Rightward shift in serum TSH range in individuals with exceptional longevity

TOPIC: TSH reference range increases with aging

Title: Extreme longevity is associated with increased serum thyrotropin.

Authors: Atzmon G, Barzilai N, Hollowell JG, Surks MI, & Gabriely I.

Reference: Journal of Clinical Endocrinology & Metabolism 94: 1251-1254, 2009



The TSH reference range, specifically the setting of the upper reference limit, is currently controversial. This is because the upper reference limit impacts the number of patients diagnosed with subclinical hypothyroidism.


To determine whether the rightward shift in the TSH population reference range found in the 80+ year old NHANES III participants extended to people of exceptional longevity.

Design, Setting and Patients

Serum FT4 and TSH of 3 populations without reported thyroid disease were investigated: Ashkenazi Jews with exceptional longevity (centenarians; median age: 98 years), younger Ashkenazi controls (median age: 72 years) and the population of thyroid disease-free 60-79 year old individuals (median age: 68 years) from the U.S. National Health and Nutrition Examination Survey 1998–2002 (NHANES controls).

Main Outcome Measures

TSH frequency distribution curves and TSH/free T4 relationships for each population were compared.


Serum TSH was significantly higher in the centenarians [1.97 (0.42–7.15) mU/L] compared with the ethnically-matched younger controls [1.55 (0.46–4.55) mU/L] and the NHANES cohort [1.61 (0.39–6.29) mU/L] (median, 2.5 and 97.5 centiles) ( P< 0.001). TSH distribution of the centenarian group was relatively similar in shape but shifted significantly to higher TSH as was the TSH at peak frequency. The TSH distribution curve of the NHANES control group was superimposable to – and not significantly different from – the Ashkenazi controls. Free T4 was similar in centenarians and the ethnically-matched controls, and a significant inverse correlation between free T4 and TSH was present in both groups.


The TSH population reference range appears to shift to higher concentrations with advancing age as a continuum. The preservation of the inverse relationship between free T4 and TSH suggests that changes in thyroid hormone negative feedback may contribute to exceptional longevity.


This study of centenarians confirms other reports that there is a shift in the TSH reference range to higher values with aging. This shift did not appear related to ethnicity because the younger group of ethnically-matched controls had an identical serum TSH distribution frequency as age-matched NHANES participants. In addition, the rightward TSH shift was not associated with increased skewing, suggesting that the shift was not due to an increased prevalence of underlying thyroid failure with aging (although specimen limitations prevented the exclusion of thyroid autoantibody positive participants). Authors suggest that the change in TSH distribution with age could result from a decline in thyroid function or a reset in the TSH/FT4 setpoint. However, other studies report a negative relationship between TSH and advancing age. The discrepancy between the current report and other studies from literature is difficult to understand and the causes may involve population selection criteria, methodological issues as well as patho-physiological factors. It should also be acknowledged that it is difficult to integrate the potential independent influences of non thyroidal conditions, such as inadequate caloric intake or multiple medications in extremely old individuals.

Circulating TSH is heterogeneous with respect to glycosylation and biologic activity, the latter being a function of TSH molecular glycosylation that is influenced by hypothalamic TRH, thyroid hormones and other factors. Current biases between TSH immunometric assays (IMA) likely result from the detection of different TSH isoforms, some of which may be biologically inert. For example, TSH values can vary by as much as 40% when different IMAs are used to measure the same serum specimen, and there is no way to know how much of the TSH detected by any given assay is biologically active. An extreme example of discordance between immunologic and biologic TSH activity is seen with pituitary disease. Specifically, patients with central hypo- or hyperthyroidism typically have paradoxically normal serum TSH because current IMA methods measure TSH molecules with reduced or enhanced biologic activity, respectively. If aging resulted in secretion of TSH molecules with intrinsically altered biological activity, any age-related trend in TSH would depend on the relative specificity of the IMA method for detecting bioactive versus bioinactive TSH isoforms.

The changes in hypothalamic-pituitary function and peripheral hormone metabolism that occur with extreme aging have uncertain pathological significance. The TRH-stimulated TSH response, as well as the TSH rise in response to hypothyroidism, becomes blunted with increasing age, whereas the 5’-deiodinase activity appears decreased in the very elderly. A growing number of studies and meta-analyses suggest that lower free T4 and higher TSH in patients over age 65 may even convey survival benefits (see the recent review article by Biondi & Cooper, Endocrine Reviews in 2008).

As discussed by the authors, there is currently controversy regarding the setting of the TSH upper limit because this impacts the proportion of patients being classified as having subclinical hypothyroidism. The debate concerning whether specific age-related TSH reference ranges are necessary overlooks the fact that TSH has a narrow within-person variability (± 0.5 mU/L over one year). A population reference range has limited value for detecting disease in the individual when the index of individuality (that is the ratio between the within-person and between-person variability) of a test is low, as it is for TSH. It follows that there should be less focus on using a fixed TSH upper limit and instead interpret TSH values relative to patient-specific factors. For example, the recent Endocrine Society guidelines (2007) suggest that a serum TSH upper limit of 2.5 mU/L should be the target for preconception and pregnancy – an upper limit that the present study clearly shows is too low for elderly patients. In addition to age, the presence of TPO antibodies as well as a myriad of other cardiovascular risk factors merit consideration when evaluating, in clinical practice, a TSH result.

In summary, studies such as the one discussed here indicate that interpretation of a serum TSH result is becoming more complex. However, the added complexity has the advantage of allowing patient-specific factors to guide the diagnostic and therapeutic response to the patient’s TSH value, and this attitude is in accord with a growing focus on personalized medicine.

Summary and Commentary prepared by Carole Spencer (Related to Chapters 4 & 6 [section a] of TDM)