Hormone Therapy and Fractures in Postmenopausal Women

Abstract

Background:

Fracture rates have been reported to be higher among older women living with HIV (WLWH) than HIV- women. Hormone therapy (HT) with estrogen can reduce vasomotor symptoms (VMS) associated with menopause and prevent fractures. Since data are limited on the benefits of HT use in WLWH, we examined associations of HT use and fractures.

Methods:

A prospective study of 1765 (1350 WLWH and 415 HIV-) postmenopausal Women’s Interagency HIV Study (WIHS) participants was performed, including self-reported HT use and fracture data from 2003-2017. Proportional hazards models determined predictors of new fractures at any site or at typical fragility fracture sites (hip, spine, wrist).

Results:

At the first postmenopausal visit, the median (IQR) age of WLWH was slightly younger than HIV- women [49.8 (46.4., 53.0) vs 50.7 (47.5, 54.0), p=0.0002] and a smaller proportion of WLWH reported presence of VMS (17% vs 26%, P<0.0001). A greater proportion of WLWH than HIV- women reported HT use (8% vs 4%, p=0.007) at the first postmenopausal visit. In multivariate analyses, white race and smoking were significant predictors of incident fracture at any site, but HT (p=0.69) and HIV status (p=0.53) were not.

Conclusion:

Our study did not find evidence of benefit or harm with regards to fracture outcomes in postmenopausal WLWH receiving HT. Further research is needed to determine whether HT has benefits beyond treatment of VMS such as prevention of adverse aging-associated outcomes.

INTRODUCTION

With successful antiretroviral therapy (ART), over half of women living with HIV (WLWH) in the United States are ≥ 50 years old; many are transitioning through menopause. Menopause is marked by declining estrogen production and often accompanied by vasomotor symptoms (VMS) such as hot flushes and night sweats.^[1] Women experience frequent VMS for 7.4 years on average, with Black women having the most enduring VMS of any racial/ethnic group.^[2] Some studies suggest that WLWH experience menopause at earlier ages ^[3–6] and VMS more commonly than women without HIV.^[7–10] Bone loss accelerates substantially during the first two years of menopause and is a major risk factor for fractures in older age.^[11] Fracture rates are 30-70% higher in older adults with HIV compared to the general population,^[12–14] and may be related to higher levels of pro-resorptive inflammatory cytokines, negative effects of specific ART on bone cells, and a higher prevalence of traditional osteoporosis risk factors.^[15]

Hormone therapy (HT), given as estrogen-only therapy after hysterectomy or as estrogen plus progestogen for women with an intact uterus, is considered the most effective treatment of VMS ^[16–18] and also prevents menopausal bone loss and decreases long-term risk of fracture.^[19–21] However, menopausal HT is uncommon in the general population and among WLWH^{[22, 23]} due to lingering concerns about cardiovascular and breast cancer risk raised by the Women’s Health Initiative studies.^[24–26] We previously found that fracture rates were higher in WLWH than HIV-seronegative women of similar age in the Women’s Interagency HIV Study (WIHS).^[27] In order to explore potential benefits of HT for WLWH, this analysis evaluates whether HT exposure reduces rate of fracture.

METHODS

Study Design: This was an observational study of time to fracture. The WIHS enrolled a prospective cohort of women in 1994–1995 from 6 sites nationally, and with additional enrollment in 2001–2002 and 2011–2012 including WLWH and HIV-seronegative women at increased risk of acquiring HIV.^{[28, 29]} Participants attended semiannual visits which included an interviewer-administered questionnaire, physical examination, and collection of laboratory specimens. Study Outcomes: Starting in 2003 (visit 17), all WIHS participants were asked whether they ever had a fracture of the hip, wrist, spine or other body site. Fracture type(s) was determined and classified as fragility (resulting from fall from standing height or less) and non-fragility. Inclusions and Exclusions: In all subsequent semi-annual study visits, participants were asked if they had new fractures (since the previous visit) with type of fracture similarly determined. We restricted analysis to participants that answered fracture questions at least once, met criteria for postmenopausal status and additionally excluded participants with a history of fracture prior to the first postmenopausal visit. Postmenopausal status was determined by participant report of hysterectomy or amenorrhea for 12 months (i.e., lack of menses in the prior 6 months at 2 consecutive semiannual visits).^[19] The first postmenopausal visit was the earliest post hysterectomy visit or otherwise of the first of 2 consecutive visits that determined postmenopausal status. Study population: In total, 1765 participants (1350 WLWH) with at least one semi-annual study visit after the first postmenopausal visit, were included in the analysis. This analysis uses data from 2003 (visit 18) through 2017 (visit 47), including all fractures at any body site. ^{[27, 30, 31]} Other covariates: Demographics, HIV treatment history, and known predictors of fracture ^{[27, 30, 31]} were quantified, including: age; self-reported race/ethnicity (non-Hispanic White, non-Hispanic Black, Other/Hispanic); body mass index (BMI) in kg/m²; current cigarette smoking; heavy alcohol (>12 drinks per week); injection drug use (IDU), opiates or cocaine use; calcium supplementation; glucocorticoid use (ever); history of diabetes; estimated glomerular filtration rate (eGFR) using the Modification of Diet in Renal Disease (MDRD) calculation; and active Hepatitis C virus (HCV) status by the presence of detectable plasma HCV RNA level at or prior to the index visit. Frequency of vasomotor symptoms (hot flashes, night sweats, cold sweats) during the previous two weeks were recorded as not at all, 1-5 days, 6-8 days, 9-13 days, or daily in WIHS semi-annual visits.^[32] Significant HT exposure was defined by use on ≥2 consecutive study visits.

STATISTICAL METHODS:

Means, medians, standard deviations, interquartile ranges, and proportions summarized study variables, depending upon whether they were continuous or categorical. This includes serial cross-sectional use of HT. P-values for comparisons of groups are from rank tests for continuous and exact tests for categorical variables. Proportional hazards models determined predictors of new fracture incidence by time since first postmenopausal visit. In these models, HT use was time updated; all other covariates were time invariant and assessed at the first postmenopausal visit. For each subject, the end of follow-up was either the visit with first fracture event, or the last visit prior to visit 48 if no fracture was reported. Only the first new fracture was included in the analysis. Bivariate analyses comparing each predictor to the outcome were considered for WLWH and women without HIV together. All variables were entered into multivariate proportional hazards models.

RESULTS:

Participant characteristics at postmenopausal visit

WLWH were younger than women without HIV (median age of 49.8 vs 50.7, p=0.0002) at the first postmenopausal visit (Table 1). WLWH were less likely to be non-Hispanic Black (68% vs 75%, p=0.001) and have BMI>30kg/m² than women without HIV (39% vs 58%, p<0.0001). WLWH were less likely to have Diabetes Mellitus (20% vs 27%, p=0.004), but more likely to have detectable HCV RNA (30% vs 22%, p=0.003) and eGFR<60 ml/min (11% vs 6%, p=0.002). WLWH were also less likely to report current smoking, heavy alcohol consumption, cocaine use than women without HIV, but more likely to report calcium supplementation (Table 1). Among WLWH at the index visit, median CD4 T cell count was 490 (288, 714 cells/mm³), 82% reported antiretroviral therapy (ART) use (including 41% on protease inhibitor and 38% on tenofovir disoproxil fumarate-containing regimens), and 65% had an HIV RNA <200 copies/ml on ART. WLWH who reported HT use at any time during the study period (n=230) were younger, more likely to be white, to be a non-smoker, and take calcium supplementation than WLWH who did not use HT (n=1123), but did not differ in other respects at the first postmenopausal visit (Table S1).

Table 1.

Characteristics at fist postmenopausal visit (Index visit) and incident fracture after index visit

Variable	Women living with HIV (N=1353)	Women without HIV (N=412)	P-value
Age at index (years, Median(IQR))	49.8 (46.4,53.0)	50.7 (47.5,54.0)	0.0002
Age > 60 years	34 (3%)	13 (3%)	0.48
Race/ethnicity			0.001
White	232 (17%)	40 (10%)
Black	925 (68%)	311 (75%)
Hispanic/Other	196 (14%)	61 (15%)
WIHS Site			0.63
Bronx/Manhattan	243 (18%)	84 (20%)
Brooklyn	174 (13%)	55 (13%)
Washington DC	162 (12%)	49 (12%)
Los Angeles	131 (10%)	29 ( 7%)
San Francisco	194 (14%)	55 (13%)
Chicago	160 (12%)	45 (11%)
Southern	289 (21%)	95 (23%)
Weight (kg, Median(IQR))	73.0 (61.7,88.0)	82.6 (68.0,96.6)	< 0.0001
BMI ≥30 kg/m2	532 (39%)	239 (58%)	< 0.0001
Smoking	653 (48%)	233 (57%)	0.003
Alcohol use	88 (7%)	63 (15%)	< 0.0001
Injection drug use	29 (2%)	14 (3%)	0.15
Opiate use	132 (10%)	43 (10%)	0.69
Cocaine use	18 (1%)	12 (3%)	0.03
Diabetes Mellitus	274 (20%)	110 (27%)	0.005
HCV status (RNA+)	402 (30%)	92 (22%)	0.003
eGFR < 60 ml/min (by MDRD)	145 (11%)	23 (6%)	0.002
Glucocorticoid use (ever)	449 (33%)	118 (29%)	0.08
Calcium supplementation	114 (8%)	17 (4%)	0.004
Vasomotor symptoms >1 time per day	224 (17%)	108 (26%)	<0.0001
Hormone Therapy	114 (8%)	17 (4%)	0.007
Incident fractures after index
Fracture at any site*	234 (17%)	62 (15%)	0.29
Hip fractures	13 (1%)	1 (0%)	0.15
Spine fractures	9 (1%)	4 (1%)	0.53
Wrist fractures	27 (2%)	6 (1%)	0.48

^*In addition to typical fragility fracture sites (hip, spine, wrist), fractures occurred in ankles, clavicles, elbows, feet, hands, knees, legs, pelvis, ribs, shoulders/arms, and unidentified sites

Vasomotor symptoms and HT use

At the index visit, fewer postmenopausal WLWH than women without HIV reported having VMS at least one time per day for the 2 weeks prior to their last visit (17% vs 26%, p<0.0001). At the index visit, 8% of postmenopausal WLWH and 4% of women without HIV reported HT use (p=0.007). Among those who used HT, the following indications were reported: menopausal symptoms or VMS (50%), post-hysterectomy (19%), osteoporosis prevention or treatment (7%), cardiovascular and neuropsychiatric reasons (1%), and other/non-identified (22%). Over the 14-year observation period, the proportion of postmenopausal women reporting HT steadily decreased. In 2003, 19% and 15% (p=0.37) of postmenopausal with and without HIV, respectively, reported HT use at their first semi-annual study visit. By 2017, the final year for this analysis, only 3% of postmenopausal with and without HIV reported HT at their first semi-annual visit.

Associations with incident fracture

During the period of observation, incident fractures at any body site occurred in 17% and 15% of women with and without HIV, respectively (p=0.09). Similarly, the incidence rate for fracture was 3.26 vs 2.73 per 100-person years for women with and without HIV, respectively (p=0.22). Fragility fractures at the hip, spine or wrist were uncommon and did not differ between women with and without HIV (Table 1). In bivariate analyses, race (white vs Black) and smoking (ever) were statistically associated with higher incident fracture (p ≤ 0.05), while HIV status and HT use were not (Table 2). In multivariate analyses, the above associations of white race and smoking with incident fracture remained statistically significant.

Table 2.

Factors Associated with Incident Fracture in Proportional Hazards Models

Variable	Univariate	P-value	Multivariate	P-value
Hormone Therapy: Yes vs. No	1.21 (0.78, 1.87)	0.39	0.98 (0.62, 1.53)	0.91
HIV Status	1.20 (0.90, 1.58)	0.21	1.12 (0.83, 1.50)	0.46
Age per-5 at baseline	0.97 (0.88, 1.07)	0.57	0.94 (0.85, 1.03)	0.19
Race: White vs. Black	1.87 (1.43, 2.46)	<0.0001	1.87 (1.40, 2.50)	<0.0001
Race: Other vs. Black	0.80 (0.56, 1.14)	0.22	0.90 (0.62, 1.30)	0.57
BMI (kg/m2)	1.00 (0.98, 1.01)	0.54	1.00 (0.98, 1.02)	0.98
Smoking: Current vs. Never	1.47 (1.05, 2.05)	0.03	1.42( 0.99, 2.05)	0.06
Smoking: Former vs. Never	1.75 (1.23, 2.49)	0.002	1.80 (1.25, 2.59)	0.0021
Alcohol (>12 drinks/wk)	1.27 (0.83, 1.93)	0.27	1.19 (0.76, 1.84)	0.45
HCV status	1.18 (0.92, 1.50)	0.19	1.06 (0.82, 1.38)	0.64
Cocaine: Former vs. Never	1.71 (0.76, 3.85)	0.20	1.57 (0.67, 3.69)	0.30
Cocaine: Current vs. Never	1.01 (0.76, 1.32)	0.96	0.93 (0.69, 1.25)	0.63

DISCUSSION

In this cohort of postmenopausal women, WLWH were less likely to report VMS and more likely to report HT use than women without HIV at the time of their first postmenopausal visit. We found an association of established predictors of fracture, white race and smoking, with incident fractures, but did not find an association of HT use with decreased incident fracture as hypothesized.

In contrast to previous fracture analyses in this cohort,^[27] we did not find a statistically significant difference between fracture incidence by HIV status. This could be explained by the smaller sample size and shorter follow-up. By restricting the analyses to postmenopausal women, the current sample size was smaller (n=1,765) than prior (n=2,375), and the median follow-up duration was reduced from 10 to 4 years; however, fracture incidence rate was higher among postmenopausal WLWH in this analysis (3.26 per 100 person-years) than in WLWH over age 40 in the prior analysis (2.19 per 100 person-years).^[27] In contrast to other studies,^[7–10] we found that VMS was less frequently reported in WLWH, but our analysis was limited to data from the first postmenopausal visit and the questionnaire utilized. A comprehensive comparison of VMS patterns and severity between WLWH and women without HIV are beyond the scope of this analysis.

Concerns related to adverse cardiovascular outcomes highlighted by the Women’s Health Initiative (WHI) studies in 2002 ^{[25, 26, 33]} led to declining HT use among women in the United States. However, subsequent analyses of the WHI found that risk of cardiovascular events depended on the timing of hormone exposure: no excess risk was observed in women under 60 years of age who were less than 10 years post-menopause.^{[34, 35]} Among women without HIV, mortality rates were actually lower in younger postmenopausal women on HT versus HT-nonusers. As a result, the North American Menopause Society (NAMS), the Endocrine Society, and the International Menopause Society have determined that benefits of HT may outweigh risks for certain populations of women, depending on age, onset of menopause, duration of HT, and comorbid conditions.^[16–18] Despite these published guidelines and the availability of safer transdermal formulations of estrogen, HT use remains low among peri- and postmenopausal women under age 60.

Our data are consistent with a 2018 report from the Metro Vancouver Study that found only 5.5% of menopausal WLWH reported being on HT.^[22] WLWH are more likely to have early menopause or primary ovarian insufficiency than women without HIV,^{[3, 4]} which are both indications for HT according to the NAMS. Hormonal changes that begin during menopause not only result in increased risk fracture, but also cardiovascular disease and neurocognitive disorders, comorbidities that disproportionately affect WLWH.^{[12–14, 36, 37]} HT may have benefits beyond treatment of VMS in reducing risk for several aging-related comorbidities relevant to HIV, but data are lacking.

The low overall utilization of HT which declined over successive years limited our ability to draw conclusions about its effect on fractures. Our study assessment of HT was also limited and did not allow differentiation of estrogen formulation, dose or route of administration. In addition, we lacked specificity on use of non-hormonal treatments for osteoporosis including bisphosphonates, teriparatide, and denosumab.

Conclusion

The potential benefits of HT for reducing fractures in postmenopausal WLWH remain unclear. Unfortunately, even well-characterized longitudinal cohorts such as the WIHS are limited by low overall HT utilization, potential indication bias of HT, and low event rates for clinically defined complications such as fracture, as well as cardiovascular events and malignancies. Defining a role for HT beyond treatment of menopausal symptoms in WLWH will therefore require a larger cohort study or randomized clinical trial with use of clinical surrogates.