A cross-sectional study to evaluate the association of hyperbilirubinaemia on markers of cardiovascular disease, neurocognitive function, bone mineral density and renal markers in HIV-1 infected subjects on protease inhibitors

Abstract

Background: Ongoing inflammation in controlled HIV infection contributes to non-AIDS comorbidities. High bilirubin appears to exhibit an anti-inflammatory effect in vivo. We therefore examined whether increased bilirubin in persons with HIV was associated with differences in markers of inflammation and cardiovascular, bone, renal disease, and neurocognitive (NC) impairment.

Methods: This cross-sectional study examined inflammatory markers in individuals with stable HIV infection treated with two nucleoside reverse transcriptase inhibitors and a boosted protease inhibitor. Individuals recruited were those with a normal bilirubin (NBR; 0–17 μmol/L) or high bilirubin (>2.5 × upper limit of normal). Demographic and anthropological data were recorded. Blood and urine samples were taken for analyses. Pulse wave velocity (PWV) measurement, carotid intimal thickness (CIT), and calcaneal stiffness (CSI) were measured. Males were asked to answer a questionnaire about sexual function; NC testing was performed using CogState.

Results: 101 patients were screened, 78 enrolled (43 NBR and 35 HBR). Atazanavir use was significantly higher in HBR. Whilst a trend for lower CIT was seen in those with HBR, no significant differences were seen in PWV, bone markers, calculated cardiovascular risk (Framingham), or erectile dysfunction score. VCAM-1 levels were significantly lower in the HBR group. HBR was associated with lower LDL and triglyceride levels. NBR was associated with a calculated FRAX significantly lower than HBR although no associations were found after adjusting for tenofovir use. No difference in renal markers was observed. Component tests of NC testing revealed differences favouring HBR but overall composite scores were similar.

Discussion: High bilirubin in the context of boosted PI therapy was found not to be associated with differences in with the markers examined in this study. Some trends were noted and, on the basis of these, a larger, clinical end point study is warranted.

Keywords:

• HIV
• Protease inhibitor
• Inflammation
• Bilirubin

Introduction

Since the introduction of combination antiretroviral therapy (cART) to treat HIV there has been a marked reduction in the incidence of AIDS events and death.¹ Persons living with HIV (PLWH) are living longer but there are concerns regarding a possible increase in age-related conditions such as heart disease, osteoporosis, NC disorders, and malignancy secondary to ongoing inflammation² in the presence of viral suppression. Studies have shown PLWH to have an increased risk of cardiovascular disease (CVD) compared to the general population^3,4; this is thought to be multifactorial and may be associated with chronic inflammation,⁵ the use of specific antiretroviral drugs, and a higher prevalence of traditional risk factors.⁶

Elevated bilirubin secondary to Gilbert’s syndrome is associated with a lower risk of cardiovascular events, possibly due to antioxidant or anti-inflammatory effects of hyperbilirubinaemia (HBR).⁷ As well as its antioxidant properties, bilirubin may have anti-inflammatory properties through a number of possible mechanisms. Bilirubin has more recently been shown to have a modulatory effect on T regulatory cell differentiation, indicating it may be protective for both autoimmune and chronic inflammatory conditions.^8–12

The use of ritonavir-boosted atazanavir is associated with unconjugated HBR via inhibition of the UGT1A1 enzyme, mimicking Gilbert’s. We postulate that HBR associated with protease inhibitor use may reduce cardiac risk and may also be relevant in protecting against renal, bone, and NC impairment in the context of HIV infection.

Methods

Ethics statement

Data analyses were conducted on anonymized data and this study was approved by the London Chelsea Ethics Committee and conducted according to Good Clinical Practice guidelines.

Study sample

This is a cross-sectional, case control study. We planned to recruit 50 cases (high bilirubin) and 50 controls (normal bilirubin) aged over 18, over the course of one year. Sample estimate was based on work looking at coronary endothelial function evaluated by assessing the change in coronary artery diameter (percent change in flow-mediated dilatation (FMD)).^13,14 Based on this work, in order to show a significant difference in mean %FMD between patients with elevated bilirubin (which is associated with reduced risk of cardiovascular events) compared to lower bilirubin levels at 5% significance level with certainty that if such a difference existed it will be shown with 80% certainty, a total sample of 98 patients were detemined to need to be recruited.

Participant selection criteria

Following written consent, the participants attended a single study visit during which they underwent the following procedures: demographic data and medical history were recorded, including details about smoking and medication; blood pressure measurement, height and weight and calculation of body mass index (BMI), waist and hip circumference measurement and waist-to-hip ratio (WHR); pulse wave velocity (PWV) from the carotid to radial artery; carotid intimal thickness (CIT); blood and urine samples; NC testing was performed using CogState. Males were asked to answer a voluntary questionnaire about sexual function.

This was a single-center, cross-sectional study. Subjects were HIV-1 positive, suppressed (HIV-1 RNA < 40 copies/mL), and stable on antiretrovirals consisting of either tenofovir/emtricitabine (TDF/FTC) or abacavir/lamivudine (ABC/3TC) plus a ritonavir-boosted protease inhibitor (darunavir or atazanavir) for more than six months and able and willing to provide informed consent. Blood drawn within the last 30 days could be used for the screening assessment. Normal bilirubin was, according to our local laboratory, at normal range with a high bilirubin level defined as greater than two-and-a-half times the upper limit of normal.

•	Group 1 - Controls (NBR): Baseline bilirubin 0–17 μmol/L
•	Group 2 – Cases (HBR): Baseline bilirubin >2.5 times the upper limit of normal (>42.5 μmol/L)

Subjects receiving any lipid-lowering therapy, anticoagulants, hormone supplements, metformin (for lipohypertrophy), or other metabolic therapies (including oral contraceptives and testosterone supplements for documented hypogonadism) had to have to be stable on these for at least three months. Those with successfully treated or spontaneously cleared hepatitis C were permitted to enter the study if infection and/or treatment had occurred at least six months beforehand.

Participant exclusion criteria

Subjects were excluded where bilirubin was greater than 42.5 μmol/L; if they had known cardiovascular disease (angina, coronary artery disease, peripheral vascular disease, stroke, congestive cardiac failure, or myocardial dysfunction), diabetes mellitus, or were receiving antihypertensive therapy or chronic non-steroidal anti-inflammatories (NSAID), including low-dose aspirin, if they had known renal, central nervous system, or other NC disease, or an HIV viral load greater than 400 copies/ml in the preceding 6 months. Using anabolic steroids was also a reason for exclusion, as was any change in antiretroviral therapy in the preceding six months. Finally, active hepatitis B (sAg +ve) or hepatitis C (detectable HCV RNA) meant that subjects could not be included.

Study objectives

The primary objectives of the study were to evaluate differences in cardiovascular and osteoporotic markers in HIV-infected patients with NBR and HBR.

•	the markers assessed for the vascular function primary endpoint included PWV, CIT, vascular endothelial function markers (iCAM and vCAM), lipid fractions/sub fractions, and homeostasis model assessment-estimated insulin resistance (HOMA-IR).
•	markers assessed for the osteoporotic primary endpoint were calcaneal bone mineral density measurement (using ultrasound was undertaken to calculate the calcaneal stiffness index (CSI), which is proven to correlate with bone mineral density).

Secondary objectives were:

•	to assess the proportion with low, medium, or high cardiovascular risk by Framingham risk score; to assess any association of HBR with IL-6, d-dimer, hsCRP, and symptoms of erectile dysfunction
•	to assess the proportion with low, medium, or high osteoporotic risk by FRAX score
•	to evaluate any association of HBR with NC function (CogState) and renal markers (uric acid, urinary protein/creatinine ratio (uPCR), retinal binding/protein ratio (RBP))

Study procedures

At the study visit, the subjects were asked to fast for 10 h prior to their attendance (except for water) and to refrain from smoking for this period, or taking recreational drugs for 24 h prior.

PWV was measured using a Complior SP machine (TensioMed, Hungary) and measurements were taken from the carotid and radial artery according to the manufacturer’s instructions. Three measurements were made and the average was recorded.

CIT was measured at the common carotid artery (CCA) using a GE-Logic ultrasound machine.

The International Index of Erectile Function (IIEF) was used to assess erectile function. For MSM a version of the IIEF previous adapted and validated for this population was used. Participants were asked a series of questions assessing erectile function over the previous four weeks. These were scored and a composite score then used to categorize those with moderate to severe erectile function.

Calcaneal ultrasound¹⁵ was undertaken using the GE-Achilles Insight machine to calculate the CSI and estimate the t-score.

NC testing was performed using a computer-based test battery (CogState). This assessment was performed on a desktop computer and comprised eight tasks. The desktop computer was situated in a quiet room where the patient was not disturbed during the procedures and took approximately 10–15 min to complete. The tasks were in the form of card games and specifically assessed simple reaction time, choice reaction time, complex reaction time, continuous performance, one-back working memory, matching, incidental learning, and associate learning. These tests allowed aspects of NC function to be assessed including visual motor function (CHASE), executive function (GML), psychomotor function (DET), visual attention, (IDN), visual learning and memory (OCL, CPAL), and working memory (ONB). An overall composite score was calculated for each subject.

Statistical methods

All statistical analyses were performed in STATA. All p-values presented are two-sided. Categorical data were compared between groups 1 (NBR) and 2 (HBR) using Fisher’s exact test. Quantitative data with hyper geometric distribution were compared between the two groups using the non-parametric Wilcoxon test. Between-group comparisons of the CogState data were performed using the Mann–Whitney U test.

We undertook further analyses to determine the importance of the nucleos(t)ide backbone, comparing those receiving ABC/3TC to those receiving TDF/FTC.

Results

Out of 101 patients who were screened, 78 were eligible and enrolled. The main reasons for exclusion between screening and enrollment were concomitant medication, changing antiretroviral regimen, or having a bilirubin value out of range. The mean age of participants was 48.3 years (sd 8.2); 93% were male and 84% white (Table 1). 43 had normal bilirubin levels (NBR; group 1) and 35 had high bilirubin (HBR; group 2). PI use differed between the two groups; atazanavir use was significantly higher in HBR compared with NBR (94 vs. 16%, p < 0.001) and darunavir use was significantly higher in the NBR compared with HBR group (79 vs. 3%, p < 0.001). Other than the expected high bilirubin and higher atazanavir use in the HBR group there were no other significant differences observed. Proportion of smokers between the two groups was similar (21% of NBR vs. 34% HBR; p = 0.209).

Table 1 Demographic characteristics

	Total = 78		NBR (n = 43)		HBR (n = 35)
	Mean	s.d.	Mean	s.d.	Mean	s.d.
Age (years)	48.3	8.2	46.8	7.8	47.9	7.9
Sex – Male (%)	94 (93%)		38 (88%)		34 (97%)
Race – White (%)	85 (84%)		33 (77%)		34 (97%)
BMI (kg/m^2)	25.3	4.4	24.8	5.7	25.4	2.9
WHR	0.97	0.07	0.98	0.08	0.97	0.05
SBP (mmHg)	129.3	13.1	128	12.4	128	14.3
DBP (mmHg)	79.5	8.7	79.0	9.2	80.1	8.5
Total bili screening (μmol/L)			10.7	3.2	67.4	22.4
Total bili baseline (μmol/L)			14.7	8.3	55.2	21.2
Time since diagnosis (years)	15.5	7.2	15.0	7.5	15.1	6.8
Time since first ART (years)	11.0	6.4	10.5	6.0	10.7	6.4

Cardiovascular

There was no difference seen in blood pressure between the two groups (Table 1). The results for all CV markers are illustrated in Table 2. No difference was seen in PWV which was 10.2 ms⁻¹ (9.6–11.4) in the NBR group compared to 10.1 ms⁻¹ (9.6–11.2) in the HBR group (p = 0.82). There was a trend towards lower (more favorable) CIT measurements in the HBR group (0.06 vs. 0.07 mm; p = 0.08). Median VCAM-1 levels were significantly lower (more favorable) in the HBR group (377 vs. 416 ng/mL; p = 0.03); of all of the soluble markers, only VCAM was shown to be significantly different between groups. No significant differences were found in insulin resistance (HOMA-IR) or insulin levels between the NBR group in median mIU/L (IQR); 7.9 (4.6–11.4) and the HBR group 8.6 (5.9–11.5) [p = 0.23].

Table 2 Cardiovascular and calcaneal markers

	NBR (n = 43)	HBR (n = 35)	P-value
	Median (IQR)	Median (IQR)
CIT (mm)	0.07 (0.06 to 0.08)	0.06 (0.05 to 0.07)	0.08
PWV (m/s)	10.2 (9.6 to 11.4)	10.1 (9.6 to 11.2)	0.82
CSI (t-score)	−0.5 (−1.3 to 0.1)	−0.6 (−1.2 to 0.1)	0.99
IL-6 (pg/mL)	1.5 (1.5 to 2.1)	1.5 (1.5 to 1.7)	0.14
D-dimer (ng/mL)	200 (190 to 330)	220 (190 to 290)	0.53
hsCRP (mg/L)	1.5 (0.6 to 2.6)	1.2 (0.6 to 2.0)	0.40
ICAM-1 (ng/mL)	287 (251 to 329)	289 (236 to 343)	0.52
VCAM-1 (ng/mL)	416 (353 to 497)	377 (319 to 416)	0.03

No differences were seen in calculated cardiovascular risk (10-year Framingham %) between the two groups. Median 10-year Framingham risk score was 8.3% in the NBR and 9.1% in the HBR groups; p = 0.706. A comparison of the proportions that fell into low, medium, and high risk-score categories also did not show significant differences (p = 0.432).

Lipids

Lipid data are presented in Table 3. Whilst there was no difference in total cholesterol between the NBR and HBR groups, triglycerides were significantly lower in the HBR group (1.7 mmol/L) as compared to the NBR group (1.4 mmol/L; p = 0.50). LDL was also significantly lower in the HBR group (2.6 vs. 3.3 mmol/L in NBR; p = 0.047). HDL was no different between the two groups and neither was the HDL:TChol ratio.

Table 3 Lipids

	NBR (n = 43)	HBR (n = 35)	P-value
	Median (IQR)	Median (IQR)
Total cholesterol (mmol/l)	N = 42	N = 34	0.272
	5.1(4.2–6.3)	4.7(4.1–5.6)
	Range: 3.0–7.4	Range: 3.0–6.8
Triglycerides	N = 42	N = 34	0.050
	1.4(0.9–2.0)	1.7(1.2–2.3)
	Range: 0.5–3.5	Range: 0.6–3.3
HDL	N = 42	N = 34	0.699
	1.2(1.0–1.3)	1.2(1.0–1.4)
	Range: 0.6–1.8	Range: 0.7–1.8
LDL	N = 42	N = 34	0.047
	3.3(2.5–4.1)	2.6(2.3–3.5)
	Range: 1.8–5.4	Range: 1.6–4.2
HDL:Chol ratio	N = 42	N = 34	0.154
	4.3(3.6–5.2)	4.0(3.4–4.7)
	Range: 2.6–8.0	Range: 2.8–7.3

Bone

No significant difference between the two groups was seen for CSI (NBR group (median, IQR) −0.5 (−1.3 to −0.1); HBR −0.6 (−1.2 to −0.1); p = 0.99; Table 2).

Based on FRAX, the HBR group had a higher fracture risk. Those in the NBR group had a 10-year calculated FRAX risk of 2.9% (2.5–5.2) and those in HBR had a risk of 5.5% (2.8–6.5); p = 0.017. Analysis by those with an “at risk” FRAX score (>3%) showed that 47.6% of those in the NBR group fell into this category compared to 68.8% in the HBR arm (p = 0.098). This difference was negated by adjustment for TDF/FTC use.

Inflammatory markers

No significance differences were seen in levels of IL6, d-dimer, and hsCRP between the NBR and HBR groups. Average IL-6 levels were 1.5 pg/mL in both groups (p = 0.14); d-dimer was 200 ng/mL in the NBR and 220 ng/mL in the HBR groups (p = 0.53); hsCRP was 1.5 mg/L in NBR and 1.2 mg/L in HBR (p = 0.40).

Erectile dysfunction

Of 91 respondents, 78% had had sex in the four weeks prior to the study visit. Of these, 8.5% reported sex with women and 90% reported sex with men (1.5% with both). Our cut-off score for moderate or severe ED was 45% of maximum possible score and/or taking ED medication. This meant a total of nine (13.6%) were classified as having ED; 0% of men having sex with women, and 14.1% of MSM. No significant differences were seen in the ED group for normal vs. high bilirubin levels, ICAM, VCAM, Framingham risk, PWV, or CIT.

NC testing

One Card Learning (OCL; visual learning) and One-Back Task (ONB; working memory) were statistically longer (less favorable) in the HBR group. Overall, composite z-scores of all NC assessments did not differ significantly between the two groups (p = 0.655). These results are illustrated in Table 4.

Table 4 CogState results

	Higher Score	NBR (n = 43)	HBR (n = 35)	P-value
Visual motor function; CHASE – mps	Better	1.14 (0.31)	1.26 (0.33)	0.08
Executive function; GML – ter	Worse	58.5 (29.0)	50.9 (15.8)	0.10
Psychomotor function; DET – lmn	Worse	2.57 (0.11)	2.53 (0.08)	0.14
Visual attention; IDN – lmn	Worse	2.75 (0.09)	2.73 (0.06)	0.11
Visual learning and memory; OCL – acc	Better	0.94 (0.11)	0.98 (0.07)	0.05*
Visual learning and memory; CPAL – err	Better	102.3 (55.3)	98.0 (68.7)	0.50
Working memory ONB – acc	Worse	1.22 (0.23)	1.34 (0.13)	0.03*

Renal

Finally, no differences were seen in any renal markers, as shown in Table 5.

Table 5 Renal markers

	NBR (n = 43)	HBR (n = 35)	P-value
	Median (IQR)	Median (IQR)
uPCR (mg/mmol)	9.5 (7–13)	9.0 (7–14)	0.86
Urinary creatinine (μmol/L)	12360 (9790–17737)	14193 (10026–17710)	0.67
uRBP/creatinine (μg/mmol)	10 (7–16)	14 (8–20)	0.99
Urinary protein (g/L)	0.14 (0.08–0.20)	0.13 (0.09–0.22)	0.49

NRTI backbone

An unplanned NRTI backbone analysis was performed. Twenty-two subjects received ABC-based therapy and 78 received TDF. No differences were seen in cardiovascular markers: Framingham (10-year risk % median, IQR): ABC 8.1, 5.6–15.3; TDF 9.5, 4.8–13.4 (p = ns); PWV and CIT also showed no significant differences. No differences were seen in bone parameters: CSI (median score, IQR): ABC −0.5, −0.8–0.8; TDF −0.5, 1.4–0.4 (p = ns); 10-year FRAX score (% median, IQR): ABC 5.0, 2.4–6.2; TDF 3.6, 2.5–5.8 (p = ns). There were statistically significant differences in uPCR (10 vs. 7; p = 0.004) and uACR (15 vs. 8; p = 0.002), with both being higher in the TDF group.

Discussion

It is worth noting some of the limitations of this study first. The cross-sectional design and small sample size mean limited conclusions can be drawn, and the small sample size meant good multivariate or adjusted analyses were difficult to perform without increasing the risk of error. Confounding by indication is possible in that there is no reason recorded for choice of antiretroviral regimen or, indeed, protease inhibitor as part of this study, and this rationale may be behind any of the differences seen.

This study did not, overall, show any significant differences for the HBR group on the majority of studied markers for cardiovascular risk, bone disease, renal disease, or NC impairment. We will discuss each in turn. It is worth noting here that the comparison is really between the actions of atazanavir and darunavir, with HBR being a result of atazanavir use. Clinical endpoint trials regarding CV disease are difficult and costly as they require large numbers of participants and long follow-up periods. Traditional screening tools to assess CVD risk such as Framingham scores have shown to be imprecise at estimating risk of CVD for PLWH.¹⁶ There is, therefore, much interest in other predictors of cardiovascular morbidity in HIV. This study aimed to compare established methods at assessing CV risk and BMD with newer markers such as carotid intimal medial thickness, PWV, erectile dysfunction, biomarkers of inflammation, and calcaneal stiffness. These have all been validated in non-HIV populations.¹⁷⁻²⁶

Arterial stiffness as measured by PWV is a well-validated non-invasive method for assessing cardiovascular risk and retains its predictive value for cardiovascular events after adjustment for classical risk factors, Framingham score and CIMT.^18,19 There are very limited data on the use of PWV as a marker of arterial stiffness in HIV-infected individuals. There was no difference in PWV (the primary endpoint). There was a statistical difference in VCAM that is consistent with published data suggesting a link between HBR or ATV/r use on vascular relaxation. However, the significant difference in mean VCAM-1 between NBR and HBR groups in Mann–Whitney U testing (see Table 1) seemed to be driven by four outliers with high VCAM-1 levels in the NBR group and there is no clear evidence from this study that ART-associated hyperbilirubinaemia is associated with lowered VCAM-1 levels. There was no evidence that VCAM-1 was different according to ABC or TDF use.

It is worth noting that the HBR group had favorable lipid markers in terms of better triglycerides and LDL. Other markers were similar in both groups, as was the calculated Framingham risk score. The CIT was not significantly lower in the HBR group, despite a small numerical difference. Data from the ACTG5260 study²⁷ demonstrated that treatment-naïve HIV-infected individuals randomized to an initial ART regimen including ritonavir-boosted atazanavir experienced slower progression of CIT than those assigned ritonavir-boosted darunavir, despite similar lipids changes and similar virologic effects. Thus, this may represent an effect or HBR or be a characteristic of ATV/r. Of note, ATV was not associated with CV risk in the D:A:D study,⁶ whereas other PIs (LPV/r, indinavir) have been. The CV safety of DRV/r has not been assessed in D:A:D.

Erectile dysfunction (ED) is also strongly associated with cardiovascular disease^20–22 and has recently been studied to explore whether presence of ED improves the risk prediction of the Framingham risk score.²³ Several of the factors that predispose to ED, such as atherosclerosis and endothelial dysfunction, are increased in HIV infected individuals and it has been clearly shown that ED occurs at a higher prevalence in this population. Initially ED was reportedly associated with protease inhibitors, however larger studies indicate that in addition to traditional risk factors such as older age and depression duration of antiretroviral treatment, but not specific ARV drugs or ARV drug classes is associated with ED.²⁸ However, the specific association between ED and markers of cardiovascular disease risk other than blood glucose and lipids has not been studied. Assessment of ED by the International Index of Erectile Function has been validated in many different populations.²⁹ More recently a version of the IIEF adapted for use in men who have sex with men (MSM) has been validated in an HIV infected population.³⁰ Low rates of ED were seen in this study and there was no significant correlation with the cardiovascular markers used. We would anticipate that ED would be correlated with CV risk markers, so the lack of association here warrants further investigation.

Individuals with HIV are also at a higher risk of low bone mineral density than their HIV-negative counterparts²⁴ and certain antiretroviral drugs have been associated with significant reductions in bone mineral density (BMD).²⁵ The GE-Achilles Insight uses calcaneal ultrasound to estimate BMD. The calcaneus is a good site for fracture risk assessment²⁶ as it is a weight bearing area, has a high trabecular content, and is easily accessible to measure. In the European guidelines BMD assessment is recommended in those with a 10-year risk of osteoporotic fracture around 5–10% according to the FRAX^® score.³¹ The gold standard to determine bone mineral density is to use dual energy X-ray absorptiometry (DEXA) this has two major inconveniences firstly it is expensive, needing to be performed in the radiology department and secondly, contains radiation. Ultrasound measurements to assess calcaneal stiffness are proven to predict hip fracture risk^32,33 and vertebral fracture status^34–36 and use of the GE-Achilles Insight method has been verified in HIV-infected individuals.¹⁵ Calculated FRAX score seems to indicate that HBR is a risk factor with those in the HBR group having a higher calculated probability of possible fracture (opposite to that which would be expected if high bilirubin had an anti-inflammatory effect), but this difference is not seen after correction for tenofovir use.

No significant differences were found in renal markers between patients with high and normal bilirubin levels.

Statistically significant differences were seen in some component NC tests. One Card Learning (OCL; visual learning) and One-Back Task (ONB; working memory) were shown to be statistically longer in the HBR group. Whilst this may warrant further consideration regarding any clinical significance, overall global NC (composite z) scores of all NC assessments did not differ significantly between patients with HBR and NBR (p = 0.655).

It is reassuring here that we saw no data to support an increase in cardiovascular risk markers with abacavir use although this is a small sample size and the study did not examine clinical endpoints. Use of TDF increased urinary protein loss, and this appeared to be a tubular pattern.

Conclusions

It does appear that high bilirubin in the context of boosted atazanavir use may have some affect on cardiovascular and bone markers but that this needs further investigation in a larger, clinical end point study.

High bilirubin in the context of ritonavir booster protease inhibitor-treated HIV has a limited impact on acute measurements of vascular function such as PWV, nor more chronic effects such as CIMT. It is difficult to comment on chronic effects fully, as our study population had a mean duration of HIV infection of around 15 years, and a mean duration of antiretroviral therapy of between 10 and 11 years. Without a full therapy history we cannot account here for the possible impact of older antiretroviral regimens.

Vascular relaxation markers including VCAM-1 may be favorably affected but do not lead to favorable erectile function differences. High bilirubin has no meaningful benefits of bone stiffness, NC function, or renal function.

In conclusion, in this small cross-sectional study, high bilirubin associated with ritonavir-boosted protease inhibitor use had some possible beneficial effects, but further elucidation in a larger, prospective study is required for more definite, clinically applicable conclusions.

Acknowledgments

The authors acknowledge the support of the St Stephen’s AIDS Trust, as well as the participants and study staff, and the contributions made by James Gibbins, Sundhiya Mandalia, and Gary Lo.