Introduction
Thyroid stimulating hormone (TSH)-suppressive therapy has been variously defined in different clinical contexts and studies. In general, exogenous thyroid hormone, usually L-thyroxine (L-T4), is used to induce subclinical thyrotoxicosis. The rationale for this treatment is that pharmacologic doses of L-T4 lead to decreased pituitary release of TSH, which, in turn, may inhibit the growth and proliferation of TSH-responsive thyroid cells. Before sensitive TSH assays were available, thyrotropin releasing hormone (TRH) suppression tests were used as an index of TSH suppression (1 ). TRH is no longer available in the United States, and serum TSH values are now used for the titration of TSH suppressive L-T4 doses. There is some disagreement about the optimal therapeutic degree of TSH suppression, but in general the goal is the maintenance of a serum TSH value below the normal range with a normal or minimally elevated serum free T4 value.

Effectiveness of TSH Suppression

For Treatment of Benign Thyroid Disease
TSH-suppressive L-T4 therapy has been commonly used since the 1960s for the treatment of benign nodular thyroid disease. The goal of such therapy is to reduce nodule size or prevent further nodule growth in order to delay any need for surgical treatment for compressive symptoms or cosmetic reasons. Although there are regional variations in treatment preferences, approximately half of endocrinologists recently surveyed in Europe and North America indicated that they would use TSH-suppressive therapy in the management of typical cases of benign nodular thyroid disease ( 2, 3, 4). However, numerous clinical studies of the efficacy of TSH suppressive therapy for nodular thyroid disease have had mixed results, and this practice has become increasingly controversial. Many of the published studies assessing the effectiveness of TSH suppressive therapy for decreasing nodule size or limiting benign nodule growth have been very short-term, had small sample sizes, have been non-randomized, have utilized imprecise outcome measures, or have lacked appropriate controls (5). Some studies have not documented appropriate TSH suppression. Difficulties in interpreting conflicting data may also stem in part from the fact that thyroid nodules have heterogeneous etiologies and that there may be substantial variability in the growth rate of individual thyroid nodules and their response to TSH suppression (6, 7). Finally, the fine needle aspiration biopsies used in most studies to rule out malignancy may themselves significantly alter nodule size (8).
The most long-term prospective randomized trial of TSH-suppressive therapy for nodular thyroid disease to date found that treated subjects had only a borderline decrease in thyroid size after 5 years of follow-up (p = 0.051), but concluded that TSH suppressive therapy prevented the thyroid growth and appearance of new nodules seen in the control group (9). In a recent trial, 27% of subjects on TSH suppressive therapy had a =50% decrease in dominant thyroid nodule size compared to 17% of controls (p = 0.04); these investigators also noted a decrease in the number of non-dominant nodules detectable by ultrasound in treated patients after 18 months (10).
Several recent meta-analyses have attempted to synthesize the results of clinical trials evaluating the effectiveness of TSH-suppressive therapy on benign solitary thyroid nodule growth. One meta-analysis of seven prospective trials found that TSH-suppressive therapy was associated with decreases in nodule size by ultrasound measurement in 17% of subjects (11). Another meta-analysis of five randomized trials found that subjects taking TSH suppressive therapy were 1.9 to 2.5 times more likely to achieve a 50% reduction in nodule size compared to controls (12). A third meta-analysis of six randomized clinical trials using ultrasonographic measurements found that the size of nodules decreased by more than 50% in subjects on TSH suppressive therapy, but these results did not reach statistical significance (13). A fourth meta-analysis of nine randomized trials similarly concluded that there was a nonsignificant trend toward decrease in nodule size by at least 50% in the L-T4 treated group (14).
The degree of TSH suppression employed has varied from study to study. A recent randomized crossover study compared the effects of high-level (serum TSH =0.01 mU/L) and low-level (TSH 0.4 – 0.6 mU/L) TSH suppression and concluded that both were equally effective at decreasing nodule size (15). The study also noted that thyroid nodules increased to their pre-treatment size in placebo-treated patients who had previously been treated with TSH-suppressive therapy. One important drawback to TSH suppression is that nodule growth typically recurs after discontinuation of L-T4 therapy (16).
There are several alternatives to TSH suppressive therapy for benign nodular thyroid disease.
Total or partial thyroidectomy is one option. In patients who undergo partial thyroidectomy for thyroid nodules, randomized trials have not shown that the use of TSH suppressive therapy reduces nodule recurrence rates (17, 18, 19), and use of TSH suppressive therapy post-operatively is not cost-effective (20). Since hypothyroidism usually occurs after surgery, replacement doses of L-T4 are given to maintain the serum TSH in the low-normal or mid-normal range. However, patients with radiation-associated benign thyroid nodules may be an exception, as post-operative nodule recurrence rates were lower in L-T4 treated patients than in controls in a nonrandomized study of 511 patients who had received local head and neck irradiation in childhood (21). Another alternative to TSH suppressive therapy in patients with benign nodular thyroid disease is watchful waiting; as long as patients are not bothered by compressive symptoms this does not entail any evident risks (22). Finally, radioactive iodine therapy may be employed; a recent study suggests that this may be more effective and have fewer side effects than TSH-suppressive therapy (23).

For Treatment of Thyroid Carcinoma
Although TSH-suppressive therapy is widely used following thyroidectomy in patients with differentiated thyroid carcinoma and is considered standard of care, its use has never been rigorously evaluated in clinical trials. Given the clear rationale for and relatively low risks of TSH-suppressive therapy, such a trial would probably not be considered ethical. Available information regarding the efficacy of TSH suppression in preventing thyroid carcinoma progression or recurrence comes from observational studies. A recent meta-analysis of 10 observational studies concluded that TSH-suppressive therapy for differentiated thyroid cancer patients resulted in improved clinical outcomes (24).
Different dosing regimens have been proposed for TSH suppression in thyroid cancer patients. Some authors have advocated the use of just enough L-T4 to suppress the serum TSH value to just below the normal reference range (25). Others feel that more aggressive TSH suppression is warranted, at least in high-risk patients. A 1996 study of differentiated thyroid carcinoma patients compared 15 subjects whose serum TSH values were consistently =1 mU/L following thyroidectomy to a group of 18 subjects whose serum TSH values were consistently <0.05 mU/L (26). There were no differences in age, sex, or initial tumor grade between the two groups. The patients with consistently suppressed serum TSH values had significantly longer disease-free survival times than the patients without TSH suppression. Another group evaluated the effects of TSH suppression on thyroid cancer in subjects from a U.S. thyroid cancer registry (27). They found no effect of the degree of TSH suppression on disease progression in low-risk patients, but a trend toward a protective effect of more aggressive TSH suppression in high-risk patients, defined as those with tumor stages III or IV. In light of this finding, it seems reasonable to adjust the degree of TSH suppression according to tumor grade as well as to other prognostic factors. For example, thyroglobulin antibody-negative patients who have undetectable stimulated (by rhTSH or L-T4 withdrawal) serum thyroglobulin levels following thyroidectomy and radioactive iodine ablation are at low risk for tumor recurrence and may require only slightly low to low serum TSH values rather than complete TSH suppression.

Adverse Effects of TSH Suppression

Cardiovascular Effects
An increased risk for atrial fibrillation has been described in patients with low serum TSH values (28). This finding was recently confirmed in a cohort of subjects not taking thyroid hormone in whom the prevalence of atrial fibrillation in patients with serum TSH values <0.4 mU/L (with normal free T3 and free T4 values) was 12.7%, compared with 2.3% in euthyroid subjects (29).
Some (30, 31, 32, 33), but not all (34), echocardiographic case control studies have noted impaired diastolic function characterized by delayed relaxation in patients on TSH-suppressive therapy. Some of these studies have reported increased left ventricular mass, particularly increased posterior wall and interventricular septum thickness (32, 33), but this finding has not been universal (30, 31). Cardiac changes in patients taking TSH-suppressive therapy appear to be associated with decreased exercise capacity. When Mercuro et al. (32) reduced subjects’ L-T4 dose to the minimal amount required to maintain the serum TSH concentration at 0.1 mU/L or lower, echocardiographic and ergometabolic parameters normalized. These studies are limited by small sample sizes, and some are further limited by the use of control subjects unmatched for body mass index and usual physical activity.

A recent study compared measurements of plasma coagulation factors in 14 thyroid cancer patients on TSH-suppressive therapy to samples obtained while the patients were hypothyroid for cancer treatment. The investigators concluded that TSH-suppressive therapy may be pro-thrombotic (35).
An important question is whether the relatively subtle cardiovascular changes seen in patients taking TSH-suppressive L-T4 doses have an effect on survival. A community-based British study analyzed thyroid status in a cohort of 1191 patients aged 60 or older and found increases in all-cause mortality at 2, 3, 4, and 5 years of follow-up for subjects with subclinical thyrotoxicosis (serum TSH values <0.5 mU/L) at baseline (36). This difference was mainly due to increases in cardiovascular mortality. No difference in mortality was seen at 10 years. Outcome ascertainment in this study was based on death certificate information, which may not have been complete. In addition, results of this study were not age-adjusted, and the group of subjects with low serum TSH values was slightly older than other groups at baseline, which may have accounted for some of the difference seen.

Skeletal Effects
Results of studies describing the effects of TSH suppression on bone have been inconsistent, in part because of differing methodologies and small sample sizes. One meta-analysis pooled data from 13 studies of bone mineral density in women on long-term TSH-suppressive therapy compared to euthyroid control subjects (37). Among premenopausal women, bone density did not differ between the L-T4 treated group and controls. However, bone density was significantly lower in treated postmenopausal women than in controls. These results were confirmed by another meta-analysis, which combined data from 41 studies examining the effects of suppressive L-T4 therapy on bone density (38). In this analysis, TSH-suppressive therapy was associated with significant bone loss at all skeletal sites in postmenopausal women, but not in premenopausal women. To date there is no strong evidence that TSH suppression causes decreases in bone density in men (39, 40).
In postmenopausal women, administration of estrogen may prevent L-T4 induced bone loss in women with suppressed serum TSH values (41). Treatment with intravenous pamidronate has been shown to increase bone density in patients receiving TSH-suppressive therapy (42); it is likely that oral bisphosphonates would also be effective although this has not been studied to date.

Thyrotoxic Symptoms
Depending on their degree of thyrotoxicosis, patients receiving TSH suppressive L-T4 doses may complain of symptoms such as anxiety, heat intolerance, tremors, sweaty skin, insomnia, forgetfulness, or mood disorders.
In one study, 24 young men were treated with 300 mcg L-T4 or placebo for 3-week periods in a double-blind crossover design (43). The L-T4 treated men required a greater effort to complete a visual search task, demonstrating effects of TSH-suppressive therapy on central information processing. However, the L-T4 doses in this study were higher than those typically used clinically for TSH suppression.

Conclusions
There are many controversies surrounding the use of TSH suppressive therapy. We believe that TSH-lowering doses of L-T4 are certainly warranted for the post-thyroidectomy treatment of differentiated thyroid cancer, and that evidence suggests that TSH suppression should be more aggressive in high-risk cancer patients than in patients with lower-risk tumors (based on tumor grade and other prognostic factors). Treatment of benign nodular thyroid disease is less straightforward, with the preponderance of evidence suggesting that TSH suppression is effective at decreasing the size or reducing the growth of at least a subset of benign thyroid nodules. However, this limited benefit must be weighed against the risks of long-term TSH-suppressive therapy, including the development of thyrotoxic symptoms, decreased bone density in postmenopausal women, and increased risk for atrial fibrillation. Patients with longstanding nodular goiter may develop functional thyroid autonomy; in those patients L-T4 therapy may cause iatrogenic overt thyrotoxicosis. Based on current evidence, we believe that TSH-suppressive therapy is not warranted for most patients with benign thyroid disease. If TSH-suppressive therapy is used for benign nodular thyroid disease, risks should be minimized by using the minimal dose of L-T4 required to decrease serum TSH values to the low but detectable range, and the TSH-suppressive therapy should be discontinued after 6 to 12 months if there is no clear therapeutic response.