Division of Endocrinology, Stanford University School of Medicine and Santa Clara Valley Medical Center, San Jose, California 95128
Subclinical hypothyroidism (SH) is a common disorder with a prevalence ranging from 110% of the adult population in most community studies (1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14). The risk of developing SH increases with female gender, advanced age, and greater dietary iodine intake. The view that most subjects with SH should be treated with L-thyroxine has gained increasing credence (15, 16, 17). A limited number of placebo-controlled randomized trials involving a small number of patients with SH have been performed (18, 19, 20, 21, 22), and several of these studies show that L-thyroxine therapy may reduce symptoms of hypothyroidism (18, 22). Other studies in subjects with SH have shown that L-thyroxine lowers low-density lipoprotein (LDL) cholesterol, improves cardiac function, and diminishes neuropsychiatric symptoms (15, 16, 17). A recent cross-sectional observational study noted an association between SH and atherosclerotic disease (13). On initial assessment then, it appears that the position advocating L-thyroxine therapy for most subjects with SH enjoys strong experimental support.
However, careful scrutiny of the entire spectrum of primary data bearing on the question of whether or not SH should be treated with L-thyroxine yields a legitimate contrary view. There are as many placebo-controlled randomized trials that observed no reduction in symptoms of SH (20, 21) as there are that note benefits of treatment. Other reports have noted the doubtful clinical significance and frequent statistical nonsignificance of L-thyroxine therapy on changes in LDL cholesterol, myocardial performance, and neuropsychiatric parameters. No association between SH and ischemic heart disease was shown in the Whickham survey (23), the most extensive longitudinal study of thyroid disease ever conducted. Such conflicting findings stem from inconsistencies of the reports in the variable definition of SH, the wide degree of thyroid failure examined, as well as the heterogeneous age, gender, and ethnicity of the subjects tested.
There is considerable evidence suggesting that the subjects with SH who would benefit most from L-thyroxine therapy are those with TSH levels exceeding 10 mU/liter. Such individuals constitute the minority of those with SH in all large-scale epidemiological studies that have stratified TSH levels (1, 2, 4, 5, 6, 7, 8, 9, 12, 14). The majority of subjects with SH, however, have slight elevations of TSH ranging between 5 and 10 mU/liter, and they have minimal, often nonsignificant, metabolic abnormalities. They are either affected by mild incipient thyroid failure for which L-thyroxine therapy has not been shown to convey recognizable benefits, or they may simply represent "euthyroid outliers" in the 2.5% tail above the upper limit of the normal TSH reference range, in which case L-thyroxine treatment would be inappropriate. The available evidence also calls into question the need to treat men with SH, who are much less prevalent than women with SH (1, 4, 6, 7, 10, 11, 14), and who manifest almost no metabolic differences compared with men with normal TSH levels.
Unfortunately no large-scale multicenter randomized trial has been performed to address and settle these issues. As a result, we are left to struggle with and sort through the conflicting results from a variety of small studies. Given this lack of definitive data, clinicians must weigh the benefit to risk ratio of initiating L-thyroxine therapy in subjects with SH, realizing that lifelong treatment is not without inconvenience or potential morbidity. In the absence of definitive outcome studies, we will conclude that the decision not to treat the majority of SH subjects is as strongly or even more strongly supported by the existent data as is the decision to treat. Thus, the burden of proof rests squarely on the shoulders of clinicians who initiate lifelong L-thyroxine treatment for SH to demonstrate that the goals of such therapy are clear and compelling.
Definition of SH
The biochemical state characterized by an elevated serum TSH level with a concomitant normal free T4 (FT4) level has received a variety of designations, including mild thyroid failure, as well as compensated, early, latent, mild, minimally symptomatic, and preclinical hypothyroidism (15, 17). The most widely applied designation for this biochemical state is subclinical hypothyroidism, and this is the one that we will use, despite the fact that its meaning is somewhat ambiguous. Not all subjects with a normal FT4 and elevated TSH are "subclinical"; some have symptoms or signs of hypothyroidism. Moreover, not all of such subjects have early thyroid failure destined for eventual overt hypothyroidism. Adding to the ambiguity, some authors have extended the definition of SH to include subjects with normal FT4 and TSH, but exaggerated TSH responses to TRH.
Despite the confusing nomenclature, we will define SH as a condition
with a normal FT4 and elevated TSH. The state of
SH may be found in at least five distinct situations (Table 1): mild unrecognized thyroid failure,
undertreated overt hypothyroidism, overtreated overt hyperthyroidism,
transient disturbances of the thyroid axis, and euthyroid outliers
(this term will be applied to the 2.5% of individuals possessing TSH
values above the 97.5 percentile of the euthyroid distribution). A
cogent treatment plan for SH should rely on awareness of the underlying
pathogenesis, because the five different clinical states clearly
require separate therapeutic strategies. There is no dispute about the
need for adjustment of medication or L-thyroxine doses as
well as closer monitoring of thyroid function tests in patients with
over- or undertreated overt thyroid dysfunction. Few would advocate
L-thyroxine treatment of subjects with transient
dysfunction of the thyroid axis. Euthyroid outliers, if distinguishable
from other subjects, clearly do not merit treatment. The controversy
surrounding treatment of SH then involves only those with incipient
thyroid failure.
|
Epidemiology of SH
In cross-sectional community studies involving at least 1000
subjects, the overall prevalence of an elevated TSH alone, or with a
concomitant normal FT4, ranges from 0.25.7% in
men and 1.213.6% in women (Table 2).
Surveys that stratified TSH levels indicate a predominance of stage b
subjects (TSH <10 mU/liter), who account for 5585% of all cases of
SH. Studies that have reported thyroid autoantibody tests on subjects
with elevated TSH demonstrate seropositivity rates from 2078%. The
widely variable scope of the surveys in Table 2
is evidenced by some
determining prevalence of elevated TSH without testing
FT4, and others lumping newly diagnosed SH
together with patients possessing iatrogenic SH. This heterogeneous
composition of subjects, along with variations in age groups, ethnic
extraction, methods of subjects selection, and degrees of iodine
repletion, all contribute to the broad spread of prevalence values.
|
The manner in which a "normal" TSH reference range is
constructed relates directly to the definition and epidemiology of SH.
Numerous studies note that the distribution of serum TSH values across
a healthy, ambulatory adult population without history of thyroid
disease or antibodies is continuous yet non-Gaussian on a linear scale,
with a marked skew toward the left (Fig. 1). Logarithmic transformation of the TSH
values results in a normal or near-normal Gaussian distribution
(25, 26). Consequently, the generally accepted method for
constructing a reference range by either nonparametrically bracketing
the central 95% of the linear TSH distribution, or by determining the
mean ± 2 SD of the logarithmic TSH distribution will
result in similar values for the upper and lower limits of normal. Such
a methodology will classify (whether justifiably or not) 2.5% of
euthyroid subjects as fulfilling biochemical criteria for SH due to TSH
values occurring in the upper tail of the distribution
(27). However, in several general population surveys, the
percentage of subjects with an elevated TSH is well below the expected
2.5% for men (7, 10, 11, 14) and women (3, 11, 14) (Table 2
). This suggests that the normal range for some TSH
assays (established in small reference samples) may not be applicable
to large general populations, especially those of men. In other words,
the methodology used to establish the upper limit of normal for a given
TSH assay will profoundly influence the prevalence of SH as well as the
percentage of euthyroid outliers.
|
Some investigators have used dynamic TRH testing in attempts to
distinguish early thyroid failure from normal thyroid function.
Although variable results have been reported, several studies note that
a substantial 2666% of subjects with TSH values from 510 mU/liter
manifest a normal, nonexaggerated TSH response to TRH
(30, 31, 32, 33). In a comprehensive systematic study of TRH
stimulation tests involving 1061 individuals with a broad range of
baseline TSH values, Spencer et al. (31) showed
that 43% of subjects with TSH values of 4.510 mU/liter manifested
normal responses to TRH. They also noted that
26% of the subjects
with TSH levels in the upper normal range (2.04.5 mU/liter) exhibited
exaggerated responses to TRH (31). Thus, it is not at all
clear how accurately the TRH stimulation test may differentiate
euthyroid subjects from those with mild thyroid failure when the TSH
value lies in the range of 210 mU/liter. Another study showed that
46% of 30 subjects with mild SH and an exaggerated TSH response to TRH
did not progress to overt hypothyroidism after follow-up to 15 yr,
suggesting that abnormal TRH-stimulation does not always predict
eventual thyroid gland failure (34). Repeat testing of TSH
levels in SH subjects demonstrated remission to normal in
15% of
individuals entering two different intervention trials (18, 19) and showed occurrences of intermittent normal TSH values in
42% of placebo-treated subjects in another study (20).
Reports from iodine-rich regions indicate that some subjects with mild
TSH elevations achieve remission after institution of a low-iodine diet
(7, 8). Thus, subjects designated as having SH with mildly
elevated TSH levels may show a negative response to TRH, a positive
response to TRH with no progression to overt hypothyroidism, or
complete remission of TSH to normal. These and other lines of evidence
suggest that a substantial proportion of subjects with mild SH may
consist of euthyroid outliers and that TRH stimulation may not
definitively distinguish such subjects from those with mild thyroid
failure.
The epidemiological data in Table 2 indicates that a preponderance of
SH subjects (72% by meta-analysis) in community studies are affected
by mild TSH elevations (<10 mU/liter), or stage b SH. Modestly
elevated TSH levels may represent the long right tail of the euthyroid
distribution, whereas dramatically elevated TSH values would more
likely portend thyroid gland failure. Do thyroid autoantibodies serve
as useful discriminants in this setting? The meta-analysis of the
studies in Table 2
determines an overall community prevalence of 6.3%
for elevated TSH or SH and that 49% of such individuals are negative
for thyroid antibodies. Large-scale community studies support the use
of high thyroid autoantibody titers as predictors for thyroid failure,
and the view that subjects with seronegative SH are at lower risk for
eventual overt hypothyroidism (6, 9). Thus, if 2.5% of
the 6.3% prevalence figure (or 40% of SH cases) were due to euthyroid
outliers, this proportion resembles the 49% seronegativity rate in SH
subjects!
Of note, epidemiological studies report that approximately 1750% of stage b SH subjects are thyroid antibody seropositive (1, 6, 7, 14). This is intermediate to reported rates of 10% in euthyroid subjects and to rates of 80% in stage c patients (1, 6). Expressed alternatively, 8095% of all seronegative individuals with elevated TSH are stage b (1, 6, 7, 14). These calculations further support the notion that a number of subjects with SH, especially those seronegative in stage b, may represent euthyroid outliers who would not progress to thyroid failure and would not need L-thyroxine.
Do symptoms improve after treatment of SH?
Five placebo-controlled blinded studies have assessed the ability
of L-thyroxine therapy to improve psychometric,
quality-of-life, or hypothyroid symptoms associated with SH (Table 3) (18, 19, 20, 21, 22). Of these five
trials, two showed statistically significant improvement (18, 22), but the average TSH level exceeded 11 mU/liter in both,
indicating large numbers of stage c patients. One study indicated
marginal, yet convincing, benefit, with a response rate difference of
24% between placebo and T4-treated subjects;
however, no baseline characteristics predicted who would benefit from
L-thyroxine (19). The remaining two studies
showed no clear benefit of L-thyroxine (20, 21); one noted some improvement of psychometric memory scores
with L-thyroxine compared with placebo (20),
but the other, which was the only study to enroll solely stage b SH
subjects noted surprising results. Significantly fewer hypothyroid
symptoms were found in both placebo and
T4-treated groups, but significantly improved
quality-of-life measures were observed only in the placebo-treated
group (21). Furthermore, stage b SH subjects were unable
to predict whether they were taking placebo or L-thyroxine
(21). Taken in concert, the conflicting results from these
studies suggest that 1) stage b SH subjects do not benefit more from
L-thyroxine than placebo, 2) only one fourth of patients
with more severe SH may benefit from L-thyroxine, and 3)
indiscriminate L-thyroxine treatment of SH would only
infrequently reverse symptoms associated with potential thyroid
deficiency, given the large proportion of community SH subjects who are
classified as stage b. Moreover, the possibility exists that
L-thyroxine may exert effects beyond that of placebo to
improve the well-being of euthyroid outliers in these trials; consider
the study that demonstrated overtly hypothyroid patients subjectively
favoring the use of L-thyroxine doses that caused
suppression of, rather than normalization of, their TSH levels
(35).
|
Although the view that overt hypothyroidism causes secondary hyperlipidemia and promotes atherosclerosis has been generally accepted (36), studies examining the relationships between hyperlipidemia, atherosclerosis, and SH have yielded less convincing results. Some cross-sectional observational surveys note a higher prevalence of SH in hyperlipidemic patients, whereas others show that SH subjects manifest moderately (mostly up to 10%) higher average total cholesterol (TC) than controls (12, 14, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48). Such evidence is far from decisive, because more than half of these noninterventional cross-sectional studies report either nonstatistically significant differences between euthyroid and SH subjects (14, 37, 40, 41, 42, 43, 47), or lower TC in SH subjects (13). Two striking themes emerge from these observational surveys: 1) stage b SH subjects manifest much milder degrees of dyslipidemia than those with stage c (12, 14, 39, 41, 43, 47, 48), with only two (12, 39) of these seven reports showing significantly elevated TC or LDL in stage b SH compared with euthyroid individuals; 2) all three studies that separately considered men with SH reported an elevation of 1% or less of TC in stage b SH compared with euthyroid controls (12, 37, 43). This result, in view of the much lower prevalence of SH in men, suggests that many of these subjects are euthyroid outliers.
Consider the findings of the large-scale epidemiological surveys with respect to thyroid dysfunction and lipid levels. The initial Whickham study observed that lipid levels were not associated with TSH elevations after age adjustment (37). The follow-up Whickham survey found no association between elevated serum TSH and increased risk of ischemic heart disease or dyslipidemia (23). A large cross-sectional survey of 3410 elderly subjects in Maryland noted a significantly elevated LDL cholesterol in subjects with SH, but only for those with TSH higher than 10 mU/liter (48); no increased frequency of diagnosed atherosclerotic diseases was found in this entire cohort of SH subjects. A report from Rotterdam noted that SH subjects actually had lower TC than controls, but manifested increased atherosclerotic vascular disease (13), suggesting that factors other than TC contribute to the increased risk of atherosclerosis, although LDL was not explicitly determined. From the conflicting data in the randomized community surveys of Whickham (23, 37), Maryland (48), and Rotterdam (13), one may question 1) whether SH is associated with dyslipidemia, 2) whether treatment of SH may reverse dyslipidemia, and 3) whether L-thyroxine therapy may prevent development of atherosclerosis if dyslipidemia is not the predominant mechanism causing presumed atherosclerotic disease in SH.
More than 20 intervention studies have assessed the effect of
L-thyroxine therapy in subjects with SH (Table 4) (18, 19, 20, 21, 22, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63) to
address the questions posed above. Unfortunately, the vast majority of
these trials were not controlled and many involved patients with
iatrogenic SH (overtreated overt hyperthyroidism or undertreated overt
hypothyroidism), which is not an appropriate comparison group for
community SH subjects, 90% of whom have no history of thyroid disease
or thyroid medication use (12).
|
What can we conclude from these small, variably defined, predominantly nonblinded, nonplacebo-controlled studies that yield conflicting results? The observation that individuals with stage b SH do not respond significantly to L-thyroxine therapy suggests that the majority of SH subjects ought not to be treated on the grounds of reversing dyslipidemia. Moreover, because the majority of subjects (>80%) in these trials were women, the evidence to use L-thyroxine therapy for hyperlipidemia in men with SH is tenuous.
Two meta-analyses that separately reviewed studies assessing the efficacy of L-thyroxine therapy on lipoproteins in SH subjects noted a mean decrease in serum TC of 0.4 mmol/liter and 0.2 mmol/liter, respectively (64, 65), although the former analysis included two studies (39, 43) that treated patients with overt hypothyroidism. The latter meta-analysis was able to attain adequate power to achieve overall statistical significance, but the stratification of study criteria found that two subsets of intervention trials did not achieve significance in the lipid response to L-thyroxine therapy: 1) studies that merited higher scores of internal validity; and 2) studies that enrolled subjects with baseline TC less than 240 mg/dl and who did not have inadequately treated overt hypothyroidism (65). These observations suggest that 1) more rigorously controlled intervention studies are needed; 2) community SH subjects, who have an overall mean TC of 224 mg/dl (12), and the vast majority of whom do not have undertreated overt hypothyroidism, would not benefit from L-thyroxine; and 3) patients with undertreated overt hypothyroidism manifest more severe metabolic dysfunction than community SH subjects and require separate therapeutic strategies.
Should treatment of SH be considered with the intent to normalize Lp(a) and homocysteine, two cardiovascular risk factors that increase in overt hypothyroidism, and both of which potentially may be invoked to explain the increased atherosclerosis seen in SH subjects in the Rotterdam study (13) who had lower TC than controls? Once again, current evidence does not favor the use of L-thyroxine in SH to treat either of these metabolic abnormalities. One study actually showed nonsignificantly increased Lp(a) levels in SH subjects after L-thyroxine treatment (59), whereas the study that solely enrolled stage b SH subjects noted no change (21). The only study that has reported a significant decrement in Lp(a) after L-thyroxine treatment of SH involved patients with an average baseline TSH of 22 mU/liter (62). There is even conflict about the ability of L-thyroxine treatment to change Lp(a) levels in patients with overt hypothyroidism (66). With respect to normalizing homocysteine levels, the only two published studies apply to L-thyroxine treatment of overt hypothyroidism (not SH) and are dichotomous in conclusions (67, 68). Although both noted significant lowering of fasting homocysteine after restoration of euthyroidism, the larger study with a control group reported that L-thyroxine alone was unable to lower postmethionine homocysteine levels, concluding that the hyperhomocysteinemia in overt hypothyroidism could not be fully reversed by L-thyroxine therapy (68).
Treatment of cardiovascular, neuromuscular, and neuropsychiatric dysfunction in SH
Features of neuromuscular dysfunction attributable to SH include myocardial dyscontractility, sensory and motor neurological impairment, and skeletal muscle dysfunction (15, 17). However, the only cross-sectional study that stratified individuals by TSH levels noted significantly altered ankle reflex times and myoglobin levels only in those patients with TSH higher than 12 mU/liter (41). Some, certainly not all, studies have determined statistically significant alterations in subjects with SH involving such indices of questionable biological significance as the stapedial reflex (69) and systolic time interval (STI) (70). Only two placebo-controlled studies have reported on L-thyroxine intervention to improve neuromuscular dysfunction, and they were in conflict about the presence of significant beneficial changes in STI after L-thyroxine (18, 19). A few noncontrolled studies showed significant improvement in neuromuscular and cardiovascular parameters after L-thyroxine treatment of SH (51, 52, 71, 72). Three of these studies enrolled patients with mean baseline TSH levels higher than 15 mU/liter (51, 52, 71). Interestingly, another study that administered triiodothyronine to euthyroid controls and SH subjects noted improved STI in both groups (73); yet another study showed significantly improved left ventricular systolic and diastolic function in T4-treated SH subjects, with some posttreatment parameters exceeding those of euthyroid controls (74). These latter results raise again the possibility that L-thyroxine may beneficially affect even euthyroid controls and emphasize the absolute need for using placebo controls and blinding participants in any studies that assess the efficacy of L-thyroxine therapy for SH. Furthermore, several other studies (not mentioned here) have correlated the diagnosis of SH with various unique measures of physiological abnormalities, but the reversal of such abnormalities with L-thyroxine therapy have not yet been demonstrated.
An abundant psychiatric literature has attempted to establish a relationship between subclinical thyroid dysfunction and affective as well as psychotic disorders (75, 76). Many such reports are hindered by not controlling for the effects of age, gender, inpatient hospitalization, and use of lithium on the prevalence of elevated TSH levels (75). Additionally, the mere association between SH and psychiatric disorders should not lead to the conclusion that SH brings about these associated disorders (76). The three large epidemiological studies that examined cognitive and affective scores in SH subjects found few (47, 77) or no (78) significant differences overall, and none in those with TSH between 5 and 10 mU/liter (47, 77). Two placebo-controlled studies assessing the effects of L-thyroxine therapy on psychometric measures noted that less than 25% of subjects responded in one study (19), and that statistically significant improvement occurred in only one parameter in the other study (20). Two nonplacebo-controlled studies noted significant reversal of only a subset of neurobehavioral and cognitive baseline abnormalities in SH subjects receiving L-thyroxine (79, 80). Given the paucity of data regarding the potential benefit of L-thyroxine for SH subjects with psychiatric disturbances, treatment is empirical and must be weighed against the risks of lifelong therapy.
Treatment of SH to prevent progression
Proponents of the view that most SH subjects should be treated with L-thyroxine advocate early initiation of therapy to prevent the future morbidity associated with development of overt hypothyroidism. The Whickham survey estimated that the progression of SH to overt hypothyroidism occurs in 25% of SH subjects per year, an increased risk affecting those with thyroid autoantibodies (9). Although the progression rate to overt thyroid failure in SH is significantly higher than in euthyroid controls, it is nonetheless necessarily modest in scale, given that the prevalence of "preclinical" thyroid autoimmunity as well as SH are each more than 510 times the magnitude of overt hypothyroidism. In addition, the clinical course of autoimmune thyroiditis (the most common single cause of SH) is marked by shifting biochemical parameters, alternating serological positivity, and fluctuating L-thyroxine dosages needed to maintain euthyroidism.
One study on the undulating course of Hashimotos disease determined that 20% of patients diagnosed with overt hypothyroidism regressed to euthyroidism after termination of T4 therapy (81). Moreover, the treatment of SH subjects would require similar, if not more intensive, biochemical and clinical monitoring than would be needed for those subjects being expectantly followed; no study has even demonstrated that early L-thyroxine intervention improves the outcomes of subjects with SH. As will be discussed in the following section, the overzealous dosing of L-thyroxine therapy with insufficient biochemical monitoring puts a substantial number of patients at risk for iatrogenic disease associated with hyperthyroidism. Once again, the benefit to risk ratio of initiating lifelong L-thyroxine therapy in the typical asymptomatic individual with stage b SH, who has minimal evidence of end-organ dysfunction, must be seriously weighed against the real morbidity associated with suppression of TSH.
Danger of treatment: overreplacement L-thyroxine therapy in hypothyroidism
Abnormal TSH values are frequently found in patients treated with L-thyroxine. Between 10% and 33% of individuals on L-thyroxine therapy have TSH values less than normal (12, 82, 83, 84, 85, 86, 87), and approximately one third to one half of these TSH levels are less than 0.1 mU/liter (83, 87). Given that most patients on L-thyroxine therapy are older individuals, the prescription of thyroid hormone for poorly established indications compounds an increasing problem of polypharmacy and drug interactions in the elderly.
Does suppression of TSH by exogenous L-thyroxine lead to a decrease in bone mineral density in postmenopausal women, the subjects who are at greatest risk for both SH and osteoporosis? An extensive recent review of this subject including a number of cross-sectional and longitudinal studies was inconclusive, with 10 studies showing a significant negative effect of L-thyroxine on skeletal integrity and a nearly equal number of studies showing no effect (88). Data from the Framingham study have shown that elderly persons 60 yr of age or older with a low TSH level have a 3-fold higher risk of developing atrial fibrillation over a 10-yr period compared with those with a normal TSH level, although only 5.7% of the subjects were taking a thyroid hormone preparation at entry (89). Left ventricular hypertrophy may be another adverse consequence of a suppressed TSH (90). Three blinded placebo-controlled studies of SH also noted overt manifestations of L-thyroxine toxicity: anxiety and tachycardia in 2 of 20 subjects (19), angina and atrial fibrillation in 2 of 18 participants (20) (all four individuals needed to be removed from these trials), and significantly increased frequency of anxiety in the L-thyroxine intervention vs. the placebo group (21).
Of additional great concern is the monitoring and management by community physicians of patients on L-thyroxine therapy. In a broad community survey of 25,862 Colorado residents, Canaris et al. (12) noted that 92% of subjects had seen a physician in the preceding year, yet 40% of the patients on L-thyroxine therapy had an abnormal TSH level, including 22% with a low TSH. It is likely that the majority of such patients were being poorly managed; two studies noted that overreplacement of primary hypothyroidism was the most common cause of a suppressed TSH in patients on L-T4 who did not have pituitary disease (91, 92). In another report, only 44% of T4-treated patients with an elevated TSH had dosages increased by their physicians who were aware of the TSH findings; even more astounding was the fact that only 11% of the patients with low TSH levels on L-thyroxine replacement had their doses decreased (86)! In view of the studies documenting the unacceptably high prevalence of low TSH levels in T4-treated patients who are managed by community physicians, it would be irrational to advocate lifelong L-thyroxine therapy for SH subjects unless the benefits are compelling and clearly outweigh the risks.
Conclusions
After reviewing a considerable spectrum of primary data, we conclude that subjects with SH should seldomly be treated with L-thyroxine. The current univariate method of measuring thyroid function is inadequate for distinguishing euthyroid outliers from subjects with mild incipient thyroid failure. Epidemiological reports observe that subjects with TSH lower than 10 mU/liter comprise a clear majority of those with SH. Observational reports suggest that a number of these stage b subjects, especially men, may actually be euthyroid outliers. Cross-sectional and interventional studies demonstrate that stage b subjects have minimal metabolic and physiological abnormalities and would be unlikely to benefit from L-thyroxine therapy. The vast majority of such studies, unfortunately, are not appropriately designed to assess the benefit of L-thyroxine treatment in SH, because they include large numbers of patients with iatrogenic SH, large numbers of patients with TSH higher than 10 mU/liter, few numbers of men, and with very few exceptions do not have placebo arms. A number of studies do illustrate interesting physiological abnormalities associated with SH, but the reversibility of such dysfunction with L-thyroxine therapy remains to be proven. The benefit to risk ratio of treating subjects with SH also remains to be determined, given the lack of outcome data, and the considerable risk of TSH suppression in patients on L-thyroxine replacement. If it is true that most subjects with SH should not be treated, then the recommendations advocating systematic screening of individuals for thyroid disease, especially in men, can also be justifiably challenged (93, 94).
With these caveats noted, we would generally limit use of L-thyroxine treatment to subjects with SH who possess a combination of the following compelling factors: TSH levels more than 10 mU/liter on repeated measurements, clear symptoms or signs (e.g. goiter) associated with thyroid failure, convincing family history of thyroid disease, state of pregnancy, strong habit of tobacco use (95), or severe hyperlipidemia not previously diagnosed. We do recognize that there exist other clinical scenarios (15), much less common than the ones noted above, and not within the scope of this discussion, in which the treatment of SH with L-thyroxine may be prudent; we leave open these possibilities in order that clinicians should consider treatment of SH on a case-by-case basis. In general, we would advocate obtaining FT4 and confirmatory TSH levels in any subject who is newly diagnosed with SH, and would recommend monitoring individuals expectantly if none of the compelling reasons for treatment are applicable.
Our position of seldomly treating SH is supported by a United Kingdom consensus panel, who advocates routine treatment of SH when the TSH exceeds 10 mU/liter (96), which is the only consistent predictor of response to L-thyroxine therapy. We propose that future studies be conducted to separately consider male subjects and to better assess the individuals who comprise the majority of cases with SH (i.e. those found on population screening to have TSH <10 mU/liter and without history of overt thyroid dysfunction). We hope that large-scale controlled intervention and outcome studies will be done to assess the potential benefit of L-thyroxine treatment in such SH subjects. We expect also that future recommendations regarding screening for thyroid disease in the community will be influenced by the results of such studies.
Acknowledgments
Received October 16, 2000. Accepted February 9, 2001.
Address all correspondence and requests for reprints to: Lawrence M. Crapo, M.D., Ph.D., Division of Endocrinology, Santa Clara Valley Medical Center, 751 South Bascom Avenue, San Jose, California 95128.
Footnotes
Abbreviations: FT4, Free T4; HDL, high-density lipoprotein; LDL, low-density lipoprotein; SH, subclinical hypothyroidism; STI, systolic time interval; TC, total cholesterol.
References