Age and the Thyrotropin Response to Hypothyroxinemia

Abstract

Context: Some studies suggest altered pituitary functioning and TSH production with aging.

Objective: Our objective was to test the hypothesis that less TSH production occurs despite comparable hypothyroxinemia with advancing age.

Design: We retrospectively studied adult outpatients of all ages with confirmed hypothyroidism and documented their TSH and free T₄ concentrations.

Participants: Two populations of 112 patients were subdivided into four age groups: 1) patients newly diagnosed with primary hypothyroidism and 2) thyroid cancer patients undergoing l-T₄ withdrawal in preparation for diagnostic or therapeutic radioiodine.

Main Outcome Measure: The relationship between paired free T₄ and TSH concentrations and patient age was studied.

Results: With spontaneous hypothyroidism, the mean TSH concentration decreased nonsignificantly in each ascending age group with comparable free T₄ (FT₄) concentrations (<35 yr, 69 mIU/liter; 35–49 yr, 49 mIU/liter; 50–64 yr, 43 mIU/liter; >64 yr, 29 mIU/liter). With iatrogenic hypothyroidism, the mean TSH concentration decreased significantly in each ascending age group (<35 yr, 156 mIU/liter; 35–49 yr, 115 mIU/liter; 50–64 yr, 74 mIU/liter; >64 yr, 46 mIU/liter; P < 0.001) despite similar FT₄ concentrations. The relationship between the log-transformed TSH and FT₄ was significantly and inversely affected by age in multivariate analyses in both spontaneous hypothyroidism (P = 0.0005) and in iatrogenic hypothyroidism (P < 0.0001).

Conclusions: Age modifies the pituitary set point or response to comparably reduced free T₄ concentrations, resulting in lesser serum TSH elevation in older individuals. This phenomenon occurs with both spontaneous and iatrogenic hypothyroidism. This may be an adaptive response in normal aging or a pathological alteration of pituitary function with age.

Age reduces the TSH response to hypothyroxinemia in both spontaneous and iatrogenic hypothyroidism.

The biochemical endpoint used to diagnosis and manage hypothyroidism is the serum TSH concentration. A patient’s physiological response to hypothyroxinemia is manifest by the degree of elevation of their serum TSH concentration.

Previous studies examining cases of spontaneous hypothyroidism have shown various effects of age on the thyrotrope response to reduced T₄ concentrations. A 1994 study showed no impact of age and found that patients older than 70 yr had similar serum TSH concentrations to patients younger than 55 yr for similar degrees of hypothyroxinemia (1). A second study showed differential responses only at extremes of age. In this study, patients older than 70 yr had lesser TSH responses to a low total T₄ level compared with patients younger than age 30 yr (2). More recently, Carlé et al. (3) reported an inverse relationship between TSH concentrations and age across the age spectrum. These authors studied cases of Hashimoto’s hypothyroidism diagnosed in the Danish community. They found that the median TSH values were significantly higher in patients under age 20 yr compared with those 80 yr and older. They also found in multivariate analyses that both total T₄ and age were significantly inversely associated with serum TSH concentrations.

Many thyroid cancer specialists have observed, anecdotally, that younger patients mount a more vigorous pituitary response to thyroid hormone withdrawal protocols, manifest by higher TSH concentrations after a specific period of abstinence from l-T₄. This impression is borne out by at least two studies. Schlumberger et al. (4) showed in 1984 that after at least 5 wk off l-T₄, the TSH concentration achieved was inversely correlated with the patient’s age. In addition, the mean TSH value was significantly higher in those younger than 45 yr compared with those 45 yr and older. A study of children with thyroid cancer undergoing withdrawal from l-T₄ without concurrent use of T₃, showed that 12 d off thyroid hormone was sufficient to produce a serum TSH concentration of at least 25 mIU/liter (5). Other studies, either with or without temporary use of T₃, have demonstrated in older individuals encompassing the adult age group that longer periods of discontinuation are generally necessary to achieve the requisite TSH concentration (6,7,8,9,10,11,12,13,14,15).

The goal of our study, therefore, was first to confirm that the TSH response to a given degree of hypothyroxinemia was affected by patient age. We then wished to determine whether the inverse relationship between serum TSH and age existed across the entire age spectrum, rather than only at extremes of age. We furthermore wished to extend the investigation to whether this phenomenon was a feature of both spontaneous and iatrogenic hypothyroidism. Finally, we speculated that the same age-TSH relationship would also hold true for subclinical hypothyroidism.

Patients and Methods

Two groups of 112 patients each were obtained through chart review after approval of the project by the Institutional Review Board. The review included medical records available from 1999 through 2009 at either Georgetown University Medical Center or Washington Hospital Center. A search was conducted for all patients with diagnoses of hypothyroidism and thyroid cancer. The paper and electronic records of these patients were reviewed to determine whether there were sufficient data to satisfy the study requirements. Any patients without the required data were excluded. All patients were outpatients, and they were excluded if the chart review suggested the possibility of any concurrent illness or confounding medications. There were no other exclusion criteria. Patients were divided into four age groups as follows: under 35 yr, 35–49 yr, 50–64 yr, and over 64 yr. These age groups were selected to divide the represented age range approximately into quarters yet ensure there were sufficient patients contained within the youngest and oldest groups.

Group with spontaneous hypothyroidism

For the group with spontaneous hypothyroidism, data were abstracted only for patients in whom the elevated TSH concentration had been confirmed with at least one confirmatory TSH reading, and l-T₄ therapy was subsequently started. However, all data were from the period before l-T₄ initiation. Patients with subclinical hypothyroidism were included if there were at least two confirmatory elevated TSH readings and if l-T₄ therapy was maintained. For the patients older than 50 and 80 yr, upper limits of 7 and 10 mIU/liter were used for defining an abnormal TSH based on the data of Surks and Hollowell (16). Patients were also excluded if they were taking any medications known to affect TSH or thyroid hormone concentrations or relationships. Examples of medications leading to exclusion were steroids, dopamine analogs, somatostatin analogs, phenytoin, carbamazepine, sertraline, rifampin, amiodarone, and lithium. The following data were collected from the records of patients with primary hypothyroidism at Georgetown University: patient age, gender, height, weight, body mass index, menopausal status, first elevated TSH with a concomitant free T₄ (FT₄) concentration, FT₄ concentration, serum creatinine concentration obtained within 2 wk of TSH determination, estimated glomerular filtration rate (GFR), year of laboratory evaluation, month of laboratory evaluation, type of hypothyroidism (Hashimoto’s hypothyroidism, surgically induced hypothyroidism, radioiodine therapy for hyperthyroidism, or external beam radiotherapy), and documentation of thyroid peroxidase antibodies. Data about thyroglobulin antibodies were not collected.

Group with iatrogenic hypothyroidism

For the group of thyroid cancer patients undergoing a protocol of withdrawal from l-T₄, and thus having iatrogenic hypothyroidism, the only exclusion criteria were missing data, confounding medications, and concurrent illness. Hospitalization for radioiodine administration only was not grounds for exclusion. Patients with all stages of thyroid cancer were included, as were patients undergoing either their first or any subsequent withdrawal protocol. To our knowledge, no patient had functioning metastases that were secreting thyroid hormone. Only patients with concomitant FT₄ concentrations that were detectable but less than 0.4 ng/dl were analyzed. The following data were collected from the records of patients with thyroid cancer at Georgetown University and Washington Hospital Center: patient age, gender, height, weight, body mass index, menopausal status, baseline TSH concentration, TSH concentration during withdrawal, FT₄ concentration during withdrawal, serum creatinine concentration obtained within 1 wk of TSH determination, estimated GFR, year of protocol, month of protocol, number of days off l-T₄ therapy, number of days off T₃ therapy, thyroglobulin concentration, thyroglobulin antibody titer, presence of distant metastases, presence of thyroid remnant, type of differentiated thyroid cancer, years elapsed since thyroidectomy, years elapsed since last radioiodine therapy, and reason for withdrawal from l-T₄ therapy (radioiodine therapy or diagnostic scanning or both). If data were collected from patients whose thyroidectomy was within the last 12 months, the years elapsed since thyroidectomy was designated as 0. If data were associated with current administration of radioiodine therapy, the years elapsed since radioiodine administration was described as 0.

Laboratory assessments and GFR calculation

Laboratory testing was performed by the clinical laboratory designated by the patient’s insurance company. TSH and FT₄ were analyzed by Quest Diagnostics (Madison, NJ), LabCorp (Burlington, NC), Georgetown University Laboratory (Washington, DC), or Washington Hospital Center Laboratory (Washington, DC). TSH assays used by these clinical laboratories employed a third-generation ultrasensitive immunochemiluminometric assay with a sensitivity of 0.01 mIU/liter (laboratory reference ranges approximately 0.4–4.5 mIU/liter). FT₄ levels were measured by the clinical laboratories using chemiluminescent immunoassays also. Over the period of this study, the reference range for FT₄ was approximately 0.8–1.80 ng/dl. GFR was estimated using the abbreviated Modification of Diet in Renal Disease study equation (GFR = 186 × serum creatinine^−1.154 × age^−0.203 × sex × race) and reported in milliliters per minute per 1.73 m² (17,18). Sex adjustments were 1 for males and 0.742 for females; adjustments for race were 1.21 for Black individuals and 1 for non-Black individuals.

Statistical analysis

Biostatistical support was provided by the Georgetown University Medical Center’s General Clinical Research Center. For pairs of continuous variables, scatter plots were constructed to visually display the respective relations. If found to be linear with both variables normally distributed, then the Pearson correlation was computed to statistically describe the significance, strength, and nature of the relation. If the assumption of linearity assumption was violated, then the Spearman rank correlation was used instead.

An independent two-samples t test was used to examine the existence of a statistically significant difference between the means of two groups if the variable being tested was normally distributed in each of the two independent groups. When the normality assumption was violated in either or both groups, the Wilcoxon-Mann-Whitney U test was used.

A one-way ANOVA was used to examine the difference in means of normally distributed variables between groups of three or more. When the normality assumption was violated, the Kruskal-Wallis test was used. The values of continuous variables whose distribution did not follow that of a normal distribution were ranked. A multifactorial ANOVA was then used to examine statistically significant differences among the ranks of the study groups. The analysis of covariance was used to test for statistically significant differences in the slopes of multiple regression lines. A multiple linear regression was used to model the relation between the dependent variable and a varying combination of predictor variables. SAS 9.1.3 was used for all analyses.

Results

The characteristics of the groups with spontaneous and iatrogenic hypothyroidism are shown in Tables 1 and 2, respectively. The spontaneously hypothyroid patients had a mean age of 46 yr and were 73% female. The cause of the hypothyroidism was Hashimoto’s hypothyroidism in 62% of patients. Elevated serum thyroid peroxidase antibodies were documented in 37%; in the remainder, the diagnosis was made on clinical grounds. The mean TSH concentration was 51 mIU/liter. The mean FT₄ was 0.76 ng/dl, with 54% of cases being due to subclinical hypothyroidism. The group of patients with iatrogenic hypothyroidism had a mean age of 47 yr and were 77% female. They had been off l-T₄ therapy for a mean of 39 d and off exogenous T₃ for a mean of 13 d. Seventy-nine percent were receiving their first radioiodine therapy shortly after their thyroidectomy. Seventy-four percent had papillary thyroid cancer, 16% had follicular thyroid cancer, and only 13% had distant metastases. The mean TSH concentration was 105 mIU/liter with a corresponding mean serum FT₄ of 0.23 ng/dl. In both the hypothyroid groups, serum creatinine progressively increased with age, whereas GFR progressively decreased with age (see Tables 1 and 2).

Table 1.

Characteristics of patients with spontaneous hypothyroidism (number of patients = 112)

Characteristic	Mean (sd)	Percentage
Age (yr)	46 (15)
Gender
Female		73
Male		27
Height (m)	1.68 (0.10)
Weight (kg)	73 (16)
BMI	26 (5)
Menopausal status
Premenopausal		43
Postmenopausal		30
Not applicable		27
TSH (mIU/liter)	51 (73)
FT4 (ng/dl)	0.76 (0.35)
Serum creatinine (mg/dl)
All ages	0.99 (0.27)
<35 yr	0.91 (0.22)
35–49 yr	0.94 (0.22)
50–64 yr	1.05 (0.28)
>64 yr	1.21 (0.36)
GFR (ml/min · 1.73 m2)
All ages	83 (30)
<35 yr	99 (36)
35–49 yr	87 (27)
50–64 yr	71 (18)
>64 yr	58 (15)
Type of hypothyroidism
Overt		46
Subclinical		54
Cause of hypothyroidism
Hashimoto’s hypothyroidism		62
Surgery		21
Radioiodine therapy		11
External beam radiotherapy		6
Presence of elevated thyroid peroxidase antibodies
Yes		37
No		35
Not determined		28

Table 2.

Characteristics of patients with iatrogenic hypothyroidism (number of patients = 112)

Characteristic	Mean (sd)	Percentage
Age (yr)	47 (15)
Gender
Female		77
Male		23
Height (m)	1.72 (0.09)
Weight (kg)	79 (19)
BMI	28 (7)
Menopausal status
Premenopausal		46
Postmenopausal		31
Not applicable		23
TSH (mIU/liter)	105 (75)
FT4 (ng/dl)	0.23 (0.09)
Serum creatinine (mg/dl)
All ages	1.22 (0.30)
<35 yr	1.09 (0.17)
35–49 yr	1.13 (0.23)
50–64 yr	1.24 (0.29)
>64 yr	1.67 (0.27)
GFR (ml/min · 1.73 m2)
All ages	64 (21)
<35 yr	75 (19)
35–49 yr	68 (19)
50–64 yr	59 (19)
>64 yr	39 (10)
Baseline serum TSH (mIU/liter)
All ages	2.0 (1.5)
<35 yr	1.7 (1.3)
35–49 yr	1.7 (1.1)
50–64 yr	2.1 (1.2)
>64 yr	3.1 (2.5)
Days off l-T4	39 (5)
Days off T3	13 (1)
Thyroglobulin (ng/ml)	85 (444), median 0
Presence of thyroglobulin antibodies
No		83
Yes		17
Distant metastases
No		87
Yes		13
Presence of thyroid remnant
Yes		86
No		14
Type of differentiated thyroid cancer
Papillary		74
Follicular		16
Other variant		10
Years elapsed since thyroidectomy	0.5 (1.2)
Years elapsed since last radioiodine therapy	0.3 (0.7)
Initial radioiodine therapy		79
Subsequent radioiodine therapy or diagnostic scanning		21

For the patients with spontaneous hypothyroidism, there was a statistically significant inverse correlation between FT₄ and log₁₀ TSH (Spearman correlation coefficient = −0.795, P value <0.0001). The slope of the relationship was −1.131 ± 0.074 with an intercept of 2.29 (TSH 194) (see Fig. 1). The relationship between FT₄ and log₁₀ TSH was significantly affected by age (P value = 0.0005). None of the following variables including gender, height, weight, BMI, menopausal status, year of laboratory evaluation, month of laboratory evaluation, type of hypothyroidism, or presence of thyroid peroxidase antibodies had any significant influence on the relationship between FT₄ and log₁₀ TSH. However, cause of hypothyroidism was a significant predictor in the model (p value = 0.042). Serum creatinine and estimated GFR were associated with TSH concentrations (P < 0.001), but were not significant predictors when age was included in the model.

Figure 1.

Relationship between FT₄ and TSH in the spontaneously hypothyroid group according to age. For all patients together, the Spearman correlation coefficient = −0.795, P value <0.0001. The inset shows the slopes within the four age groups: dots and dashes, under 35 yr; dashes, 35–49 yr; dots, 50–64 yr; solid line, over 64 yr.

Patients were divided into 4 age groups as follows: <35 yr (n = 32), 35–49 yr (n = 38), 50–64 yr (n = 30), and >64 yr (n = 12). For each group there was a statistically significant inverse correlation between FT₄ and log₁₀ TSH (Spearman correlations: coefficient = −0.799, P value <0.0001; coefficient = −0.850, P value <0.0001; coefficient = −0.867, P value <0.0001; and coefficient = −0.606, P value = 0.036, respectively). The slope of the relationship was not statistically different for each age group. Separate intercepts on the y-axis could be visualized for each age group, but these were not statistically different (see inset in Fig. 1). When employing the same four age groups, neither the mean TSH nor the mean FT₄ were statistically different for each group (see Table 3).

Table 3.

Differences in mean TSH and mean FT₄ values with patients divided into four age groups

Age groups (yr)	Number	TSH				FT4
		Mean (mIU/liter)	Log-transformed mean TSH	sd (mIU/liter)	P value (for differences between age groups)	Mean (ng/dl)	sd (ng/dl)	P value (for differences between age groups using Kruskal-Wallis test)
Spontaneous hypothyroidism
<35	32	69	1.83	82	0.45a	0.82	0.41	0.16a
35–49	38	49	1.69	92		0.81	0.31
50–64	30	43	1.63	39		0.67	0.35
>64	12	29	1.46	22		0.68	0.30
Iatrogenic hypothyroidism
<35	28	156	2.19	98	<0.0001b	0.23	0.09	0.94a
35–49	40	115	2.06	70		0.24	0.11
50–64	30	74	1.86	31		0.23	0.07
>64	14	46	1.66	20		0.22	0.09

^aKruskal-Wallis test. 

^bANOVA. 

For the patients with iatrogenic hypothyroidism, there was a very narrow range of FT₄ values, and there was no significant inverse correlation between FT₄ and log₁₀ TSH. Results for the group with iatrogenic hypothyroidism showed a statistically significant association between log₁₀ TSH levels and FT₄ after adjusting for age (P value <0.0001). Only age was a significant predictor in the model (P value <0.0001). All other variables including gender, height, weight, BMI, baseline TSH concentration before withdrawal from l-T₄, menopausal status, year of laboratory evaluation, month of laboratory evaluation, days off l-T₄, days off T₃, thyroglobulin concentration, presence of thyroglobulin antibodies, distant metastases, presence of thyroid remnant, type of differentiated thyroid cancer, years since thyroidectomy, and years since last radioiodine therapy were not significant predictors in the model. Serum creatinine and estimated GFR, however, were again associated with TSH concentrations (P = 0.005 and P = 0.003, respectively) but were not significant predictors when age was included in the model.

When patients with iatrogenic hypothyroidism were again divided into four age groups [<35 yr (n = 28), 35–49 yr (n = 40), 50–64 yr (n = 30), and >64 yr (n = 14)], there was a statistically significant difference in the mean TSH concentrations among the four age groups (ANOVA based on ranks P value <0.0001) (see Table 3). After multiplicity adjustments, statistically significant differences in mean TSH concentrations were found between age groups under 35 and 35–49 yr (P value = 0.0476), age groups under 35 and 50–64 yr (P value <0.0001), and age groups under 35 and over 64 yr (P value <0.0001). There were also significant differences between age groups 35–49 and 50–64 yr (P value = 0.0028), and age groups 35–49 and over 64 yr (P value <0.001). Finally, there were also significant differences between age groups 50–64 and over 64 yr (P value = 0.019). On the other hand, there were no statistically significant differences in mean FT₄ levels between the four age groups (Kruskal-Wallis test P value = 0.939).

Finally, when the FT₄ values and TSH values from the patients with both spontaneous and iatrogenic hypothyroidism were combined, there was a significant inverse relationship between FT₄ and log₁₀ TSH (Spearman correlation coefficient = −0.681, P value <0.0001). The slope of this relationship was −1.063 ± 0.043, and the intercept was 2.1 (TSH 125) (see Fig. 2). When these combined data were again divided into the same four age groups used previously [<35 yr (n = 60), 35–49 yr (n = 78), 50–64 yr (n = 60), and >64 yr (n = 26)], there were again significant relationships between FT₄ and log₁₀ TSH (P value <0.0001 in each group). The slopes of the relationships were not statistically different between age groups; neither were the intercepts on the y-axis statistically different (see inset in Fig. 2).

Figure 2.

Relationship between FT₄ and TSH for all hypothyroid patients (spontaneous and iatrogenic causes combined) according to age. For all patients together, the Spearman correlation coefficient = −0.681, P value <0.0001. Most of the values for the group with iatrogenic hypothyroidism fall within the outlined circle. The inset shows the slopes within the four age groups: dots and dashes, under 35 yr; dashes, 35–49 yr; dots, 50–64 yr; solid line, over 64 yr.

Discussion

In the present study, we confirm that the magnitude of the pituitary response to hypothyroidism, as judged by the serum TSH concentration, is affected by patient age. In contrast to several previous studies, we found this relationship to be present across the entire age spectrum, including in the middle-aged years. We found this relationship to hold true regardless of whether the etiology of the primary hypothyroidism was spontaneous or iatrogenic. Moreover, the cause of the spontaneous hypothyroidism (Hashimoto’s hypothyroidism vs. other causes) did not appear to obscure the finding of a relationship. As a novel finding, we also showed that the relationship extended to cases of confirmed subclinical hypothyroidism.

We did not find that the mean TSH values seen in cases of spontaneous hypothyroidism were different for different age groups, as has previously been found by others. We speculate that this could be explained by the fact that these patients represent a diverse group in whom other factors such as duration of disease, severity of hypothyroidism, and cause of hypothyroidism may all have influenced the magnitude of the TSH response achieved. Patients with overt hypothyroidism would be expected to have had unrecognized disease for a much longer period than patients with subclinical disease, with patients thus coming to attention at different points in the course of their disease. The inclusion of patients with hypothyroidism caused by surgery, radioiodine therapy, and radiation treatment may have also contributed to differing presentation of hypothyroidism. However, when examining the patients with iatrogenic hypothyroidism, we demonstrated significantly different mean TSH values within the different age groups. We believe this occurred, in contrast to the finding in the group with spontaneous hypothyroidism, because of the controlled conditions of the hypothyroidism, in which all patients underwent the same period of l-T₄ deprivation. This uniformity may have allowed the effect of age to become manifest. It is well documented that there may be a delay in the ability of the pituitary to signal a low T₄ concentration with a high TSH value after previous suppression by high T₄ levels (19,20). This effect did not seem to be evident in our group with iatrogenic hypothyroidism, because there was no relationship between baseline TSH concentrations and hypothyroid TSH levels. However, 79% of our patients were undergoing their initial radioiodine therapy after thyroidectomy, and their most recent TSH value was normal.

TSH is known to be cleared through both hepatic metabolism and renal excretion (21,22). It is interesting that older patients had lower TSH concentrations, despite the fact that based on animal data, both these avenues of clearance would be expected to be reduced in older individuals (23). It is known that increased serum creatinine levels can be a consequence of severe hypothyroidism (24,25,26). This has also been documented in the setting of the iatrogenic hypothyroidism used in the management of thyroid cancer patients, where serum creatinine elevations have been reported in approximately 90% of patients (24,27). Such hypothyroidism-induced declines in renal function have been associated with metformin toxicity (28), lithium toxicity (29), and rhabdomyolysis (30) in case reports. In our study, we were not able to document baseline creatinine levels in our patients, but the serum creatinine concentrations measured when the patients were hypothyroid appeared to be higher than the reported age-related reference ranges (31,32). Similarly, GFR values were lower than those typically seen in euthyroid patients of similar ages (33,34). In our patients, in addition to evidence of decreased renal function with increasing degrees of hypothyroidism, there was also a decline in renal function with age. Because our older patients could potentially have augmentation of their TSH concentrations on the basis of their diminished renal function, this suggests that the attenuation in their pituitary response to hypothyroxinemia may be even more profound than is apparent from our data.

The serum FT₄ concentrations seen in the hypothyroid state in our study were not different between age groups, regardless of the cause of the hypothyroidism. It has also previously been shown that FT₄ levels do not change with age in euthyroid individuals. For example, FT₄ concentrations did not differ between young and old in 22,116 outpatients without thyroid disease in the United States (35) or between young and old in 387 subjects without thyroid disease studied in Europe (36). In addition, FT₄ levels also were not different in younger control groups and a group comprised entirely of healthy centenarians (37). These data would suggest that TSH alterations in response to hypothyroidism are not secondary to subtle alterations in FT₄ levels themselves. The daily l-T₄ dose required to maintain euthyroidism is also known to decrease with age (38,39,40,41,42). It is possible that this reduced requirement with age may be partially due to altered body composition and changes in l-T₄ metabolism or clearance with age. Alternatively, there could be an altered pituitary response to circulating thyroid hormones.

It is unclear from these data whether this phenomenon of age-related decrease in TSH concentrations is due to altered responsiveness of the pituitary to reduced FT₄ concentrations, a modified pituitary set point, a reduced capacity for production of TSH, or a combination of these factors. Each of these changes could either represent a normal adaptation to age or a pathological change with its onset in older age groups. For example, there could be less vigorous secretion of TSH in older individuals due to a protective down-regulation of the pituitary-thyroid axis. Alternatively, there could be impaired production of TSH due to pathological deterioration of pituitary function or reserve with age. It has been shown that TSH pulse amplitude is decreased at night in the elderly subjects compared with younger subjects, suggestive of alteration of the pituitary-thyroid axis with age (43). TSH responses to TRH have been reported to be normal in the elderly in some studies (43) but diminished in other studies (36).

TSH reference intervals are impacted by age, with a shift to higher TSH concentrations in older age groups (16,35). These latter findings, if they imply that higher TSH concentrations are normal in older age groups, might also suggest that a lesser TSH response to hypothyroxinemia in older individuals may be protective. This would certainly be in concert with the findings that higher TSH concentrations are associated with longevity in older individuals (44). An alternative contradictory or opposite conclusion that could be drawn from our data are that the serum TSH may not completely reflect the degree of hypothyroxinemia experienced by older patients.

The major shortcomings of this study are due to its retrospective design. In addition to the statistical limitations inherent in any retrospective study, the design of the study predicated that assays from four different clinical laboratories were run at different points in time and were then pooled. This is obviously not an ideal situation and could be avoided if funding was secured to conduct a prospective study using uniform assays for both TSH and FT₄ measurement. A prospective study would also have permitted tighter processing of patients and more sophisticated manipulations to examine whether TSH levels tended to plateau, or reach a plateau earlier, in different age groups. This would have increased the robustness of the findings. As a counterpoint to these weaknesses, however, the strengths of the study are that subjects had confirmed hypothyroidism of two different causes (spontaneous and iatrogenic) and were of ages spanning the age spectrum. In addition, we included patients with subclinical hypothyroidism and were able to collect fairly extensive clinical data on our subjects.

The clinical consequences of these findings are clear with respect to the l-T₄ withdrawal protocols used to prepare thyroid cancer patients for diagnostic scanning or radioiodine therapy. If this method of preparation is selected, the patient’s age should be considered, because the period of withdrawal, and attendant symptomatic hypothyroidism, can likely be considerably shortened in younger patients, with the desired TSH concentration of 30 mIU/liter (11) still being achieved. With respect to interpreting thyroid function tests in cases of spontaneous hypothyroidism, it is apparent that in older individuals, a particular TSH value may be associated with a greater degree of hypothyroxinemia. Thus, the possibility of significant hypothyroxinemia in older individuals should be entertained despite a modest degree of TSH elevation. However, the implication of this observation for the treatment of hypothyroidism in older patients has yet to be determined.