Multiple Hormonal Deficiencies in Anabolic Hormones Are Found in Frail Older Women: The Women's Health and Aging Studies

Abstract

Background

Alterations in anabolic hormones are theorized to contribute to aging and frailty, with most studies focusing on the relationship between individual hormones and specific age-associated diseases. We hypothesized that associations with frailty would most likely manifest in the presence of deficits in multiple anabolic hormones.

Methods

The relationships of serum levels of total IGF-1, DHEAS, and free testosterone (T) with frailty status (nonfrail, prefrail, or frail) were analyzed in 494 women aged 70–79 years enrolled in the Women's Health and Aging Studies I or II. Using multivariate polytomous regression, we calculated the odds of frailty for deficiency in each hormone (defined as the bottom quartile of the hormone) individually, as well as for a count of the hormones.

Results

For each hormone, in adjusted analyses, those with the deficiency were more likely to be frail than those without the deficiency, although this did not achieve statistical significance (IGF-1: odds ratio [OR] 1.82, confidence interval [CI] 0.81–4.08; DHEAS: OR 1.68, CI 0.77–3.69; free T: OR 2.03, CI 0.89–4.64). Compared with those with no hormonal deficiencies, those with one deficiency were not more likely to be frail (OR 1.15, CI 0.49–2.68), whereas those with two or three deficiencies had a very high likelihood of being frail (OR 2.79, CI 1.06–7.32), in adjusted models.

Conclusions

The absolute burden of anabolic hormonal deficiencies is a stronger predictor of frailty status than the type of hormonal deficiency, and the relationship is nonlinear. These analyses suggest generalized endocrine dysfunction in the frailty syndrome.

AGE-RELATED changes have been described in several endocrine pathways under hypothalamic-pituitary control: a decline in output from the growth hormone insulin-like growth factor-1 (IGF-1) axis, decreased dehydroepiansterone-sulfate (DHEAS) output from the adrenal axis, and declines in the sex steroids estradiol and testosterone (T) from the gonadal axis (1). Indeed, individual “syndromes” have been named after alterations in each of these pathways, with significant overlap in the postulated adverse effects, ranging from changes in body composition to effects on overall well-being. Because of the adverse sequelae related to sarcopenia, much research has concentrated on the anabolic hormones IGF-1, DHEAS, and T. Observational studies focusing on the relationship between individual hormones and isolated physiological systems have been performed. Some studies have shown adverse associations between lower hormonal levels and clinical outcomes, but these associations have not been consistent across studies (2–5). Clinical replacement trials of growth hormone, DHEA, and T in healthy older people have had largely disappointing results (6–10), leaving the field with lingering unanswered questions about the clinical impact of age-related hormonal changes.

One explanation for the lack of consensus is that the underlying paradigm of the one-hormonal-deficiency-one-replacement-hormone model may not be the best model in an older person, particularly in the subgroup of older people who are experiencing poor health, frailty, or advanced old age (11). Frailty represents a biological syndrome of increased vulnerability to stressors that results from decreased physiological reserves (12,13). The physiological underpinnings of frailty are not well defined but are likely to include an accelerated decline in function across multiple physiological systems. In support of this, we and others have reported the association between individual factors, such as cortisol diurnal rhythm, inflammatory factors, micronutrient deficiencies, and anemia, and the frailty syndrome (14–18).

We sought to examine the associations of individual anabolic hormonal deficiencies from the growth hormone, adrenal, and gonadal axes with frailty, and to assess their combined effects as well. We had three hypotheses: frail women would be more likely to have relative deficiencies, individually, in the hormones IGF-1, DHEAS, and free T than nonfrail women; frail women would be more likely to exhibit multiple deficiencies than nonfrail women; and the number of hormonal deficiencies would be a stronger predictor of frailty status than the type of hormonal deficiency. Although the role of T in older women is controversial, we included this hormone in our analyses because it is a hormone whose anabolic properties have been well studied in men, and we sought to examine its association with frailty in women.

METHODS

Participants

Study participants were community-dwelling women 70–79 years of age who were enrolled in one of two longitudinal, population-based companion studies: Women's Health and Aging Studies (WHAS) I and II (19,20). Women from both studies were recruited from a random sample of community-dwelling women 65 years or older selected from the Health Care Financing Administration's Medicare eligibility list for Baltimore, MD. Women who reported difficulty with one or more tasks in two or more of four domains of functioning (mobility or exercise tolerance, upper extremity activities, basic self-care, and household management tasks) were eligible for WHAS I. In 1992, 1,002 women representative of the one third most disabled community-dwelling older women enrolled. WHAS II was designed to be a companion study to WHAS I and comprised a cohort of women aged 70–79 with difficulty in zero or one domain of physical function. In 1994, 436 women representative of the two thirds least disabled older women living in the community enrolled. Aside from disability status, the studies’ eligibility criteria were identical, except that WHAS I required a Mini-Mental State Examination (21) score above 17, whereas WHAS II required a score above 23. All evaluations, interviews, and physical examinations were conducted using the same rigorously standardized methods. The Johns Hopkins University Institutional Review Board approved both studies, and all participants gave informed consent.

Biochemical Measurements

Blood samples were collected between 9 am and 2 pm in a nonfasting state, processed, frozen, and sent the same day to Quest Diagnostic Laboratories (Teterboro, NJ). The serum total IGF-1 level was measured using a radioimmunoassay with ethanol extraction (Nichols Institute Diagnostics, San Juan Capistrano, CA). The overall coefficient of variation (CV) was less than 10% and the assay sensitivity 0.1 μg/L. DHEAS, total T, and sex hormone–binding globulin (SHBG) assays were later performed in duplicate on additional serum stored at −80°C. DHEAS was measured by enzyme-linked immunosorbent assay (American Laboratory Products, Co, Windham, NH). The interassay CV was 3.3%; assay sensitivity was 0.02 μg/L. Total T was measured using liquid chromatography with mass spectrometry after nonpolar solvent extraction, and SHBG was measured by immunoradiometric assay (Esoterix, Inc, Calabasas Hills, CA). The minimum detection limit for total T was 3 ng/dL and intra-assay CV was 5.7% at low values and 1.7% at high values. The intra-assay CV was 2.4% for SHBG. Free T was estimated from total T, SHBG, and albumin levels using established methods based on the law of mass action (22).

Frailty Status

Frailty status was defined as originally operationalized by Fried and associates (12) in the Cardiovascular Health Study (CHS) and validated by Bandeen-Roche and associates (13) in the WHAS. Five characteristics of frailty were used: shrinking (body mass index [BMI] <18.5 kg/m² or lost ≥10% of weight since age 60), weakness (grip strength equivalent to the lowest quartile in CHS, by gender and BMI strata), poor endurance (self-report of exhaustion), slowness (walking speed equivalent to the lowest quartile in CHS, by height), and low activity (activity level in kcal/wk equivalent to the lowest quartile in CHS). Those with none of the five characteristics were considered to be nonfrail, those with one or two were deemed prefrail, and those with three, four, or five were considered to be frail.

Covariates

Sociodemographic characteristics included age, race, education, and smoking status. Current cigarette smoking was assessed, and the participant was categorized as a non-, former, or current smoker (19). BMI (kg/m²) was computed from objective measures. Seventeen chronic diseases were ascertained at baseline with disease-specific standardized algorithms (19). Disease categories used in analysis included a count of the following diseases: congestive heart failure, diabetes mellitus, peripheral arterial disease, stroke, coronary heart disease (angina or myocardial infarction), chronic obstructive pulmonary disease, hip fracture, osteoarthritis, rheumatoid arthritis, Parkinson's disease, and malignant neoplasms (excluding basal cell cancer).

Statistical Analysis

To be included in these analyses, participants were required to have IGF-1, DHEAS, total T, SHBG, and albumin measurement and frailty status at the initial study visit (N = 494). There were no women taking preparations that contained T or DHEA at that study visit. Baseline characteristics were compared between WHAS I and II, using the chi-square test for binary outcomes and student's t tests for continuous outcomes. To appropriately reference inferences derived from the combined data back to the sampling population of community-dwelling women aged 70–79 years, study-specific probability weights were incorporated into all comparisons and regression analyses (19,23). Due to the skewed distribution of all the hormonal data, geometric means were used to describe all hormonal levels. Because there are no commonly used cutoffs to describe hormonal deficiency in this age group, we used the lowest quartile of each hormone as the cut point for defining risk groups in regression analyses. This is congruent with other studies in which we, and others, have examined risks associated with relative hormonal deficiency in older individuals (24–26). Geometric means and prevalences of hormonal deficiency were compared by frailty status, using trend tests to assess statistical significance. We also examined the prevalence of abnormal hormones (0–3) by frailty status. We subsequently performed polytomous logistic regression analyses to examine the relationship between individual hormones and frailty status, using nonfrail women as the reference group. All the regression models included adjustment for age, education, race, smoking status, BMI, oral estrogen use, oral corticosteroid use, and a count of baseline adjudicated chronic conditions. Additional analyses were performed to examine the relationship between the number of hormonal deficiencies and frailty status. Due to the small number of women with all three deficiencies (n = 18), those with two or three deficiencies were categorized as one group. Analyses were performed using Stata/SE version 9.2 for Windows.

RESULTS

As shown in Table 1, women aged 70–79 enrolled in WHAS I were less likely to be white and to have higher education compared with women of the same age range in WHAS II. They also carried a greater disease burden, had a higher BMI, reported worse health, and were more likely to be frail than their less disabled counterparts. Oral corticosteroid and estrogen use were low in both studies.

Table 1.

Baseline Characteristics by Study (N = 494)

Characteristic	WHAS I (age 70–79), N = 176	WHAS II (age 70–79), N = 318	p value
Age (y), mean (SD)	74.2 (2.7)	73.8 (2.7)	.11
White race (%)	68.8	82.7	<.01
Education (grade), mean (SD)	10.1 (7.0)	12.8 (3.2)	<.01
Smoking status (%)
Nonsmoker	42.6	53.5	.06
Former smoker	42.6	35.9
Current smoker	14.8	10.7
BMI (kg/m2), mean (SD)	29.2 (7.8)	26.6 (5.1)	<.01
Number of diseases, mean (SD)	1.9 (1.4)	0.9 (0.9)	<.01
Self-reported health (%)
Excellent	2.8	15.6	<.01
Very good/good	45.4	74.3
Fair/poor	51.8	10.2
Corticosteroid use (%)	5.1	0.9	.01
Estrogen use (%)	9.1	13.5	.19
Frailty (%)
Frail	26.1	2.5	<.01
Prefrail	61.9	33.3
Nonfrail	11.9	64.2

Note: SD = standard deviation; WHAS = Women's Health and Aging Study; BMI = body mass index.

When the weighted geometric mean level of each individual hormone—IGF-1, DHEAS, and free T—was assessed by frailty status, frail women had lower average hormonal levels of each hormone than nonfrail women, with intermediate levels in prefrail women (Table 2). We subsequently defined the bottom quartile of each hormone as indicative of relative deficiency. In Figure 1, the prevalence of relative hormonal deficiency is shown by frailty status. For each hormone studied, there was a stepwise increase in the prevalence of hormonal deficiency with increasing frailty burden.

Table 2.

Weighted Geometric Mean (SD) of Hormone Levels by Frailty Status

Hormone	Nonfrail (N = 225)	Prefrail (N = 215)	Frail (N = 54)	p Value
IGF-1 (μg/mL)	114.5 (1.5)	111.5 (1.7)	104.2 (1.6)	.18
DHEAS (μg/dL)	0.41 (2.21)	0.31 (2.43)	0.28 (2.72)	<.01
Free testosterone (pg/mL)	1.16 (2.42)	1.14 (2.24)	0.88 (3.02)	.09

Note: SD = standard deviation; IGF-1 = insulin-like growth factor-1; DHEAS = dehydroepiandrosterone-sulfate.

Figure 1.

Prevalence of hormone deficiency by frailty status.

In adjusted analyses examining each hormone individually, women with relative hormonal deficiencies were more likely to be frail than nonfrail, although none of these relationships achieved statistical significance (IGF-1: odds ratio [OR] 1.82, confidence interval [CI] 0.81–4.08; DHEAS: OR 1.68, CI 0.77–3.69; free T: OR 2.03, CI 0.89–4.64) (Table 3). Likewise, women in the bottom quartile were more likely to be in the prefrail group than the nonfrail group, achieving statistical significance for IGF-1 and DHEAS (IGF-1: OR 1.89, CI 1.10–3.26; DHEAS: OR 1.70, CI 1.93–3.10; free T: OR 1.40, CI 0.78–2.52).

Table 3.

Odds ratio of Individual Hormones Predicting Frailty Status

	Age Adjusted Odds Ratio		Multivariate Adjusted Odds Ratio*
Hormone	Prefrail vs Nonfrail	Frail vs Nonfrail	Prefrail vs Nonfrail	Frail vs Nonfrail
IGF-1 deficiency† (<87.8 μg/L)	1.50 (0.95–2.37)	1.75 (0.87–3.50)	1.89 (1.10–3.26)	1.82 (0.81–4.08)
DHEAS deficiency† (<0.20 μg/L)	1.70 (1.04–2.78)	2.22 (1.10–4.48)	1.70 (0.93–3.10)	1.68 (0.77–3.69)
Free testosterone deficiency† (<0.7 pg/mL)	1.07 (0.66–1.74)	1.80 (0.90–3.58)	1.40 (0.78–2.52)	2.03 (0.89–4.64)

Notes: IGF-1 = insulin-like growth factor-1; DHEAS = dehydroepiandrosterone-sulfate.

^*Adjusted for age, race, education, smoking status, body mass index, number of diseases, corticosteroid use, and estrogen use (n = 485 with complete data on covariates).

^†For each hormone, deficiency was defined as the bottom quartile.

When the data were examined using a count of abnormal hormone levels, women in the nonfrail group were more likely to have no hormonal abnormalities, whereas women in the frail group were more likely to have two or three hormonal abnormalities (p = .06 for chi-square test of independence) (Figure 2). In adjusted analyses examining the ability of the hormonal deficiency count to predict frailty, women with one hormonal deficiency were not more likely to be frail than women with no hormonal deficiencies (OR 1.15, CI 0.49–2.68) (Table 4). However, women with two or more deficiencies were significantly more likely to be frail than their counterparts with no hormonal deficiencies (OR 2.79, CI 1.06–7.32), suggesting a nonlinear relationship between the burden of hormonal deficiencies and frailty. The relationship between hormonal deficiency and prefrail status was somewhat different. Those with one hormonal deficiency were more likely to be prefrail than those with no hormonal deficiencies (OR 1.71, CI 1.03–2.85), with a similar likelihood of prefrail status in those having two or three deficiencies compared with those with no deficiencies (OR 2.25, CI 1.12–4.53).

Figure 2.

Number of abnormal hormones by frailty status.

Table 4.

Odds ratio of Number of Hormonal Deficiencies* Predicting Frailty Status

	Age adjusted odds ratio		Multivariate adjusted odds ratio†
Number	Prefrail vs nonfrail	Frail vs nonfrail	Prefrail vs nonfrail	Frail vs nonfrail
0	1	1	1	1
1	1.55 (0.98–2.45)	1.18 (0.53–2.64)	1.71 (1.03–2.85)	1.15 (0.49–2.68)
2 or 3	1.61 (0.93–2.79)	2.73 (1.28–5.86)	2.25 (1.12–4.53)	2.79 (1.06–7.32)

Notes: *For each hormone, deficiency was defined as the bottom quartile.

^†Adjusted for age, race, education, smoking status, body mass index, number of diseases, corticosteroid use, and estrogen use (n = 485 with complete data on covariates).

DISCUSSION

Our data demonstrate that both prefrail and frail older women tend to have lower levels of anabolic hormones and are more likely to exhibit relative deficiency in multiple hormones than similarly aged nonfrail counterparts. They also show that the aggregate burden of hormonal deficiencies is an independent predictor of frailty and that the hormonal burden is more strongly associated with frailty than the type of hormonal deficiency is. Interestingly, the hormones studied were from three distinctly regulated hormonal axes that in younger individuals fail simultaneously only in rare cases of pituitary dysfunction. Furthermore, IGF-1 and T have different receptors, each present in muscle, and a DHEAS receptor has never been cloned. However, there may be a biological link among these hormones beyond their individual axes, as muscle IGF-1 production is affected by T and DHEAS in part acts through its conversion to androgens (27).

The etiology of the gradual decline in production of IGF-1, DHEAS, and T with increasing age is not known, although population-based data suggest that the onset in our study population predated our hormonal measurements by more than 50 years (1). Within the context of this overall decline, we were able to show that a relative deficiency in each hormone could be defined.

Our data also suggest that the regulation of hormonal axes differs between prefrail and frail women. There was little difference between prefrail and frail women in the magnitude of the association between individual deficiencies in IGF-1 or DHEAS and frailty status, suggesting that these relative deficiencies do not, individually, distinguish prefrail women from frail women. In addition, in the prefrail women, the magnitude of association differed little by hormonal burden. In contrast, in the frail women, the magnitude of association for any one abnormality was substantially smaller than that seen for any of the individual deficiencies, whereas the magnitude of association for two or three abnormalities was substantially greater than that seen for any of the individual deficiencies. This nonlinearity suggests that it is not simply a dose-response effect occurring but rather that a threshold has been crossed of multisystem dysregulation and disrupted homeostasis.

We are aware of only two studies that have examined the relationship between any of the hormones analyzed in our study—IGF-1, DHEAS, or T—and the frailty syndrome in women, either singly or in combination. In the Longitudinal Aging Study Amsterdam, low IGF-1 levels were associated with a greater number of frailty indicators (28). In a small case-control study, Leng and associates (29) found that frail women had lower levels of IGF-1 and DHEAS than nonfrail women.

In men, one study has examined the association between T and frailty, and several others have examined the relationship between multiple hormonal abnormalities and functional outcomes. In the Massachusetts Male Aging Study, there was no relationship between free T level and frailty (30). In a study of exceptionally healthy men, Morley and associates (31) found that bioavailable T, DHEAS, and IGF-1/GH were each associated with functional measures. Recent analyses in men enrolled in the invecchoiare in CHIANTI study, using the same three anabolic hormones as we did, showed a stepwise association between the number of hormonal deficiencies and mortality in older men, even after adjustment for comorbidity (26). Of note, they did not see an association between any of these hormones individually and mortality. Similarly, in a study of men affected by congestive heart failure, those with deficiencies in all three anabolic hormones had the lowest 3-year survival (27% vs 83% in those with no deficiencies) (32). Analogous studies examining more than two anabolic hormones and any outcome have not been performed in women.

Strengths of our study include number of women studied, their representativeness of women aged 70–79 across the full spectrum of function and health in the community, the quality of the assays used, and the wealth of data available for covariate adjustment. Our major limitation is that we cannot determine in this cross-sectional analysis whether frailty results from or initiates the disruption in multiple hormonal axes, although our model of frailty would suggest that both could occur in a vicious cycle (12). In addition, our analyses were performed in women only, limiting our ability to generalize to frail men.

The therapeutic implications of our data cannot be established from this study. Data from the surgical literature suggest that in response to acute physical stress, there is an appropriate compensatory decline in anabolic hormones (33,34), to divert energy away from building muscle and toward healing. It is possible that in those predisposed to frailty, after an acute illness or other stressor, there is failure to return to baseline levels of anabolic hormones, or that the severity of coexistent diseases elicits a prolonged lowering of anabolic hormones. This interpretation would suggest that these lower levels of anabolic hormones are a marker of a physiological state of stress. Alternatively, given the known mechanisms of each of these hormones in building muscle, a reasonable hypothesis is that age-related low levels of anabolic hormones contribute over time to sarcopenia and frailty (27,35). If so, our data suggest that multiple small effects in aggregate lead to adverse clinical sequelae and that, if replacement was to occur, it would require low doses of multiple anabolic hormones. An added benefit to this approach would be fewer side effects from the use of low hormone doses. Furthermore, our data suggest that the benefits of multiple anabolic hormone replacement might have a synergistic effect exclusively in frail women. As a next step, confirmation of the joint contributions of dysregulation in multiple anabolic hormones to adverse outcomes should be performed in other study populations of older women.