Arch Dis Child 98:258-264 doi:10.1136/archdischild-2012-302633
  • Original article

Metabolic abnormalities and body composition of HIV-infected children on Lopinavir or Nevirapine-based antiretroviral therapy

  1. Louise Kuhn1,2
  1. 1Gertrude H. Sergievsky Center, Columbia University, New York, New York, USA
  2. 2Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, New York, USA
  3. 3Faculty of Health Sciences, Rahima Moosa Mother and Child Hospital, University of the Witwatersrand, Johannesburg, South Africa
  4. 4Department of Pediatrics, College of Physicians and Surgeons, Columbia University, New York, New York, USA
  1. Correspondence to Dr Stephen Arpadi, Gertrude H. Sergievsky Center, College of Physicians and Surgeons, Columbia University Medical Center, 630 W. 168th Street, New York, NY 10032, USA; sma2{at}
  • Accepted 15 October 2012
  • Published Online First 5 December 2012


Background Few studies have assessed metabolic and body composition alterations in perinatally HIV-infected African children on antiretroviral therapy (ART). We compared metabolic profiles and regional fat of children on ritonavir-boosted lopinavir (lopinavir/ritonavir), lamivudine and stavudine to those switched to nevirapine, lamivudine and stavudine.

Methods This study evaluated metabolic and body composition outcomes in 156 HIV-infected children completing a randomised trial that assessed the continued use of lopinavir/ritonavir-based ART or switch to nevirapine-based ART in Johannesburg, South Africa (2005–2010). Fasting total cholesterol (TC), high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides total and regional body fat (BF) were measured. A clinical assessment for lipodystrophy (LD) was conducted.

Results 156 children (mean age 5.1±0.8 years, mean duration of treatment 4.2±0.7 years, mean time since randomisation 3.4±0.7 years) were enrolled. 85 were randomised to the lopinavir/ritonavir group and 71 to the nevirapine group. The lopinavir/ritonavir group had lower mean HDL (1.3±0.4 vs 1.5±0.4 mmol/l, p<0.001) and higher mean TC (4.4±1.0 vs 4.1±0.8 mmol/l, p=0.097), LDL (2.6±0.9 vs 2.3±0.7 mmol/l, p=0.018) and triglycerides (1.1±0.4 vs 0.8±0.3 mmol/l, p<0.001). The lopinavir/ritonavir group had more total BF by mean skinfold sum (43±11.1 vs 39±10.1 mm, p=0.031) and BF% by bioelectrical impedance analysis (17.0±7.0 vs 14.1±8.0%, p=0.022). Thirteen (8.4%) met criteria for LD.

Conclusions Unfavourable alterations in lipid profile and triglycerides, and differences in fat are detectable in young HIV-infected South African children receiving lopinavir/ritonavir-based regimens versus those switched to nevirapine-based regimens. Interventions to mitigate these alterations are warranted to reduce long-term cardiovascular disease risk.

What is already known on this topic

  • Changes in lipid, glucose and triglyceride metabolism, and regional fat are well described among older children and adolescents with perinatally acquired HIV-infection receiving antiretroviral therapy (ART) in high income countries.

  • There are limited studies of younger HIV-infected children in sub-Saharan Africa, where >90% of HIV-infected children live and where there is increasing availability of ART.

What this study adds

  • Hypercholesterolemia and hypertriglyceridemia were detectable in HIV-infected children who were initiated on ART before the age of 2 years.

  • These metabolic markers were significantly higher among patients on lopinavir/ritonavir-based than nevirapine-based ART.

  • Lipodystrophy is visible in 8.4% of children and was confirmed by simple anthropometry.


Unfavourable alterations in lipid and glucose metabolism, well established risk factors for cardiovascular disease (CVD), are reported among children and adolescents with perinatally acquired HIV infection receiving antiretroviral therapy (ART).18 Significant proportions of ART-treated children have elevated total cholesterol (TC), low density lipoprotein (LDL) and triglycerides, and abnormal glucose and insulin homeostasis.18 Multiple factors underlie the aetiology of these metabolic derangements, including specific ART drugs, particularly protease inhibitors (PIs) which are an important cause of dyslipidemia and impaired glucose homeostasis.13 ,710 Exposure to nucleoside (NRTI) and non-nucleoside reverse transcriptase inhibitors (NNRTIs) has also been associated with dyslipidemia.7

Abnormal body fat (BF) distribution, or lipodystrophy (LD), also occurs in HIV-infected children both alone and in combination with metabolic abnormalities, and may also contribute to increased CVD risk.2 ,1116 A number of drugs, including thymidine analogue NRTIs and PIs, are associated with LD.17 With greatly improved survival due to effective therapies, the development of treatment strategies for children that optimise long-term viral suppression and limit long-term adverse effects is an important goal.

Few studies have evaluated lipid profiles and insulin-glucose metabolism or BF distribution in young children receiving ART in sub-Saharan Africa, where over 90% of HIV-infected children live.18 The objectives of this study are to examine the lipid profiles, insulin sensitivity, markers of inflammation, and regional fat distribution of HIV-infected children in South Africa who were initiated on a PI-based regimen prior to age 2 years and randomly assigned to continue on a PI-based regimen or switch to an NNRTI-based regimen. These parameters are also compared in children with and without LD.

Patients and methods

Study design

This cross-sectional study evaluated HIV-infected children at the final study visit of a randomised clinical trial (NEVEREST 2) that assessed continued ritonavir-boosted lopinavir (lopinavir/ritonavir)-based therapy versus switching to nevirapine-based therapy, conducted from 2005 to 2010.19 In NEVEREST 2, children who maintained a viral load (VL) of <400 copies/ml for ≥3 months while on their primary lopinavir/ritonavir-containing regimen were randomised to one of two groups. Children randomised to the lopinavir/ritonavir group continued to receive lopinavir/ritonavir (230 mg/m2), lamivudine (4 mg/kg) and stavudine (1 mg/kg), while the children in the nevirapine group switched to nevirapine (120 mg/m2 once per day for the first 2 weeks and thereafter 200 mg/m2 every 12 h), lamivudine and stavudine, the two regimens recommended for paediatric use in South Africa.20 The study was conducted at the Empilweni Clinical Research Unit at Rahima Moosa Mother and Child Hospital in Johannesburg, South Africa and was approved by the Institutional Review Boards of Columbia University and the University of the Witwatersrand. Each child's guardian provided signed informed consent.


Weight (kg) was measured using a digital scale and height (cm) was measured using a stadiometer. Body mass index (BMI) was calculated as weight (kg) divided by squared height (m2). Weight-for-age (WAZ), height-for-age (HAZ) and BMI-for-age z scores (BAZ) were determined using WHO standards.21 Blood pressure was measured using an electronic sphygmomanometer with a paediatric cuff (Angelus, model number SN-AJ710150) and compared with age and sex reference values.22 A clinical assessment for LD was completed by two trained physicians (RS, LM) and children were classified as having LD (LD+), possible LD (possible LD) or no signs of LD (LD−) if ≥2, 1 or 0 features of lipoatrophy (ie, sunken cheeks, temporal wasting, skinny limbs or wasting of buttocks) or lipohypertrophy (ie, increased abdominal girth, dorsal cervical enlargement or breast enlargement) were present.12

Following an overnight fast (≥8 h), TC, LDL, high-density lipoprotein (HDL), triglycerides, C-reactive protein (CRP) and insulin were measured using the Roche COBAS INTEGRA 400 system and venous blood glucose with a handheld glucometer. VL by quantitative HIV-1 RNA PCR (Cobas Ampliprep Taqman V2, Roche), absolute CD4 count and CD4 percentages (Beckman Coulter Flow Analyzer) were also determined. TC categories were set as <4.4 mmol/l (<170 mg/dl) for acceptable, 4.4–5.1 mmol/l (170–199 mg/dl) for borderline, and ≥5.2 mmol/l (≥200 mg/dl) for elevated.23 A cut-off value of 0.9 mmol/l (<35 mg/dl) was used for abnormal HDL concentrations.24 LDL categories were set as <2.9 mmol/l (<110 mg/dl) for acceptable, 2.9–3.3 mmol/l (110–129 mg/dl) for borderline, and ≥3.4 mmol/l (≥130 mg/dl) for elevated. Triglycerides >1.69 mmol/l (>150 mg/dl) and CRP >3 mg/l were considered elevated.2325 Impaired fasting glucose was defined as ≥6.11 mmol/l (≥110 mg/dl).26 Homeostasis model assessment of insulin resistance (HOMA-IR) >3.16 was used to determine insulin resistance.27 All impaired fasting glucose measurements were repeated for confirmation.

Anthropometrics, including circumferences and skinfold thicknesses, were measured by two trained physicians (RS, LM) unblinded to treatment using the average of triplicate standardised measurements. Circumferences were measured to the nearest 0.1 cm using a flexible tape measure with a spring tension attachment at the mid-upper arm (MUAC), mid-upper thigh (MTC), mid-waist (MWC) and maximum hip (MHC). Skinfolds, including biceps skinfold (BSF), triceps skinfold (TSF), subscapular skinfold (SSF), suprailiac skinfold (SISF), umbilical skinfold (USF) and mid-thigh skinfold (MTSF), were measured with a Harpendon caliper (Baty International, Burgess Hill, UK). Cumulative skinfold sum (SFS) was calculated by adding BSF, TSF, SSF, SISF, USF and MTSF.

Extremity fat areas (cm2), including upper arm fat area (AFA), percentage of fat in upper arm (%AF), upper leg fat area (LFA) and percentage of fat in upper leg (%LF), were calculated.28 Regional (arm, trunk, leg) fat as proportions of total BF were calculated (arm: (BSF + TSF)/SFS; trunk: (SSF + SISF + USF)/SFS; and leg: MTSF/SFS). The trunk–extremity skinfold ratios were calculated as indicators of central subcutaneous fat as compared to extremity fat (trunk–arm: (SSF + SISF)/(BSF + TSF + SSF + SISF); and trunk–leg: (SSF + SISF)/(MTSF + SSF + SISF)).29 Total body resistance was measured by single-frequency bioimpedance analysis (BIA) (Quantum II, RJL Systems, Clinton Township, Michigan, USA) and used in a prediction equation to estimate %BF.30

Statistical analysis

Modified intent-to-treat analyses of the treatment groups (lopinavir/ritonavir vs nevirapine) were conducted using the Wilcoxon rank sum test for non-normally distributed and the t test for normally distributed continuous variables. The χ2 or Fisher's exact test were used for categorical variables. These tests were also used for comparisons across LD categories. Multiple linear regression was performed to assess differences in arm, trunk and leg fat across LD groups, adjusting for total fat, sex and age. All statistical tests were two-sided and a p value <0.05 was considered statistically significant. Analyses were performed using SAS V.9.1.3 (SAS Institute, Cary, North Carolina, USA).


Of the 195 children randomised in the NEVEREST 2 clinical trial, 156 completed the final study visit and were included in this analysis (figure 1). Six (3.1%) children died, 28 (14.4%) were lost to follow-up, and five (2.6%) transferred out. At the time of this assessment, subjects were between 3.3 and 7.3 years old, 52% were male, and 85 were previously randomised to the lopinavir/ritonavir group and 71 to the nevirapine group (table 1). There were no differences in total time on ART or time since randomisation between the groups. WAZ, HAZ or BAZ were similar between groups at randomisation and at the time of this assessment. At the final study visit, five (5.9%) in the lopinavir/ritonavir group and seven (9.9%) in the nevirapine group had moderate or severe malnutrition, that is, WAZ <−2.21 Sixteen (18.8%) children in the lopinavir/ritonavir group and 13 (18.3%) in the nevirapine group had a HAZ <−2. No differences in the proportion of children with abnormal systolic (38.8 vs 26.8%, p=0.112) or diastolic (42.3 vs 35.2%, p=0.363) blood pressure were detected. Most children (88%) had a VL of <50 copies/ml. There were no virological differences between the groups. Compared to the lopinavir/ritonavir group, the nevirapine group had a higher median CD4 cell count. There were no differences in LD classification between the lopinavir/ritonavir and nevirapine groups.

Table 1

Characteristics at the time of assessment of 156 perinatally HIV-infected South African children previously suppressed on ritonavir-boosted lopinavir (LPV/r)-based treatment and randomised to either maintain the LPV/r-based regimen or switch to a nevirapine (NVP)-based regimen

Figure 1

Flow diagram of study participants in NEVEREST 2 (2005–2010) in Johannesburg, South Africa detailing the number of subjects excluded from the analysis, the reason for their exclusion, and distribution of included subjects by randomisation group. Time intervals are provided as mean±SD. ART, antiretroviral therapy; LPV/r, ritonavir-boosted lopinavir; NVP, nevirapine; VL, viral load.

Several metabolic measurements differed between the treatment groups (table 2). The mean TC was greater in the lopinavir/ritonavir group, as was the proportion with elevated TC. Significantly lower mean HDL concentrations and higher mean LDL concentrations were observed in the lopinavir/ritonavir group. Additionally, a significantly higher ratio of TC:HDL was observed in the lopinavir/ritonavir group. The mean triglyceride concentration and proportion with abnormal triglycerides were also greater in the lopinavir/ritonavir group. Mean CRP levels and the proportion with elevated CRP were lower for the lopinavir/ritonavir group. There were no differences in recent (ie, past 1 month) or acute illnesses between the groups.

Table 2

Metabolic and other laboratory assessments of 156 perinatally HIV-infected South African children suppressed on ritonavir-boosted lopinavir (LPV/r)-based treatment and randomised to either maintain the LPV/r-based regimen or switch to a nevirapine (NVP)-based regimen

In addition, multiple abnormalities were more common in the lopinavir/ritonavir group. More children in the lopinavir/ritonavir group had elevated TC and LDL (13.2% vs 5.5%, p = 0.054) and elevated TC and triglycerides (5.9% vs 0.0%, p = 0.038) compared to children in the nevirapine group. Mean glucose and HOMA-IR did not differ between the groups. A repeat impaired fasting glucose was detected in one (1.2%) child in the lopinavir/ritonavir group and one (1.4%) child in the nevirapine group. None in the lopinavir/ritonavir group had insulin resistance compared to three (4.3%) in the nevirapine group (p=0.09).

Complete body composition measurements were available for 139 children (table 3). Children in the lopinavir/ritonavir group had a greater amount of BF than children in the nevirapine group, as measured by mean SFS. A similar pattern in total BIA-determined %BF was observed. LFA and percentage of fat in the upper leg were both significantly greater in the lopinavir/ritonavir group. In addition, while the arm fat proportion of total fat was not different between the groups, the trunk fat proportion of total fat was lower for the lopinavir/ritonavir group and the leg fat proportion of total fat was significantly greater for the lopinavir/ritonavir group. Although there was no difference in trunk–arm skinfold ratios between the groups, the lopinavir/ritonavir group had a lower trunk–leg skinfold ratio compared to the nevirapine group.

Table 3

Body composition measures of 156 perinatally HIV-infected South African children suppressed on ritonavir-boosted lopinavir (LPV/r)-based treatment and randomised to either maintain the LPV/r-based regimen or switch to a nevirapine (NVP)-based regimen (mean±SD)

Thirteen (8.3%) children were classified as LD+ and 18 (11.5%) as possible LD. Of the 13 LD+ children, all had signs of lipoatrophy and six also had signs of lipohypertrophy. There were no differences in age, sex or age at ART initiation, total time on ART, randomisation group, WAZ, HAZ, BAZ or proportion with VL<50 copies/ml between the three LD groups.

No differences in TC, HDL, LDL, TC:HDL ratio, CRP, glucose or HOMA-IR were detected across LD groups (data not shown). However, mean triglycerides and proportion with abnormal triglycerides was higher for LD+ than LD− children, as well as for possible LD than LD− children.

Overall, LD+ children had significantly less BF compared to LD− children as measured by SFS. A similar pattern was observed using BIA-based estimates of total BF%, although this did not achieve statistical significance. As presented in figure 2, no differences were seen in arm fat; however, LD+ children had a greater trunk fat proportion of total fat and less leg fat as a proportion of total fat compared to LD− children. A similar pattern of differences in arm (p=0.04), trunk (p=0.01) and leg (p=0.05) fat between LD+ and LD− children was detected after adjusting for total fat, sex and age by means of multiple linear regression. LD+ children also had significantly greater trunk–arm and trunk–leg skinfold ratios compared to LD− children.

Figure 2

Regional fat (as % of total) by skinfolds of children with lipodystrophy (LD+), possible LD or no LD (LD−), as determined by clinician assessment. Children were classified as LD+, possible LD or LD− if >2, 1 or 0 features of lipoatrophy (ie, sunken cheeks, temporal wasting, skinny limbs, wasting of buttocks) or lipohypertrophy (ie, increased abdominal girth, dorsal cervical enlargement, breast enlargement) were present. BSF, biceps skinfold; LD, lipodystrophy; MTSF, mid-thigh skinfold; SFS, skinfold sum; SISF, suprailiac skinfold; SSF, subscapular skinfold; TSF, triceps skinfold; USF, umbilical skinfold.


In this study of young perinatally HIV-infected South African children initiated and suppressed on PI-based ART and then randomised to either remain on the initial PI-based regimen or switch to an NNRTI-based regimen, the group remaining on the lopinavir/ritonavir-based regimen had unfavourable concentrations of lipoproteins and triglycerides compared to those switching to the nevirapine-based regimen. These results confirm and extend our prior, shorter term observations and other published studies.1 ,3 ,4 ,31 ,32 With extended survival and the requirement for lifelong ART, these metabolic alterations may pose long-term risk with respect to CVD, as CVD often results from exposure to risk factors that have their origins in childhood.33 Serum lipid concentrations, insulin resistance, hypertension and obesity early in life contribute to the development of atherosclerotic lesions in children and adolescents.34 ,35 In addition, studies among adults contracting HIV indicate that HIV infection per se may be associated with greater CVD, possibly due to chronic immune activation.36 Also, increased intima-media thickness, an indicator of atherosclerosis, has been reported in studies of HIV-infected children and adolescents receiving ART.37 The constellation of risk factors seen during childhood and adolescence among those with perinatally-acquired HIV suggests that strategies for CVD risk reduction during childhood should be developed.

The elevated CRP reported in this study has also been previously reported.37 While CRP might have been expected to cluster with alterations in lipids and triglycerides or regional fat depots, our finding of elevated CRP in the NNRTI group compared to the PI group is not without precedent nor is the long-term significance known.38 ,39

In this cohort of children receiving stavudine-backbone ART, continuation on a PI-based regimen was associated with greater overall subcutaneous fat, and less central (trunk) fat in relation to leg fat in comparison to children switched to an NNRTI-based regimen. This difference in subcutaneous regional fat was detectable even though the groups had a similar overall WAZ, BMI and proportion of children with LD. Since children in our study were randomly assigned to continue on a PI-based regimen or switch to an NNRTI-based regimen, our findings provide less confounded evidence that PI-based regimens are associated with greater amounts of subcutaneous fat compared to NNRTI than available from previous non-randomised studies.2 ,13 ,40 ,41 It is not possible to determine which of the patterns of changes in adipose tissue we observed is a departure from normal, since our study did not include a healthy comparison group and appropriate skinfold standards are not available for young South African children. In the present study no relationship was detected between adipose and metabolic measures. Whether the clustering of these features will evolve with additional time and exposure to ART remains to be determined.

LD is cosmetically stigmatising and can adversely affect adherence to ART.42 The finding that a substantial proportion of young children had LD has implications for future adherence, especially during adolescence when awareness of physical appearance is heightened. Most studies indicate the prevalence of LD increases with age, duration of therapy, and puberty.5 ,15 In addition, stavudine, possibly due to inhibition of mitochondrial DNA polymerase, has been implicated in lipoatrophy, and for this and other reasons, WHO advises phasing out or discontinuing stavudine.43 ,44 Of interest, anthropometry appears to confirm the distinctive regional fat pattern of LD observed by clinician assessment. Standardised clinical and anthropometric assessments for early detection of changes in regional fat are a valuable addition to clinical studies.

There are a number of limitations to this study. The metabolic and body composition measurements did not assess visceral adiposity, were obtained at the final visit of this trial, and were not available prior to randomisation. For those switched back to nevirapine, the observation time was relatively short. In addition, a family history of lipid disorders, diet and exercise may be relevant to our findings but were not assessed. In addition, although significant proportions of subjects had abnormal blood pressure, interpretation is difficult as the norms applied were from US children. Furthermore, the measurement of glucose-insulin homeostasis used is relatively insensitive compared to more invasive tests.8 Finally, a single CRP measurement may not be reflective of long-term inflammatory status.

Lopinavir/ritonavir is highly effective in suppressing HIV and an essential part of treatment for children in resource-limited settings. WHO includes lopinavir/ritonavir as part of first-line treatment for HIV-infected children with prior exposure to NNRTI used for prevention of mother-to-child transmission and for those who experience treatment failure with NNRTIs.43 The long-term impact of the BF and lipid alterations bears further study, as do efforts to devise safe and effective approaches for the prevention and management of these abnormalities. It is also essential to expand access to newer antiviral agents with potentially fewer metabolic effects to children in resource-limited settings.43 ,45


  • Contributors SA, LK, EJA and AC designed the study. SA, LK, EJA, AC, LM and RS developed the protocol. AC, RS, LM and FP were responsible for clinical oversight. LK, SS and SA carried out the data analysis. All authors were involved in the interpretation of the findings and the writing of the report.

  • Funding This study was supported in part by the National Institute of Child Health and Human Development (grant numbers HD 47177, HD 61255) and Secure the Future Foundation (grant number RES 219).

  • Competing interests None.

  • Ethics approval Columbia University and the University of the Witwatersrand approved this study.

  • Provenance and peer review Not commissioned; externally peer reviewed.


Free sample

This recent issue is free to all users to allow everyone the opportunity to see the full scope and typical content of ADC.
View free sample issue >>

Don't forget to sign up for content alerts so you keep up to date with all the articles as they are published.