Testosterone therapy effects adipose distribution in older females post hip-fracture: The STEP-HI study

Abstract

Background

With aging and injury, females experience ectopic redistribution of appendicular adipose tissue (AAT) into the visceral compartment, where adipose tissue (VAT) becomes highly inflammatory and increases risk of reinjury and chronic illness. Therefore, strategies that can disrupt this unhealthy adipose redistribution after hip fracture injury are of great interest. We examined the effects of testosterone therapy on total adipose tissue (TAT) and adipose distribution in older females recovering from hip fracture.

Methods

This was a sub-analysis of the STEP-HI study, a multi-site randomized clinical trial in which older females recovering from hip fracture were assigned to a 6-month exercise intervention combined with either topical testosterone gel (EX + T, n = 35, age = 79 ± 9 years) or placebo gel (EX + P, n = 31, age = 76 ± 7 years). Changes in TAT, AAT, and VAT mass and percentage of TAT in each region (%AAT and %VAT) were measured using dual x-ray absorptiometry, and changes over the 6-month intervention were compared between groups.

Results

Over the intervention, changes were similar in TAT (EX + P: 298 ± 2002 g, EX + T: 419 ± 2086 g, p = 0.810), AAT (EX + P: 52 ± 1007 g, EX + T: 39 ± 1078 g, p = 0.810), %AAT (EX + P: 0.42 ± 1.40% of TAT, EX + T: 0.52 ± 1.67% of TAT, p = 0.792) and VAT (EX + P: 45 ± 232 g; EX + T: −44 ± 151 g; p = 0.073). However, relative changes in %VAT from pre-intervention (EX + P: Δ3.51 ± 18.42%; EX + T: −Δ10.57 ± 17.13%; p = 0.004) marked favorable effects of testosterone on relative visceral adiposity.

Conclusion

While testosterone did not decrease overall adipose stores compared to exercise alone in older females recovering from hip fracture, it did promote a healthy pattern of adipose distribution away from the viscera.

Clinical trial #

NCT02938923.

Graphical abstract

1. Introduction

Growing rates of obesity are globally impacting public health issues across the lifespan [1]. Aging is associated with both increases in total body adiposity and a redistribution of adipose from more innocuous subcutaneous stores ectopically into the metabolically stressful visceral compartment, increasing the risk of many obesity-related illnesses [2,3]. The increases in total adiposity and a redistribution of adipose into the visceral compartment observed with aging have been linked to decreases in sex hormones during both menopause (estrogen and progesterone) and andropause (testosterone) [4,5]. Such changes in body composition are rapidly compounded in older females after suffering a hip fracture and significantly increase their risk of reinjury and rehospitalization [[6], [7], [8], [9]]. Recently, a multi-site clinical trial in older females recovering from hip fracture investigated the combined effects of pairing a structured exercise program with exogenous testosterone on functional recovery and changes in muscle and bone mass; however, changes in total adiposity and adipose distribution were not reported [10]. Therefore, the purpose of the present study is to compare changes in total adipose tissue (TAT), visceral adipose tissue (VAT), and distribution of adipose in the visceral compartment (%VAT) observed during this multi-site clinical trial to examine the effects of exercise with and without testosterone supplementation on total body adiposity and adipose distribution.

Hip fractures are 2.9 times more prevalent in females than males and represent one of the leading causes for loss of independence in older females [11,12]. Low lean tissue (muscle and bone mass) and high adiposity contribute to an increased risk of falls and hip fractures [[13], [14], [15]]. People who are recovering from a hip fracture commonly experience rapid unfavorable alterations in body composition due to decreased physical activity, loss of function, and pain throughout the recovery process [6,7]. This not only increases the risk of a subsequent hip fracture but also increases the risk of other injuries, illnesses, frailty, disability, and death [8,9]. The standard of care for recovering from hip fracture involves physical therapy, but such treatments fail to elicit full recovery of function in nearly two-thirds of these females [[16], [17], [18]]. The additive effect of exogenous testosterone with physical therapy for recovery from hip fracture has gained little attention in older females despite testosterone's dual roles as both an anabolic and lipolytic hormone [19,20]. Nonetheless, despite strong biological plausibility, few studies have evaluated testosterone use in older females. Furthermore, none have included structured resistance training despite the combined beneficial effects of these two therapies that has been observed in younger populations and older males [10,[21], [22], [23], [24]].

In addition to its anabolic and lipolytic function, testosterone has also been shown to play an important role in adipose distribution [2,4,5,25]. In males, circulating testosterone concentrations are associated with a healthy pattern of adipose distribution with decreased relative VAT (%VAT). Decreasing %VAT creates a healthier pattern of adipose distribution as VAT is more inflammatory and metabolically stressful than the subcutaneous adipose stores, which make up the majority of TAT [26]. Furthermore, measurements of relative visceral adiposity are of clinical interest given that they been shown to be stronger predictors of cardiometabolic health and cardiovascular disease risk than TAT or VAT alone [27,28]. Additionally, relative visceral adiposity measurements have been noted to be predictive of survivability in certain medical conditions [29,30], however, their relationship to survivability in older females recovering from hip fracture is currently unknown. Testosterone is believed to reduce %VAT because VAT stores are more sensitive to testosterone than subcutaneous adipose tissue stores due to their higher androgen receptor content and greater cellularity, innervation and blood flow [[31], [32], [33]]. These differences in regional adipose stores are supported by observations testosterone's effects on increased lipolytic activity and reduced adipocyte differentiation are greater in VAT than subcutaneous adipose tissue [33].

In older males, age-related reductions in circulating testosterone levels have been proposed as a primary cause in of increased VAT and %VAT that is observed with aging and exogenous testosterone supplementation has been shown to encourage selective decreases in VAT that improves adipose distribution [34,35]. Similarly, females experience a detrimental pattern of adipose redistribution with aging, particularly during and after menopause, that results in increased VAT. However, in females these detrimental patterns of adipose redistribution can be attributed to decreases in estrogen, and the effects of testosterone are unknown [10,36,37]. In observational studies of post-menopausal females, endogenous testosterone has demonstrated beneficial correlations to various cardiovascular health biomarkers as well as reduced incidence of experiencing a major cardiac event, however the effects of exogenous testosterone on cardiovascular health and disease risk are presently unknown as is the role that adipose distribution plays in these risk [[38], [39], [40]].

While testosterone is believed to play a similar lipolytic role in females as in males, previous studies have observed that in adult females, biologically available testosterone is positively associated with VAT [[41], [42], [43]]. Furthermore, females presenting with elevated testosterone due to polycystic ovarian syndrome and females undertaking testosterone therapy to transition to males have both presented with selective increases in VAT, suggesting a possible differential role of testosterone on adipose distribution in females and males [44,45]. These sex-specific differences in the effects of testosterone have been attributed to differing roles that testosterone and estrogen play within the hypothalamic-pituitary-gonadal pathway and potential conversion of testosterone into estradiol via aromatase [32,33,46]. However, as these processes vary with aging, the effects of testosterone on adipose distribution may differ across age groups [46]. Unfortunately, research into the effects of testosterone in older females is limited, with one study in post-menopausal females reporting that 3 months of testosterone use decreased lipolytic signaling in subcutaneous adipose tissue that was not observed in the comparison group given estradiol, suggesting that testosterone differentially influences regional adipose stores [47]. However, in this study, similar changes were seen in both groups in body mass, and no measures of VAT, adipose distribution or lean mass were provided [35]. Furthermore, no studies have directly observed the effects of testosterone supplementation in post-menopausal women recovering from a hip fracture, nor how testosterone replacement may interact with a structured exercise program.

Because of the potential benefit of testosterone therapy paired with exercise in older females recovering from hip fracture, it is important to understand how this therapy may affect visceral adipose accumulation and distribution in this population. This sub-study performed secondary analysis of deidentified data from the “Starting a Testosterone and Exercise Program after Hip Injury” (STEP-HI, NCT#02938923) study [10], to examine the effects of exercise with and without testosterone therapy on adipose distribution in older females recovering from a hip fracture. We hypothesized that while both exercise groups would experience a reduction in TAT and VAT, females receiving testosterone therapy would experience a greater reduction in VAT and a healthier pattern of adipose distribution as evidenced by reduced %VAT.

2. Methods

2.1. Overview of study design

The present study is a sub-analysis of deidentified data from the STEP-HI Study, a multi-site, phase three, double-blinded, placebo-controlled randomized clinical trial that was completed across eight sites in the United States between December 2018 and August 2023 (NCT02938923). The aim of the STEP-HI study was to evaluate the effects of supervised exercise training combined with testosterone therapy on functional outcomes in older women recovering from hip fracture. More specifically, the study randomized participants to either usual care for recovery from hip fracture, which included a home exercise program, or a supervised exercise program (twice a week in-person multi-modal exercise) and either a topical testosterone gel (EX + T) or a placebo gel (EX + P). The detailed methodology and primary outcomes of this trial have previously been published and are briefly described below [10,48]. Our analytic plan was approved by the STEP-HI publication committee and involves both data and analyses which have not been previously reported from the two supervised exercise cohorts. The enhanced usual care cohort included in the original study was not included in the present study because it was not directly related to the proposed research question and had substantially smaller sample size than the other intervention groups. The STEP-HI study was approved by the Institutional Review Board at Washington University in St. Louis (IRB#: 201609077) with facilitated review approved by the Institutional Review Board at the University of Connecticut Health Center (IRB#: 018-075-2).

2.2. Participants

Females aged 65 or older who were recovering from a recent hip fracture were recruited from eight clinical research sites across Northeastern, Central, Western, and Southern regions of the United States [10]. Prior to enrollment in the parent clinical trial, participants were screened for inclusion and exclusion criteria. Inclusion criteria were older females who were within 22 weeks of their surgical repair date, were either community dwelling or in assisted living prior to their hip fracture, were mildly to moderately frail as indicated by a modified Physical Performance Test (mPPT) score between 12 and 28, and had a baseline total serum testosterone level of 60 ng/dL or less. Exclusion criteria included having cognitive impairment as determined by a Short Blessed Test score greater than 10, being unable to provide informed consent for any reason, and any other factors that could have either affected participant safety or interfered with study participation by discretion of the researcher [49].

For this sub-study, additional inclusion criteria were: randomization to a supervised exercise group and availability of dual x-ray absorptiometry (DXA)- reported VAT, as this was a primary outcome of the study. Across the eight clinical trial sites, three different DXA systems were used (four sites used the Hologic Horizon A DXA System, two sites used the Lunar GE iDXA system, and two sites used the GE iDXA system). Of these, the two sites using the GE iDXA system were excluded due to VAT measurements not being included in the reports. Thus, this sub-study consisted of six of the eight total sites from the main STEP-HI study [10].

Participants’ characteristics [age, height, body mass, race and ethnicity, days removed from surgery to baseline testing, physical function using the score from the Short Physical Performance Battery (SPPB), mPPT, Functional Status Questionnaire (FSQ), and total number of comorbidities] at baseline assessment are reported in Table 1 [10].

Table 1.

Descriptive table of participants characteristics and various outcomes by intervention groups.

Characteristic	EX + T N = 35	EX + P N = 31	p-value [1]
Age (years)	79 (9); 77 (70, 86)	76 (7); 76 (70, 83)	0.253
Ethnicity			0.318
Hispanic or Latino	1 (2.9%)	2 (6.5%)
Not Hispanic or Latino	34 (97%)	27 (87%)
Unknown	0 (0%)	2 (6.5%)
Race			0.205
Black or African American	1 (2.9%)	3 (9.7%)
White	34 (97.1%)	27 (87.1%)
Other	0 (0%)	1 (3.2%)
Comorbidities	4 (2); 4 (2, 6)	5 (3); 5 (3, 7)	0.166
Time from Surgery to Baseline Testing (days)	93 (28); 92 (69, 107)	109 (31); 98 (88, 132)	0.076
SPPB	7 (2); 7 (6, 9)	6 (2); 7 (5, 8)	0.302
mPPT	21 (7); 22 (18, 27)	21 (7); 22 (16, 25)	0.714
FSQ Score	24 (5); 24 (21, 28)	21 (5); 23 (18, 26)	0.049∗
DXA (Lunar or GE)			0.805
Hologic iDXA	20 (57%)	16 (52%)
Lunar iDXA	15 (43%)	15 (48%)
Height (cm)	160 (7); 158 (155, 165)	161 (7); 160 (154, 166)	0.463
Body Mass (kg)	64 (14); 65 (54, 72)	70 (15); 69 (58, 78)	0.114
BMI (kg/m2)	25.2 (5.2); 25.5 (20.8, 27.0)	27.3 (5.8); 26.5 (23.8, 29.9)	0.140

Data presented as mean (standard deviation); median (Q1, Q3) for continuous variables and count (percentage) for categorical variables. SPPB = Short Physical Performance Battery, mPPT = Modified Physical Performance Battery, FSQ=Functional status questionnaire, DXA = Dual X-Ray Absorptiometry, BMI=Body mass index. [1]Wilcoxon rank sum test; Fisher's Exact Test for Count Data with simulated p-value (based on 2000 replicates); Fisher's exact test; Wilcoxon rank sum exact test. ∗ Indicates statistical significance between conditions (P < 0.05).

2.3. Body composition assessment

Body composition was assessed using DXA prior to and at the end of the 24-week intervention. DXA imaging body composition reports were generated at each clinical trial location from manufacturer provided algorithms and were centrally managed by the Investigations in Metabolism, Aging, Gender and Exercise (IMAGE) center at the University of Colorado Denver who also ensured identical procedures, oversaw credentials of DXA technicians, and performed image reviews. From each scan, body mass, %BF, TAT, VAT, and appendicular adipose tissue (AAT) were recorded. For VAT and AAT which are regional adipose measures both absolute and relative values as a percentage of TAT were reported (i.e. %VAT and %AAT).

2.4. Exercise program

All participants in this sub-study took part in a supervised exercise program consisting of two non-consecutive days of multimodal exercise each week over the 24-week intervention. Exercise sessions consisted of functional movements, mobility and balance exercises, and after the first month, high-intensity strength training exercises. In addition to the supervised exercise, participants were also expected to perform home exercises three other days each week, consisting of progressive walking challenges and three resistance exercises. Exercise adherence did not differ significantly between the EX + T and EX + P groups. Session attendance and home-exercise completion were monitored at all sites, and adherence was comparable across conditions, reducing the likelihood that differences in adipose distribution were driven by differential exposure to the exercise intervention.

2.5. Testosterone and placebo gel interventions

As described previously, participants were randomly assigned to either testosterone (EX + T) or placebo (EX + P) gel with both participants and researchers blinded to their assigned intervention [10]. The testosterone intervention used a generic 1.0% testosterone gel; the placebo gel was chemically identical but without the testosterone. In both conditions participants were provided a topical gel that was self-administered. Serum testosterone levels were assessed in all participants after two weeks and then monthly after the baseline until the completion of the intervention. Dosage of the testosterone gel was adjusted with the goal of achieving serum levels of 110–160 ng/dL (reference range was 12–78 ng/dL) as prescribed by an unblinded physician. To ensure blinding was maintained, whenever a person in the EX + T required a dosage adjustment, a matched adjustment was made to an individual in the EX + P cohort.

2.6. Statistical analysis

A descriptive analysis was conducted to summarize baseline participant characteristics and outcomes at different time points separately for the EX + P and EX + T groups. The two groups were compared with respect to each variable using Wilcoxon rank-sum tests for continuous variables and Fisher's exact tests for categorical variables, given the distributional skewness and sparse cells observed for some variables. Outcome changes that showed approximately bell-shaped distributions were compared between groups using unpaired t-tests. In addition, the mean difference between groups with its 95% confidence interval was plotted for each outcome, along with the observed data at each time point within each group.

Linear mixed effects models were used to model the six-month outcomes, including a random intercept per subject and fixed effects for intervention group, and baseline outcome level. Models were run both unadjusted and adjusted for baseline covariates, including comorbidities, FSQ score, age, and height. Models for AAT and VAT were additionally adjusted for body mass. Both adjusted and unadjusted mean differences between groups at six months were reported, with corresponding 95% confidence intervals and p-values. All tests were two-sided, and statistical significance was evaluated at the 5% level. Analyses were conducted in R version 4.4.2.

3. Results

The entire STEP-HI trial enrolled 104 participants from eight clinical trials sites who were randomized into the supervised exercise groups (53 in the EX + T group and 51 in the EX + P group) and completed physical functioning testing at three months, six months, or both three and six months. For this adiposity-focused analysis, only women randomized to supervised exercise and with complete baseline and 6-month DXA data from scanners capable of reporting VAT were eligible. Because two of the eight sites used GE iDXA systems that do not provide VAT measurements, 28 participants were excluded due solely to scanner capability. An additional 10 participants lacked either baseline or six-month DXA reports and were excluded. These exclusions were entirely measurement-related rather than participant-related, resulting in a final analytic sample of 66 females from the six VAT-capable sites (35 in the EX + T group and 31 in the EX + P group).

Participant characteristics and outcomes at baseline are summarized by group, with between-group comparison results presented in Table 1. No significant differences between groups were observed in baseline participant characteristics (age, race and ethnicity, height, body mass, time removed from surgery, body mass index, number of comorbidities, or DXA model used for analysis). While no differences were observed in direct assessments of physical function via the SPPB or mPPT, self-reported function measured by FSQ Score, significantly differed between groups (EX + T: 25 ± 5 vs. EX + P: 21 ± 5, p = 0.049).

Body composition and adipose distribution values are provided in Table 2. At baseline, no significant differences were present in any selected outcome (body mass, %BF, TAT, AAT, VAT, %AAT, or %VAT). At the 6-month follow-up, only TAT was significantly different between conditions, being greater in EX + P than EX + T.

Table 2.

Descriptive table of participants characteristics and various outcomes by intervention groups.

Characteristic	EX + T (N = 35)	EX + P (N = 31)	p-value [1]
Outcomes at Baseline
Body Mass (kg)	64 (14); 65 (54, 72)	70 (15); 69 (58, 78)	0.114
%BF (%)	38 (8); 39 (32, 44)	41 (7); 43 (39, 46)	0.075
TAT (g)	24,924 (9420); 25,776 (16,554, 28,630)	28,874 (9570); 28,687 (23,902, 36,031)	0.057
AAT (g)	11,896 (4188); 12,247 (7,867, 14,736)	13,349 (4482); 12,973 (9,543, 16,228)	0.191
%AAT (% of TAT)	49 (7); 50 (43, 54)	47 (7); 48 (43, 52)	0.219
VAT (g)	781 (578); 583 (377, 1036)	898 (618); 764 (453, 1163)	0.323
%VAT (% of TAT)	2.99 (1.61); 2.53 (1.85, 3.74)	2.96 (1.56); 2.36 (2.15, 3.86)	0.949
Outcomes at 6-month Follow-up
Body Mass (kg)	66 (15); 65 (54, 74)	71 (15); 72 (59, 82)	0.147
%BF (%)	38 (7); 38 (35, 44)	41 (7); 42 (38, 46)	0.081
TAT (g)	25,343 (9802); 25,057 (17,551, 28,919)	29,172 (9566); 28,540 (23,083, 36,800)	0.049∗
AAT (g)	11,935 (4296); 11,632 (8,628, 14,438)	13,401 (4480); 13,557 (10,093, 16,881)	0.219
%AAT (% of TAT)	48 (8); 50 (42, 54)	47 (7); 47 (41, 51)	0.300
VAT (g)	736 (614); 519 (329, 950)	943 (723); 714 (406, 1303)	0.136
%VAT (% of TAT)	2.74 (1.80); 2.27 (1.75, 3.25)	3.08 (1.81); 2.63 (1.93, 4.00)	0.271
Outcomes Changes from Baseline to 6-month Follow-up
Body Mass (kg)	1.66 (2.86); 1.89 (0.20, 3.65)	0.84 (2.47); 0.98 (−0.60, 2.70)	0.218
%BF (%)	−0.16 (2.19); 0.00 (−2.00, 1.80)	−0.08 (1.76); 0.00 (−1.60, 1.10)	0.861
TAT (g)	419 (2086); 303 (−1,129, 2001)	298 (2002); 377 (−857, 1499)	0.810
AAT (g)	39 (1078); 148 (−615, 553)	52 (1007); −144 (−688, 648)	0.960
%AAT (% of TAT)	−0.52 (1.67); −0.36 (−1.09, 0.11)	−0.42 (1.40); −0.74 (−1.59, 0.80)	0.792
VAT (g)	−44 (151); −48 (−136, 15)	45 (232); 14 (−57, 123)	0.073
%VAT (% of TAT)	−0.26 (0.51); −0.25 (−0.55, 0.02)	0.12 (0.52); 0.03 (−0.29, 0.41)	0.004∗

Data presented as mean (standard deviation), Median (Q1, Q3).

EX + T = Exercise plus Testosterone Treatment, EX + P: Exercise plus Placebo Treatment, BMI=Body mass index, %BF=Percent body fat, TAT=Total adipose tissue, AAT=Appendicular adipose tissue, VAT=Visceral adipose tissue. [1]Wilcoxon rank sum test; ∗ Indicates statistical significance between conditions.

Both groups exhibited similar trends, characterized by increases in TAT and AAT, and decreases in %BF and %AAT (Fig. 1). Interestingly, in the EX + T group, VAT and %VAT decreased, and the trends were opposite in the EX + P group (Fig. 1). Changes from baseline to six months were similar between groups across outcomes for absolute VAT (EX + T: 44 ± 151 g, EX + P: 45 ± 232 g, p = 0.073) but differed for %VAT (EX + T: 0.26 ± 0.51% of TAT vs. EX + P: 0.12 ± 0.52% of TAT, p = 0.004) (Fig. 2, Table 2).

Fig. 1.

Trends in total adipose tissue (TAT), appendicular adipose tissue (AAT), visceral adipose tissue (VAT), % body fat (%BF), and relative AAT (%AAT) and VAT (%VAT), in response to a 6-month resistance training intervention with (EX + T) and without (EX + P) topical testosterone supplementation.

Fig. 2.

Individual changes in absolute visceral (VAT) appendicular (AAT) and relative visceral (%VAT) and appendicular (%AAT) adipose tissue in response to a 6-month resistance training intervention with (EX + T) and without (EX + P) testosterone supplementation. Values are depicted by a percentage change from baseline with means and 95% confidence intervals. Unadjusted differences between means (EX + T – EX + P) are reported with mean and standard error of mean reported. ∗ Indicates statistical significance between conditions.

Relative changes in a regional adiposity from baseline are presented in Fig. 2. Relative changes from baseline were similar between conditions in AAT (EX + T: Δ0.49% ± 8.7%, EX + P: Δ0.41% ± 6.96%, p = 0.966) and %ATT (EX + T: 1.18% ± 3.21%, EX + P: 0.91% ± 2.94%, p = 0.792). However, relative differed between conditions for VAT (EX + T: Δ-8.97% ± 19.35%, EX + P: Δ5.09% ± 20.71%, p = 0.006) and %VAT (EX + T: Δ-10.57% ± 17.13%, EX + P: Δ3.51 ± 18.42%, p = 0.002).

In the multivariate analysis, no significant between-group differences were observed at six months in body mass, BMI, %BF, TAT, AAT, or %AAT after controlling for baseline levels and other baseline covariates (Table 3). Notably, comparing outcomes at six months while adjusting for baseline values provides an estimate that is statistically comparable to analyzing changes from baseline. The mean VAT and %VAT were 59 g and 0.30% of TAT were lower, respectively, in the EX + T group compared with the EX + P group; however, only the difference in %VAT reached statistical significance (adjusted model: p = 0.029, un-adjusted model: p = 0.003).

Table 3.

Adjusted mean differences in body composition and adipose tissue distribution outcomes between the EX + T and EX + P groups, estimated using linear mixed-effects models.

Outcome	Adjusted 6-month Mean (EX + T)	Adjusted 6-month Mean (EX + P)	Adjusted Between-Group Mean Difference (95 % CI)	p-value (Adjusted/Unadjusted)
Body Mass (kg)	68.53	67.80	0.73 (−0.70, 2.16)	0.321/0.265
%BF (%)	39.27	39.40	−0.13 (−1.22, 0.96)	0.819/0.650
TAT (g)	27,264	27,002	262 (−852, 1377)	0.646/0.827
AAT (g)	12,672	12,568	104 (−441, 649)	0.710/0.890
%AAT (%TAT)	47.49	47.31	0.18 (−0.61, 0.98)	0.652/0.718
VAT (g)	806	865	−59 (−162, 44)	0.264/0.091
%VAT (%TAT)	2.76	3.05	−0.30 (−0.55, −0.04)	0.029/0.003∗

Adjusted means differences in body composition and adipose distribution outcomes between EX + T (Exercise plus Testosterone Treatment) and EX + P (Exercise plus placebo treatement), estimated using linear mixed-effects models. All models were adjusted for baseline comorbidities, functional status questionnaire score, age, and heigth with AAT and VAT models further adjusted for body mass. %BF=Percent body fat, TAT=Total adipose tissue, AAT=Appendicular adipose tissue, VAT=Visceral adipose tissue. ∗ Indicates statistical significance between conditions (P < 0.05).

4. Discussion

Strategies to mitigate selective VAT accumulation that occurs in older females recovering from hip fracture have clinical value but are minimally studied. In the present study we observed that, while changes in TAT did not differ between EX + T and EX + P, females randomized to testosterone supplementation experienced a healthier pattern of adipose redistribution, with mitigation of visceral adipose accumulation, as indicated by a reduced %VAT. These results suggest that supplementing exercise with carefully managed exogenous testosterone may also help improve metabolic health following hip fracture in older females by decreasing adipose tissue accumulation viscerally, and thus potentially diminishing the unfavorable metabolic consequences and associated long-term negative health outcomes. However, additional research is needed to investigation the potential clinical benefits of these changes.

Hip fractures can be devastating injuries older adults, where 76% of patients report long-term detrimental changes in health and physical function that in some cases persist indefinitely [[50], [51], [52]]. Early progressive exercise interventions, which are started soon after an initial recovery period, have been shown to be the most effective method of recovering physical function after hip fracture by helping patients regain strength and physical function [[16], [17], [18]]. However, the reduced physical capacity and inflammatory burden after hip fracture can lead to rapidly deteriorating body composition and accumulation of VAT that increases the risk of falls and subsequent hip fracture as well as decreases overall health and long-term health outcomes. Therefore, during this critical time, pairing a structured exercise program with testosterone therapy, is an attractive therapeutic intervention for recovering from hip fracture to promote long-term health, by improving body composition. However, the observation that structured resistance exercise was ineffective at decreasing %BF or TAT was unanticipated, as resistance exercise has been shown to provide marginal reductions in each of these components in healthy adults across the age-span [53,54]. These results may have been influenced by the limited physical capacity of the target population and the design of the exercise intervention that focused on functional strength and balance over caloric expenditure. Most notably, older females recovering from hip fracture are likely to experience increases in %BF and TAT during their recovery due to physical limitations or regaining lost weight that resulted from the initial injury [6,7]. Therefore, it is possible that the exercise intervention mitigated increases in these that would have been otherwise present; however, this cannot be concluded as pre-injury assessments were not available for comparison. While it cannot be concluded that resistance exercise improved TAT or %BF in the present study, it did not worsen in either condition and may have mitigated increases that could have inherently occurred as part of the recovery from hip fracture [6,7].

The primary finding of the present study was that EX + T was effective in stimulating a healthy redistribution of adipose tissue with reduced %VAT compared to EX + P. These findings were in line with our hypothesis and previous observations in healthy younger and older males [34,35]. However, the present study is the first to directly investigate the changes in adipose distribution in older females using testosterone. The results from this randomized control trial contrast with indirect evidence of potential sexual dimorphism in the role of testosterone in VAT accumulation that was proposed based on observational data that is subject to additional biases and confounding [[41], [42], [43], [44], [45]]. Therefore, these results may ease concerns raised from this indirect evidence and demonstrates similar benefits of adipose redistribution seen in older males as seen in older females. However, the clinical significance of these results remains unclear as while decreased relative visceral adiposity (visceral to subcutaneous adipose ratio) is a stronger predictor of cardiometabolic health and disease than TAT or VAT alone and has been linked to survivability rates with some medical conditions, the clinical effects of the magnitude of changes observed in this population is unknown [[27], [28], [29], [30]]. Given these uncertainties, along with the mixed results from the parent STEP-HI study, where EX + T reported greater improvements in the short physical performance battery than EX + P, but similar improvements in 6-min walk distance, leg press strength, and appendicular lean mass [10], additional research is needed determine the potential long-term clinical benefits of EX + T for older females recovering from hip fracture surgery along with the potential risks of such a treatment.

Additionally, further research is needed to better understand how the effects of exogenous testosterone on adipose distribution may differ between age groups and if the effects of testosterone in older females are specific to the structured exercise and injury conditions of the present study. In conclusion, the present study provides evidence of the positive effects of testosterone therapy by demonstrating improvements in regional metabolically stress adipose stores but also expands upon our previous understanding of the role that testosterone plays in adipose redistribution in older females.

5. Limitations

In addition to the limitation noted previously regarding lack of pre-injury data, which is generally unavoidable in a study of people recovering from hip fracture, variance in participants’ baseline body composition, and lack of a non-exercise control group, the present study has several other noteworthy limitations. Most notably, the present study utilized DXA scans to assess adipose measurements rather than criterion Computed Tomography or Magnetic Resonance Imaging measurements. However, previous studies have demonstrated a high degree of agreement between DXA and these criterion technologies, and DXA provides an advantage in the assessment of adipose distribution as Computed Tomography and Magnetic Resonance Imaging scans are usually single slice or segmental and measurements are therefore not inclusive of the entire body. Additionally, two different DXA systems were used (Hologic Horizon A and the Lunar GE iDXA) which while strongly correlated (TAT; r = 0.993, VAT: r = 0.938) use different analysis methods [55]. Another limitation of the present study is that while increases in TAT and VAT commonly occur in those recovering from a hip fracture, the present study was not designed as a weight loss study and included individuals regardless of baseline BMI classification. While the majority of participants in both treatment groups were overweight, both groups included individuals who were obese and therefore are likely to experience amplified effects of a lipolytic intervention and those with a healthy weight status that are likely. While changes in body mass did not differ between groups, both groups experienced non-significant mean increases in body mass and decreases in %BF that were two fold greater in EX + T than EX + P (body mass: Δ1.66 kg vs 0.84 kg, %BF: Δ-0.16% vs −0.08%) and very modest mean reductions in %BF. As such, it is possible that testosterone elicited lipolytic effects on TAT or VAT that were either masked by changes in body mass or outside of the detection of the present study. Similarly, results from this study were limited by the small sample size available for analysis, therefore clinically significant differences may have been present that were not statistically significant given lack of statistical power.

A final limitation of the present study is that results from the present study should be restricted to the specific doses of testosterone provided and study population included. Specifically, the present study provided variable testosterone doses intended to elicit a slightly supraphysiological level (110–160 ng/dL compared to a reference range of 12–78 ng/dL) in older females with <60 ng/dL¹⁰. However, the selected thresholds for these values are not well established for older females recovering from hip fracture and may need to be reevaluated along with initial adverse event reports that are described in detail previously [10]. Additionally, the potential risks of providing these doses of testosterone must be weighed against more objective long-term health outcomes such as 5-year survival, rehospitalization rate, or incidence of major cardiac event before the efficacy of such a pharmacological intervention can be objectively assessed.

6. Conclusion

The present study provides the first data from an RCT on the combined effects of a resistance training program with and without testosterone on total adiposity and adipose distribution in older females recovering from hip fracture. We observed that TAT and %BF did not change in either group (EX + T or EX + P), indicating neither exercise nor testosterone effectively reduced total adiposity during hip fracture recovery. However, adipose distribution improved in EX + T as indicated by reduced %VAT compared to EX + P. Given adipose stored in the visceral compartment has a disproportional effect on metabolic health, these results suggest that supplemental testosterone therapy in older females recovering from a hip fracture could promotes a healthy redistribution of adipose. Additional research is required to confirm these findings in other settings, e.g in older females following other injuries or without injury. Given the high five-year mortality rate following a hip fracture, even modest improvements in metabolic health or reduction in comorbidities could provide clinically significant effects; therefore, further research examining these effects on long-term health outcomes is warranted.

Key Take-Aways

• •After suffering a hip-fracture, increases in visceral adiposity can lead to detrimental long-term health consequences in older females.
• •Supplementing exercise with testosterone therapy did not reduce total adipose tissue, but reduced relative adiposity of the visceral compartment in older females recovering from hip-fracture.
• •Sex-hormones play an important role in adipose distribution, as demonstrated by a beneficial redistribution of adipose in older females when using testosterone that is paired with a structured exercise program.

Ethics

This parent STEP-HI study was approved by the Institutional Review Board at Washington University in St. Louis at Schwitalla Hall, Suite M238, 1402 S Grand Blvd., St. Louis, MO 63104 under protocol #: 201609077. A separate facilitated review for this study was approved by the Institutional Review Board at the University of Connecticut Health Center located at the L Building, 5th floor at UConn Health, 263 Farmington Avenue, Farmington, CT 06030 under protocol #: 018-075-2.

Author contribution

The study conception was performed by JEE. Study methodology and validation was performed by JEE, SZ, FX, CLK, JMB, RHF, JK, COR, and GAK. Formal analysis and data curation was performed by SZ and CLK. Data visualization was performed by SZ, CLK, and JEE. Investigation was performed by JMB, RHF, EB, and JSL. JEE wrote the initial draft with all authors contributing to review and editing. Project administration was performed by JEE, EB and GAK.

Declaration of artificial intelligence (AI) and AI-assisted technologies

AI was not used in the preparation of this work for publication.

Source of funding

Pilot testing and data collection efforts from this study were funded by the National Institute on Aging through the following grants R21 AG023716 (PI: EB), R34 AG040257 (PI: EB), R01 AG051647 (CO-PIs: EB, Dr. Kenneth Schechtman, and Dr. Jay Magaziner) and organizational support was provided by National Institute of Aging grants P30 AG067988 (PI: GAK), P30 AG024832 (PI: Dr. Elena Volpi), and P30 AG028747 (PI: Dr. Jay Magaziner) from the NIA. Support for STEP-HI at the Baltimore site was provided by the Baltimore Veterans Affairs Medical Center Geriatric Research, Education, and Clinical Centers (GRECC). Time and publication support for this manuscript was provided by the National Institute of Aging through a career development award K01 AG092901 (PI: JEE).

Declaration of competing interest

All authors have no conflicts of interest related to this study or its results.