Dyslipidemias and Elevated Cardiovascular Risk on Lopinavir-Based Antiretroviral Therapy in Cambodia

Background Lopinavir/ritonavir (LPV/r) is widely used in Cambodia with high efficacy but scarce data exist on long-term metabolic toxicity. Methods We carried out a cross-sectional and retrospective study evaluating metabolic disorders and cardiovascular risk in Cambodian patients on LPV/r-based antiretroviral therapy (ART) for > 1 year followed in Calmette Hospital, Phnom Penh. Data collected included cardiovascular risk factors, fasting blood lipids and glucose, and retrospective collection of bioclinical data. We estimated the 10-year risks of coronary heart disease with the Framingham, Ramathibodi-Electricity Generating Authority of Thailand (Rama-EGAT), and the Data Collection on Adverse Effects of Anti-HIV Drugs (D:A:D) risk equations. We identified patients with LDL above targets defined by the French expert group on HIV and by the HIV Medicine Association of the Infectious Disease Society of America and the Adult AIDS Clinical Trials Group (IDSA-AACTG). Results Of 115 patients enrolled—mean age 40.9 years, 69.2% male, mean time on LPV/r 3.8 years—40 (34.8%) had hypercholesterolemia (> 2.40 g/L), and 69 (60.0%) had low HDL cholesterol (< 0.40 g/L). Twelve (10.5%), 28 (24%) and 9 (7.7%) patients had a 10-year risk of coronary heart disease ≥ 10% according to the Framingham, D:A:D, and Rama-EGAT score, respectively. Fifty one (44.4%) and 36 (31.3%) patients had not reached their LDL target according to IDSA-AACTG and French recommendations, respectively. Conclusion Prevalence of dyslipidemia was high in this cohort of HIV-infected Cambodian patients on LPV/r. Roughly one third had high LDL levels requiring specific intervention.


Introduction
HIV-related mortality has been dramatically reduced by the widespread use of antiretroviral treatment (ART). However, in developed countries the range of morbidity has increased, due to the emergence of cardiovascular and other non-AIDS related diseases. The role of dyslipidemia in cardiovascular morbidity is now well documented as well as the role of cardiovascular diseases as an important cause of death in HIV-infected patients [1][2][3]. These multifactorial complications are related to traditional risk factors including tobacco smoking, to the action of HIV itself via immune activation, as well as the toxicity of some antiretroviral drugs [4]. The role of several protease inhibitors in the occurrence of dyslipidemia is now well established and cumulative exposure to protease inhibitors has been robustly associated with a higher risk of myocardial infarction than in the general population [5].
In Cambodia, expanded access to ART has dramatically reduced morbidity associated with opportunistic infections [6], with excellent adherence and treatment outcomes [7,8]. However, data on metabolic morbidity are limited. The expending use of ritonavir-boosted lopinavir (LPV/r) as second-line ART regimen could increase the risk of lipid disorders and contribute to an increased cardiovascular risk. Data on metabolic disorders in other South-East Asian patients on ART is scarce; high levels of dyslipidemias have been reported in Thai adults [9]. This study sought to assess the frequency of glucose and lipid metabolism disorders in Cambodian HIV-infected patients on LPV/r, and to assess their 10-year risk of developing coronary heart disease (CHD) according to the Framingham, D:A:D and Rama-Egat scoring systems [10][11][12].

Study design, settings and patients
We conducted a cross-sectional study from November, 2010 to May, 2011, in the HIV cohort of the Calmette Hospital, Phnom Penh, Cambodia. Patient follow-up and management in the cohort has been described elsewhere [7,13]. Routine metabolic monitoring included yearly fasting lipids. Fenofibrate was available for patients with triglycerides 5 g/l. Patients with either virologically confirmed ART-failure or repeated toxicities were switched from nonnucleoside reverse transcriptase inhibitor (NNRTI) to protease inhibitor-containing regimen. At the time of the study, LPV/r was the only PI routinely available in Cambodia.
Patients aged 18 years were eligible for inclusion if they had been on LPV/r for 12 months at the time of assessment. We excluded patients receiving statins in the private sector.

Study procedures and measurements
After written informed consent, patients underwent fasting lipids and glucose blood tests, complete physical examination with anthropometric measures, and a standardized questionnaire on demographic data and cardiovascular risk factors. Previous lipids and glucose measurements, CD4 count and plasma HIV RNA viral load, history of ART and other treatments were collected retrospectively from the patient's medical chart and/or the cohort database.

Ethical considerations
The study was approved by the Cambodian National Ethics Committee for Health Research. This study was conducted in accordance with the Declaration of Helsinki [14] and all patients signed the informed consent form prior to inclusion.

Variables
Hypercholesterolemia was defined as total cholesterol 2.4 g/l, low high-density lipoproteins cholesterol (HDL) as < 0.40 g/l, severe hypertriglyceridemia as triglycerides > 5g/l or receiving fenofibrate [15]. Elevated low-density lipoproteins cholesterol (LDL) was defined as 1.60 g/l, globally, and based on cardiovascular risk, individually. Hypertension was defined either as systolic blood pressure 140 mmHg and/or diastolic blood pressure 90 mmHg measured on both arms lying after 10-minute rest, or being on antihypertensive treatment. Diabetes mellitus was defined as either fasting glucose 1.26 g/l or being on antidiabetic treatment. Abdominal obesity was defined as waist circumference 90 cm in men and 80 cm in women which are measures recommended for Asian populations [16].
Individual 10 year risk of coronary heart disease (CHD) were calculated for each subject using the Framingham, D:A:D, and Rama-EGAT scoring systems, elaborated to predict the risk of angina pectoris, or myocardial infarction (MI) for the Framingham score, invasive coronary artery procedure, MI, or death from other CHD for the D:A:D score, and MI or invasive coronary artery procedure for the Rama-EGAT score [10][11][12] (Table 1). For the D.A.D score, Family history -Coronary vascular disease -Exposure to antiretrovirals the 10-year cardiovascular risk was derived from the 1-year risk score, as proposed by the authors. 10-year cardiovascular risk was considered high when 20%. We also evaluated the proportion of patients who had not reached LDL levels recommended by the HIV Medicine Association of the Infectious Disease Society of America and the Adult AIDS Clinical Trials Group (IDSA-AACTG) [15], and the French expert group, as well as LDL levels requiring immediate initiation of statins, according to this latter. As recommended by the HIV Medicine Association of the Infectious Disease Society of America and the Adult AIDS Clinical Trials Group, patients were classified according to their level of cardiovascular risk based on their cardiovascular history, cardiovascular risk factors, and 10 year CHD risk calculated by the Framingham equation in four groups: 1) highest risk patients with either established CHD or CHD risk equivalent defined as diabetes mellitus or 10 year risk superior to 20% according to the Framingham score; 2) patients with 2 or more cardiovascular risk factors and a 10 year CHD risk of 10 to 20%; 3) patients with 2 or more cardiovascular risk factors and a 10 year CHD risk of 0 to 10%; 4) patients with 0 or 1 cardiovascular risk factors [15]. We estimated the proportion of patients having an LDL level above or equal to the optimal target and to the treatment decision threshold recommend for their cardiovascular risk group, respectively: 1) LDL goal 1.0 g/l and treatment decision threshold 1.3 g/l; 2) LDL goal and treatment decision threshold 1.3 g/l; 3) LDL goal 1.3 g/l and treatment decision threshold 1.6 g/l; 4) LDL goal of 1.6 and treatment decision threshold of 1.9 in patients with 0 or 1 CVRF [15]. We also evaluated the proportion of patients having reached the LDL target based on the cardiovascular risk classification proposed by the French recommendations [17]: 1) LDL 1.0 in highest risk patients with either established CHD or 3 CVRF; 2) 1.3 in patients with 2 CVRF; 3) 1.6 g/l in patients with 1 CVRF; 4) 1.9 in patients with 0 CVRF.
Lipodystrophy was defined as the presence of at least one sign graded as moderate or severe by the investigator among all the following signs, collected independently: fat loss in face, buttocks, arms and legs for lipoatrophy; the presence of "buffalo hump"-type fat deposits in upper back, increased breast size, and abdominal obesity for lipohypertrophy.

Statistical analyses
Comparisons between groups of patients were performed by Student t-test or non-parametric Kruskal-Wallis test for continuous variables, and Pearson chi-square or Fisher's exact test for categorical variables. Paired continuous variables were compared using Student's t test. Agreement between categorical classifications was measured by the kappa coefficient. All statistical comparisons were two-tailed, and P<0.05 was considered to be statistically significant. All analyses were performed using SAS version 11.1 (SAS Institute Inc, USA).

Patients characteristics
Of 1281 patients enrolled in the cohort, 173 (13.5%) were on LPV/r and 115 (66.5%) were enrolled in the study (Fig 1). The remaining patients were excluded because 41 had been on LPV/r 1 year and 15 were not seen during the study period. Additionally, 2 were excluded because they received statins.
Patients had been on LPV/r for a median time of 4.0 years (Interquartile range (IQR) 2.5-5.0) ( Table 2). None reported a history of CHD, peripheral vascular disease, or cerebrovascular accident. A total of 31 patients (27.0%) were taking fenofibrate, 11 (9.6%) were on antidiabetic, and 11 (9.6%) on antihypertensive medications. Demographic and HIV-related characteristics did not differ significantly between patients on fenofibrate or not.

Dyslipidemias on lopinavir/ritonavir
Lipid profiles and prevalence of dyslipidemia are displayed in Table 2 At ART initiation, 58 (54.2%) patients out of 105 with HDL available had a low HDL. Only 70 patients had lipid levels available at ART initiation and ART switch from NNRTI to LPV/r based ART. LDL levels were measurable in 64 at ART initiation and 60 at ART switch. Comparisons between lipid levels at ART initiation and evaluation on LPV/r and between ART switch and evaluation on LPV/r are displayed in Table 3. Globally, from ART initiation to evaluation, total cholesterol increased by 0.60 ± 0.61 g/l, and LDL increased by 0.51 ± 0.62. On LPV/r, total cholesterol increased by 0.46 ± 0.68 g/l, LDL increased by 0.48 ± 0.64 g/l, and HDL decreased by 0.13 ± 0.41 g/l. In the 84 patients without fenofibrate, total cholesterol and LDL were increased compared to ART initiation levels (p <0.0001 and p <0.0001); however HDL and triglycerides were unchanged (p = 0.22 and p = 0.44).

Cardiovascular risk and target LDL levels
Major cardiovascular risk factors are displayed in Table 2. When considering all risk factors exclusive of LDL, the median number of risk factor per patient was 1 (IQR: 0 to 2). 47 patients had only risk factor and 38 of them (80.9%) had isolated low HDL.  When assessing 10-year risk of CHD, 1 (0.9%), 11 (9.6%) and 0 (0%) patients were considered at high CHD risk by the Framingham, D:A:D and Rama-EGAT score, respectively. Additionally, following US recommendations, 20 patients with diabetes should be considered at high cardiovascular risk, raising the number to 21 (18.3%).

Discussion
Prevalence of dyslipidemias was high in this cohort of HIV-infected Cambodian patients on LPV/r based-regimen for a median duration of 4 years. Up to 42% of the patients were considered as having LDL higher than desired and 30% need immediate intervention including dietetic rules and treatment by statins.
Our findings were consistent with a previous report in Cambodia showing an overall prevalence of dyslipidemias of 73% after 4 years on ART, in a cohort with 10% of patients on LPV/r based ART. Total cholesterol is particularly high in our patients compared to other cohorts of HIV-infected patients on LPV/r [2,19] but consistent with early report from the French APROCO cohort of patients on protease inhibitor-based ART before use of ritonavir boosting [20]. A more recent study in 170 Thai patients on PI-based ART reported a mean total cholesterol of 2.6 g/L and worsened lipid profiles anomalies in patients on ritonavir-boosted PI regimens [9]. Analysis of data collected routinely in cohort follow-up showed that total cholesterol, triglycerides, and LDL levels increased on LPV/r, a well known effect of this therapeutic regimen [19,21,22]. Our study, with 3 time points confirmed these results in a South-East Asian population.
HDL levels were low in our patients at ART initiation, which was expected, low HDL being a common HIV-induced dyslipidemia and possibly a main contributor to CHD risk in HIVinfected population [23]. Low HDL levels have been described in ART-naïve HIV-infected   Table 4.  individuals in caucasian and other populations [24][25][26]. Surprisingly, in our patients, HDL levels decreased on LPV/r whereas other studies reported an increase on ART [27]. This is of particular concern, even in the absence of other cardiovascular risk factors, as isolated low HDL has been shown to be associated to elevated cardiovascular risk in various studies [28] and specifically in Asian populations [29]. Hypercholesterolemia was 5 to 10-fold higher than estimated in a recent survey in the Cambodia general population which showed a global and urban hypercholesterolemia prevalence of 3.2% and 7%, respectively, using similar threshold or the presence of lipid lowering treatments [30]. Considering other risk-factors, reported tobacco smoking was much lower in our study compared to the general population survey which reported a 49.3% and 4.8% current smoking rate in men and women, and prevalence of hypertension and diabetes in the much higher than the 3.1 and 3.2% found in the survey, respectively. The prevalence observed in our study may have been overestimated due to the cross-sectional design, but the frequency of diabetes or antihypertensive medications intake confirmed the importance of these disorders.
The RAMA-EGAT score, specifically designed for a South-East Asian population, showed that cardiovascular risk was higher in this study than reported previously in Thai patients of similar age and duration on ART, including 60% on protease inhibitor-based ART [12]. Score performance for the prediction of 10-year cardiovascular risk differed from the study in Thailand which showed a good agreement between the D:A:D and the RAMA-EGAT scores, where in our study D:A:D predicted a higher cardiovascular risk in patients while Framingham and RAMA-EGAT scores had much lower estimates, which may be due to the high prevalence of diabetes.
The IDSA-AACTG and French scores agreed to identify roughly 1/3 of patients needing immediate treatment for LDL levels above the desired target. However statins are expensive and not widely available in setting such as Cambodia. Diet, lifestyle changes, and change of ART if 10-year CV risk > 20% seem to be the most appropriate solution [31]. Atazanavir, a newer protease inhibitor associated with only modest changes in lipid profile [32], is also becoming increasingly available in resource limited-setting and is recommended as first choice for second-line regimen since 2012 in Cambodia.
There are limitations to this study. First, patients from this single site may not be representative of the general HIV-infected population in Cambodia. As shown in other studies in Asia, lipid levels and cardiovascular risk in urban population may be different from rural population [33,34]. We may also have overestimated hypertension and diabetes due to the essentially cross-sectional design of our study. Second, 27% of patients were on fenofibrate which may have led to decreased LDL and increased HDL levels. However the poor control of lipids emphasizes the interest for a better management of metabolic disorders in these patients. At last, none of the risk equations used in our study were validated specifically in HIV-infected South-East Asian populations. More epidemiological studies are needed to document incidence of cardio-vascular and other non-AIDS related morbidity in HIV-infected populations in South-East Asia and Africa. This would help validating risk equations and decision-making on risk reduction at individual level. Dyslipidemias were frequent in HIV-infected Cambodian patients on LPV/r for more than one year, contributing to an elevated cardiovascular risk. The particularly high prevalence of low HDL warrants further investigation and better evaluation of cardiovascular risk in Asian population. In Cambodia, the progressive switch from LPV/r to atazanavir will certainly contribute to an improvement in lipid profiles and will need further evaluation. However, this study highlights the necessity to better take into account non-AIDS related co-morbidities in limited-resources settings as HIV care becomes chronic care and as these countries will face major challenges on management of cardiovascular morbidity in general [35].
Supporting Information S1 Data. This is the S1 Data, which correspond to the entire data's table of the study. (XLS)