Introduction
In 2015, there were an estimated 2.1 million [confidence interval (CI): 1.7–2.6 million] children (<15 years) living with HIV globally, with East and Southern Africa contributing 1.3 million (CI: 1.1–1.6 million) and South Africa having 320,000 (CI: 260,000–400,000).(1) An increasing percentage of these children is accessing antiretroviral therapy (ART), 51% (CI: 37–63) in East and Southern Africa and 55% (CI: 45–70) in South Africa.(1) ART has converted HIV infection into a chronic illness.(2) However, despite increased survival, there is growing concern about the metabolic complications of ART that include dyslipidaemia, especially, in children in whom the longest exposure to ART is anticipated.(3)
In both adults and children, HIV infection itself causes an abnormal lipid profile by resulting in hypertriglyceridaemia, increased very low-density lipoproteins, decreased total cholesterol and high-density lipoproteins.(4–6) ART in children initially corrects this effect towards a more normal lipid profile.(4) However, continued exposure may result in a dyslipidaemic trend in the lipid profile.(7,8) Some children demonstrate transient dyslipidaemia but in those in whom there is persistent dyslipidaemia, lipid levels increase to a plateau around 2 years on treatment.(7,8)
Paediatric studies on dyslipidaemia in HIV-infected children have largely investigated ART-associated lipodystrophy syndrome (9–12) for which the emerging consensus definition includes peripheral lipoatrophy, visceral adipose tissue accumulation, hypercholesterolaemia, hypertriglyceridaemia, insulin resistance and diabetes mellitus.(12)
Protease inhibitors (PIs) have consistently proven to be the most dyslipidaemic of the antiretroviral drugs in children.(4,7,10,13,14) The PI ritonavir or PIs that are co-formulated with ritonavir are highly dyslipidaemic.(9,15–17) This is noteworthy as all PIs are now routinely used with ritonavir. The reported prevalence of dyslipidaemia in children on PIs has ranged between 5% and 90%.(18) Paediatric studies conducted prior to 2009 on lipodystrophy syndrome, including dyslipidaemia, investigated a range of PIs, including dual PI-containing regimens.(7,10,15) However, these studies are not comparable as different definitions were applied, lipids were measured at varying time points, non-contemporary PIs were included and even dual PIs were prescribed. Furthermore, the dynamic background of HIV infection and ART-induced changes over time, as well as the effects of physiological-age- and maturation-dependent changes, impact on the comparability of these studies.(15)
There are only two studies that have been conducted on PI-associated dyslipidaemia in African children. Both studied very young children (<6 years of age) with the latter focusing on the lipodystrophy syndrome.(4,19) The prevalence of dyslipidaemia was reported to be 36% and 40%, respectively. Therefore, we sought to determine the prevalence of dyslipidaemia in African children (0–19 years) treated with second-generation PIs and the risk factors associated with development of dyslipidaemia.
Methods
We conducted a retrospective study, between 2004 and 2015, at the Harriet Shezi Children's Clinic (HSCC), Chris Hani Baragwanath Academic Hospital, Johannesburg, South Africa. Since 2004, children less than 3 years old and/or under 10 kg are initiated on a lopinavir/ritonavir (LPV/r)-based first-line regimen, while those older than 3 years and over 10 kg are initiated on a non-nucleoside reverse-transcriptase-inhibitor (NNRTI)-based first-line regimen. Patients failing an NNRTI-based first-line regimen are switched to an LPV/r-based second-line regimen. In the first guideline in 2004, the NRTI-backbone was lamivudine and stavudine for first-line therapy, and didanosine and zidovudine for second-line therapy. However, from 2010 to date, the first-line NRTI backbone was changed to lamivudine and abacavir. The tenofovir–emtricitabine combination was introduced in 2013 and became an NRTI option for those children weighing at least 35 kg. The South African Antiretroviral Treatment Guidelines stipulate that all patients on LPV/r should be monitored for dyslipidaemia with non-fasting total cholesterol and triglycerides at baseline and annually as part of routine care.(20–23)
Patient-level data were extracted from the clinic's Therapy Edge database. Demographic characteristics (age and sex), anthropometry [weight, height and body mass index (BMI)], non-fasting cholesterol and triglycerides levels, CD4 cell count, CD4 per cent, viral load (VL), co-medications and co-morbid conditions associated with dyslipidaemia were collected from the ART initiation visit to 24-month post-ART initiation. Additional laboratory data were obtained from the National Health Laboratory Service.
There were four cross-sectional points of analysis in this study: ART start (date of ART initiation), LPV/r start (LPV/r-based regimen was initiated; either as a first- or second-line regimen), 12- and 24-month post-LPV/r start. For patients who initiated ART on an LPV/r regimen, the ART start and LPV/r start visit were the same. The visit date that corresponded to the patient being on an LPV/r-based regimen for 9–15 months was defined as the 12-month visit; similarly, the visit date that corresponded to the patient being on an LPV/r-based regimen for 21–27 months was defined as the 24-month visit. Immune status was defined as normal (CD4 cell count ≥500 cells/mm3 for children older than 5 years and CD4% ≥ 25% for children younger than 5), moderate immune suppression (CD4 cell count ≥200 and <500 cells/mm3 for children older than 5 years and CD4% ≥ 15% and <25% for younger children) and severe immune suppression (CD4 cell count <200 cells/mm3 for children older than 5 years and CD4% < 15% for younger children). Anthropometric z-scores were calculated using the WHO macro for Stata.(24) Underweight was defined as a weight-for-age z-score (WAZ) less than −2 (WAZ < −2SD), while stunted was defined as a height-for-age z-score (HAZ) less than −2 (HAZ < −2SD). We used weight-for-height to assess wasting in children ≤5 years, defined as weight-for-height z-score (WHZ) less than −2 (WHZ < −2SD). For children >5 years, we used BMI to assess wasting, defined as a BMI z-score (BMIZ) less than −2 (BMIZ < −2SD). Viral suppression was defined as a VL less than 400 copies/ml.(20–23)
Dyslipidaemia was defined as hypercholesterolaemia (total cholesterol ≥5.13 mmol/l) and/or hypertriglyceridaemia (total triglycerides ≥1.69 mmol/l).(25) This definition applied to two previous South African studies,(4,19) and the studies on American children that derived these values used non-fasting lipids.(26,27) At 12 and 24 months, we chose a log VL less than 4 (10,000 copies/ml) as a proxy for significant drug pressure which was inclusive of VLs that were suppressed. This was chosen to denote reasonable adherence in order to investigate the effect of LPV/r on the measured lipids. Young children (receiving LPV/r-based ART) with higher baseline VLs take longer to achieve virologic suppression.(28)
Descriptive statistics were used to summarize the clinical and demographic characteristics of children included in the study at ART start, LPV/r start, 12 and 24 months. Chi-squared tests were used to identify categorical variables associated with dyslipidaemia. Variables associated with dyslipidaemia in the bivariate logistic models (at 20% significance, p < 0.2) were included in the multivariate logistic regression models. Stepwise selection and backwards elimination were used to determine the most parsimonious model. A p-value <0.05 was considered significant in the selection of the final model. Statistical analyses were performed using Stata 13.1 (StataCorp, College Station, TX, USA).
In order to determine whether children who had lipids measured were significantly different to those who did not have lipids measured at each of the analysis time points, a sensitivity analysis was performed. We compared demographic and clinical characteristics among those who did not have lipids measured and those who did have lipids measured, but did not have any dyslipidaemia. Ethical approval to conduct this retrospective study was obtained from the Human Research Ethics Committee of the University of the Witwatersrand.
Results
Of the 7189 HIV-infected children since April 2004, 6058 (84%) were initiated on ART; 56% (3426) of whom were treated with an LPV/r-based regimen. Patients on LPV/r for less than 9 months (757), those who initiated LPV/r 6 months prior to their first visit at HSCC (98), patients initiated on ART prior to 1 April 2004 (146) and patients with no information at 12 months (137) were excluded, leaving 2145 children included in the analysis (Fig 1).
Children treated with LPV/r1
HSCC = Harriet Shezi Children's Clinic.
1LPV/r = lopinavir/ritonavir.
By 12- and 24-month post-LPV/r initiation, children had moved to another site, had a regimen change, were lost to follow up or demised leaving 1905 and 1522 children who had visits at these time points, respectively (Fig 1). Children treated with PIs other than LPV/r were also excluded.
At ART initiation, of the 2145 children who initiated an LPV/r regimen, half (52%) were males, with 76% younger than 5 years of age. Only 17% had no immune suppression at ART start, with 51% presenting with severe immune suppression. At ART start, the median overall anthropometric z-scores were normal. However, children younger than 5 years had significantly lower z-scores compared to older children; median WAZ −2.22 [interquartile range (IQR): −3.47; −0.88] and median HAZ −2.51 (IQR: −3.64; −1.17) (Table 1).
Demographic and clinical characteristics of 2145 children ever started on an LPV/r1-based regimen at ART initiation.
| Characteristics | Overall | Age group <5 years, n (%N) | Age group 5–9 years, n (%N) | Age group 10–14 years, n (%N) | Age group 15–19 years, n (%N) | p-Value |
|---|---|---|---|---|---|---|
| n (%N) or median (IQR) | ||||||
| Total (N) | 2145 | 1635 (76.2) | 283 (13.2) | 210 (9.8) | 17 (0.8) | |
| Males | 1104 (51.5) | 840 (51.4) | 146 (51.6) | 107 (51.0) | 11 (64.7) | 0.761# |
| CD4 data available2 | 1596 (74.4) | 1201 (73.5) | 230 (81.3) | 153 (72.9) | 12 (70.6) | |
| CD4% | – | 14.8 (10; 22) | – | – | – | |
| CD4 cell count | – | – | 231 (103; 413) | 140 (28; 246) | 112 (57; 303) | <0.001* |
| Immune suppression3 | ||||||
| • Severe | 816 (51.1) | 607 (50.5) | 99 (43.0) | 103 (67.3) | 8 (66.7) | <0.001# |
| • Moderate | 511 (32.0) | 375 (31.2) | 87 (37.8) | 45 (29.4) | 4 (33.3) | |
| • None | 269 (16.9) | 220 (18.3) | 44 (19.1) | 5 (3.3) | 0 (0) | |
| Viral loads available2 | 1496 (69.7) | 1117 (68.3) | 214 (75.6) | 153 (72.9) | 12 (70.6) | |
| Log viral load in copies/ml | 5.5 (4.8; 6.1) | 5.7 (5.0; 6.2) | 5.0 (4.5; 5.4) | 5.0 (4.3; 5.4) | 5.0 (4.2; 5.5) | <0.001* |
| WAZ4 | −1.81 (−3.25; −0.35) | −2.22 (−3.47; −0.88) | 0.52 (−0.6; 2.06) | −0.15 (−0.67; 0.20) | – | <0.001* |
| HAZ5 | −1.85 (−3.30; −0.46) | −2.51 (−3.64; −1.17) | 0.45 (−0.77; 2.56) | −0.76 (−1.64; 0.63) | −1.56 (−1.99; −0.80) | <0.001* |
| WHZ in children ≤5 years6 | −0.95 (−2.23; 0.19) | −0.95 (−2.24; 0.19) | – | – | – | – |
| BMIZ in children >5–19 years7 | 0.26 (−0.56; 0.99) | – | 0.37 (−0.46; 1.09) | 0.17 (−0.64; 0.86) | −1.27 (−1.80; 0.59) | <0.001* |
LPV/r = lopinavir/ritonavir.
Chi-squared test.
Data were not available for the rest of the children at ART initiation; these values were used as the denominators for frequency percentages provided in this section of the table.
Immune suppression categories defined as follows: severe immune suppression (CD4% < 15% if age ≤5 years or CD4 cell count <200 if age >5 years); moderate immune suppression (CD4% 15%–24% if age ≤5 years or CD4 cell count 200–500 if age >5 years); no immune suppression (CD4% ≥ 25% if age ≤5 years or CD4 cell count >500 if age >5 years).
Kruskal–Wallis test.
Fisher's exact test.
WAZ = weight-for-age z-score.
HAZ = height-for-age z-score.
WHZ = weight-for-height z-score.
BMIZ = body-mass-index z-score.
Few children had lipids (either triglycerides or cholesterol) measured over the follow-up period, increasing from 7% (146/2146) at ART initiation to 24% (365/2146) after 24 months on LPV/r. All analyses were based on children who had measured lipids [n = 146 (ART start), n = 161 (LPV/r start), n = 335 (12 months post-LPV/r) and n = 365 (24 months post-LPV/r)] (Table 2).
Demographic and clinical characteristics of children with measured lipids at ART start, LPV/r start, 12 and 24 months on LPV/r1
| Variable | ART start | LPV/r start | 12 Months on LPV/r | 24 Months on LPV/r | |||||
|---|---|---|---|---|---|---|---|---|---|
| Triglycerides | Cholesterol | Triglycerides | Cholesterol | Triglycerides | Cholesterol | Triglycerides | Cholesterol | ||
| Lipids | |||||||||
| Total with measured lipids (N) | 133 | 146 | 150 | 161 | 335 | 332 | 363 | 365 | |
| Normal lipids n (%N) | 62 | 139 | 76 | 151 | 219 | 298 | 257 | 306 | |
| Elevated n (%N) | 71 (53%) | 7 (5%) | 74 (49%) | 10 (6%) | 116 (35%) | 34 (10%) | 106 (29%) | 59 (16%) | |
| Demographic and clinical characteristics n (%N) | |||||||||
| Sex | M | 67 (50%) | 69 (47%) | 69 (46%) | 71 (44%) | 157 (47%) | 158 (48%) | 188 (52%) | 188 (52%) |
| Age groups | <10 Years | 123 (93%) | 135 (93%) | 118 (79%) | 126 (78%) | 245 (73%) | 238 (72%) | 271 (75%) | 272 (75%) |
| Immune suppression | <25%/500 | 94 (71%) | 102 (70%) | 98 (65%) | 107 (67%) | 120 (36%) | 114 (34%) | 82 (23%) | 82 (23%) |
| Severe immune suppression | <15%/200 | 58 (44%) | 57 (39) | 56 (37%) | 57 (35%) | 26 (8%) | 26 (8%) | 15 (4%) | 16 (4%) |
| Log VL | <5 log | 75 (56%) | 80 (55%) | ||||||
| Log VL | <4 log | – | – | 23 (34%) | 24 (33%) | 264 (79%) | 261 (79%) | 271 (75%) | 272 (75%) |
| WAZ2 | Normal | 40 (30%) | 42 (29%) | 36 (24%) | 42 (26%) | 183 (55%) | 175 (53%) | 218 (60%) | 219 (60%) |
| HAZ2 | Normal | 40 (30%) | 32 (22%) | 47 (31%) | 48 (30%) | 178 (53%) | 174 (52%) | 188 (52%) | 191 (52%) |
| WHZ2 and BMI2 (<10)3 | Normal | 61 (50%) | 71 (53%) | 53 (35%) | 59 (37%) | 203 (83%) | 195 (82%) | 233 (86%) | 233 (86%) |
| BMI2 (≥10 years)3 | Normal | 6 (60%) | 6 (55%) | 22 (69%) | 22 (63%) | 71 (79%) | 75 (80%) | 67 (73%) | 67 (72%) |
| ART duration | <24 Months | – | – | 122 (81%) | 131 (81%) | 233 (70%) | 226 (68%) | 122 (34%) | 124 (34%) |
| 24–60 Months | – | – | 17 (11%) | 18 (11%) | 68 (20%) | 71 (21%) | 198 (54%) | 198 (54%) | |
| ≥60 Months | – | – | 11 (8%) | 12 (8%) | 34 (10%) | 35 (11%) | 43 (12%) | 43 (12%) | |
| Regimen | (NRTI)4 + efv5 | 17 (13%) | 21 (14%) | – | – | – | – | – | – |
| (NRTI) + LPV/r | 90 (68%) | 94 (65%) | 119 (79%) | 126 (78%) | 311 (93%) | 308 (93%) | 349 (96%) | 351 (96%) | |
| (NRTI) + lpvr + rtv6 | 26 (19%) | 31 (21%) | 27 | 31 | 15 | 15 | 7 | 7 | |
| LPV/r first line | 108 (81%) | 117 (80%) | 110 (73%) | 119 (74%) | 218 (65%) | 210 (63%) | 242 (67%) | 242 (66%) | |
| LPV/r second line | 25 (19%) | 29 (20%) | 40 (27%) | 42 (26%) | 117 (35%) | 122 (37%) | 121 (33%) | 123 (34%) | |
LPV/r = lopinavir/ritonavir.
WAZ = weight-for-age z-score, HAZ = height-for-age z-score, WHZ = weight-for height z-score, BMIZ = body-mass-index z-score.
The respective number of patients less and over 10 years of age were used as denominators for frequency percentages.
NRTI = nucleoside reverse transcriptase inhibitor.
Efaviranz.
Ritonavir.
Amongst those who had measured lipids, the prevalence of dyslipidaemia at ART start was 47%, decreasing to 36% at 24 months. The prevalence of hypercholesterolaemia was 5% at ART start, increasing to 16% by 24 months, while the prevalence of hypertriglyceridaemia decreased from 53% at ART start to 29% by 24 months (Fig 2).
Prevalence of dyslipidaemia (hypertriglyceridaemia or hypercholesterolaemia)
1LPV/r = lopinavir/ritonavir.
Very few patients had measured lipids at all four time points; 16 and 19 for triglycerides and cholesterol, respectively. The median age of these patients was 1.2 (IQR 1.2; 2) years; thus, the ART start and LPV/r start visits were the same. Amongst these patients, the rates of dyslipidaemia were 43%, 39% and 68% at ART start, 12 and 24 months, respectively (Fig 3). In this small cohort, we observed a high proportion with hypertriglyceridaemia at ART start (44%) which decreased slightly (31%) at 12 months but increased again (47%) at 24 months. In contrast, hypercholesterolaemia increased from 12% at ART start to 37% by 24 months (Fig 3).
Prevalence of dyslipidaemia, hypertriglyceridaemia and hypercholesterolaemia among children with lipid measurements at all three time points (ART start, on LPV/r1 for 12 and 24 months)
1LPV/r = lopinavir/ritonavir.
Univariate analyses of the effect of demographic and clinical factors at ART initiation showed that children older than 10 years were less likely to have dyslipidaemia [odds ratio (OR) = 0.10, 95% CI: 0.01–0.81] and those with a higher VL (OR = 2.5, 95% CI: 1.09–5.67) were more likely to have dyslipidaemia. In multivariate analysis, only older age remained negatively associated with dyslipidaemia (OR = 0.10, 95% CI: 0.01–0.81) (Table 3).
Univariate and multivariate analysis at ART start and LPV/r1 start.
| Variable | ART start | LPV/r start | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Univariate analysis | Multivariate analysis | Univariate analysis | Multivariate analysis | ||||||
| Unadjusted OR (95% CI) | p-Value | Adjusted OR (95% CI) | p-Value | Unadjusted OR (95% CI) | p-Value | Unadjusted OR (95% CI) | p-Value | ||
| Males | F | 1 | 1 | ||||||
| M | 1.3 (0.68–2.35) | 0.460 | 1.18 (0.65–2.15) | 0.58 | |||||
| Age groups | <10 years | 1 | 1 | 1 | |||||
| 10–19 years | 0.1 (0.01–0.81) | 0.031 | 0.1 (0.01–0.81) | 0.031 | 0.28 (0.12–0.66) | 0.004 | |||
| Immune suppression | ≥25%/500 | 1 | 1 | ||||||
| <25%/500 | 1.64 (0.61–4.41) | 0.330 | 1.75 (0.77–4.06) | 0.179 | |||||
| Severe immune suppression | ≥15%/200 | 1 | 1 | ||||||
| <15%/200 | 1.80 (0.90–3.61) | 0.098 | 2.02 (1.04–3.93) | 0.037 | |||||
| Log VL | <5 log | 1 | 1 | ||||||
| ≥5 log | 2.49 (1.09–5.67) | 0.030 | 1.05 (0.44–2.51) | 0.912 | |||||
| WAZ2 | Normal | 1 | 1 | ||||||
| Underweight | 1.80 (0.87–3.76) | 0.115 | 1.45 (0.68–3.10) | 0.336 | |||||
| HAZ2 | Normal | 1 | 1 | ||||||
| Stunted | 1.51 (0.73–3.13) | 0.266 | 1.57 (0.80–3.10) | 0.193 | |||||
| WHZ3 and BMIZ3 (<10) | Normal | 1 | 1 | ||||||
| Wasted | 1.53 (0.70–3.32) | 0.285 | 1.81 (0.78–4.17) | 0.164 | |||||
| BMIZ (>10) | Normal | 1 | – | 1 | |||||
| Wasted | –4 | 0.9 (0.08–9.97) | 0.932 | ||||||
| ART duration | <24 Months | – | 1 | 1 | |||||
| 24–60 Months | – | 0.19 (0.05–0.77) | 0.010 | 0.19 (0.05–0.77) | 0.010 | ||||
| ≥60 Months | – | 0.18 (0.04–0.88) | 0.034 | 0.18 (0.04–0.88) | 0.034 | ||||
| Regimen | (NRTI)3 + efv5 | – | – | ||||||
| (NRTI) + LPV/r | – | 1 | |||||||
| (NRTI) + LPV/r + rtv6 | – | 1.66 (0.79–3.48) | 0.180 | ||||||
| LPV/r as first or second line | First line | – | 1 | ||||||
| Second line | – | 0.31 (0.15–0.67) | 0.003 | ||||||
LPV/r = lopinavir/ritonavir.
WAZ = weight-for-age z-score, HAZ = height-for-age z-score, WHZ = weight-for height z-score, BMIZ = body-mass-index z-score.
NRTI = nucleoside reverse transcriptase inhibitor.
Not evaluable (only six measured lipids in the normal BMIZ category).
Efaviranz.
Ritonavir. Bold values represent statistical significance
Children who were older than 10 years of age (OR = 0.28, 95% CI: 0.12–0.66), on ART for more than 24 months (OR = 0.19, 95% CI: 0.05–0.77) and on LPV/r as a second-line regimen (OR = 0.31, 95% CI: 0.15–0.67), were less likely to have dyslipidaemia, while those presenting with severe immune suppression (OR = 2.02, 95% CI: 1.04–3.93) were more likely to have dyslipidaemia at LPV/r start (Table 3). However, on multivariate analyses, only ART duration greater than 24 months remained negatively associated with dyslipidaemia (OR = 0.19, 95% CI: 0.05–0.77) (Table 3).
At 12 months after LPV/r start, univariate analyses of the effect of demographic and clinical factors on dyslipidaemia showed that being over 10 years (OR = 0.37, 95% CI: 0.22–0.65), having a high VL (log VL ≥ 4) (OR = 0.38, 95% CI: 0.18–0.79), being on ART for 24–60 months (OR = 0.53, 95% CI: 0.30–0.95) and being on LPV/r as a second-line regimen (OR = 0.48, 95% CI: 0.30–0.76) were protective against dyslipidaemia. In multivariate analyses, only age older than 10 and high VLs remained associated with dyslipidaemia (OR = 0.41, 95% CI: 0.22–0.75 and OR = 0.45, 95% CI: 0.21–0.95, respectively) (Table 4).
Univariate and multivariate analysis at LPV/r1 12 and 24 months.
| Variable | 12 Months | 24 Months | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Univariate analysis | Multivariate analysis | Univariate analysis | Multivariate analysis | ||||||
| Unadjusted OR (95% CI) | p-Value | Adjusted OR (95% CI) | p-Value | Unadjusted OR (95% CI) | p-Value | Adjusted OR (95% CI) | p-Value | ||
| Sex | F | 1 | 1 | ||||||
| M | 0.89 (0.58–1.38) | 0.611 | 1.17 (0.77–1.79) | 0.46 | |||||
| Age groups | <10 Years | 1 | 1 | 1 | |||||
| 10–19 Years | 0.37 (0.22–0.65) | 0.000 | 0.41 (0.22–0.75) | 0.004 | 0.39 (0.23–0.67) | 0.001 | |||
| Immune suppression | ≥25%/500 | 1 | 1 | ||||||
| <25%/500 | 0.64 (0.40–1.03) | 0.063 | 0.77 (0.45–1.29) | 0.325 | |||||
| Severe immune suppression | ≥15%/200 | 1 | 1 | ||||||
| <15%/200 | 0.58 (0.24–1.42) | 0.232 | 0.58 (0.18–1.85) | 0.361 | |||||
| Log VL | <4 log | 1 | 1 | 1 | 1 | ||||
| ≥4 log | 0.38 (0.18–0.79) | 0.009 | 0.45 (0.21–0.95) | 0.036 | 0.33 (0.14–0.78) | 0.011 | 0.33 (0.14–0.79) | 0.012 | |
| WAZ2 | Normal | 1 | 1 | ||||||
| Underweight | 0.99 (0.52–1.89) | 0.971 | 1.31 (0.68–2.45) | 0.433 | |||||
| HAZ2 | Normal | 1 | 1 | ||||||
| Stunted | 1.07 (0.69–1.67) | 0.756 | 1.23 (0.81–1.89) | 0.328 | |||||
| WHZ2 and BMIZ2 (<10) | Normal | 1 | 1 | ||||||
| Wasted | 0.28 (0.06–1.37) | 0.116 | 2.15 (0.59–7.82) | 0.245 | |||||
| BMIZ (>10) | Normal | 1 | 1 | ||||||
| Wasted | 0.63 (0.13–3.16) | 0.578 | 0.87 (0.21–3.49) | 0.849 | |||||
| ART duration | <24 Months | 1 | |||||||
| 24–60 Months | 0.53 (0.30–0.95) | 0.033 | 1 | ||||||
| ≥60 Months | 0.47 (0.21–1.05) | 0.066 | 0.15 (0.05–0.44) | 0.001 | 0.19 (0.06–0.56) | 0.002 | |||
| Regimen | (NRTI)3 + lpv/r | 1 | 1 | ||||||
| (NRTI) + lpvr + rtv4 | 1.91 (0.67–5.39) | 0.223 | 0.28 (0.03–2.39) | 0.247 | |||||
| LPV/r as first or second line | First line | 1 | 1 | ||||||
| Second line | 0.48 (0.30–0.76) | 0.002 | 0.41 (0.26–0.67) | <0.000 | |||||
LPV/r = lopinavir/ritonavir.
WAZ = weight-for-age z-score, HAZ = height-for-age z-score, WHZ = weight-for height z-score, BMIZ = body-mass-index z-score.
NRTI: nucleoside reverse transcriptase inhibitor. Bold values represent statistical significance
Similarly, at 24 months after LPV/r start, univariate analyses showed that age older than 10 (OR = 0.39, 95% CI: 0.23–0.67), high VLs (OR = 0.33, 95% CI: 0.14–0.78), ART duration of at least 60 months (OR = 0.15, 95% CI: 0.05–0.44) and LPV/r as a second-line regimen (OR = 0.41, 95% CI: 0.26–0.67) were protective against dyslipidaemia. Only high VLs and ART duration of at least 60 months remained protective of dyslipidaemia in the multivariate model (OR = 33, 95% CI: 0.14–0.79 and OR = 0.19, 95% CI: 0.06–0.56, respectively) (Table 4).
Discussion
This study found a high prevalence of dyslipidaemia among children who were ART-naive, and this pattern is consistent with other studies in the literature.(4,29) Our findings suggest a trend towards a decreasing prevalence for hypertriglyceridaemia (53%–29%) from ART start to 24 months, while the prevalence of hypercholesterolaemia increased from 5% to 16%, respectively.
Viraemia-induced dyslipidaemia at baseline has been described with ART, initially decreasing triglycerides and increasing cholesterol towards normal. Further lipid increases occur with continued ART, notably with PIs.(4,6) The recommended NRTI backbone has varied in the various iterations of the guidelines. An initial unreported analysis based on different NRTI backbones found no association with dyslipidaemia at all four time points; hence, the effect of the different NRTIs on dyslipidaemia was not studied. A transient dyslipidaemia has also been described on ART, including PIs, with stabilization expected to occur by 24 months. Hence, the high prevalence of dyslipidaemia still present at 24 months (36%) is concerning.
In this study, we used a log VL less than 4 as a proxy for significant drug pressure. Notwithstanding, it is important to note that at 12 months, 77% of the patients with a log VL less than 4 were by definition virally suppressed (VL < 400) and at 24 months 82% were virally suppressed.
Children younger than 10 years of age had a significant association with dyslipidaemia on univariate analysis at all four time points. Furthermore on multivariate analysis, scrutiny of the PI (LPV/r) effects at 12 months, found that children younger than 10 years with near suppressed/suppressed VLs (<4 logs) were more likely to have dyslipidaemia. At 24 months on LPV/r, ART duration greater than 60 months was an additional protective factor on multivariate analysis.
The long-term cardiovascular and metabolic consequences of PI-induced dyslipidaemia in these young children is largely unknown. However, several studies suggest that this poses a potential cardiovascular risk.(30–33) Children initiating ART at less than 3 years of age can be expected to remain on LPV/r for a very long time. The findings from NEVEREST III, which showed no significant difference in viral rebound and viral failure post switch to an efavirenz-based regimen, offer hope of ameliorating this risk.(34)
Typical of retrospective studies, not all the variables of interest were comprehensively collected in this routine-care setting; notably, lipids were particularly poorly investigated at routine clinical visits. The authors hypothesize that the decreased lipid measurement in children younger than 10 years old was possibly related to difficulties with drawing large volumes of blood or the limited therapeutic options for young children with dyslipidaemia. In addition, due to the retrospective and routine nature of the database, we were not able to measure the effect of co-medications and co-morbid conditions on dyslipidaemia as very few patients had both measured lipids and information on co-morbidities and co-medications available. Nevertheless, we believe that the information from this article is valuable as it describes the largest set of paediatric lipid profiles in children on PIs.
Similar to other studies, we did not obtain lipids from children in the fasting state.(4,7,8,35) Nevertheless, triglycerides are affected by the non-fasting state, while cholesterol is not affected by the non-fasting state.(36) There are no cholesterol and triglycerides reference values available for South African children; thus, the effect of the use of reference values from North American children on the presented results is unknown. Lastly, this study could not evaluate the effect of diet, smoking or family history of lipid disorders as these data were not routinely collected.
The strength of this study lies in the multiple cross- sectional points for the evaluation of the prevalence of dyslipidaemia suggesting the increasing prevalence of hypercholesterolaemia over time and the decreasing prevalence of hypertriglyceridaemia, thereby confirming findings from other studies.(4,14) This study highlights the importance of adherence to guidelines and the creation of quality assurance measures to audit clinical practice.
Conclusion
The high prevalence of dyslipidaemia, notably, amongst young children expected to be treated with PIs such as LPV/r for long durations is concerning. This underscores the need for increasing safer drug options for this population in low-income countries who have limited access to newer drug classes like integrase inhibitors. Furthermore, long-term evaluation for dyslipidaemia and its cardiovascular complications is essential for those already on therapy.