SHBG, Bone Mineral Density, and Physical Function Among Injection Drug Users With and Without HIV and HCV

Abstract

Context

Sex hormone–binding globulin (SHBG) is a glycoprotein that regulates the bioavailability of sex hormones and is higher in people with HIV (PWH) and hepatitis C virus (HCV). SHBG is associated with aging-related diseases, including osteoporosis and frailty in the general population. However, the relationship between SHBG concentration and bone mineral density (BMD) and physical function among PWH and HCV is unclear.

Objective

This study aimed to evaluate the association between chronic infection with HIV and HCV and SHBG, and to assess the relationship of circulating SHBG concentrations with low BMD, physical function impairment, and frailty.

Methods

A cross-sectional study was conducted of 278 HCV-exposed (HCV antibody positive) adults enrolled with and without HIV and HCV from the AIDS Linked to the IntraVenous Experience cohort study into 4 groups: HCV–/HIV–, HCV–/HIV+, HCV+/HIV–, and HCV+/HIV+. We evaluated the association between SHBG concentrations and grip strength, gait speed, Short Physical Performance Battery score, frailty (Fried Frailty Phenotype), and BMD (lumbar spine, total hip, and femoral neck T-score) by using adjusted multivariable regression stratified by sex.

Results

SHBG concentrations were higher in women, in those with HIV RNA greater than 400 copies/mL (P = .02) and HCV RNA greater than 15 IU/mL (P < .001). In adjusted models, higher SHBG concentrations among women were statistically significantly associated with lower grip strength (–0.43 [95% CI, –0.77 to –0.081] kg/10 nmol/L, P < .05), higher odds of frailty (odds ratio, 1.49 [95% CI, 1.07 to 2.08], P < .05), and lower T-scores at the lumbar spine (–0.070 [95% CI, –0.15 to –0.001] SD/10 nmol/L T-score BMD, P < .05). Similar associations were not observed among men.

Conclusion

Higher SHBG concentrations are associated with the presence of HIV and HCV viremia. Among women, but not men, higher SHBG concentrations were associated with lower grip strength, higher odds of frailty, and lower lumbar spine BMD. The underlying mechanisms of these associations require further investigation.

Sex hormone-binding globulin (SHBG) is a glycoprotein that transports steroid hormones and is produced predominantly in the liver (1), but is also expressed in many sex steroid–responsive tissues (2, 3). Most of the circulating total testosterone and estradiol is bound to SHBG, and approximately 1% to 3% of sex steroids circulate freely (4). SHBG has the highest affinity for androgens, particularly with 5-dihydrotestosterone (DHT) (5). In addition, in preclinical studies SHBG can also bind to its membrane receptor (SHBG-R), stimulate 3′,5′-cyclic adenosine monophosphate (cAMP) production (6, 7), and produce downstream effects on cellular proliferation in the breast (8, 9) and prostate (10), although this mechanism in humans has not been clearly established.

Our group and others have shown that SHBG concentrations are higher among people with HIV (PWH) and hepatitis C virus (HCV) compared to those without (11-13), a phenomenon independent of liver disease severity. In addition, SHBG concentrations have been studied as a potential biomarker of aging (8) due to age-related increases, although SHBG concentrations trajectories differ by sex: a linear trajectory in men and a U-shaped trajectory in women (14). In the general population, higher SHBG concentrations have been associated with poor bone health and frailty (15, 16), a critical aging-related syndrome characterized by vulnerability to stressors (17) and also predictive of frailty (15) and all-cause mortality in men (18). Chronic HIV infection has been hypothesized to accelerate the aging process (19, 20) given that a higher prevalence of geriatric conditions, such as sarcopenia, falls, and frailty, has been shown in PWH compared to people without HIV (21, 22). Thus, the higher SHBG concentrations in aging and in chronic viral infection such as HIV and HCV may suggest shared underlying pathophysiological mechanisms.

As the older population continues to grow at an unprecedented rate, it is important to identify and better characterize aging-related biomarkers. Such measures may help to identify those at the highest risk of aging-related conditions, such as osteoporosis, sarcopenia, and falls, all hallmarks of frailty. Given that these aging-related conditions may be more prevalent in individuals with active HCV and/or HIV, these biomarkers may be particularly useful to identifying vulnerable subgroups for targeted intervention among those aging with HIV and HCV. Our study had 2 aims: 1) to evaluate the association between chronic infection with HIV and HCV and SHBG, and 2) to assess the relationship of circulating SHBG concentrations with low bone mineral density (BMD), physical function impairment, and frailty.

Materials and Methods

Study Population

In this cross-sectional study from March 2016 to December 2019, we enrolled HCV-exposed (HCV antibody positive) adults with and without HIV infection from the AIDS Linked to the IntraVenous Experience (ALIVE) cohort. The ALIVE cohort consists of individuals who have been prospectively followed on a semiannual basis since 1988 with a history of injection drug use in the community (23). For this study, we enrolled African American adults participating in ALIVE into 4 groups: HCV–/HIV–, HCV–/HIV+, HCV+/HIV–, and HCV+/HIV+. HIV+ was defined by the presence of a positive HIV antibody test. Those HIV+ participants with HIV RNA greater than 400 copies/mL were considered to have clinically significant viremia. HCV+ was defined as having an HCV RNA greater than 15 IU/mL. Participants with HCV RNA less than or equal to 15 IU/mL (ie, below the limit of quantification of the assay) cleared the virus without receiving any HCV therapy; none of the participants received direct-acting antivirals. Individuals with end-stage renal disease were excluded. The study was approved by the Johns Hopkins Institutional Review Board, and all participants provided written informed consent.

Data Collection

At each semiannual visit ALIVE participants completed standardized questionnaires and underwent clinical examination. Detailed information obtained at each follow-up visit included socioeconomic, behavioral, and clinical parameters for the prior 6-month period. Substance use including alcohol, tobacco, and illicit injection and noninjection drug use were assessed by participant self-report of behaviors in the prior 6-month period. Blood samples for this substudy were collected at the same visits at which whole-body dual-energy x-ray absorptiometry (DXA) scans were performed. Comorbid conditions ascertained included participant self-report of any provider diagnosis for the following conditions:1) depressive symptoms (Centers for Epidemiologic Studies Depression Scale score ≥ 16) (24); (2) self-reported hypertension or receiving antihypertension medications; 3) diabetes mellitus (DM): self-reported DM and use of DM medications vs no DM); 4) self-reported kidney disease; 5) self-reported chronic lung disease; 6) self-reported stroke; and 7) liver disease as assessed by the use of a FibroScan machine (Echosens) by transient elastography that measures the velocity of a shear wave propagating through the liver (25), providing a marker of liver stiffness in kPa units. Liver disease or fibrosis was defined by a fibrosis score cutoff greater than or equal to 9.3 kPa. Hazardous alcohol use was assessed using the Alcohol Use Disorders Identification Test (AUDIT) (26). Alanine transaminase (ALT) and aspartate transaminase (AST) levels were examined as continuous variables, and as 2.5 times the upper limit of the normal (ULN) reference range the upper limit of normal for AST level was 37 U/L or (AST level > 2.5 × ULN) and for ALT the level was 40 U/L or (ALT level > 2.5 × ULN) (25).

At each visit, ALIVE participants without HIV received HIV serology testing and participants with HIV had CD4 + cell counts measured by standard flow cytometry and HIV-1 plasma RNA levels assessed by a COBAS AmpliPrep/COBAS TaqMan HIV-1 Monitor with a lower limit of detection of 50 copies/mL (Roche Diagnostics). HCV RNA plasma quantification was conducted at the Johns Hopkins Pathology laboratory by COBAS 6800/8800 systems (Roche Diagnostics) with a lower limit of detection of 15 IU/mL (27), and HCV immunoglobulin G antibody testing was performed as previously described (28).

Sex Hormone Measures

Phlebotomy was performed on all participants before noon. Samples were frozen at –80 °C until the time of sex hormone measurement. Free testosterone, total testosterone, estradiol, and SHBG were measured at the Brigham Research Assay Core Laboratory at Brigham and Women’s Hospital, Boston, Massachusetts (Dr Shalender Bhasin, director). Free testosterone levels were measured by equilibrium dialysis (29-31). Total testosterone and estradiol were extracted by solid-phase extraction and eluted by high-performance liquid chromatography. The determination was performed by mass spectrometry in an electrospray ionization source. Deuterated stable isotope was used for the calibration of the assay. The testosterone assay had a sensitivity of 1 ng/dL, with an interassay coefficient of variation (CV) for men less than 7% and for women less than 5%) and an intra-assay CV for men less than 2% and women less than 5%. The estradiol assay had a sensitivity of 1 pg/mL with interassay CVs less than 12%, and intra-assay CVs less than 5%. SHBG concentrations were measured using a chemiluminescent assay (Beckman Coulter) with an interassay CV between 5.2% and 5.5%, intra-assay CVs between 4.5% and 4.8% and a sensitivity of 0.33 nmol/L.

Body Composition and Bone Mineral Density Measurements

Body mass index (BMI) was calculated for all participants. Whole-body DXA scans were performed on a Hologic Horizon machine (Hologic Inc) to obtain measurements of whole-body total lean mass (in kg) and whole-body total fat mass (in kg). The participants had BMD assessed at the lumbar spine and hip (total and femoral neck). T-scores were calculated from the site-specific BMD measures using the White, young, female, database as a reference population for T-scores per International Society for Clinical Densitometry recommendations (32).

Frailty Assessment

Frailty was assessed as previously described using the 5 original Fried physical frailty phenotype criteria: slow gait speed, decreased grip strength (weakness), poor endurance (exhaustion), low physical activity, and physical shrinking (weight loss) (17, 33). Each frailty parameter was considered as a binary variable (0, 1) and summed to obtain a frailty score; 3 or greater was considered frail, 1 or 2 considered prefrail, and scores of 0 considered robust (34).

Short Physical Performance Battery Score

The Short Physical Performance Battery (SPPB) was developed by the National Institute on Aging as an objective measure of lower-extremity function and consists of 3 components: timed standing balance tests, timed walking test (gait speed), and timed chair rise (35). For balance, participants were asked to stand and hold their feet in side by side, semi-tandem, and tandem positions for 10 seconds each. Gait speed was measured in two 4-m walks at a normal pace, with the faster of the 2 walks used for analysis. The timed chair-rise task was the time required to rise from a sitting to fully standing position 5 times as quickly as possible. Each component was scored as 0 to 4 points based on predetermined cutoffs to generate a total score from 0 to 12; higher scores correspond to better physical performance. A score lower than 10 indicates one or more mobility limitations. ALIVE research assistants, trained by the National Institute on Aging staff (36), administered the SPPB.

Statistical Analysis

Baseline characteristics of PWH and HCV were compared using nonparametric Wilcoxon test or chi-square or Fisher exact test as appropriate. Assessment of the normality of the SHBG concentrations was performed by creating a histogram to examine the frequency distribution and by performing the Shapiro-Wilk test.

The following variables were assessed age, sex (reference, male), whole-body lean mass, whole-body total fat, HIV status defined by HIV serology testing, HCV infection defined by the presence of detectable HCV RNA, education (reference < 12th grade), current alcohol use (reference, no), current smoking (reference, no), current injection drug use (reference, no), free testosterone, estradiol, and comorbidity count (ie, sum of comorbid conditions). HIV-specific factors were nadir CD4+ T-cell count, current CD4+ T-cell count, uncontrolled HIV viremia (< 400 HIV RNA copies/mL), and current antiretroviral therapy use.

First, we used univariate and multivariable linear regression to evaluate factors that were associated with SHBG concentrations. We also assessed for interaction by sex and HIV and sex and HCV. The regression coefficients (β) in the statistical models represent the SHBG concentration per one-unit difference of the independent variable (covariate). Second, stratified analyses by sex were also performed. Linear regression was performed to evaluate the relationship between the SHBG concentrations and continuous outcomes (grip strength; gait speed; T-score at lumbar spine, total hip, and femoral neck). Logistic and multinomial logistic regression were used to binary and multilevel categorical outcomes, including SPPB score less than 10 (37) and frailty (robust, prefrail, and frail), respectively. The models assessing factors associated with SHBG concentrations were adjusted for the following parameters: demographics (age, sex, education), body composition (whole-body lean mass, whole-body total fat), HCV viremia and HIV status; DM status, AST, ALT, AST level greater than 2.5 × ULN, ALT level greater than 2.5 × ULN, and liver stiffness (modeled both as a continuous and categorical variable). The models assessing the second aim of our study (ie, bone and physical function outcomes) were adjusted for the following parameters: demographics (age, education), body composition (whole-body lean mass, whole-body total fat), HCV viremia and HIV status, sex hormones (free testosterone, estradiol) alcohol use, smoking, injection drug use, and comorbidity count, stratified by sex. Analyses were performed using STATA (version 13; Stata Corp).

Results

Participant characteristics by sex are shown Table 1. Among 278 participants exposed to HCV (176 men and 102 women), the median (Q1-Q3) of age by sex was 54 years (range, 50-57 years) for women and 58 years (range, 54-62 years) for men. Women had higher BMI, higher whole-body fat mass, higher prevalence of liver disease, higher SHBG concentrations, higher estradiol, and were more likely to have undetectable HCV RNA. In addition, women had lower education levels; lower free and total testosterone; lower grip strength; lower T-score at the lumbar spine, total hip, and femoral neck; and lower whole-body lean mass compared to men.

Table 1.

AIDS Linked to the IntraVenous Experience (ALIVE) bone participant characteristics

	Men (N = 176)	Women (N = 102)	P
Age, y	58 (54 to 62)	54 (50 to 57)	.23
Education ≥ high school,%	48%	36%	.04
BMI, kg/m2	24.5 (21.4 to 28.7)	27.7 (22.9 to 33.5)	< .001
Current smoker,%	76%	77%	.92
Current alcohol use,%	62%	40%	<.001
Current injection drug use,%	42%	31%	.08
HCV RNA ≤ 15 IU/mL, %	27%	39%	.03
HIV positive, No.	53	35
CD4 + T cell count, cells/mm3	559 (340 to 731)	420 (248 to 588)	.33
On ART, %	92%	85%	.40
HIV RNA ≤ 400 copies/mL, %	72%	70%	.84
Hypertension, %	47%	54%	.29
Diabetes, %	10%	10.6%	.88
Depressive symptoms, CES-D, %	28%	28%	≥ .99
Renal disease, %	1%	1%	.91
Chronic lung disease, %	12%	40%	< .001
Stroke, %	2%	3.0%	.72
Liver disease, %	4%	7%	< .001
AST, U/L	34(26 to 55)	26 (19 to 42)	.01
ALT, U/L	27 (19 to 46)	19 (13 to 32)	.003
AST level > 2.5 × ULN	44%	29%	.02
ALT level > 2.5 × ULN	30%	10%	.04
Walk speed, m/s	0.97 (0.81 to 1.13)	0.96 (0.81 to 1.14)	.71
Grip strength, kg	40 (34 to 46)	27 (22 to 32)	.025
SPPB score	11 (10 to 12)	9 (9 to 11)	.82
Robust, %	21%	22%	.88
Prefrail, %	52%	50%	.76
Frail, %	11%	11%	.97
T-score lumbar spine (SD)	0.21 (–0.52 to 1.16)	–0.34 (–1.46 to 55)	.0013
T-score total hip (SD)	0.20 (–0.52 to 1.14)	–0.33 (–1.45 to 0.54)	.0002
T-score femoral neck (SD)	0.21 (–0.52 to 1.15)	–0.33 (–1.45 to 0.54)	.0001
Wb. lean mass, kg	51.8 (47.8 to 58.9)	41.2 (38.2 to 47.4)	< .001
Wb. fat mass, kg	20.7 (15.4 to 28.9)	31.4 (21.8 to 43.8)	< .001
Free testosterone, pg/mL	87 (56 to 142)	58 (43 to 79)	< .001
Total testosterone, ng/dL	439 (276 to 590)	19.6 (12.9 to 28.6)	< .001
Estradiol, pg/mL	11 (7 to 17)	3 (2 to 7)	.026
SHBG, nmol/L	76 (53 to 102)	93 (64 to 138)	.005

Data are expressed as median (Q1-Q3) and percentage.

Abbreviations: ALT, alanine transaminase; ART, antiretroviral therapy; AST, aspartate transaminase; BMI, body mass index; CES-D, Centers for Epidemiologic Studies Depression scale score ≥ 16; HCV, hepatitis C virus; SHBG, sex hormone–binding globulin; SPPB, Short Physical Performance Battery; Wb., whole body.

Among the PWH, the median (Q1-Q3) of CD4 cell count was 420 cells/mm³ (248 588 cells/mm³) for women and 559 cells/mm³ (340 731 cells/mm³) for men, and the prevalence of HIV RNA less than or equal to 400 copies/mL was not different by sex. Participant characteristics by each of the 4 study groups are shown in Supplementary Table 1 (38).

Supplementary Fig. 1 (38) shows SHBG median concentrations by HIV and HCV status, stratified by sex, women (Supplementary Fig. 1A), and men (Supplementary Fig. 1B). Higher SHBG concentrations were observed among women compared to men. SHBG concentrations among people with HIV monoinfection were similar for women (65 ± 8 nmol/L) and (58 ± 6 nmol/L) men and compared to individuals without HIV or HCV infection (77 ± 10 nmol/L and 56.0 ± 4 nmol/L). SHBG concentrations among people with HCV monoinfection were higher (105 ± 8 nmol/L) for women and (90 ± 4 nmol/L) for men (P < .05, P < .001) compared to women and men without HCV infection (65 ± 8 nmol/L and 58 ± 6 nmol/L). Among people with HIV and HCV coinfection, higher SHBG concentrations were observed (133 ± 9 nmol/L) for women (101 ± 6.5 nmol/L) and for men (P < .001, P < .001) compared to those without HIV and HCV.

Factors Associated With Sex Hormone–binding Globulin Concentrations

Univariate and multivariate were performed to determine factors that were associated with SHBG concentrations in all participants (Table 2) and stratified by sex (Tables 3-5). In univariate linear regression (see Table 2), higher SHBG concentrations were associated with lower whole-body fat and lean mass, female sex, HIV RNA greater than 400 copies/mL, HCV RNA greater than 15 IU/mL, higher AST, higher ALT, higher liver stiffness by FibroScan, higher AST level greater than 2.5 × ULN and higher ALT level greater than 2.5 × ULN. Multivariable analysis (model 1), unadjusted both for whole-body fat and lean mass, showed that higher SHBG concentrations were associated with female sex, HIV RNA greater than 400 copies/mL, and HCV RNA greater than 15 IU/mL. Multivariable analysis (model 2), adjusted for covariates including whole-body fat and lean mass, showed that female sex and whole-body fat were no longer associated with SHBG. Multivariable analysis (model 3), adjusted for covariates including AST and ALT, showed similar associations as model 1 and that higher SHBG concentrations was associated with low ALT. Multivariable analysis (model 4), adjusted for covariates including liver stiffness as a continuous variable, showed similar associations as model 1 and that higher SHBG concentrations were associated with greater liver stiffness. These models show that higher SHBG was associated with HIV and HCV viremia, independent of liver disease, regardless of the marker used. In addition, we did not find any association between SHBG concentrations and liver disease (liver stiffness ≥ 9.3 kPa). Multivariable analysis adjusted for all covariates in model 1 and sex interaction with HIV and HCV showed no interaction between sexxHIV viremia, (P = .96) or sex*HCV viremia, (P = .17). Table 3 shows univariable analyses for women and men. Higher SHBG concentrations were associated with lower whole-body fat and lean mass, HCV RNA greater than 15 IU/mL, and higher AST in women and men; and HIV RNA greater than 400 copies/mL, higher ALT, and greater liver stiffness in men.

Table 2.

Factors related to circulating sex hormone–binding globulin concentration in AIDS Linked to the IntraVenous Experience (ALIVE)

	Univariable		Model 1		Model 2		Model 3		Model 4
Factors	β (95% CI)	P	β (95% CI)	P	β (95% CI)	P	β (95% CI)	P	β (95% CI)	P
Age, y	0.020 (–0.064 to 0.106)	.63	0.11 (0.025 to 0.19)	.011	0.054 (–0.033 to 0.14)	.26	0.13 (0.043 to 0.21)	.003	0.10 (–0.008 to 0.21)	.07
Whole-body fat mass, kg	–0.059 (–0.10 to –0.018)	.006	–	–	0.0015 (–0.056 to 0.059)	.96	–	–	–	–
Whole-body lean mass, kg	–0.174 (–0.23 to –0.12)	< .001	–	–	–0.14 (–0.23 to –0.058)	< .001	–	–	–	–
Female sex	1.60 (0.48 to 2.71)	.005	2.37 (1.29 to 3.46)	.000	0.64 (–1.11 to 2.39)	.47	2.68 (1.61 to 3.75)	< .001	2.24 (0.88 to 3.60)	< .001
HIV-serostatus	Reference	–	Reference		Reference	–	Reference		Reference
HIV + serostatus
HIV RNA (≤ 400 copies/mL)	0.42 (–0.88 to 1.72)	.52	0.36 (–0.81 to 1.52)	.55	0.27 (–0.93 to 1.47)	.66	0.34 (–0.81 to 1.49)	.56	0.53 (–0.91 to 1.96)	.47
HIV RNA (> 400 copies/mL)	2.44 (0.59 to 4.30)	.01	2.76 (0.96 to 4.55)	.03	2.90 (1.00 to 4.80)	.002	2.29 (0.49 to 4.08)	.013	2.89 (0.64 to 5.15)	.012
HCV- serostatus	Ref	–	Ref	–	Ref	-	Ref	–	Ref	–
HCV RNA (> 15 IU/mL)	3.74 (2.65 to 4.83)	< .001	4.16 (3.09 to 5.24)	.000	3.50 (2.37 to 4.63)	.000	3.58 (2.47 to 4.69)	< .001	3.52 (2.18 to 4.87)	< .001
Diabetes	–0.53 (–2.36,1.30)	.57	–1.01(-2.66 to 0.64)	.23	–0.12 (–1.86 to 1.62)	.89	–1.13 (–2.72 to 0.46)	.16	–0.20 (–2.91 to 2.50)	.88
AST, U/L	0.038 (0.023 to 0.053)	< .001	–	–	–	–	0.031 (–0.030 to 0.024)	.83	–	–
ALT, U/L	0.027 (0.009 to 0.045)	.003	–	–	–	–	–0.003 (–0.030 to 0.024)	.013	–	–
Liver stiffness, kPa	0.18 (0.062 to 0.31)	.003	–	–	–	–	–	–	0.13 (0.020 to 0.25)	.02
AST level > 2.5 × ULN	2.78 (1.70 to 3.86)	< .001
ALT level > 2.5 × ULN	1.52 (0.27 to 2.77)	.017
Liver disease (≥ 9.3 kPa)	1.36 (–0.18 to 2.90)	.08	0.35 (–1.10 to 1.80)	.64	0.27 (–1.22 to 1.75)	.72	–	–	–	–

Univariable linear regression; unadjusted.

Model 1: adjusted for age, sex, HIV RNA (HIV seronegative = reference, ≤ 400 copies/mL, >400 copies/mL), HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, liver disease: based on FibroScan, fibrosis was ≥ 9.3 kPa.

Model 2: adjusted for age, sex, whole-body total lean mass, whole-body total fat mass, HIV RNA (HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL), HCV RNA (≤ 15 IU/mL = reference, >15 IU/mL), diabetes, liver disease: based on FibroScan, fibrosis was ≥ 9.3 kPa.

Model 3: adjusted for age, sex, HIV RNA (HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL), HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, AST, ALT.

Model 4: adjusted for age, sex, HIV RNA (HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL), HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, liver stiffness as continuous.

Abbreviations: β, regression coefficient; ALT, alanine transaminase; ART, antiretroviral therapy; AST, aspartate transaminase; BMI, body mass index; HCV, hepatitis C virus; ULN, upper limit of normal.

Table 3.

Factors related to circulating sex hormone–binding globulin concentration in men and women

	Men		Women
	Univariate		Univariate
Factors	β (95% CI)	P	β (95% CI)	P
Age, y	0.039 (–0.06 to 0.14)	.44	0.091 (–0.081 to 0.263)	.29
Whole-body fat mass, kg	–0.11 (–0.168 to –0.055)	< .001	–0.074 (–0.143 to 0.0063)	.03
Whole-body lean mass, kg	–0.22 (–0.30 to –0.15)	< .001	–0.13 (–0.23 to –0.026)	.014
HIV– serostatus	Reference	–	Reference	–
HIV+ serostatus
HIV RNA (≤ 400 copies/mL)	0.33 (–1.16 to 182)	.66	0.050 (–1.94 to 2.93)	.69
HIV RNA (> 400 copies/mL)	2.67 (0.48 to 4.86)	.017	1.91 (–1.43 to 5.25)	.26
HCV RNA (> 15 IU/mL)	3.40 (2.07 to 4.69)	.001	4.84 (3.007 to 6.69)	< .001
AST, U/L	0.041 (0.024 to 0.05)	.001	0.048 (0.014 to 0.08)	.005
ALT, U/L	0.031 (0.013 to 0.05)	.001	0.036 (–0.0086 to 0.08)	.11
Liver stiffness, kPa	0.21 (0.078 to 0.34)	.002	0.055 (–0.22 to 0.33)	.69
Liver disease	1.10 (–0.15 to 2.35)	.08	–1.65 (–3.27 to 0.94)	.28
Diabetes	–0.058 (–2.19 to 2.07)	.96	–1.44 (–4.77 to 1.88)	.39

HCV RNA (undetectable ≤ 15 IU/mL = reference, detectable > 15 IU/mL), HCV RNA (undetectable ≤ 15 IU/mL = reference, detectable > 15 IU/mL); liver disease: based on liver stiffness, fibrosis was ≥ 9.3 kPa.

Abbreviations: β, regression coefficient; ALT, alanine transaminase; AST, aspartate transaminase; BMI, body mass index; HCV, hepatitis C virus; ULN, upper limit of normal.

Table 4 shows multivariable analyses for women. In women, the fully adjusted multivariable analysis showed that higher SHBG concentrations were associated with lower whole-body lean mass, HIV RNA greater than 400 copies/mL, and HCV RNA greater than 15 IU/mL. Table 5 shows multivariable analyses for men. In men, the fully adjusted multivariable analysis showed that higher SHBG concentrations were associated with lower whole-body lean mass, HCV RNA greater than 15 IU/mL, and higher liver stiffness.

Table 4.

Factors related to circulating sex hormone–binding globulin concentration in women

	Women
	Model 1		Model 2		Model 3
Factors	β (95% CI)	P	β (95% CI)	P	β (95% CI)	P
Age, y	0.091 (–0.081 to 0.263)	.29	0.21 (0.03 to 0.40)	.02	0.11 (–0.20 to 0.42)	.48
Whole-body fat mass, kg	–0.074 (–0.143 to 0.0063)	.03	0.007 (–0.09 to 0.10)	.89	0.039 (–0.13 to 0.21)	.64
Whole-body lean mass, kg	–0.13 (–0.23 to –0.026)	.014	–0.084 (–0.23 to 0.06)	.25	–0.40 (–0.77 to –0.04)	.03
HIV– serostatus	Reference	–	Reference	–	Reference	–
HIV+ serostatus
HIV RNA (≤ 400 copies/mL)	0.050 (–1.94 to 2.93)	.69	–0.79 (–3.23 to 1.64)	.52	–0.13 (–3.71 to 3.46)	.94
HIV RNA (> 400 copies/mL)	1.91 (–1.43 to 5.25)	.26	1.96 (–1.77 to 5.69)	.30	3.72 (–1.27 to 8.70)	.14
HCV RNA (> 15 IU/mL)	4.84 (3.007 to 6.69)	< .001	4.95 (2.63 to 7.27)	.001	2.60 (–0.85 to 6.07)	.14
Diabetes	–1.44 (–4.77 to 1.88)	.39	–0.05 (–3.17 to 3.07)	.97	–0.39 (–6.58 to 5.80)	.90
AST, U/L	–	–	0.072 (0.001 to 0.14)	.05	–	–
ALT, U/L	–	–	–0.08 (–0.18 to 0.18)	.11	–	–
Liver stiffness, kPa	0.055 (–0.22 to 0.33)	.69	–	–	0.045(–0.24 to 0.33)	.75
Liver disease, < 9.3 kPa	–1.65 (–3.27 to 0.94)	.28	–	–	–	–

HCV RNA (undetectable ≤ 15 IU/mL = reference, detectable > 15 IU/mL), HCV RNA (undetectable ≤ 15 IU/mL = reference, detectable > 15 IU/mL); liver disease: based on liver stiffness, fibrosis was ≥ 9.3 kPa.

Model 1: adjusted for age, sex whole body lean mass, whole-body total fat, HIV RNA ((HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL, HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, liver disease: based on liver stiffness, fibrosis was ≥ 9.3 kPa.

Model 2: adjusted for age, sex whole-body lean mass, whole-body total fat, HIV RNA ((HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL, HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, AST, and ALT.

Model 3: adjusted for age, sex whole-body lean mass, whole-body total fat, HIV RNA ((HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL, HCV RNA (≤ 15 IU/mL = reference, >15 IU/mL), diabetes, and liver stiffness continuous.

Abbreviations: β, regression coefficient; ALT, alanine transaminase; AST, aspartate transaminase; HCV, hepatitis C virus.

Table 5.

Factors related to circulating sex hormone–binding globulin concentration in men

	Men
	Model 1		Model 2		Model 3
Factors	β (95% CI)	P	β (95% CI)	P	β (95% CI)	P
Age, y	–0.03 (–0.13 to 0.075)	.58	–0.024 (–0.14 to 0.086)	.66	–0.026 (–0.15 to 0.10)	.69
Whole-body fat mass, kg	0.004 (–0.07 to 0.08	.91	0.003 (–0.072 to 0.078)	.09	0.011 (–0.09 to 0.11)	.83
Whole-body lean mass, kg	-0.21 (–0.32 to –0.10)	< .001	–0.18 (–0.29 to –0.069)	.002	–0.20 (–0.33 to –0.70)	.003
HIV– serostatus	Reference	–	Reference	–	Reference
HIV+ serostatus
HIV RNA (≤ 400 copies/mL)	0.42 (–1.00 to 1.85)	.56	0.49 (–0.92 to 1.91)	.49	0.22 (–1.39 to 1.83)	.79
HIV RNA (> 400 copies/mL)	2.02 (–0.42 to 4.47)	.11	1.87 (–0.54 to 4.28)	.13	2.26 (–0.65 to 5.17)	.13
HCV RNA (> 15 IU/mL)	2.30 (0.86 to 3.74)	.002	2.22 (0.81 to 3.64)	.002	2.52 (0.88 to 4.16)	.003
Diabetes	0.77 (–1.39 to 2.94)	.48	0.54 (–1.61 to 2.70)	.62	–0.17 (–3.66 to 3.32)	.93
AST, U/L	–	–	0.017 (–0.009 to 0.044)	.19	–	–
ALT, U/L	–	–	0.0031 (–0.024 to 0.031)	.82	–	–
Liver stiffness, kPa	–	–	–	–	0.14 (0.022 to 0.27)	.021
Liver disease	0.73 (–0.59 to 2.06)	.28	–	–	–	–

HCV RNA (undetectable ≤ 15 IU/mL = reference, detectable > 15 IU/mL), HCV RNA (undetectable ≤ 15 IU/mL = reference, detectable > 15 IU/mL); liver disease: based on FibroScan, fibrosis was ≥ 9.3 kPa.

Outcome: SHBG (per 10 nmol/L); multivariate linear regression; unadjusted.

Model 1: adjusted for age, sex whole-body lean mass, whole-body total fat, HIV RNA ((HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL, HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, liver disease: based on FibroScan, fibrosis was ≥ 9.3 kPa.

Model 2: adjusted for age, sex whole-body lean mass, whole-body total fat, HIV RNA ((HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL, HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, AST, and ALT.

Model 3: adjusted for age, sex whole-body lean mass, whole-body total fat, HIV RNA ((HIV seronegative = reference, ≤ 400 copies/mL, > 400 copies/mL, HCV RNA (≤ 15 IU/mL = reference, > 15 IU/mL), diabetes, and liver stiffness continuous.

Abbreviations: β, regression coefficient; ALT, alanine transaminase; AST, aspartate transaminase; HCV, hepatitis C virus; SHBG, sex hormone–binding globulin.

Association of Sex Hormone–binding Globulin With Grip Strength, Gait Speed, Short Physical Performance Battery Score, and Frailty

Table 6 and Supplementary Fig. 2A-2D (38) show the association between SHBG concentration and grip strength, gait speed, SPPB score, and frailty among women and men in separate fully adjusted multivariable analyses. Supplementary Table 2 (38) shows similar results; however, these models were adjusted for total testosterone instead of free testosterone. Among women, in the fully adjusted model, higher SHBG concentrations were statistically significantly associated with lower grip strength (–0.43 [95% CI, –0.77 to –0.081] kg/10 nmol/L; P < .05) and greater odds of frailty (odds ratio [OR], 1.49 [95% CI, 1.07 to 2.08]; P < .05). Gait speed and SPPB score were not associated with SHBG concentrations. In an exploratory analysis, the exclusion of 26 premenopausal women did not substantially alter the point estimate of these relationships, therefore we used the full sample of women for the analysis. Among men, in the fully adjusted model, SHBG was not associated with grip strength, gait speed, and SPPB score or frailty, but higher whole-body lean mass was associated with higher grip strength, gait speed, SBBP score, and frailty. We explored sex interaction with SHBG in the full population and no interaction was found for grip strength, walk speed, SPPB score, or frailty. In addition, in women and men, we conducted simple mediation analysis and did not find evidence that lean mass was a mediator of the association between SHBG concentration and grip strength (data not shown).

Table 6.

Association between sex hormone–binding globulin concentration and grip strength, walk speed, Short Physical Performance Battery score, and frailty among women and men

Outcome	Grip strength, kg	Gait speed, m/s	SPPB score (< 10)	Frailty
	β (95% CI)	β (95% CI)	OR (95% CI)	RR (95% CI)
Women
SHBG (per 10 nmol/L)	–0.43 (–0.77 to –0.081) a	0.004 (–0.011 to 0.018)	0.99 (0.87 to 1.13)	1.49 (1.07 to 2.08) a
Age y	0.035 (–0.30 to 0.23)	–0.004(–0.015 to 0.007)	1.04 (0.93 to 1.15)	1.04 (0.82 to 1.33)
HCV status	2.86 (–0.68 to 6.41)	–0.086 (–0.23 to 0.063)	1.13 (0.32 to 3.96)	0.094 (0.005 to 1.94)
HIV status	–2.37 (–5.60 to 0.86)	–0.072 (–0.21 to 0.06)	1.05 (0.33 to 3.33)	0.008 (0.003 to 2.55)
Wb. total lean mass, kg	0.030 (–0.18 to 0.24)	–0.0040 (–0.013 to 0.0051)	1.06 (0.95 to 1.19)	1.12 (0.93 to 1.36)
Wb. total fat, kg	0.018 (–0.13 to 0.16)	–0.0013 (–0.007 to 0.005)	0.98 (0.93 to 1.04)	1.01 (0.89 to 1.15)
Free testosterone, ng/dL	–10.0 (–37.32 to 17.30)	0.94 (–0.20 to 2.07)	0.04 (2.90e–6 to 545)	2.90e–15 (1.22e–36 to 0.69)
Estradiol (per 10 pg/mL)	0.18 (–0.12 to 0.47)	0.01 (–0.002 to 0.022)	0.87 (0.73 to 1.02)	1.03 (0.81 to 1.32)
Comorbidity count	–0.25 (–1.70 to 1.20)	–0.02 (–0.08 to 0.04)	1.34(0.78 to 2.31)	2.49 (0.84 to 7.40)
Men
SHBG (per 10 nmol/L)	–0.12 (–0.53 to 0.28)	0.006 (–0.007 to 0.02)	1.00 (0.90 to 1.11)	1.05 (0.85 to 1.37)
Age, y	–0.14 (–0.38 to 0.01)	–0.003 (–0.01 to 0.005)	1.002 (0.94 to 1.07)	0.90 (0.79 to 1.02)
HCV status	0.34 (–3.18 to 3.86)	0.093 (–0.022 to 0.21)	0.37 (0.14 to 0.99) a	0.86 (0.14 to 5.44)
HIV status	0.58 (–2.49 to 3.66)	–0.011 (–0.11 to 0.09)	0.29 (0.076 to 1.11)	0.50 (0.094 to 2.62)
Wb. total lean mass, kg	0.36 (0.072, 0.66) a	0.011 (0.002 to 0.02) a	0.92 (0.85 to 0.99) a	0.82 (0.70 to 0.95) a
Wb. total fat, kg	–0.015 (–0.23 to 0.20)	–0.008 (–0.014 to –0.001) a	1.04 (0.98 to 1.10)	1.10 (0.98 to 1.23)
Free testosterone, ng/dL	–0.006 (–0.29 to 0.28)	–0.005(–0.014 to 0.004)	1.10 (0.49 to 2.46)	2.46 (0.83, 7.24)
Estradiol (per 10 pg/ml)	0.056 (–1.68 to 1.71)	0.01(–0.04 to 0.066)	0.67 (0.38 to 1.19)	0.93 (0.82 to 1.05)
Comorbidity count	–1.43 (–3.10 to 0.25)	–0.032 (–0.090 to 0.025)	1.56 (0.97 to 2.50)	5.27 (1.92 to 14.47)

^a P-value ≤ .05.

^b P-value ≤ .01.

Multivariable linear, logistic regression, and multinomial logistic regression models were performed.

Model 1: adjusted for age, education, whole-body total lean mass, whole-body total fat mass, alcohol use, smoking, injection drug use, HIV status, HCV status, free testosterone, estradiol, and comorbidity count.

Abbreviations: HCV, hepatitis C virus; OR, odds ratio; RR, relative risk; SHBG, sex hormone–binding globulin; SPPB, Short Physical Performance Battery; Wb., whole-body.

Association of Sex Hormone–binding Globulin With Bone Mineral Density at the Lumbar Spine, Total Hip, and Femoral Neck

Table 7 and Supplementary Fig. 3A and 3B (38) show the association between SHBG concentration and BMD at 3 sites (lumbar spine, total hip, and femoral neck) among women and men in separate fully adjusted multivariable analyses. Supplementary Table 3 (38) shows similar results; however, these models were adjusted for total testosterone instead of free testosterone. Among women, higher SHBG concentrations were associated with lower T-score BMD SD/10 nmol/L at the lumbar spine (–0.070 [95% CI, –0.15 to –0.001]) but not at the total hip and femoral neck. Among men, SHBG concentrations were not associated with BMD in the lumbar spine, total hip, and femoral neck T-score. We explored sex interaction with SHBG in the full population and no interaction was found for grip strength, walk speed, SPPB score, or frailty. We did not find evidence that lean mass was a mediator of the association between SHBG concentration and frailty.

Table 7.

Association between sex hormone–binding globulin concentration and spine, total hip, and femoral neck T-scores among women and men

Outcome	T-score spine	T-score total hip	T-score femoral neck
	coeff (95% CI)	coeff (95% CI)	coeff (95% CI)
Women
SHBG (per 10 nmol/L)	–0.070 (–0.15 to –0.001) a	–0.039 (–0.088 to 0.011)	–0.040 (0.016 to –0.06)
Age, y	–0.058 (–0.11 to –0.006) a	–0.069 (–0.11 to -0.031) a	–0.069 (–0.11 to –0.023) a
HCV status	0.023 (–0.68 to 0.73)	0.021 (–0.49 to 0.53)	0.060 (–0.52 to 0.65)
HIV status	–0.16 (–0.80 to 0.47)	0.0084 (–0.50 to 0.52)	–0.31 (–0.85 to 0.22)
Wb. total lean mass, kg	–0.022 (–0.063 to 0.019)	0.006 (–0.025 to 0.036)	0.014 (–0.021 to 0.050)
Wb. total fat, kg	0.031 (0.0022 to 0.060)	0.025 (0.00040 to 0.045) a	0.016 (–0.009 to 0.042)
Free testosterone, ng/dL	3.69 (–1.65 to 9.03)	0.90 (–3.01 to 4.82)	–0.21 (–4.77 to 4.35)
Estradiol (per 10 pg/mL)	0.042 (–0.064 to 0.10)	0.029 (–0.013 to 0.072)	0.0032 (–0.046 to 0.052)
Comorbidity count	0.46 (0.17 to 0.75) b	0.14 (–0.063 to 0.35)	0.046 (–0.19 to 0.28)
Men
SHBG (per 10 nmol/L)	–0.028 (–0.10 to 0.043)	0.022 (–0.06 to 0.054)	0.03 (–0.028 to 0.089)
Age, y	0.017 (–0.024 to 0.058)	–0.013 (–0.046 to 0.020)	–0.043 (–0.077 to –0.0091) a
HCV status	0.26 (–0.36 to 0.88)	–0.035 (–0.53 to 0.46)	0.19 (–0.33 to 0.71)
HIV status	0.16 (–0.38 to 0.70)	–0.099 (–0.53 to 0.33)	–0.08 (–0.52 to 0.37)
Wb. total lean mass, kg	0.078 (0.029 to 0.13) a	0.061 (0.022 to 0.10) a	0.066 (0.023 to 0.11) a
Wb. total fat, kg	–0.014 (–0.05 to 0.021)	0.009 (–0.020 to 0.037)	0.016 (–0.013 to 0.045)
Free testosterone, ng/dL	0.28 (–0.21 to 0.77)	0.22 (–0.17 to 0.61)	0.28 (–0.12 to 0.68)
Estradiol (per 10 pg/mL)	0.18 (–0.13 to 0.5)	0.15 (–0.088 to 0.38)	0.086 (–0.16 to 0.32)
Comorbidity count	–0.16 (–0.45 to 0.14)	–0.14 (–0.37 to 0.10)	–0.32 (-0.56 to –0.08) b

^a P-value ≤ .05.

^b P-value ≤ .01.

Multivariable linear models were performed.

Model 1: adjusted for age, education, whole-body total lean mass, whole-body total fat mass, alcohol use, smoking, injection drug use, HIV status, HCV status, free testosterone, estradiol, and comorbidity count.

Abbreviations: HCV, hepatitis C virus; SHBG, sex hormone–binding globulin; Wb., whole-body.

Discussion

To our knowledge, this is the first study to establish a link between SHBG concentrations with physical function and BMD in people with and without HCV and HIV. We found that the people with chronic, uncontrolled HIV and HCV infection, particularly women, had higher SHBG concentrations compared to people without HIV or HCV. In addition, our stratified analysis by sex showed that, among women but not in men, higher SHBG concentrations were associated with lower grip strength, greater odds of frailty, and lower BMD T-score at the lumbar spine in models adjusted for multiple confounding factors including liver disease (39, 40). These findings suggest that SHBG concentrations may be an important aging-related biomarker in this population who are at higher risk of aging-related conditions, including physical function impairment and osteoporosis.

Women: Sex Hormone–binding Globulin, Physical Function, and Osteoporosis

Among women, we found that higher SHBG concentrations were associated with lower grip strength and increased the odds of being frail, independent of liver disease and other confounders. To our knowledge, no study in the general population or any other population has explored the relationship between circulating SHBG concentrations and grip strength and frailty in women.

We also found that higher SHBG concentrations were associated with lower BMD T-score in the lumbar spine, independent of sex hormones (free testosterone and estradiol) and liver disease. These findings are consistent with other studies in postmenopausal women showing that higher SHBG concentrations are associated with increased risk of bone fracture (16, 41-43). Also, SHBG concentrations were reported to correlate negatively with BMD at the distal radius (44), proximal femur (45, 46), trochanter, and lumbar spine (45). Furthermore, there is evidence that postmenopausal women with SHBG polymorphisms (A allele compared to the GG genotype for rs1799941) and resultant increases in SHBG concentrations have significantly lower BMD at the proximal femur (47). It has been hypothesized that SHBG may have an antiestrogenic effect, which may mediate its negative effect on bone metabolism (48). However, the present study revealed that high SHBG was independently associated with BMD even after adjustment for estradiol and free testosterone.

The pathophysiology of our findings is not entirely clear. There is evidence showing that SHBG is not only a carrier protein that affects the bone indirectly, but also a mediator that modulates membrane receptors including estrogen receptors (9, 10, 49). However, in transgenic mice that overexpress the human SHBG transgene (SHBG-Tg), no clear bone phenotype was observed, arguing against a direct effect on skeletal health. It should be noted that wild-type mice lack expression of hepatic SHBG, so it is unclear whether this model is generalizable to human physiology (5). Further investigation is required to better understand the direct effect of SHBG on bone in humans.

An alternative hypothesis is that the effect of SHBG on bone may be confounded by the degree of adiposity. Lower BMI is a well-known risk factor for low BMD (44). Given the reciprocal relationship of SHBG and adiposity (8), our observed association between higher SHBG and lower lumbar spine BMD could be explained by decreased adiposity in individuals with SHBG concentrations. However, we adjusted for whole-body fat mass by DXA in multivariable analysis and the association persisted.

Men: Sex Hormone–binding Globulin, Physical Function, and Osteoporosis

Unlike what we found in women, we did not find associations between SHBG concentrations and physical function, frailty, and T-score at the spine or hip in men. Previous studies have reported that higher SHBG concentrations in men were associated with greater odds of frailty (OR, 1.25-3) (15, 50) or frailty phenotypic components, such as weight loss and exhaustion (49). Moreover, in men, higher SHBG concentrations have also been associated with increased risk of bone fracture and osteoporosis (16, 43, 50). Although we did not confirm these prior observations in men, it is possible that our study was underpowered to identify associations between SHBG and grip strength, frailty, and BMD as we showed in women, or perhaps there is a biological effect driving these associations that requires further investigation.

Free testosterone and total testosterone have been the focus of the role of hormones associated with preserved physical function in men (51, 52). In fact, there are limited studies that have examined the relationship between circulating SHBG concentrations and physical function; most showed no association or contradictory findings in men (51, 53, 54). Previous studies have shown that SHBG concentrations in men were not associated with physical function measures such as chair stands, gait speed, lower-extremity power, and grip strength (51, 53). In other studies, higher SHBG concentrations were associated with an attenuated decline in grip strength and appendicular lean mass, but these associations were not independent of sex hormone levels in men (54).

The lack of consistency between our study and others may stem from inherent differences between the study populations under investigation. The age range was higher in the aforementioned prior studies, and the majority of participants in those studies were White, a contrast to our younger, all African American cohort. In earlier studies, investigators also did not account for whole-body lean mass.

Sex and Sex Hormone–binding Globulin

In our study, SHBG concentrations were higher in women, but these sex-based differences did not persist after the model was adjusted for body composition (whole-body lean and whole-body fat mass). This suggests that the association of higher SHBG concentrations with sex may be confounded by body composition. In fact, we also found that higher SHBG concentrations were associated with lower whole-body lean mass and lower whole-body fat mass in the fully adjusted model in both sexes. The mechanisms for this are unknown. Our findings are consistent with the well-established inverse relationship between central fat distribution, BMI, and SHBG (55, 56). In addition, it has been previously reported in cross-sectional studies that higher SHBG concentrations were associated with lower muscle mass and muscle volume in older adult women (57) and with whole-body lean mass in healthy men (58).

HIV/Hepatitis C Virus and Sex Hormone–binding Globulin

Notably, in our study, higher SHBG concentrations were also associated with detectable HCV and HIV after the adjustment of several markers of liver damage including AST, ALT, and liver stiffness. Recently, we and others have reported that SHBG concentrations are higher in people with HIV and HCV (11, 12, 59). The mechanisms underlying the association of SHBG concentrations with chronic viral infections such as detectable HIV and HCV are not clear. We speculate that systemic inflammation may influence circulating SHBG concentrations with HIV; even with effective antiretroviral therapy, heightened systemic inflammation persists compared to matched controls. Among people with HCV, it been shown that HCV viral clearance reduces SHBG levels (12). Systemic inflammation is also associated with geriatric conditions such as frailty among people with HIV and HCV (40, 60). Higher circulating SHBG concentrations may be a compensatory mechanism that protects against inflammatory cytokines and systemic inflammation associated with HIV and HCV, with preclinical studies demonstrating that SHBG decreases inflammation (61). The relationship between inflammation, SHBG production, and regulation requires further investigation.

Our study has some strengths and limitations. First, our study is cross-sectional, did not include younger adults, and all the study cohort was African American, so it may not be generalizable to younger people or other racial groups. Also, owing to our cross-sectional data we could not determine if SHBG mediates the effect of HIV and HCV on bone, muscle, and frailty.

At the same time, there were several strengths of the study. Namely, it included women and men with detailed exploration and quantification of several aging-related outcomes. This study also included evaluation of several key aging-related outcomes, and detailed phenotyping including ascertainment of sex hormones using state-of-the art methodology, body composition measurements, assessment of liver disease, and vulnerable populations across 4 distinct groups with and without HIV and HCV infection.

Conclusion

Our data demonstrated that HIV and HCV viremia are associated with higher SHBG and also that a sex-specific association in women with and without HIV and HCV with higher SHBG concentrations are associated with lower grip strength, greater odds of frailty, and lower BMD T-score at the spine. This finding may support the notion that SHBG is a biomarker that should be examined for the early detection of physical function decline and osteoporosis, particularly in a high-risk female population. Investigating the mechanisms underlying associations among SHBG concentrations and chronic viral infections, aging, sex, and aging-related comorbidities may help inform our understanding of an accelerated aging phenotype among individuals with HIV and HCV.

Glossary

Abbreviations

ALIVE

AIDS Linked to the IntraVenous Experience

ALT

alanine transaminase

AST

aspartate transaminase

BMD

bone mineral density

BMI

body mass index

CV

coefficient of variation

DM

diabetes mellitus

DXA

dual-energy x-ray absorptiometry

HCV

hepatitis C virus

OR

odds ratio

PWH

people with HIV

SHBG

sex hormone–binding globulin

SPPB

Short Physical Performance Battery

ULN

upper limit of normal

Financial Support

This work was supported by the National Institutes of Health/National Institute of Standards and Technology (NIH/NISA; No. U01DA036297), NIH/National Institute of Allergy and Infectious Diseases (NIAID; Nos. R01AI120733 and K24AI120834). Additional support was provided by the Johns Hopkins Institute for Clinical and Translational Research (ICTR), which is funded in part by the NIH/National Center for Advancing Translational Sciences (NCATS; No UL1 TR003098.). Its contents are solely the responsibility of the authors and do not necessarily represent the official view of the Johns Hopkins ICTR, NISA, NIAID, NCATS, or NIH.

Author Contributions

Acquisition, analysis, and data analysis: J.P.D.; critical revision of the manuscript for important intellectual content, editing of the manuscript: all authors; editing of data analysis: R.V.; study concept, design, and interpretation of the data and editing of the manuscript: T.T.B.

Disclosures

T.T.B. has served as a consultant to ViiV Healthcare, Janssen, Merck, Gilead Sciences, and Theratechnologies. The other authors have nothing to disclose.

Data Availability

Some or all data sets generated during and/or analyzed during the present study are not publicly available but are available from the corresponding author on reasonable request.