Comorbidities in Women Living With HIV

A Systematic Review

Sonia Raffe; Caroline Sabin; Yvonne Gilleece


HIV Medicine. 2022;23(4):331-361. 

In This Article


Study Selection

Our initial database search identified 3050 documents. An additional 35 articles were identified through other sources. After removal of duplicates, 2075 articles remained for abstract review, with 153 full-text articles assessed for eligibility and 38 included in the final review (Figure 1). Article exclusion was predominately due to inadequate women in the cohort studied or lack of sex-based analysis.

Figure 1.

Flow chart of study selection process

Cardiovascular Disease

Seventeen articles relating to CVD met eligibility criteria (Table 1).

Comparison of CVD Risk Factors Between WLWH and HIV-negative Women. Three studies compared prevalence of traditional risk factors between WLWH and HIV-negative women. The first, a cross-sectional analysis of the Cardiovascular Risk Evaluation and Antiretroviral Therapy Effects (CREATE) cohort, compared the burden of traditional CVD risk factors in a cohort of ethnically and socioeconomically diverse PLWH in London, with a control population from UK general practice. The 253 WLWH in the cohort were more likely to smoke (16% vs. 12.7%) and have a waist circumference > 88 cm (64% vs. 40.1%) than women in the general population. Prevalence rates of hypertension (9% vs. 13%) and diabetes (2% vs. 3%) as well as mean total cholesterol (TC; 4.74 vs. 5.2mmol/L) and high-density lipoprotein cholesterol (HDL-C) (1.54 vs. 1.5 mmol/L) were similar regardless of HIV status. However, WLWH were younger than the women in the general population (mean age 38.8 vs. 52.1 years) and, while 9% of the WLWH were white, ethnicity data were unavailable for the general population.[16]

In a US study, the 710 WLWH in the cohort were more likely to have elevated triglycerides (> 150 mg/dL, 33.6% vs. 23.4%), reduced HDL-C (< 50 mg/dL, 53.8% vs. 41.1%) and to be smokers (59.2% vs. 40.4%) compared with HIV-negative women, but were less likely to have a body mass index (BMI) > 30 kg/m2 (25.3% vs. 44.5%) or to be hypertensive (22.9% vs. 28%).[17] By contrast, Tariq et al.[18] found no difference in prevalence of hypertension (30.2% vs. 26.6%; p = 0.69), BMI > 30 kg/m2 (37.2% vs. 25.7%; p = 0.12), TC:HDL-C ratio > 5 mg/dL (7% vs. 8.3%; p = 0.95) or glucose > 5.5 mmol/L (12.8% vs. 16.5%; p = 60) between WLWH and HIV-negative women aged > 50 years. Notably, this UK-based cohort was smaller (86 WLWH and 109 HIV-negative women), the WLWH were younger (median age 54 vs. 57 years), more likely to be black African (67.4% vs. 20.2%; p = 0.0001), single (79.1% vs. 48.6%; p = 0.0001) and unemployed (62.8 vs. 41.3%; p = 0.005). A fourth study, without an HIV-negative control group, reported a high prevalence of modifiable CVD risk factors in 161 African-American WLWH in the USA (diabetes, 33%; dyslipidaemia, 40%; hypertension, 42%; current smokers, 42%; overweight/obese, 63%)[19]

Cortés et al. reported that among postmenopausal Hispanic and African-American women, the median Framingham Risk Score (FRS) did not differ between the 109 women living with, and 43 women without, HIV[20] (14.6 vs. 15.5; p = 0.73) although the WLWH were younger and more likely to be African-American (39.4% vs. 20.9%). In WLWH, older age at HIV diagnosis was associated with higher FRS.

Comparison of CVD Risk Factors Between WLWH and men Living With HIV (MLWH). Four studies compared the prevalence of CVD risk factors between WLWH and MLWH. In a US cohort of 7436 (25% female) PLWH aged > 50 years,[21] WLWH were significantly more likely to be black (age 50–64 years: 62% vs. 33%; age ≥ 65 years: 58.6% vs. 29.7%) and live-in poverty (50–64 years: 59.9% vs. 35%; ≥ 65 years: 47.3% vs. 27.1%) than MLWH. WLWH were less likely to have private health insurance (50–64 years: 22% vs. 36.1%; ≥ 65 years: 19.5% vs. 41.1%) or to have educational attainment beyond high school (50–64 years: 37.7% vs. 58.3%; ≥ 65 years: 28.8% vs. 53.3%). Following adjustment for these factors, WLWH aged 50–64 years were more likely to have a BMI > 24 kg/m2 (adjusted prevalence difference = 8.4%), hypertension (3.9%) or TC > 200 mg/dL (9.9%); WLWH aged ≥ 65 years were more likely to be diabetic (13.1%) and to have a high TC (18.8) than were MLWH.

By contrast, the international D:A:D cohort of > 49 000 PLWH found many CVD risk factors to be more common in MLWH, including hypertension (7% vs. 10.8%) and diabetes (1.7% vs. 2.7%).[22]

When comparing overall CVD risk between 128 WLWH and 633 MLWH in Germany, WLWH had lower rates of moderate and high 10-year cardiovascular risk, although the women in this cohort were significantly younger (mean age 40.4 vs. 45 years).[23]

A cross-sectional analysis of the HIV Outpatient Study (HOPS) cohort used excess heart age [chronological age – heart age (age of a person with the same predicted risk but with all other risk factors in the normal range)] to estimate the physiological age of the vascular system of 3086 (20% female) PLWH.[24] The excess heart age of the WLWH in the cohort of 13.1 years was greater than that of both the MLWH (11.5 years) and women in the general US population (5.4 years). Although the MLWH and WLWH had a similar age, the WLWH were more likely to be non-Hispanic/Latino black (54.6% vs. 23.6%), coinfected with HCV (25.7% vs. 14.9%) and have a BMI > 30 kg/m2 (41.8% vs. 20%) but were less likely to be educated beyond high school (27.9% vs. 54.4%).

Incidence of CVD Events. Six studies considered the incidence of CVD events. Womack et al. found a higher incidence of CVD [acute myocardial infarction (MI), unstable angina, ischaemic stroke or congestive heart failure] among 710 WLWH (13.5/1000 person-years) compared with 1477 HIV-negative women (5.3/1000 person-years).[17] This risk remained after adjusting for potential confounders [hazard ratio (HR) = 2.8, 95% confidence interval (CI): 1.7–4.6]. WLWH with HIV RNA > 500 copies/mL were at greatest risk (HR = 4.4, 95% CI: 2–9.9). The mortality risk was also higher in WLWH than in HIV-negative women (HR = 2.6, 95% CI: 1.7–3.9).

A 12-year Italian cohort study of 3766 PLWH found a two-fold increased risk of CVD events overall [standardized incidence ratio (SIR) = 2.02] in PLWH as compared with the general population,[25] although the pattern of events appeared to differ by sex, with the relative risk of coronary artery disease being higher in WLWH than in MLWH (WLWH: SIR = 2.91; MLWH: SIR = 1.82) but that of stroke being lower in WLWH (WLWH, 2.07; MLWH, 2.3). Triant et al. also found an increased risk of major adverse cardiac events (acute MI, stroke, angina or coronary revascularization) in PLWH in a cohort of a similar size in the USA compared with HIV-negative controls [incidence rate ratio (IRR) = 1.56], with a greater risk in WLWH (IRR = 2.19) than in MLWH (IRR = 1.35).[26] By contrast with the Italian cohort, WLWH were at greater risk of both MI (IRR = 2.35 vs. 1.33) and stroke (IRR = 2.1 vs. 1.32) than were MLWH. A further US-based cohort study including 1350 WLWH reported that HIV was a risk factor for stroke in women, even after adjusting for established risk factors (HR = 1.21; p < 0.001)[27] and a French cohort also found an increased risk of MI in PLWH compared with the general population [standardized morbidity ratio (SMR) = 1.5, 95% CI: 1.3–1.7], particularly in WLWH (SMR = 2.6, 95% CI: 1.8–3.9).[28]

Peripheral artery disease (PAD) was seen more frequently in Danish women (OR = 1.49, 95% CI: 1.19–1.87) and particularly WLWH (OR = 2.24, 95% CI: 1.06–4.73), compared with the Danish general population. The PLWH were less likely to be of Scandinavian ancestry (76% vs. 89%), more likely to smoke (28% vs. 13%) and have a lower mean BMI (25 vs. 27 kg/m2) and less likely to be hypertensive (48% vs. 61%) than the general population although risk of PAD remained after adjustment.[29]

Management of CVD Risk Factors and Acute Events. Little evidence was available regarding how successfully traditional cardiovascular risk factors are managed in WLWH. Tariq et al. found that regardless of HIV status, few women eligible for lipid-lowering or antihypertensives were receiving treatment, although the numbers in this study were small.[18] One study found that female sex was associated with successful modification of blood pressure among PLWH but with no difference between women and men in smoking cessation or cholesterol reduction.[30]

Ogunbayo et al. evaluated sex differences in management of MI in PLWH in a large US-based cohort.[31] Among those with an acute MI, WLWH were younger (53.1 vs. 54.3), more likely than MLWH to be black (57.2 vs. 36%), to have diabetes (38.5% vs. 24.1%), to be obese (14.1% vs. 5.2%) and to be anaemic (26.7% vs. 16.9%). The WLWH were less likely to have an ST-segment elevation MI (23.4% vs. 34.6%; p < 0.001) and were significantly less likely to have received a percutaneous coronary intervention than were MLWH (54.2% vs. 69.7%; p < 0.01). They were more likely to have no revascularization intervention (37.8% vs. 21.8%; p < 0.01) although the rate of coronary artery bypass grafting was similar.

Pathophysiology of CVD in WLWH. Fitch et al. used cardiac computed tomography (CT) angiography and markers of immune activation to examine the atherosclerotic plaque features and detailed indices of immune activation among 60 WLWH compared with 30 well-matched female HIV-negative controls.[13] The prevalence of coronary plaque was similar between WLWH and controls (37% vs. 38%; p = 0.88) but WLWH demonstrated a significantly higher prevalence of noncalcified coronary plaque (35% vs. 12%; p = 0.04) and a lower prevalence of calcified plaque (6% vs. 26%; p = 0.01) than did the controls. A calcium risk score > 100 was also less frequent in WLWH (15% vs. 2%; p = 0.02). Markers of inflammation (hsCRP and hsIL-6) did not differ by HIV status but markers of immune activation [sCD163 (p = 0.006), sCD14 (p = 0.5), CXCL10 (p = 0.002), percentage CD8+ ( < 0.0001), CD8+ HLA-DR+ (p = 0.0004), HLA-DR+CD38+CD4+ (p < 0.0001) and CD14+CD16+ (p = 0.008)] were all significantly higher in WLWH.

Renal Disease

No studies looking specifically at renal disease in WLWH, and only six studies that presented data disaggregated by sex were identified (Table 2).

Incidence of Renal Disease. Only one eligible study compared the incidence of kidney disease in PLWH with data from an HIV-negative population. The North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD) group reported that the incidence of end-stage renal disease (ESRD)[32] was higher in all PLWH (179/100 000 person-years) but particularly in WLWH (259/100 000 person-years) than in the general population (35/100 000).

An analysis of the ATHENA cohort assessed the impact of ethnicity on the development of chronic kidney disease (CKD) in PLWH,[33] finding a significantly higher prevalence of CKD in those born in sub-Saharan Africa (adjusted OR = 1.49) or Asia and the Pacific region (adjusted OR = 2.27) than in those born in Western Europe. However, the risk of developing CKD over time was similar regardless of country of birth (adjusted HR =1.00, 95% CI: 0.63–1.59). Although no sex-specific data were presented, those born in sub-Saharan Africa were more likely to be female (50.1% vs. 8.9%), younger (36.9 vs. 44 years) and to have lower CD4 count (380 vs. 460 cells/μL).

A 2018 study of 4337 (19% female) PLWH in Spain found that WLWH were significantly more likely to have mild renal impairment than MLWH (29.2% vs. 23.9%; p = 0.002).[34] WLWH were older (47 vs. 44 years), more likely to be black (5.3% vs. 1.4%), had a higher prevalence of HCV coinfection (32.8% vs. 19.4%), had lived with HIV for longer (17.5 vs. 9.2 years) and had a prior AIDS diagnosis. However, female sex remained a risk factor for mildly decreased renal function on multivariate analysis (OR = 1.02–1.48; p = 0.031).

Several large cohort studies of PLWH, while primarily looking at longer-term outcomes in PLWH with impaired renal function also found an association between female sex and renal impairment. A baseline analysis of the UK Collaborative HIV Cohort (UK CHIC) Study found that WLWH were over-represented among PLWH with an estimated glomerular filtration rate (eGFR) between 44 and 30 mL/min (29.2% vs. 70.8%) and an eGFR < 30 mL/min (42.5% vs. 57.5%).[35] The EuroSIDA study group also found a disproportionate number of women with a baseline eGFR < 60 mL/min (28.2%; p = 0.0004)[36] and, in a sub-study, female sex was associated with an increased incidence of developing CKD [relative hazard (RH) = 1.68; p = 0.0013].[37] In the development of a validated long-term risk score model for CKD in PLWH, the D:A:D Study also identified female sex as a significant predictor of CKD.[38]

Bone Disease

Six studies looking at bone health in WLWH were included (Table 3).

Bone Mineral Density in WLWH Compared With HIV-negative Women. Dual-energy X-ray absorptiometry (DXA) was used in two studies to compare bone mineral density (BMD) between women with and without HIV. Among 128 postmenopausal Hispanic and African-American women,[39] WLWH had a lower baseline BMD at many sites (lumbar spine, total hip and distal radius) than HIV-negative women. The study also reported a 2.4-fold greater annualized percentage decrease in BMD in WLWH at the lumbar spine (p = 0.0009), 3.7-fold at the distal radius (p = 0.006) and 1.7-fold at the ultra-distal radius (p = 0.02), even after adjustment for traditional risk factors. HIV status was associated with reduced BMD in an analysis of 440 women from the Women's Interagency HIV Study (WIHS) cohort, although the WLWH in this cohort were older (44 vs. 37 years), more likely to be HCV-coinfected (32% vs. 14%) and to be postmenopausal (26% vs. 3%) than the HIV-negative comparison group.[40]

Fracture Incidence. Yin et al. found no significant difference in fracture incidence between postmenopausal African-American women living with or without HIV (10% vs. 8%).[39] Similarly, an analysis of 5826 (21% female) PLWH from the HOPS cohort found no significant difference in crude fracture rate by HIV status,[41] although authors reported a higher incidence of vertebral (18% vs. 4%; p < 0.01) and femoral neck (5% vs. 1%; p = 0.004) fractures in WLWH but significantly fewer fractures at non-fragility sites (69% vs. 86%; p < 0.01) than among HIV-negative women. By contrast, among 2375 women in the WIHS cohort, although the overall incidence of fracture was higher (2.19 vs. 1.54/100 person-years; p = 0.002) in WLWH than in HIV-negative women, the incidence of fragility fractures was similar (0.56 vs. 0.39/100 person-years; p = 0.13) despite the WLWH being older (40 vs. 35 years), of lower weight (74.5 vs. 79.7kg), more likely to be HCV-coinfected (24% vs. 15%) and to be postmenopausal (19% vs. 11%) than the HIV-negative women.[42] Gedmintas et al. found no significant difference in incidence rate of either any (IRR = 1.00) or fragility (IRR = 1.26) fractures between 869 WLWH and 2292 MLWH,[43] although no adjustment was made for potential confounders.

HIV-related Factors and Bone Disease. HIV-related risk factors associated with increased fracture incidence included CD4 T-cell count nadir < 200 cells/μL and an AIDS-defining diagnosis.[41,42] Traditional risk factors such as being postmenopausal, HCV coinfection, current smoking, diabetes, substance misuse and decreased total lean mass were also associated with decreased BMD and fracture incidence regardless of HIV status.[40,41]

Libois et al. reported a high prevalence of osteopenia/osteoporosis (31.5%) among 89 premenopausal WLWH,[44] with no difference between ART-naïve women and women on ART, regardless of drug class received. Findings were consistent at both the lumbar spine (13.5% vs. 16% vs. 25.9%) and femoral neck (21.6% vs. 32% vs. 33.3%). After adjustment, only low BMI was a risk factor for osteopenia/osteoporosis (p = 0.004), although notably only two women in the cohort had received ART containing tenofovir disoproxil fumarate (TDF). Sharma et al. reported a significant association between use of nonnucleoside reverse transcriptase inhibitor (NNRTI) and increased BMD at the hip and femoral neck but no association between cumulative use of TDF, protease inhibitors (PIs) or other ART with BMD loss at any site.[40] Among postmenopausal Hispanic and African-American women, 78% of the WLWH were on ART with a mean duration of exposure of 4.5 years,[39] of whom 21% were on a TDF-containing regimen. After exclusion of those not on ART, the annualized rate of bone loss remained greater in WLWH than in HIV-negative women, although this did not differ between those on PI-based vs. NNRTI-based ART at any site. The rate of bone loss was, however, higher in those on TDF than those on non-TDF-containing ART at the lumbar spine (–2.8 vs. 0.7; p = 0.001), the distal radius (–2.3 vs. –0.8; p = 0.02) and the ultra-distal radius (–2.2 vs. –1; p = 0.001), with these differences remaining after confounder adjustment. Bone loss did not differ by TDF exposure at the total hip or femoral neck.

Neurocognitive Disease

Nine studies reporting aspects of neurocognitive disease in WLWH were identified (Table 4).

Risk Factors for Neurocognitive Disease. Only one study, a cross-sectional analysis of cardiometabolic risk factors and cognitive impairment in a cohort of 795 (20% female) older PLWH (> 40 years), included a male comparison group.[45] Although a greater proportion of WLWH than MLWH met the criteria for cognitive impairment (36% vs. 26%; p = 0.003) there were significant differences between the women and men in terms of ethnicity (52% women vs. 25% men were black; p < 0.001) and educational attainment (median 12 vs. 14 years of education; p < 0.001). Increased physical activity (OR = 0.33 for > 3 days/week of physical activity; p = 0.003) and higher HDL-C (OR = 0.78 for every 10 mg/dL higher; p = 0.28) were protective against cognitive impairment in the female participants but not among men. No significant associations between other traditional cardiovascular risk factors or socioeconomic factors were reported in this cohort.

By contrast, Gustafson et al. found that a BMI < 18.5 kg/m2 was associated with poorer performance across a limited battery of cognitive tests in WLWH compared with those with a BMI in the healthy range (18.5–25 kg/m2).[46] This relationship was not seen in HIV-negative women. Obesity (BMI > 30 kg/m2) was related to better performance on some aspects of neurocognitive testing but a worse performance on others in both WLWH and HIV-negative women. No consistent relationships between waist circumference or waist-to-hip ratio and cognitive performance were seen. Of note, most women (69%) in the study had BMI > 25 kg/m2 with only 3.1% of WLWH and 1% of HIV-negative women having a BMI < 18.5 kg/m2.

Pattern of Cognitive Impairment in WLWH. In 2015, Maki et al.[47] examined the association between HIV status and cognition in the WIHS cohort and explored the impact of HIV on the pattern and magnitude of cognitive impairment. In this cohort of 1521 women (13% of the 1019 WLWH had CD4 T-cell count < 200 cells/μL and 53% had an undetectable plasma HIV RNA), HIV status had a small but significant (0.05–0.09 SD units) negative impact on verbal learning and memory, speed of information processing and attention. Of note, the impact of HIV on cognitive function was less than the effect of years of education, age, race/ethnicity, household income and reading level. WLWH with a low CD4 count, high viral load, low education or an AIDS-defining illness were more vulnerable to cognitive deficit.

Drugs and Neurocognitive Disease in WLWH. Two studies examined the impact of drugs on cognitive performance. The first considered medications with known neurocognitive-adverse effects (NC-AE) on 1037 WLWH and 521 HIV-negative women.[48] The WLWH were older (47 vs. 43 years), more likely to have positive HCV antibody (Ab) serology (23% vs. 13%) but less likely to report recent heavy alcohol (14% vs. 22%), marijuana (15% vs. 22%) or recreational drug (6% vs. 8%) use. The WLWH reported using more NC-AE medications than HIV-negative women (p < 0.05), although opioid and anticonvulsant use was similar in both groups.

The WLWH performed worse than HIV-negative women on global function (p = 0.01), memory (p = 0.04), attention/working memory (p = 0.02) and executive function (p = 0.02), but measures of fluency and motor skills were similar. NC-AE medication use was not associated with worse cognitive performance nor did it moderate the association between HIV and performance. However, anticholinergic burden was negatively associated with learning and executive function; a greater association in WLWH than in HIV-negative women suggested an increased cognitive vulnerability.

Meyer et al. investigated the impact of illicit drug use on 1003 women with and 496 women without HIV.[49] Reported drug use was high regardless of HIV status, with recent recreational drug use reported by 9.6% of WLWH and 11% HIV-negative women, and former use by 48.6% and 42.4%, respectively. Regardless of reported drug use, WLWH performed worse than HIV-negative women on total learning, learning slope, delayed recall and recognition (p-values < 0.05). While recent drug users performed worse on learning slope (p = 0.04), delayed recall (p = 0.007) and recognition (p = 0.002), former users did not perform differently to never-users on any measure. Recent drug use (compared with never-use) had a negative impact on verbal learning and memory in WLWH but not in HIV-negative women, suggesting a potential synergistic neurotoxicity between HIV serostatus and recent drug use which remained following adjustment for HIV-specific characteristics.

Stress and Neurocognitive Disease in WLWH. Four studies from the WIHS cohort examined the impact of stress on cognitive impairment. The first of these by Rubin et al. considered the relationship between perceived stress and cognitive performance,[50] demonstrating similar proportions of WLWH and HIV-negative women reporting high perceived stress (38% vs. 36%; p = 0.41). Regardless of HIV status, those with higher perceived stress performed worse on all indices of the Hopkins Verbal Learning test. In WLWH, those with higher levels of perceived stress performed worse on the verbal memory domain (p < 0.001), specifically driven by delayed recall (p < 0.001). This was not seen among the HIV-negative women, where there were no significant differences in either the verbal domain (p = 0.99) or delayed recall (p = 0.76) between those with higher and lower perceived stress. Among the WLWH, those with higher perceived stress had lower CD4 counts and higher plasma HIV RNA; however, the relationship with poor verbal memory and delayed recall persisted after controlling for these factors.

A further analysis of the WIHS cohort examined the relationship between post-traumatic stress disorder (PTSD) and verbal learning and memory in women with and without HIV.[51] Again, the proportions of women with PTSD (defined by the PTSD Checklist-Civilian version) were similar in WLWH and HIV-negative women (18% vs. 16%; p = 0.49) and, regardless of HIV status, PTSD was significantly and inversely associated with cognitive performance, particularly verbal learning (p < 0.001), memory (p < 0.001) and psychomotor speed (p < 0.001). There was no relationship between HIV status and the impact of PTSD on verbal learning or memory, although WLWH meeting criteria for PTSD performed worse on assessment of fine motor skills (p = 0.003). Of note, in this cohort the WLWH were significantly older (p < 0.001), had a higher minority representation, and were more likely to be HCV-Ab-positive (31% vs. 19%; p < 0.001) and to use antidepressants (17% vs. 10%; p < 0.001) and less likely to report excessive alcohol (15% vs. 25%; p < 0.001) or marijuana use (15% vs. 22%; p < 0.001) compared with HIV-negative women.

A small sub-study of 38 predominately African-American WLWH used structural magnetic resonance imaging (MRI) to examine the association between stress, verbal memory and brain volume.[52] In this cohort, 26% reported depressive symptoms, 26% had an elevated PTSD burden and 84% reported having experienced abuse (sexual, physical or domestic). Women with a higher perceived stress score performed worse on testing of verbal memory (p < 0.005). As established in the HIV-negative population, high perceived stress score was associated with smaller volumes bilaterally in the medial temporal region (parahippocampal gyri) and prefrontal cortex regions, both involved in verbal memory. A further study, utilizing functional MRI to explore cortical activation during verbal memory tasks in 36 WLWH, found that patterns of brain activation during recognition, but not encoding, were associated with perceived stress score.[53] Women with higher perceived stress score demonstrated greater deactivation in medial prefrontal cortex.