Association of single nucleotide polymorphisms with dyslipidemia in antiretroviral exposed HIV patients in a Ghanaian population: A case-control study

Dyslipidemia is a potential complication of long-term usage of antiretroviral therapy (ART) and also known to be associated with genetic factors. The host genetic variants associated with dyslipidemia in HIV patients on ART in Ghana have not been fully explored. The study constituted a total of 289 HIV-infected patients on stable ART for at least a year. Fasting blood was collected into EDTA tube for lipids measurement. Lipid profiles were used to define dyslipidemia based on the NCEP-ATP III criteria. HIV-infected subjects were categorized into two groups; those with dyslipidemia (cases) (n = 90; 31.1%) and without dyslipidemia (controls)(n = 199; 68.9%). Four candidate single nucleotide polymorphism (SNP) genes (ABCA1-rs2066714, LDLR-rs6511720, APOA5-rs662799 and DSCAML1-rs10892151) were determined. Genotyping was performed on isolated genomic DNA of study participants using PCR followed by a multiplex ligation detection reaction (LDR). The percentage of the population who had the rare homozygote alleles for rs6511720 (T/T), rs2066714 (G/G), rs10892151 (T/T) and rs662799 (G/G) among case subjects were 5.5%, 14.4%, 6.6% and 10.0% whiles 2.0% 9.1%, 6.5% and 4.0% were observed among control subjects. There were statistically significant differences in the genotypic prevalence of APOA5 (p = 0.0357) and LDLR polymorphisms (p = 0.0387) between case and control subjects. Compared to the AA genotype of the APOA5 polymorphisms, individuals with the rare homozygote genotype [aOR = 2.38, 95%CI(1.06–6.54), p = 0.004] were significantly associated with an increased likelihood of developing dyslipidemia after controlling for age, gender, treatment duration, CD4 counts and BMI. Moreover, individuals with the rare homozygous genotype of ABCA1 (G/G) [aOR = 10.7(1.3–88.7), p = 0.0280] and LDLR (rs6511720) G>T [aOR = 61.2(7.6–493.4), p<0.0001) were more likely to have high levels of total cholesterol levels. Our data accentuate the presence of SNPs in four candidate genes and their association with dyslipidemia among HIV patients exposed to ART in the Ghanaian population, especially variants in APOA5-rs662799 and LDLR rs6511720 respectively. These findings provide baseline information that necessitates a pre-symptomatic strategy for monitoring dyslipidemia in ART-treated HIV patients. There is a need for longitudinal studies to validate a comprehensive number of SNPs and their associations with dyslipidemia.

Introduction various forms of genetic polymorphisms, similar to that in non-HIV adults [27,28]. Several single nucleotide polymorphisms (SNPs) that could account for a significant portion of the variation of blood lipoprotein concentrations have been identified through recent candidate gene studies and genome-wide association studies (GWAS) [15,29].
The association between gene polymorphisms that may signal a predisposition to lipid abnormalities and clinical progression of HIV infection has not been thoroughly studied in Africa where the prevalence of HIV is on the increase. SNP prevalence differs by population and at present, the majority of the SNP-associated dyslipidemic studies among HIV patients have come from non-African countries with only a few of these studies emanating from Africa [15,24,30]. To the best of our knowledge, no published study has explored the genetic variants and markers associated with dyslipidemia in HIV-infected individuals on HAART in a Ghanaian population. This study, therefore, investigated the distribution of SNPs in four candidate genes that have had significant published lipid associations and their resultant associations with plasma lipid levels in Ghanaian HIV-infected patients on HAART. An understanding of the impact of host genetic factors on the prevalence of dyslipidemia in a cohort of HIVinfected individuals on HAART would promote interventions in the scaling up of treatment regimen.

Study design and subjects
This study comprised HIV-1 infected patients who were on ART regimen for at least one year, with either a protease inhibitor (PI) or non-nucleoside reverse-transcriptase inhibitor (NNRTI) backbone without any history of dyslipidemia, hypertension, and diabetes. The NCEP-ATP III criteria were used to defined dyslipidemia among the HIV seropositive [31]. The HIV seropositive subjects were categorized into two groups; those with dyslipidemia (cases) and without dyslipidemia (controls). The adjuvant antiretroviral drugs were stavudine, lamivudine, and zidovudine with priorities for inclusion being given to participants who consented to undergo biochemical and genetic testing. Pregnant women, patients being treated with lipid-lowering drugs and those with neurological conditions that prevented them from understanding the concept of the research were excluded from the study.

Sample size determination
Based on previous report from a study conducted on the burden of dyslipidemia among adults in Ghanaian a population [32] and the lack of knowledge of the frequency of polymorphisms in the population, we assumed an expected proportion of 0.1 for exposure in HIV seropositive subjects with dyslipidemia, an assumed odds ratio of 2, a confidence interval of 95%, a power of identifying a significant difference between two groups, and 1:3 ratio, a total of 289 subjects were recruited.

Data collection and biochemical analysis
A structured questionnaire was administered to each patient to obtain demographic information. Details on ARVs, time of diagnosis, duration on ARVs, CD4 counts were obtained from the medical folders of the patients. Fasting blood samples were collected for the analysis of lipid parameters and genomic DNA.
Blood samples were taken after an overnight (12-14 hours) fast into EDTA tubes for biochemical analysis. Fasting lipid panel including total cholesterol (TC), HDL-cholesterol and triglycerides (TG) were measured using Flexor junior (Vital Scientific, Dieren, Netherlands) chemistry autoanalyzer. LDL-C was calculated from the Friedewald's formula, LDL-C = TC-(HDL-C-TG/2.2). Patient with triglycerides above 4.52 mmol/L was not included in the study since LDL-C was not directly measured and due to the deficit of the Friedewald's equation, which overestimates LDL-C levels when triglycerides are high. This is to reduce any bias that might affect the relationship found between LDL-C and dyslipidemia. The NCEP-ATP III criteria were used to define dyslipidemia as reduced HDL (<1.03 mmol/L in males; <1.29 mmol/ L in females), raised TG �1.7 mmol/L, TC >6.2 mmol/L and LDL-C >3.37 mmol/L or specific treatment for such lipid abnormalities [31].

Anthropometric and hemodynamic measures
Anthropometric measures such as weight and height were performed using an automated weighing scale. Portable height rod stadiometers were used for height measurements to the nearest centimeter. Body mass index (BMI) was defined as weight (kg)/height (m) 2 . Blood pressure was measured using an automated sphygmomanometer (Omron M7 Intelli IT). Three consecutive readings of blood pressure measurements were taken from the patients' right arm and the mean of two closest values was recorded.

Single nucleotide polymorphisms selection and genotyping
The four candidate SNPs (rs2066714, rs6511720, rs662799, and rs10892151) selected for this study have been shown in previous studies to be significantly associated with lipid serum abnormalities following a review of GWAS and PubMed reports of SNPs associated with dyslipidemia among HIV-infected individuals [24,[33][34][35]. Genomic DNA was isolated from EDTA-collected whole blood samples of study participants using the Qiagen midi kit prep as per manufacturer's protocol. Genotyping was carried out on isolated genomic DNA of study participants using a multiplex ligation detection reaction (LDR), a sequence-specific genotyping method that has been used efficiently in polymorphism typing and detection of mutations in disease genes [36]. Triplex multiplex reaction assays were setup with products being run on 10% polyacrylamide gel for LDR genotype observation (58 to 90 base pairs). The LDR products differed in sizes of about 8 base pairs for each triplex (Fig 1). The runs were controlled with Amelogenin XY (AME XY) to determine a successful reaction and gender-confirming for study participants.
Allele specific LDR products are assigned by color and size following 10% polyacrylamide gel electrophoresis. Each band as shown in the image represents a specific genotype of each allele. At the bottom of the gel are leftover LDR probes from the reaction mix as represented by shared common bands (Fig 1). Association of single nucleotide polymorphisms with dyslipidemia in antiretroviral exposed HIV patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227779 January 13, 2020

Ethical consideration
The study protocol was approved by the Committee of Human Publication and Research Ethics of the School of Medical Sciences, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana. All participants gave written informed consent and were assured that the information gathered was to be used strictly for research and academic purpose only. In addition, respondents were given the freedom to opt-out at any time they thought they could not continue with the study.

Data management and statistical analysis
The NCEP-ATP III criteria were used to defined dyslipidemia among the HIV seropositive [32]. The HIV seropositive subjects were categorized into two groups; those with dyslipidemia (cases) and without dyslipidemia (control). Microsoft Excel software was used to set up a database, and to avoid entry error, the double-entry method was used, and data were analyzed using SPSS version 25 and R program where appropriately. Parametric continuous were analyzed with a t-test and expressed as mean ± standard deviation (SD) after checking for normality with the Kolmogorov-Smirnov test.
ANOVA was used to compare continuous variables from more than two groups with Yates post-test, the chi-square test was to compare differences between more than three groups for categorical variables while Fisher exact was used to assess differences between two groups for categorical variables Allele frequencies were estimated by gene counting. Deviations in the genotype frequencies from the Hardy-Weinberg equilibrium (HWE) were tested using the chi-square (χ 2 ) analysis. Initial univariate binary logistic models were performed to determine the association between SNPs and dyslipidemia, followed by adjusted multivariate binary logistic models controlling for age, gender, BMI, CD4 counts and duration of HIV infection. Adjusted multivariate binary logistic models were used also to identify SNPs independently associated with lipid abnormalities holding confounding variables constant. A p-value of less than 0.05 was considered statistically significant. Table 1 shows a comparison of demographic, hemodynamic indices, lipid parameters, dyslipidemic indices between cases and control subjects. Dyslipidemia was found in 31.1% (90/289). A statistically significant difference between case and control subjects was observed with regards to age (p = 0.0328). There were more females than males in the study population (p>0.05). The average duration of treatment among the subjects was 4 years. There were statistically significant differences between the case and control subjects in relation to CD4 counts (p = 0.0010) and BMI (p<0.0001). The case subjects had significantly higher levels of TC (p<0.000) LDL-C (p<0.0001) and CR (p<0.0001) compared to the control subjects. Compared to HIV seropositive subjects with dyslipidemia, those without dyslipidemia had significantly increased levels of HDL-C (0.7 ±0.3 vs. 1.2±0.5, p<0.0001). Table 2 shows the frequency distribution of the genotypes and alleles of the four SNPs. The percentage of the population who had the rare homozygote alleles for rs6511720 (T/T), rs2066714 (G/G), and rs10892151 (T/T) and rs662799 (G/G) among the case subjects were 5 .5%, 14.4%, 6.6% and 10.0% whiles 2.0% 9.1%, 6.5% and 4.0% were observed among the control subjects. Statistically significant differences in allelic frequency were observed for DSCAML1 (p = 0.0008)and APOA5 (p = 0.0251) among case subjects, and ABCA1 (p = 0.0010) and DSCAML1 (p = 0.0084) among control subjects. Moreover, chi-square analysis reveals significant differences in the genetic frequencies of LDLR (p = 0.0387) and APOA5 (p = 0.0353) polymorphisms respectively.

Total Cholesterol (TC)
The case subjects who had the rare homozygous genotype for LDLR (p<0.0001), ABAC1 (p = 0.0287) DSCAML1 (p = 0.0003) and APOA5 (p = 0.0151) had a significantly higher level of TC compared to the combined heterozygous and non-carriers' genotypes [ Table 3].

Triglycerides (TG)
Statistically significant differences were observed in TG levels for DSCAML1(p = 0.0006) and LDLR (p = 0.0178) polymorphisms between homozygous rare genotype and heterozygous and non-carriers' genotypes among the case subjects. Similar patterns were observed for DSCAML1 polymorphisms (p = 0.0133) among the control subjects [ Table 3].

HDL Cholesterol (HDL-C)
There were significant differences in HDL-C levels with regards to APOA5 polymorphisms (p = 0.0075) in case subjects. Thus, the combined heterozygous and non-carriers genotype had high levels of HDL-C compared to the homozygous genotypes [ Table 3].

LDL Cholesterol (LDL-C)
There were significant increased LDL-C levels in the rare-allelic subjects for LDLR (p<0.0001) and DSCAML1 (p = 0.0078) polymorphisms compared to participants with heterozygous and homogenous genotypes together among the case subjects [  Table 4]. Table 5 shows the association between SNPs with lipid abnormalities after controlling for age, gender, duration of treatment, CD4 counts and BMI respectively.  Association of single nucleotide polymorphisms with dyslipidemia in antiretroviral exposed HIV patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227779 January 13, 2020 Association of single nucleotide polymorphisms with dyslipidemia in antiretroviral exposed HIV patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227779 January 13, 2020 S1 Table shows

Discussion
Long term usage of ART has been implicated in several metabolic effects including dyslipidemia. However, this side effect varies among individuals on ART with comparable clinical and demographic characteristics. As such, inherited predispositions and genetic factors have been implicated in the cause of this metabolic alteration. The present study, therefore, assessed the prevalence of SNPs and their associations with dyslipidemia in HIV patients on ART in Ghana. Four candidate SNPs (rs2066714, rs6511720, rs662799, and rs10892151) reviewed from PubMed were analyzed for associations with dyslipidemia among HIV-infected individuals.
Several studies have shown the association between dyslipidemia and HIV patients on ART [37,38]. The prevalence rate of dyslipidemia among HIV patients on ART in this current study was 31.1% with low HDL-C and high TC levels being the commonest lipid abnormalities. The observed prevalence of dyslipidemia is comparable to the range of reports from previous studies by Obirikorang et al. [9] and Ngala et al. [10] in Ghana and Kodogo et al. [10] in Zimbabwe. However, it is lower compared to 78.9% prevalence rate reported by Limas et al. [39] in a cross-sectional study among Brazilian HIV individuals on ART. Furthermore, our observed results for low HDL-C, high TC and LDL-C levels among HIV patients on ART are consistent with other studies [40,41]. Results from the present study revealed that subjects who received PI-based treatment had significantly increased levels of TC, LDL cholesterol and triglycerides levels compared to the non-PI-based subjects. These findings are consistent with  for the rare recessive polymorphism of rs6511720, rs2066714, rs10892151 and rs662799 genotypes using a combination of carriers and homozygotes as a referent genotype, aOR = Adjusted odds ratio, CI = Confidence interval. A multivariate logistic model was performed controlling for age, gender, duration of treatment, CD4 counts and BMI † p<0.05 several case reports [42,43] and cross-sectional studies [44,45] which have reported that PI exposures are associated with hypercholesterolemia and hypertriglyceridemia. Our study demonstrated that the presence of the homozygous recessive/mutant genes for LDLR, ABCA1, DSCAML1 and APOA5 genotypes among our study participants were very low [ Table 2]. Similar low frequencies in APOA5 and LDLR homozygotes mutant genes were reported by Lazzaretti et al. [15] in a cross-sectional study among HIV-infected patients on ART in the Brazilians population. Genome-wide associations studies have identified LDL-R SNP rs6511720 (G>T), which is located in intron-1 of the gene, to be associated with lower plasma levels of LDL-C and a lower risk of CHD [35]. Data from the GLGC consortium suggested that LDLR rs6511720 minor allele is prevalent in about 10% of the population and have established that the allele is protective, being associated with lower levels of LDL-C [46].
Notwithstanding that, the frequency of recessive genes was low in this study population, these findings are consistent with previous reports by Wang et al., [19] and Aragones et al. [24] who did not find any significant differences in their studies with respect to prevalence of the DSCAML1 and ABCA1 genotypes. Our findings demonstrated that the allelic frequencies for ABCA1 and APOA5 polymorphisms were in not equilibrium with the Hardy-Weinberg equation among the case subjects as well as ABCA1 and DSCAML1 polymorphisms among the control subjects. The low frequencies of the homozygotes of the rare alleles of these SNPS may have contributed to the observed deviations.
A longitudinal study conducted by Rotgar et al. [47] validated the contribution of 42 SNPs to dyslipidemia among the HIV-infected population treated with ART. The authors reported that the degree of the contribution of SNPs and ART to dyslipidemia are similar and therefore genetic information should be considered in addition to the dyslipidemic effect of ART agents [47]. Results from the present study revealed that individuals with rs6511720 T/T genotype recorded increased levels of TC, TG, and LDL-C compared to the combination of non-carriers and heterozygous genotypes (GG/GT) [ Table 3]. A further multivariate logistic model analysis showed a strong association of LDLR polymorphisms with high levels of TC holding all other confounding variables constants. A study by Lazzaretti et al. [15] reported that LDLR intron 19G>T (rs6511720) did not contribute to the plasma lipid levels in their dataset, and hence may reflect a limited effect of this SNPs in HIV-infected patients. The inconsistency could be due to different geographical settings and thereby calls for more research in these SNPs among African descendants.
In this study, ABCA1 (rs2066714) G/G genotype was associated with an increased probability of having dyslipidemia. Furthermore, subjects with the homozygotes of the rare alleles of ABCA1 have increased levels of TC compared to the common allelic (AA/AG) genotypes. As a further matter, subjects carrying the minor allelic variant of ABCA1 were more likely to have low levels of HDL cholesterol among the HIV population in the present study though no statistical significance was observed. These observations are congruent with previous reports demonstrated in literature that ABCA1 is associated with familial HDL deficiency. We should mention here that low HDL-C was one of the commonest lipid abnormalities observed in this study. Moreover, ABCA1 mediates the efflux of cellular cholesterol and phospholipids unto ApoA-I and thereby plays a central role in regulating cellular cholesterol homeostasis, and forming HDL [48,49].
Individuals with T/T genotype for the DSCAML1 (rs10892151) polymorphisms among the cases in this study had increased levels of TC, TG, and LDL-c compared with the other combined genotypes. Further analysis indicated that individuals with the homozygous minor recessive gene (carriers) (T/T) were more likely to have increased levels of TC and LDL-C. These observed findings are in parallel with previous reports by Aragones et al. [24] who reported a strong association between the expression of the rs10892151 T allelic variant and dyslipidemia, mostly hypertriglyceridemia and decreased HDL-cholesterol levels. In contrast, Pollin et al., [23] observed that rs10892151 T carriers had lower fasting and postprandial serum triglycerides values than non-carriers, and they found a linkage disequilibrium with an APOC3 null mutation, which was likely the result of a founder effect in their high-fat feeding intervention study. Evidence provided in the literature shows that carriers of this null mutation have low circulating apolipoprotein (Apo) C-III levels and reduced fasting and post-prandial triglyceride concentrations [23], which is likely due to the well-established function of Apo C-III as an inhibitor of lipoprotein lipase [50].
This current study showed that carriers of the minor allelic variant of the APOA5 (rs662799) A>G SNP gene, were 2 times more likely to developed dyslipidemia. Moreover, individuals with homozygotes (G/G) have significantly increased levels of TC and LDL cholesterol compared to wild type and heterozygote combined (AA/AG). Previous studies have found rs662799 to be associated with elevated plasma triglyceride levels as well as HDL-C and total cholesterol [21]. Two polymorphisms in the APOA5 gene, −1131T>C and S19W (56C>G), have already been shown to be associated with elevated triglyceride levels in different populations [51,52]. This study, however, considered the APOA5 (rs662799) A>G variant and found that HIV individuals with at least one G allele had higher TC and LDL cholesterol levels.
Lazzaretti et al. [15] reported that APOA5 −1131T>C (rs662799) was associated with plasma triglycerides (TG) and low-density lipoprotein cholesterol levels (LDL-C) as well as high-density-lipoprotein cholesterol levels. To the best of our knowledge, this study is the first to investigate the APOA5 (rs662799) A>G variant in HIV patients in a Ghanaian population. Another study by Echeverria et al. [53] reported that polymorphisms in genes associated with the development of atherogenic dyslipidemia, especially variants in the APOA5 gene, can influence the circulating CD4 T-cell levels in chronically HIV-infected patients. Although a previous study has reported the effect of APOA5 on CD4 levels, individuals with the minor allelic variant of APOA5 (rs662799) A>G have a significant likelihood of developing dyslipidemia in HIV subjects on ART after controlling for CD4 counts as well as age, gender, duration of treatment and BMI.
This study has strength in it possibly being the first study to assess the prevalence of SNPs and their associations with dyslipidemia in the Ghanaian population as a proof of concept. Although previous reports have shown an uneven continuous rate of the incidence of HIV in the Ghanaian population with more than 60% of people living with HIV are females [54], the proportion of HIV-infected females in this study population was much higher than males which could affect the generalizations of our findings to the HIV population in Ghana. In addition, our study is limit by the fact that the number of investigated polymorphisms is not comprehensive, notwithstanding, this is a baseline study for future exploratory analysis of SNPs and their associations with dyslipidemia among HIV patients on ART in Ghana.

Conclusion
This study has highlighted the evidence that SNPs in four candidate genes are present in HIV patients exposed to ART in the Ghanaian population. Dyslipidemia remains prevalent among HIV patients. SNPs were found to associate with dyslipidemia especially variants in APOA5-rs662799 and LDLR-rs651172 respectively. This finding provides baseline information that necessitates a pre-symptomatic strategy for monitoring dyslipidemia in ART-treated HIV patients Findings from this study should be validated in a longitudinal case-control study considering the disease and its therapeutic implications. These candidate SNPs if validated would help in serving as potential biomarkers to detect individuals at risk for dyslipidemia.
Supporting information S1 Dataset. SPSS sheet of a dataset on which conclusions of this manuscript were made. (SAV) S1