate in Graves disease? Results from a longterm prospective study. Clinical Endocrinology Reinwein D, enker G, Lazarus JH Alexander WD & and The European Multicentre Trial Group of the Treatment of Hyperthyroidism with Antithyroid Drugs, A prospective randomized trial of antithyroid drug dose in Graves disease therapy. Journal of Clinical Endocrinology and Metabolism Jorde R, Ytre-Arne K, Stromer J & Sundsfjord J. Short-term treatment of Graves disease with methimazole in high versus low doses. Journal of Internal Medicine Rittmaster RS, Abbott EC, Douglas R, Givner ML, Lehman L, Reddy S, Salisbury SR, Shlossberg AH, Tan MH & York SE. Effect of methimazole, with or without L-thyroxine, on remission rates in Graves disease. Journal of Clinical Endocrinology and Metabolism Tuncel E, Ertturk E & Imamoglu S. Effects of antithyroid drugs alone and plus L-thyroxine on the relapse of Graves disease. Journal of Endocrinological Investigation (7 Suppl) A Hashizume K, Ichikawa K, Sakurai A, Suzuki S, Takeda T, Kobayashi M, Miyamoto T, Arai M & Nagasawa T. Administration of thyroxine in treated Graves disease. Effects on the level of antibodies to thyroid-stimulating hormone receptors and on the risk of recurrence of hyperthyroidism. New England Journal of Medicine Raber W, Kmen E, Waldhausi W & Vierhapper H. Medical therapy of Graves disease: effect on remission rates of methimazole alone and in combination with triiodothyronine. European Journal of Endocrinology Pfeilschifter J & Zeigler R. Suppression of serum thyrotropin with thyroxine in patients with Graves disease: effects on recurrence of hyperthyroidism and thyroid volume. European Journal of Endocrinology Hoermann R, Quadbeck, Roggenbuck U, Szabolcs I, Pfeilschifter J, Meng W, Reschke K, Hackenberg K, Dettmann J, Prehn, Hirche H, Mann K & the asedow Study Group. Relapse of Graves disease after successful outcome of antithyroid drug therapy: results of a prospective randomized study on the use of levothyroxine. Thyroid Quadbeck, Hoermann R, Janssen OE & Mann K. Medikamentose behandliung der immunehyperthreose (typ morbus asedow) patientselektion, langzeitverlauf und rezidivprophylaxe. Internist Mastorakos G, Doufas AG, Mantzos E, Mantzos J & Koutras DA. T4 but not T3 administration is associated with increased recurrence of Graves disease after successful medical therapy. Journal of Endocrinological Investigation Glinoer D, de Nayer P, ex M elgian CS & Group on GD, Effects of L-thyroxine administration, TSH-receptor antibodies and smoking on the risk of recurrence in Graves hyperthyroidism treated with antithyroid drugs: a double-blind prospective randomized study. European Journal of Endocrinology Glinoer D. Epidemologie clinique de la maladie de asedow en elgique. Revue Medicale de ruxelles A296 A Toft A. Thyroxine suppression therapy in Graves disease. aillieres Clinical Endocrinology and Metabolism Solomon, Glinoer D, Lagasse R & Wartofsky L. Current trends in the management of Graves disease. Journal of Clinical Endocrinology and Metabolism Azizi F, Ataie L, Hedayati M, Mehrabi Y & Shiekholeslami F. Effect of long-term continuous methimazole treatment of thyrotoxicosis: comparison with radioiodine. European Journal of Endocrinology Received 12 May 2005 Accepted 6 July 2005