CONTROVERSY IN CLINICAL ENDOCRINOLOGY The Thyrotropin Reference Range Should Remain Unchanged

Transcription

1 X/05/$15.00/0 The Journal of Clinical Endocrinology & Metabolism 90(9): Printed in U.S.A. Copyright 2005 by The Endocrine Society doi: /jc CONTROVERSY IN CLINICAL ENDOCRINOLOGY The Thyrotropin Reference Range Should Remain Unchanged Martin I. Surks, Gayotri Goswami, and Gilbert H. Daniels Division of Endocrinology and Metabolism, Department of Medicine (M.I.S., G.G.), and Department of Pathology (M.I.S.), Montefiore Medical Center and Albert Einstein College of Medicine, Bronx, New York 10467; and Thyroid Unit and Department of Medicine, Massachusetts General Hospital and Harvard Medical School (G.H.D.), Boston, Massachusetts Context: Recent recommendations to decrease the upper limit of the TSH reference range from 4.5 to 2.5 miu/liter, based on the high proportion of normal people whose serum TSH is less than 2.5 miu/ liter and the observation that those with TSH between 2.5 and 4.5 miu/liter [upper reference range (URR)] have increased risk of progression to overt hypothyroidism (Whickham, 20-yr data), have not been subjected to critical analysis. Study Subjects: The study subjects were from the Reference Group of NHANES III, 14,333 people more than 12 yr old, without known thyroid disease or antithyroid antibodies; 85% had TSH levels below 2.5 miu/liter, and 2.3% had subclinical hypothyroidism (SCH). An additional 9.7% had URR TSH, representing 20.6 million Americans, who would also be identified as SCH if the upper TSH limit were decreased. Many with URR TSH do not have thyroid disease. Intervention: The time of phlebotomy is important, because the TSH PRIMARY HYPOTHYROIDISM IS one of the most common endocrine disorders, affecting many millions around the world. Although experienced clinicians may suspect severe hypothyroidism, more often the symptoms and signs of this disease are subtle and neither sensitive nor specific for the disorder. Therefore, the diagnosis of hypothyroidism requires measurement of serum TSH in conjunction with the serum free T 4 concentration (1). The inverse logarithmic relationship between free T 4 and TSH means that small changes in the serum free T 4 concentration can, in turn, lead to much more dramatic changes in serum TSH. Hypothyroidism ranges from mild to severe. Overt primary hypothyroidism is characterized by decreased serum free T 4 and increased serum TSH levels. The term subclinical hypothyroidism (SCH), also called mild hypothyroidism, refers to a laboratory abnormality characterized by an elevated serum TSH with a normal free T 4 concentration (2 4). Patients with these findings may or may not be symptomatic. Other causes of elevated serum TSH should be excluded. Abbreviations: DF-RF, Disease-free without risk factors; SCH, subclinical hypothyroidism; TPO, thyroid peroxidase; URR, upper reference range. JCEM is published monthly by The Endocrine Society ( endo-society.org), the foremost professional society serving the endocrine community. level varies throughout the day, with early morning values greater than later ones, and is accentuated by sleep deprivation, strenuous exercise, or working during the night or evening shifts. Repeated measurements in the same individual vary considerably over months. Results: About half of those with URR TSH probably have thyroid disease, but most with thyroid disease, antithyroid peroxidase antibodies, have TSH below 2.5 miu/liter. Those with URR TSH with thyroid disease probably have minimal thyroid deficiency, without any reported adverse health consequences or benefit of treatments with levothyroxine. Conclusion: Because routine levothyroxine treatment is not recommended for SCH, it is certainly not warranted in individuals with URR TSH. For all patients with URR TSH, it is reasonable to determine serum TSH every 1 2 yr. (J Clin Endocrinol Metab 90: , 2005) There is a continuum between the euthyroid state and hypothyroidism, as there is between normal and elevated serum TSH concentrations. Therefore, the distinction between a normal and an elevated serum TSH level is somewhat arbitrary. However, it has been used to distinguish individuals with normal thyroid function, who may or may not have underlying thyroid pathology, from those with SCH. In current assays, the upper limit of the TSH reference range is approximately 4.5 miu/liter (5). In 2003, the American Association of Clinical Endocrinologists issued a statement encouraging doctors to consider treatment for patients who test out of the boundaries of a narrower margin based on a target TSH level of miu/liter (6). Respected authorities have echoed this recommendation and even suggested that the range be further contracted to an upper limit of normal of 2.5 miu/liter (7 11). However, decreasing the upper limits of normal for serum TSH may have enormous health and economic implications and should not be undertaken without extensive discussion and consideration. Comprehensive analyses, discussions, and debate preceded changes in criteria for diagnosing diabetes mellitus (12), hypertension (13), and hypercholesterolemia (14). In each of these instances the guiding principle for change was the firm knowledge that interventions employing these guidelines would have important health benefits that outweigh potential risks. Such data are 5489

2 5490 J Clin Endocrinol Metab, September 2005, 90(9): Surks et al. Thyrotropin Reference Range not available for individuals whose serum TSH is between 2.5 or 3.0 and 4.5 miu/liter. Moreover, many individuals with TSH concentrations in this range do not have thyroid disease. Therefore, we do not support a lowering of the upper limit of the normal range for serum TSH at this time and detail our reasons in the following discussion. Normal Range for Serum TSH The normal or reference range for any parameter should provide a framework for physicians to categorize patients as 1) normal, 2) those who may require closer observation or intervention to detect and then treat a specific adverse health outcome, and 3), those with disease who might benefit from treatment. For example, using glucose tolerance testing, individuals are characterized as normal, prediabetic, or diabetic (12); for blood pressure measurements, normal, prehypertensive, and hypertensive populations have been defined (13). Health interventions (usually lifestyle) are indicated in the intermediate groups. Defining a normal range for serum TSH implies that anyone outside of this range must be abnormal and, by inference, is a candidate for therapy. Several important questions must be considered when defining a normal range for serum TSH. 1) What is the distribution of TSH measurements in the population? 2) What is the reliability and reproducibility of a given TSH measurement within and between assays? 3) What is the stability of the measurement? Does it vary over the course of the day and from day to day? 4) Is a serum TSH concentration between 3.0 and 4.5 miu/liter highly sensitive and specific for thyroid disease? 5) What is the natural history of a person with a serum TSH between 3.0 and 4.5 miu/liter? Does it inevitably rise to levels seen in overt hypothyroidism or may it return to normal? 6) What are the risks and benefits of levothyroxine treatment in individuals with serum TSH between 3.0 and 4.5 miu/liter? 7) What are the consequences of narrowing the TSH normal range? TSH Distribution NHANES III (5) provides data on the distribution of TSH, thyroid hormones, and antithyroid antibodies in 17,353 people (total population) that represent the weighted racial distribution of 205,562,185 individuals at or over the age of 12 yr and designed to represent the United States population. Individuals who reported thyroid disease, goiter, or taking thyroid medication (820 people) were excluded from the total population, leaving 16,533 people classified as the diseasefree population. A third group of 13,334 people (the reference population) further excluded people who had antithyroid antibodies; were pregnant, were taking estrogens, androgens, or lithium; and were without laboratory evidence of overt hypothyroidism or hyperthyroidism. Free T 4 was not measured in this study. Because people who are pregnant, or treated with estrogen or androgen ( 1,000 in NHANES III) 1 do not have significant serum TSH elevations, their exclusion makes the reference population less than optimal for determination of TSH distribution in a population without thyroid 1 Unpublished analyses of NHANES III database (5) kindly provided by Joseph G. Hollowell. disease or risk factors (possibly excluding family history) for thyroid disease. The disease-free group represents most patients who are evaluated for hypothyroidism, because clinicians generally do not measure anti-thyroid peroxidase (anti-tpo) antibodies before measurement of TSH. Thus, the disease-free population is most appropriate for assessment of the range of TSH that is relevant to doctors in practice. We, therefore, present analysis of the distribution of TSH in this population. We also present data for a new subgroup from NHANES III without risk factors for thyroid dysfunction [disease-free without risk factors (DF-RF)]. For the latter group, we obtained unpublished population data for the disease-free population, excluding only the subjects with antithyroid antibodies, those taking lithium, or the few discovered biochemically to have overt hyperthyroidism or hypothyroidism (see Footnote 1). This new group represents a population without self-reported thyroid disease and no risk factors for thyroid disease, except, possibly, a family history of thyroid disease. Similar to the disease-free population, the non-log-transformed distribution of TSH values of this new group is significantly skewed toward higher TSH values (Fig. 1). When log-transformed, the histogram is closer to a Gaussian curve, but is still skewed toward higher TSH values (Fig. 2). The geometric mean is ( sem) miu/liter, the median is 1.40 miu/liter, and the 95% reference interval (2.5th to 97.5th percentiles) is miu/liter. For the disease-free population not adjusted for risk factors for thyroid disease, the geometric mean is ( se) miu/liter, the median is 1.49 miu/liter, and the 2.5th and 97.5th percentiles are 0.5 and 5.52 miu/liter, respectively (5). The upper limit is clearly higher in the disease-free population than in the DF-RF, probably due to the inclusion of some patients with thyroid autoantibodies. Based on the serum TSH distribution for the disease-free and disease-free population without risk factors, representing approximately 200 million individuals in the United FIG. 1. Data for a new subgroup from NHANES III without risk factors for thyroid dysfunction (DF-RF). Similar to the disease-free population, the non-log-transformed distribution of TSH values of this new group is significantly skewed toward higher TSH values.

3 Surks et al. Thyrotropin Reference Range J Clin Endocrinol Metab, September 2005, 90(9): miu/liter. We cannot find support for these statements in the NHANES III database. Unpublished data from NHANES III (see Footnote 1) indicate that only 81.5% of the disease-free group, 85.9% of the DF-RF group, and 85.9% of the reference group had TSH levels below 2.5 miu/liter (Table 1). FIG. 2. When log-transformed, the histogram is closer to a Gaussian curve, but is still skewed toward higher TSH values. States, an additional % of the U.S. population at or above age 12 yr ( million) would be diagnosed with hypothyroidism if the upper limit of normal for serum TSH was 3.0 miu/liter (Table 1). If the upper limit were further decreased to 2.5 miu/liter, an additional % of the population, approximately million individuals would be considered hypothyroid. Thus, a total of million additional individuals would be diagnosed if the upper limit of the TSH range were decreased to miu/ liter. By comparison, % of the population ( million individuals) has SCH when defined as a serum TSH level of 5 10 miu/liter. Lowering the upper limit to miu/liter in these populations would result in a % increase in individuals considered hypothyroid. Even in the reference population, which excludes thyroid antibody-positive subjects as well as those who are pregnant or taking estrogens or androgens, a decrease in the upper limit of the TSH range to miu/liter would result in an approximately 3-fold increase in the number of people newly designated as hypothyroid (12 million). These projections are supported by an analysis of TSH values in patients without a history of thyroid disease in a tertiary care practice (15). Decreasing the upper limit of the TSH reference range to 3.0 miu/liter resulted in more than a 4-fold increase in patients classified as hypothyroid. Some of those who suggest that the upper limit of the TSH range should be 2.5 miu/liter state that 95% of the subjects in the reference group (9) or rigorously screened normal euthyroid volunteers (8) had TSH levels between 0.5 and 2.5 Reproducibility of TSH Measurements TSH assay characteristics depend on the reagents, protocols, and technical performance of the assay. These were defined and described previously (8). High-quality laboratories should have intraassay variation of less than 5% for TSH in the range of miu/liter. Fewer data are available concerning interassay variation in TSH concentration performed on the same sample. Although it is somewhat greater than the intraassay variation, interassay variation should still be less than 10% (8, 16 18). Thus, in appropriately calibrated assays, a single TSH determination should reasonably reflect the TSH concentration in that sample. Stability of TSH over Time in the Same Individual The reference range applies to populations, but clinicians must apply the results of a single TSH determination to an individual patient. It is particularly important for clinicians to know whether a single TSH measurement reflects the individual s average TSH measurement over time. Therefore, the stability or variability of repetitive TSH determinations over short periods of time is critically important. Although such measurements are not performed in clinical practice, several studies provide data showing significant variation in repeated TSH measurements over time in the same individuals. Fish et al. (19) measured TSH 13 times within a 6-d interval in five patients taking the same daily dose of levothyroxine for at least 3 months. The initial serum TSH was normal for these patients. Considerable variation was found when TSH was measured repeatedly, with four of the five patients having at least one TSH value move out of the normal range. Others have made similar observations (20 22). Andersen et al. (22) measured thyroid hormones and TSH in 16 healthy men monthly for 1 yr. Each man seemed to have a specific and unique set point for thyroid hormone concentrations. TSH measurements in any individual varied within 50% of the TSH distribution of the entire group s samples. The variation in serum TSH concentration in repeated samples from the same individual was large and clinically significant. For example, with a mean TSH of 1.4 miu/liter, repeated measurements in the same subject could vary between 0.7 and 2.1 miu/liter; with a mean TSH of 2.4 TABLE 1. Serum TSH distribution (% population) within small intervals between 2.5 and 4.5 miu/liter [courtesy of Joseph G Hollowell; data from NHANES III database (5)] Population No. TSH intervals (miu/liter) Combined Total 17, Disease free 16, Disease free risk free 14, Reference 13, The total, disease-free, and reference populations are defined (5). The Disease Free Risk Free population is the disease-free population minus those with anti-tpo antibodies, treated with lithium or found to have hyperthyroidism or hypothyroidism. The last column (Combined) provides the sum of the smaller intervals between 3.0 and 4.5 miu/liter.

4 5492 J Clin Endocrinol Metab, September 2005, 90(9): Surks et al. Thyrotropin Reference Range miu/liter, the variation could be between 1.2 and 3.6 miu/ liter; and for TSH of 3.6 miu/liter, the variation could be between 1.8 and 5.4 miu/liter. Even with a mean normal TSH of miu/liter, some repeated TSH measurements would be expected to exceed the newly proposed upper limits of normal for TSH ( mu/liter) (6 11). Those values could classify patients incorrectly as abnormal, although the mean concentration over time is quite normal. The time of phlebotomy is also important. In a clinical practice, serum TSH obtained at h in fasting patients was significantly higher in 97 of 100 consecutive individuals than TSH determined at h in the postprandial state (23). The average decline between 0730 and 0900 and 1030 and 1200 h was 26.4%. Of 10 patients classified as having SCH based on the fasting sample, eight were reclassified as normal (TSH, 4.5 miu/liter) based on the late morning sample. Ten of 15 individuals with fasting early morning TSH between 3.0 and 4.5 mu/liter had a late morning nonfasting TSH level below 3.0 mu/liter. 2 Without addressing the etiology of the TSH variation, this study does confirm that the time of day of blood TSH sampling may influence the serum TSH concentration, resulting in misclassification as either abnormal or normal. The more narrow the reference range, the more likely that such misclassification will occur. A nocturnal surge in serum TSH is well recognized. Serum TSH approximately doubles near bedtime and begins to decrease to lower values at the time of sleep (24 27). The decline in serum TSH concentration may not be complete at 0800 h, when blood sampling generally begins, resulting in a higher serum TSH than at later times during the day. Mean TSH between 0800 and 0945 h was between 2.47 and 3.47 miu/liter in five healthy subjects without thyroid disease who were sampled every 15 min between 0800 and 1800 h. Average TSH deceased 20 32% in one or more subsequent 2-h intervals in every subject. 3 This effect is amplified after sleep deprivation (28, 29) and exercise (30). Furthermore, the peak in pulsatile TSH secretion occurs in the morning hours in night-shift workers and also may be variable and delayed in those who work during the evening shift, e.g h (31). The night-shift and evening-shift workers whose blood samples are obtained for TSH measurement h would have higher TSH concentrations than their 24-h mean levels, and some could fall into the miu/liter range. In large numbers of subjects whose time of sleep is delayed to between 0400 and 1400 h, blood sampling could occur at their bedtime peak TSH. In the United States, 3 million people work during the conventional night shift, and 4 million work evening shifts, e.g h (32). Thus, 7 million people, 5% of the U.S. workforce over age 16 yr, would probably have their phlebotomy at a time when serum TSH is at or near its highest level of the day. Results based on that determination might result in misclassification. 2 Unpublished data from Ref. 23 kindly provided by Ronald R. Scobbo and Thomas W. vondohlen. 3 Unpublished data on TSH circadian rhythms kindly provided by Mary Samuels. Specificity of TSH between 3.0 and 5.0 miu/liter for Hypothyroidism Before making any clinical decisions based on a single serum TSH determination between 3.0 and 4.5 miu/liter, TSH determination should be repeated within several weeks to several months to exclude transient thyroid dysfunction or other disorders (2, 3). Common causes of transient TSH elevations in a disease-free group include subacute, lymphocytic, or postpartum thyroiditis and recovery from a nonthyroidal illness (2). The presence of anti-tpo antibodies is prima facie evidence of autoimmune thyroid disease and predicts an increased risk for development of SCH or overt hypothyroidism when serum TSH is greater than 2.0 miu/liter (33). However, many TPO antibody-positive individuals never develop overt or subclinical hypothyroidism. We concede that many patients with serum TSH between 3.0 and 4.5 mu/liter have anti-tpo antibodies and may have the very earliest stage of subclinical hypothyroidism. However, among the 1719 individuals in the disease-free population who had anti-tpo antibodies, 62.2% of them had serum TSH below 3.0 miu/ liter (5) (see Footnote 1). Therefore, autoimmune thyroid disease and thyroid dysfunction are not synonymous. Anti-TPO antibodies were present in 17.5%, 24.9%, and 30.0% of patients whose TSH was between 3.0 and 3.49, 3.5 and 3.99, and 4.0 and 4.49 miu/liter, respectively (see Footnote 1). Overall only 22.2% of those subjects with TSH between 3.0 and 4.49 miu/liter in the disease-free group had anti-tpo antibodies, whereas 77.8% did not. It is possible that the true number of antibody-positive individuals was underestimated in the NHANES III study, because the assays were less sensitive than currently available tests. However, even if the prevalence of anti-tpo antibodies were doubled, the positive predictive value for autoimmune thyroid disease with a TSH level between 3.0 and 4.5 miu/liter would still be less than 0.5. More than 50% of individuals within this TSH range would be anti-tpo antibody negative and mislabeled as having thyroid disease or hypothyroidism. In contrast, 56.8% of those with serum TSH between 5 and 10 miu/liter had TPO antibodies. Because 77.8% of the diseasefree population whose serum TSH was between 3.0 and 4.49 miu/liter did not have antithyroid antibodies, they would probably not be at significant risk for hypothyroidism. Given the inverse log-linear relationship between free T 4 and serum TSH, the TSH response to even minor thyroid hormone changes is markedly amplified (17). It is therefore possible that some individuals with serum TSH near the reference TSH median value (1.4 mu/liter) may have minimal thyroid dysfunction, because this value may represent a small deviation from their own set point. A logical corollary is that some individuals whose TSH is in the upper portion of the reference range may not have thyroid disease, but may actually be at their set points. Some recognized causes of mildly elevated serum TSH concentrations in euthyroid individuals are circulating TSH variants of decreased biological potency (34) and TSH resistance syndromes (35 37). Spurious TSH elevations due to assay interference can also occur with circulating heterophilic antibodies (38 40) or antimouse IgG (41, 42).

5 Surks et al. Thyrotropin Reference Range J Clin Endocrinol Metab, September 2005, 90(9): Natural History and Consequences of Serum TSH between 3.0 and 5.0 miu/liter The potential consequences of a serum TSH in the range of miu/liter will be reviewed briefly to provide context for the range of miu/liter. For individuals with a serum TSH of 5 10 miu/liter, there is an increased risk of progression to overt hypothyroidism. The risk (33) is estimated to be 2.6%/yr in the absence of anti-tpo antibodies and 4.3%/yr in their presence. After 20 yr, hypothyroidism developed in 33% of patients with mildly raised TSH and in 55% of similar patients who also had antithyroid antibodies (33). Comparable results were reported in a study of patients with elevated serum TSH levels after treatment with radioactive iodine or thyroidectomy (43). Over a mean 9.2-yr interval, overt hypothyroidism developed in 42.8% of those with initial TSH levels greater than 6 12 miu/liter, but in none of the patients with initial TSH levels between 4 and 6 miu/liter. Not all individuals with SCH progress to overt hypothyroidism or even continue to have SCH. In one study of SCH in individuals over age 55 yr, serum TSH returned to normal in 37% during a mean follow-up period of 32 months (44). TSH returned to normal in 61% of antibody-negative individuals and in 30% of those with positive antithyroid antibodies. Of patients over the age of 60yr with TSH between 5 and 10 miu/liter, serum TSH was normal 12 months later in 4% of antibody-positive individuals and 11% of those without antithyroid antibodies (45). In a recent report, 11 of 21 octogenarians with subclinical hypothyroidism (TSH, 4.7 miu/liter) had normal thyroid function 3 yr later (46). Thus, in treating individuals in an attempt to prevent hypothyroidism, many patients who would spontaneously return to normal would be inappropriately treated. Limited natural history data are available for individuals with serum TSH between 2.5 and 5.0 miu/liter. We obtained unpublished data for rates of progression to overt hypothyroidism for such patients from the Whickham study. 4 Table 2 shows the 20-yr risk of developing hypothyroidism in women within TSH levels from 1 5 miu/liter in the presence and absence of antithyroid antibodies. When serum TSH was between 3.0 and 5.0 miu/liter in adults yr of age, the probability of developing hypothyroidism in 20 yr was less than 10%, whereas in adults yr old, the 20-yr probability increased to 5 15%. When antithyroid antibodies were present as well, the 20-yr prevalence increased to 15 30% in the younger adults and to 25 50% in the older age group. Therefore, even in older patients, who have increased probability of having SCH (2), the probability of developing hypothyroidism during their life expectancy is 25 50%, and normalization of TSH over time is common (44 46). Other adverse health consequences have not been confirmed for individuals with serum TSH between 5 and 10 miu/liter (2). An increased risk for development of atherosclerotic heart disease in SCH has been suggested by a crosssectional study that did not stratify serum TSH (47). No progression of atherosclerosis was found in the longitudinal 4 Unpublished data from the 20-yr follow-up of the Whickham study (31) kindly provided by Mark P. Vanderpump and Joyce M. French. TABLE 2. The probability (%) of developing hypothyroidism in women in 20 yr calculated from age, serum TSH, and antithyroid antibody status TSH (miu/liter) Age (yr) Neg. Pos. Neg. Pos. Neg. Pos. Neg. Pos. Neg. Pos Neg., No antimicrosomal titer greater than 1/100; Pos., positive antimicrosomal titer greater than 1/100. [Courtesy of Mark Vanderpump and Joyce M. French; unpublished data from the 20-yr follow-up of the Whickham Study (33)]. component of this study. Another study (48) did not confirm an increased risk for atherosclerotic heart disease. Several studies of surrogate markers for heart disease suggest a possible increased risk for coronary heart disease in SCH (49 52); other studies did not confirm significant abnormalities when TSH was less than 10 miu/liter (48, 53 55). Using surrogate markers to predict outcome is fraught with hazard, as was made painfully clear with recent data suggesting increased risk for coronary disease with estrogens when most surrogate studies pointed to a decreased risk. It is therefore critically important to study the incidence of coronary heart disease and mortality over time to properly assess cardiovascular risk. Five prospective cohort datasets are available for analysis. One 10-yr study showed an increase in cardiovascular mortality only in yr 4 6 in men with SCH, but no increase in women (56). Three studies with 4.6, 10, and 20 yr of observation did not show an increase in adverse cardiac health outcomes when TSH was elevated (47, 57, 58). One study in octogenarians demonstrated increased survival in patients with untreated hypothyroidism and SCH (46). Although designed specifically to observe individuals with low serum TSH, a 10-yr cohort study of individuals over the age of 60 yr did not show an increase in mortality in those with serum TSH over 5 miu/liter; a nonsignificant trend toward decreased mortality was found (46). No data are available for incident atherosclerotic heart disease or mortality in patients with TSH in the range of miu/liter. However, it seems logical to conclude that the consequences of TSH in this range would be even less than that for those with TSH of 5 10 miu/liter. Benefits and Risks of Levothyroxine Treatment in Patients When TSH Is miu/liter There is no compelling evidence that treatment with levothyroxine improves symptoms compared with placebo in individuals with TSH in the miu/liter range (59). Treatment of subclinical hypothyroidism when serum TSH is between 5 and 10 miu/liter generally does not have a beneficial effect on serum lipid profiles (60). The only documented adverse health outcome for individuals with TSH levels between 3.0 and 5.0 mu/liter is progression to overt hypothyroidism. Levothyroxine treatment would clearly prevent that outcome, but at what price? As noted above,

6 5494 J Clin Endocrinol Metab, September 2005, 90(9): Surks et al. Thyrotropin Reference Range many such individuals do not have hypothyroidism or autoimmune thyroid disease, and many of those who do will spontaneously return to normal thyroid function (44 46). Moreover, there are well-established consequences to being labeled with an illness (61). Furthermore, levothyroxine therapy is often not optimally controlled and may not be innocuous. Individuals in the community who take levothyroxine feel less well than controls who do not (62). Although the explanation for this requires additional study, the observation is robust. Twenty percent of levothyroxine-treated hypothyroid patients are undertreated, with TSH concentrations greater than 5.0 miu/liter. Twenty-one percent of levothyroxinetreated patients are overtreated, causing subclinical hyperthyroidism with serum TSH below 0.4 miu/liter. Only about 60% of levothyroxine-treated patients have serum TSH in the target range (5, 63). These less than optimal results of levothyroxine treatment may be due in part to its narrow therapeutic range (64). Thus, small changes in absorption may result in subclinical hypothyroidism or subclinical hyperthyroidism. Although undertreatment results only in lack of full benefit of the drug in those who need it, overtreatment may be associated with adverse health outcomes. Patients with subclinical hyperthyroidism (particularly the elderly) have a 3- to 6-fold increased risk for atrial fibrillation (65, 66). Woman with subclinical hyperthyroidism, particularly those who are estrogen deficient, suffer an increased rate of loss of bone mineral and fracture (67, 68). Subclinical hyperthyroidism in the elderly has also been associated with increased risk of all cause as well as cardiovascular mortality (57). Current estimates are that approximately 500,000 people are at increased risk for these adverse outcomes of levothyroxine treatment (69). The recent evidence that subclinical hypothyroidism in the very old decreases mortality cannot be ignored, but cannot yet be extrapolated to other age groups. Some might suggest that educating physicians could substantially reduce these risks. However, experience teaches us that practice patterns are very difficult to change. For example, a recommendation not to employ the T 3 resin uptake test was made many years ago (70), but the test is still in wide use. It is often interpreted incorrectly, leading to increased costs of additional unnecessary testing and referral, and anxiety for the patient and family. Consequences of Narrowing the TSH Normal Range In the disease-free group, 4.3% (representing 8.4 million people) had subclinical or overt hypothyroidism (TSH, 4.49 miu/liter; Table 1). Decreasing the upper limit of the reference range to 3.0 miu/liter results in characterizing an additional 7.9% of the population (15.4 million subjects) as hypothyroid. Even when all risk factors are removed from consideration (disease-free minus risk factors), 2.3% (3.9 million subjects) had subclinical hypothyroidism (Table 1); using a suggested revised upper limit of 3.0 miu/liter, an additional 6.4% of the population (10.8 million subjects) would be considered hypothyroid. Thus, decreasing the upper limit of the normal range to 3.0 miu/liter would markedly increase the number of individuals identified as hypothyroid. Although labeling a patient with a TSH of 3.1 miu/ liter as hypothyroid would not necessitate therapy, in practice many or most nonspecialist clinicians and many specialists are inclined to treat abnormal laboratory tests. The low positive predictive value for thyroid disease of a TSH level between 3.0 and 4.5 miu/liter, the absence of demonstrated benefit with therapy, and the potential for harm with therapy lead us to conclude that changing the normal TSH range at this time is premature, unjustified, and unwise. Conclusion We conclude from the data at hand that the generally accepted reference range for serum TSH, miu/liter (5), should also be considered the normal range, because the TSH range must 1) have clinical utility, 2) not result in inappropriate labeling of normal individuals as having SCH, 3) not result in unnecessary T 4 treatment and potential adverse effects of treatment, and 4) must predict a high likelihood of potential benefit with therapy. We concede that some individuals with TSH between 3.0 and 4.5 miu/liter have the very earliest stage of hypothyroidism, but at the present time, appropriate follow-up is the most prudent way to manage these patients. Recommendations 1) If serum TSH is between 3.0 and 4.5 miu/liter, the determination should be repeated within several weeks to months, depending on the clinical circumstances. 2) When TSH remains in this range on repeated testing, the patient should be carefully examined for an abnormal thyroid gland. A family history and history of previous treatment for thyroid dysfunction should be obtained. These data help define the risk of developing subclinical or overt hypothyroidism over the long term. 3) Determination of anti-tpo antibodies may be helpful in defining the risk of progression. However, even in their presence, the rate of progression is 25 50% over 20 yr, depending on age. 4) With or without anti-tpo antibodies, patients with TSH between 3.0 and 4.5 miu/liter could be considered at risk for hypothyroidism and should have repeat TSH determination in 6 months to 1 yr, then, if unchanged, about once a year. 5) Other than avoiding clinical manifestations of overt hypothyroidism in those who would have progressive decline in thyroid function, current data do not support routine levothyroxine treatment in patients with TSH between 5.0 and 10 miu/liter. We do not, therefore, recommend routine levothyroxine treatment of individuals with TSH levels between 3.0 and 4.5 miu/liter. Acknowledgments We thank the following colleagues who provided unpublished data and analyses used in this publication: Joseph G. Hollowell, Mary Samuels, Ronald R. Scobbo, Thomas W. vondohlen, Mark R. Vanderpump, and Jane M. French. Received January 26, Accepted April 6, Address all correspondence and requests for reprints to: Dr. Martin I. Surks, Division of Endocrinology and Metabolism, Montefiore Medical Center, 111 East 210th Street, Bronx, New York msurks@westnet.com.

7 Surks et al. Thyrotropin Reference Range J Clin Endocrinol Metab, September 2005, 90(9): References 1. Ladenson PW, Singer PA, Ain KB, Bagchi N, Bigos ST, Levy EG, Smith SA, Daniels GH 2000 American Thyroid Association guidelines for detection of thyroid dysfunction. Arch Intern Med 160: Surks MI, Ocampo E 1996 Subclinical thyroid disease. Am J Med 100: Surks MI, Ortiz E, Daniels GH, Sawin CT, Col NF, Cobin RH, Franklyn JA, Hershman JM, Burman KD, Denke MA, Gorman C, Cooper RS, Weissman NJ 2004 Subclinical thyroid disease: scientific review and guidelines for diagnosis and management. JAMA 291: Col, NF, Surks MI, Daniels GH 2004 Subclinical thyroid disease: clinical applications. JAMA 291: Hollowell JG, Staehling NW, Flanders WD, Hannon WH, Gunter EW, Spencer CA, Braverman LE 2002 Serum TSH, T 4, and thyroid antibodies in the United States population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III). J Clin Endocrinol Metab 87: Baskin HJ, Cobin RH, Duick DS, Gharib H, Guttler RB, Kaplan MM, Segal RL 2002 American Association of Clinical Endocrinologists medical guidelines for clinical practice for the evaluation of treatment of hyperthyroidism and hypothyroidism. Endocr Pract 8: Stathatos N, Wartofsky L 2002 Managing subclinical hypothyroidism in women. Women Health Primary Care 5: National Academy of Clinical Biochemistry NACB laboratory medicine practice guidelines Lee SL, When is the TSH normal? New criteria for diagnosis and management. Proc 12th Annual Meeting of the American Association of Clinical Endocrinologists, San Diego, CA, 2003 (thyroidtoday.com) 10. Haugen, BR 2003 When isn t the TSH normal and why? Clinical implications and causes. 12th Annual Meeting of the American Association of Clinical Endocrinologists, San Diego CA, May 14, 2003 (thyroidtoday.com) 11. Stephens PA 2004 The Endocrine Society: current issues in thyroid disease management. Endocr News 29: Expert Committee on the Diagnosis and Classification of Diabetes Mellitus 2003 Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care 26: Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo Jr JL, Jones DW, Materson BJ, Oparil S, Wright Jr JT, Roccella EJ 2003 Seventh report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High Blood Pressure. Hypertension 42: Cleeman JI 2001 Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) expert panel on detection, evaluation and treatment of high blood cholesterol in adults (Adult Treatment Panel III). JAMA 285: Fatourechi V, Klee GG, Grebe SK, Bahn R, Brennan MD, Hay ID, McIver B, Morris III JC 2003 Effects of reducing the upper limit of normal TSH values. JAMA 290: Spencer CA, Takeuchi M, Kazarosyan M 1996 Current status and performance goals for serum thyrotropin (TSH) assays. Clin Chem 42: Andersen S, Pedersen KM, Bruun NH, Laurberg P 2002 Narrow individual variations in serum T 4 and T 3 in normal subjects: a clue to the understanding of subclinical hypothyroidism. J Clin Endocrinol Metab 87: Browning MCK, Ford RP, Callaghan SJ, Fraser CG 1986; Intra- and interindividual biological variation of five analytes used in assess thyroid function: implications for necessary standards of performance and the interpretation of results. Clin Chem 32: Fish LH, Schwartz HL, Cavanaugh J, Steffes MW, Bantle JP, Oppenheimer JH 1987 Replacement dose, metabolism, and bioavailability of levothyroxine in the treatment of hypothyroidism: role of triiodothyronine in pituitary feedback in humans. N Engl J Med 316: Feldt-Rasmmussen U, Petersen PH, Blaabjerg O, Horder M 1980 Long-term variability in serum thyroglobulin and thyroid related hormones in healthy subjects. Acta Endocrinol (Copenh) 95: Maes M, Mommen K, Hendrickx D, Peeters D Hondt P, Ranjan R, De Meyer F, Scharpe S 1997 Components of biological variation, including seasonality, in blood concentrations of TSH, TT 4,FT 4, PRL, cortisol and testosterone in healthy volunteers. Clin Endocrinol (Oxf) 46: Andersen S, Bruun NH, Pedersen KM, Laurberg P 2003 Biologic variation is important for interpretation of thyroid function tests. Thyroid 13: Scobbo RR, vondohlen TW, Hassan M, Islam S 2004 Serum TSH variability in normal individuals: the influence of time of sample collection. West Virginia Med J 100: Azukizawa M, Pekary AE, Hershman JM 1976 Plasma thyrotropin, thyroxine and triiodothyronine relationships in man. J Clin Endocrinol Metab 43: Parker DC, Pekary AE, Hershman JM 1976 Effect of normal and reversed sleep-wake cycles upon nyctohereral rhythmicity of plasma thyrotropin: evidence suggestive of an inhibitory influence of sleep. J Clin Endocrinol Metab 43: Weeke J, Gundersen HJG 1978 Circadian and 30 minutes variations in serum TSH and thyroid hormones in normal subjects. Acta Endocrinol (Copenh) 89: Samuels MH, Veldhuis JD, Henry P, Ridgway EC 1990 Pathophysiology and co-pulsatile release of thyroid stimulating hormone, luteinizing hormone, follicle stimulating hormone and subunit. J Clin Endocrinol Metab 71: Brabant G, Prank K, Ranft U, Schuermeyer T, Wagner TOF, Hauser H, Kummer B, Feistner H, Hesch RD, von zur Muhlen A 1990 Physiological regulation of circadian and pulsatile thyrotropin secretion in normal man and woman. J Clin Endocrinol Metab 70: Gary KA, Winokur A, Douglas SD, Kapoor S, Zaugg L, Dinges DF 1996 Total sleep deprivation and the thyroid axis: effects of sleep and waling activity. Aviat Space Environ Med 67: Buxton OM, Frank SA, L Hermite-Baleriaux M, Leproult R, Turek FW, van Cauter E 1997 Roles of intensity and duration of nocturnal exercise in causing phase delays of human circadian rhythms. Am J Physiol 36:E536 E Weibel L, Brandenberger G, Goichot B, Spiegel K, Ehrhart J, Follenius M 1995 The circadian thyrotropin rhythym is delayed in regular night workers. Neurosci Lett 187: Bureau of Labor Statistics, Department of Labor Employment situation summary Vanderpump MP, Tunbridge WM, French JM, Appelton D, Bates D, Clark F, Evans JG, Hasan DM, Rodgers H, Tunbridge F, Young ET 1995 The incidence of thyroid disorders in the community: a twenty-year follow-up of the Whickham Survey. Clin Endocrinol (Oxf) 43: Beck-Peccoz P, Amr S, Menezes-Ferreira MM, Faglia G, Weintraub BD 1985 Decreased receptor binding of biologically inactive thyrotropin in central hypothyroidism. Effect of treatment with thyrotropin-releasing hormone. N Engl J Med 312: Xie J, Pannain S, Pohlenz J, Weiss RE, Moltz K, Morlot M, Asteria C, Persani L, Beck-Peccoz P, Parma J, Vassart G, Refetoff S 1997 Resistance to thyrotropin (TSH) in three families is not associated with mutations in the TSH receptor or TSH. J Clin Endocrinol Metab 82: Alberti L, Proverbio MC, Costagliola S, Romoli R, Boldrighini B, Vigone MC, Weber G, Chiumello G, Beck-Peccoz P, Persani L 2002 Germline mutations of TSH receptor gene as cause of nonautoimmune subclinical hypothyroidism. J Clin Endocrinol Metab 87: Tonaccherra M, Perri A, De Marco G, Agretti P, Banco ME, Di Cosmo C, Grasso L, Vitti P, Chiovato L, Pinchera A 2004 Low prevalence of thyrotropin receptor mutations in a large series of subjects with sporadic and familial nonautoimmune subclinical hypothyroidism. J Clin Endocrinol Metab 89: Brennan MD, Klee GG, Preissner CM, Hay ID 1987 Heterophilic serum antibodies: a cause for falsely elevated serum thyrotropin levels. Mayo Clin Proc 62: Wood JM, Gordon DL, Rudinger AN, Brooks MM 1991 Artifactual elevation of thyroid-stimulating hormone. Am J Med 90: Ward G, McKinnon L, Badrick T, Hickman PE 1997 Heterophilic antibodies remain a problem for the immunoassay laboratory. Am J Clin Pathol 108: Howantz PJ, Howantz JH, Lamberson HV, Ennis KM 1982 Incidence and mechanism of spurious increases in serum thyrotropin. Clin Chem 28: Zweig MH, Csako G, Benson CC, Weintraub BD, Kahn BB 1987 Interference by anti-immunoglobulin G antibodies in immunoradiometric assays of thyrotropin involving mouse monoclonal antibodies. Clin Chem 33: Huber G, Staub J-J, Meier C, Mitrache C, Guglielmetti M, Huber P, Braverman LE 2002 Prospective study of the spontaneous course of subclinical hypothyroidism: prognostic value of thyrotropin, thyroid reserve, and thyroid antibodies. J Clin Endocrinol Metab 87: Diez JJ, Iglesias P 2004 Subclinical hypothyroidism may progress to overt hypothyroidism or disappear. J Clin Endocrinol Metab 89: Parle JV, Franklyn JA, Cross KW, Jones SC, Sheppard MC 1991 Prevalence and follow-up of abnormal thyrotrophin (TSH) concentrations in the elderly in the United Kingdom. Clin Endocrinol (Oxf) 34: Gussekloo J, van Exel E, de Craen AJM, Meinders AE, Frolich M, Westendorp RGL 2002 Thyroid status, disability and cognitive function, and survival in old age. JAMA 292: Hak AE, Pols HA, Visser TJ, Drexhage HA, Hofman A, Witteman JC 2000 Subclinical hypothyroidism is an independent risk factor for atherosclerosis and myocardial infarction in elderly women: the Rotterdam Study. Ann Intern Med 132: Lindeman RD, Romero LJ, Schade DS, Wayne S, Baumgartner RN, Garry PJ 2003 Impact of subclinical hypothyroidism on serum total homocysteine concentrations, the prevalence of coronary heart disease (CHD), and CHD risk factors in the New Mexico Elder Health Study. Thyroid 13: Caraccio N, Ferrannini E, Monzani F 2002 Lipoprotein profile in subclinical hypothyroidism: response to levothyroxine replacement, a randomized placebo-controlled study. J Clin Endocrinol Metab 87: Monzani F, Caraccio N Kozakowa M, Dardano A, Vittone F, Virdis A, Taddei S, Palombo C, Ferrannini E 2004 Effect of levothyroxine replacement on lipid profile and intima-media thickness in subclinical hypothyroidism: a doubleblind, placebo-controlled study. J Clin Endocrinol Metab 89: Taddei S, Caraccio N, Virdis A, Dardano A, Versari D, Ghiadoni L, Salvetti A, Ferrannini E, Monzani F 2003 Impaired endothelium-dependent vasodilatation in subclinical hypothyroidism: beneficial effect of levothyroxine therapy. J Clin Endocrinol Metab 88:

8 5496 J Clin Endocrinol Metab, September 2005, 90(9): Surks et al. Thyrotropin Reference Range 52. Danese MD, Ladenson PW, Meinert CL, Powe NR 2000 Effect of thyroxine therapy on serum lipoproteins in patients with mild thyroid failure: a quantitative review of the literature. J Clin Endocrinol Metab 85: Aldasouqi S, Nkansa-Dwamena D, Bokhari S, Alzahrani AS, Khan M, Al- Keffi A, Merdad A 2004 Is subclinical hypothyroidism associated with hyperhomocysteinuria? Endocr Pract 10: Christ-Crain M, Meier C, Guglielmetti M, Huber PR, Riesen W, Staub J-J, Muller B 2003 Elevated C-reactive protein and homocysteine values: cardiovascular risk factors in hypothyroidism? A cross-sectional and a double-blind placebo-controlled trial. Atherosclerosis 166: Perez A, Cubero JM, Sucunza N, Ortega E, Arcelius R, Rodriguez-Espinosa J, Ordonez-Llanos J, Blanco-Vaca F 2004 Emerging cardiovascular risk factors in subclinical hypothyroidism: lack of change after restoration of euthyroidism. Metabolism 53: Imaizumi M, Akahoshi SI, Ichimaru S, Nakashima E, Hida A, Soda M, Usa T, Ashizawa K, Yokoyama N, Maeda R, Nagataki S, Eguchi K 2004 Risk for ischemic heart disease and all-cause mortality in subclinical hypothyroidism. J Clin Endocrinol Metab 89: Parle JV, Maisonneuve P, Sheppard MC, Boyle P, Franklyn JA 2001 Prediction of all-cause and cardiovascular mortality in elderly people from one low serum thyrotropin result: a 10-year cohort study. Lancet 358: Vanderpump MP, Tunbridge WM, French JM, Appelton D, Bates D, Clark F, Grimley EJ, Rodgers H, Tunbridge F, Young ET 1996 The development of ischemic heart disease in relation to autoimmune thyroid disease in a 20-year follow-up study of an English community. Thyroid 6: Kong WM, Sheikh MH, Lumb PJ, Freedman DB, Crook M, Dore CJ, Finer N 2002 A 6-month randomized trial of thyroxine treatment in women with mild subclinical hypothyroidism. Am J Med 112: Ineck BA, Ng TMH 2003 Effects of subclinical hypothyroidism and its treatment on serum lipid profiles. Ann Pharmacother 37: MacDonald LA, Sackett Dl, Haynes RB, Taylor DW 1984 Labelling in hypertension: a review of the behavioural and psychological consequences. J Chron Dis 37: Saravanan P, Chau W-F, Roberts N, Vedhara K, Greenwood R, Dayan CM 2002 Psychological well-being in patients on adequate doses of l-thyroxine: results of a large, controlled community-based questionnaire study. Clin Endocrinol (Oxf) 57: Canaris GJ, Manowitz NR, Mayor G, Ridgway EC 2000 The Colorado Thyroid Disease Prevalence Study. Arch Intern Med 160: Carr D, Mcleod DT, Parry G, Thornes HM 1988 Fine adjustment of thyroxine replacement dosage: comparison of the thyrotrophin releasing hormone test using a sensitive thyrotophin assay with measurement of free thyroid hormones and clinical assessment. Clin Endocrinol (Oxf) 28: Sawin CT, Geller A, Wolf PA, Bleanger AJ, Baber E, Bacharach P, Wilson PW, Benjamin EJ, D Agostino RB 1994 Low serum thyrotropin concentrations as a risk factor for atrial fibrillation in older persons. N Engl J Med 331: Auer J, Scheiber P, Mische T, Langsteger W, Eber O, Eber B 2001 Subclinical hyperthyroidism as a risk factor for atrial fibrillation. Am Heart J 142: Bauer DC, Ettinger B, Nevitt M, Stone KL 2001 The Study of Osteoporotic Fractures Research Group. Risk for fracture in women with low serum levels of thyroid-stimulating hormone. Ann Intern Med 134: Vestergaard P, Mosekilde L 2002 Fractures in patients with hyperthyroidism and hypothyroidism: a nationwide follow-up study in 16,249 patients. Thyroid 12: Cooper DS, Ridgway EC 2002 Thoughts on prevention of thyroid disease in the United States. Thyroid 12: Surks MI, Chopra IJ, Mariash CN, Nicoloff, JT, Solomon, DH 1990 American Thyroid Association Guidelines for use of laboratory tests in thyroid disorders. JAMA 263: JCEM is published monthly by The Endocrine Society ( the foremost professional society serving the endocrine community.