Anthropometric definitions for antiretroviral-associated lipodystrophy derived from a longitudinal South African cohort with serial dual-energy X-ray absorptiometry measurements

Abstract

The development of lipodystrophy is associated with the long-term use of antiretroviral therapy (ART). We assessed agreement between patient-reported lipodystrophy and body composition measures using dual-energy X-ray absorptiometry (DXA) and developed objective measures to define lipoatrophy and lipohypertrophy in black South Africans. One hundred and eighty-seven ART-naïve HIV-infected adults were enrolled in a 24-month longitudinal study. Self-reported information on regional fat loss and fat gain, anthropometry, and DXA measures were collected at baseline, three, six, 12, 18, and 24 months after starting ART. Receiver operating characteristic curves were used to describe the performance of anthropometric variables using change in limb and trunk fat measured by DXA, as the reference standard. The proportion of men and women who developed lipoatrophy and lipohypertrophy increased over the 24-month period, with lipoatrophy occurring more frequently in men (21% versus 10%). In women, lipoatrophy was best determined by thigh skinfold thickness (80.3% correctly classified) and mid-arm circumference (77.6% correctly classified). None of the anthropometric measures performed well for defining lipoatrophy in men. Anthropometric measures performed well for defining lipoatrophy in women, but not lipohypertrophy.

Keywords

HIV antiretroviral therapy lipoatrophy lipohypertrophy lipodystrophy

Introduction

Long-term use of antiretroviral therapy (ART) is associated with a number of metabolic complications, including fat redistribution or lipodystrophy. Lipodystrophy is an all-encompassing term used to refer to lipoatrophy (loss of subcutaneous fat, particularly in the face and limbs), lipohypertrophy (increase in fat, particularly around the abdomen and breast), or a combination of both.¹ As both lipoatrophy and lipohypertrophy are associated with an increased risk of cardiovascular disease,^2–5 it is important that simple, objective measures to diagnose lipodystrophy be developed for HIV care and treatment.

Lipodystrophy has been diagnosed using both subjective (self-report and examination by a clinician)^6–8 and objective methods (computed tomography, magnetic resonance imaging, and dual-energy X-ray absorptiometry [DXA]).^9–12 Diagnosing lipodystrophy using costly objective methods is problematic, especially in resource-limited settings¹³ where self-report using standardised questionnaires and clinical examination are the most commonly used assessment methods. While some studies found a high level of agreement between self-report and clinical examination,^14,15 others reported poor agreement.^16,17 A cross-sectional study comparing patient and physician report to DXA-measured limb fat showed reasonable levels of agreement between lipoatrophy scores as measured by questionnaire and DXA-measured limb fat.¹⁸ Studies on the diagnosis of lipodystrophy have mostly been done in men from high-income countries. There are minimal data on diagnosing lipodystrophy in sub-Saharan Africa, which bears the brunt of the HIV epidemic and where more women are infected than men. There is a high prevalence of obesity in black South African women and they have less visceral fat than white South African women.¹⁹ We previously developed anthropometric cut-points to diagnose lipodystrophy in a cross-sectional study, but the reference standard was patient report.¹²

Therefore, the aim of our study was, first, to develop objective measures to define lipoatrophy and lipohypertrophy by comparing change in limb fat and trunk fat as measured by DXA to anthropometric variables; and second, to assess agreement between patient-reported lipoatrophy and lipohypertrophy with objective measures derived by DXA in a sample of black South Africans starting ART.

Methods

Participants

A convenience sample of ART-naïve HIV-infected men and women presenting at Crossroads Community Health Centre in Cape Town were enrolled in a 24-month longitudinal study. Patients were initiated on ART after enrolment into the study and completion of baseline evaluations. During the course of the study there was a change in the nucleoside reverse transcriptase inhibitors used in first-line ART regimens in South Africa, from stavudine or zidovudine to tenofovir, together with lamivudine, and efavirenz or nevirapine. The second-line ART regimen consisted of zidovudine, lamivudine, and lopinavir/ritonavir.²⁰

Ethics approval

The study proposal was submitted and approved by the Research Ethics Committee of the Faculty of Health Sciences at the University of Cape Town. Written informed consent was obtained from all participants prior to participation in the study, at baseline and again at follow-up.

Testing procedures

Socio-demographic information was collected from participants at the beginning of the study using an interviewer-administered questionnaire. The Lipodystrophy Case Definition questionnaire⁶ was used to collect self-reported information on fat gain or fat loss. Self-reported lipoatrophy was defined as moderate or severe fat loss in two or more regions, and self-reported lipohypertrophy was defined as moderate or severe fat gain in two or more areas.²¹

Anthropometry and DXA were performed at baseline, three, six, 12, 18, and 24 months. All anthropometric measurements were performed by the same anthropometrist. Participants wore light clothing without shoes. Weight was measured with an electronic load scale in kilograms, to the nearest two decimals. Height was measured with a portable stadiometer and recorded to the nearest millimetre. Body circumferences (waist, hip, mid-upper arm, and mid-thigh) were measured with a non-elastic tape measure and recorded to the nearest millimetre. Sagittal abdominal diameter (SAD) was recorded at the umbilical level as the height of the abdomen measured from the examination table with the participant lying down with straightened legs. Skinfold callipers were used to measure skinfold thicknesses (biceps, triceps, subscapular, abdomen, suprailiac, thigh, and calf), on the right side of the participant, while they were seated. Anthropometric variables were measured three times and the mean used. DXA (Discovery-W®, software version 12.7.3.7; Hologic, Bedford, MA) was used to measure fat mass according to standard procedures by an independent observer. The in vivo coefficient of variation was 1.67% for fat mass. DXA cut-off lines positioned at anatomical markers were used to obtain fat mass for the whole body as well as for the various regions of interest (arms, legs, and trunk). The trunk comprised the region between the neck and waist, excluding the arms. Limb composition was determined by summing the arms and legs. Clinical records obtained from health facilities were reviewed to obtain information on ART regimens, viral loads, and CD4 cell count.

Data analysis

Data analysis was carried out using the STATA/SE statistical software package version 14.1 (StataCorp., College Station, TX, USA). A sample size calculation based on a population of 500, 5% margin for error, and 95% confidence interval indicated that a sample size of 184 participants was needed. Continuous variables were described as medians and interquartile ranges and were compared using Wilcoxon Rank Sum tests. Binary variables were described using frequency and percentages and compared using Chi square tests. The Jonckheere–Terpstra test for ordered variables was used to measure trends over time.

Two DXA-defined definitions of lipoatrophy were explored for the development of anthropometric measures: (1) ≥20% loss of limb fat from baseline²² and (2) ≥10% loss of limb fat (all limbs combined) from baseline by DXA scan.²³ Lipohypertrophy was defined as ≥20% gain in trunk fat from baseline by DXA scan.^24,25 Receiver operating characteristic (ROC) curves were used to describe the performance of a number of anthropometric variables in men and women using DXA-defined lipoatrophy and lipohypertrophy as the reference standard at 12, 18, and 24 months on ART. Anthropometric variable selection was based on associations with lipoatrophy or lipohypertrophy. The area under the ROC curve (AUC) was used to assess the diagnostic performance of each variable. In addition, sensitivity, specificity, likelihood ratios, and predictive values were calculated for variables with ROC AUCs of ≥0.70 at the optimum cut-points. Cut-point selection was based on maximising both sensitivity and specificity in order to correctly classify the greatest proportion of participants.

Results

ART-naïve participants (n = 187; 29.4% men and 70.6% women) were enrolled into this study (Figure 1). There was no difference in baseline characteristics between those lost to follow-up and those that remained in the study. Baseline characteristics are shown in Table 1. Men were significantly older than women (35 versus 31 years; p = 0.008), and more men (71%) than women (56%) were initiated on a stavudine-based regimen. The proportion of participants on stavudine and tenofovir did not differ between commencement of ART and 12 months later. However, after 24 months on ART, the proportion of participants on a stavudine-based regimen had decreased and those on a tenofovir-based regimen had increased. More women were initiated onto nevirapine (69.8%) than efavirenz (30.2%). Only one participant was exposed to a drug regimen containing a protease inhibitor. Viral suppression (viral load <50 copies/ml) was achieved in 73% of participants after six months of ART.

Figure 1.

Consort diagram showing the number of participants lost to follow-up at each time point. ART: antiretroviral therapy.

Table 1.

Characteristics of participants at commencement of ART.

	Total Median (IQR)n = 187	Women Median (IQR)n = 132	Men Median (IQR)n = 55	P-value
Age	33.0 (27.0–39.0)	31.0 (26.0–38.0)	35.0 (29.0–41.0)	0.008
CD4 cell count (cells/μl)	156 (110–196)	161 (110–204)	137 (112–183)	0.126
	n (%)	n (%)	n (%)
Smoker	60 (32.3)	24 (18.3)	36 (65.5)	<0.001
Antiretroviral drugs
NRTI
Stavudine	110 (60.8)	71 (56.4)	39 (70.9)	0.065
Zidovudine	20 (11.1)	19 (15.1)	1 (1.8)	0.009
Tenofovir	51 (28.2)	36 (28.6)	15 (27.3)	0.858
NNRTI
Efavirenz	66 (36.5)	38 (30.2)	28 (50.9)	0.008
Nevirapine	115 (63.5)	88 (69.8)	27 (49.1)	0.008

ART: antiretroviral therapy; IQR: interquartile range; NNRTI: non-nucleoside reverse transcriptase inhibitor; NRTI: nucleoside reverse transcriptase inhibitor.The bold values identify p-values≥0.05.

Change in anthropometric measures over the 24-month period is shown in Table 2 for women, Table 3 for men. In women, all anthropometric measures, except waist/hip ratio, showed a significantly increasing trend over time. Only weight and BMI showed significant increases at all time points up until 18 months on ART when paired t-tests were used to describe the change. In men, weight, BMI, mid-upper arm circumference, mid-thigh circumference, and skinfold thicknesses showed a significantly increasing trend over time. None of the anthropometric measures showed significant increases at all time points when paired t-tests were used to describe the change. Weight gain in women was two-fold higher than in men (9.1 kg versus 4.2 kg).

Table 2.

Comparison of anthropometric measures in women at baseline, three, six, 12, 18, and 24 months.

	BaselineMedian (IQR)n = 132	Three months Median (IQR)n = 103	P-value^a (102 pairs)	Six months Median (IQR)n = 102	P-value^b (97 pairs)	12 months Median (IQR) n = 88	P-value^c (84 pairs)	18 months Median (IQR) n = 77	P-value^d (70 pairs)	24 months Median (IQR) n = 66	P-value^e (58 pairs)	P-value^f (increasing trend)
Height (m)	1.58 (1.55–1.62)	1.57 (1.54–1.62)	0.419	1.58 (1.55–1.62)	0.856	1.58 (1.55–1.62)	0.411	1.57 (1.54–1.62)	0.542	1.58 (1.55–1.63)	0.081	0.340
Weight (kg)	62.0 (54.8–71.3)	65.8 (57.5–73.6)	<0.001	68.1 (58.2–77.0)	0.006	69.2 (59.3–82.8)	0.003	69.8 (59.8–81.8)	0.036	71.1 (57.6–84.4)	0.433	<0.001
BMI	24.8 (21.1–28.7)	26.3 (22.4–30.0)	<0.001	27.1 (22.8–30.9)	0.019	27.8 (23.9–33.4)	<0.001	28.3 (22.4–31.8)	0.035	27.7 (22.7–33.9)	0.411	<0.001
Sagittal abdominal diameter (cm)	18.5 (16.5–21.0)	19.0 (16.5–21.5)	0.742	20.3 (16.5–22.3)	0.237	20.0 (17.6–22.6)	0.126	21.0 (18.0–23.8)	0.038	21.0 (18.0–24.8)	0.760	<0.001
Circumferences
Waist (cm)	80.8 (74.5–90.5)	84.5 (76.0–93.0)	0.002	84.0 (76.0–93.5)	0.102	86.9 (79.0–96.8)	0.015	85.5 (79.0–97.0)	0.097	89.5 (76.0–100.5)	0.029	<0.001
Hip (cm)	98.0 (91.8–104.0)	100.5 (93.0–108.0)	<0.001	104.0 (94.0–110.0)	<0.001	105.0 (94.0–113.5)	0.011	104.5 (94.0–113.0)	0.088	105.0 (93.0–112.5)	0.933	<0.001
Waist/hip ratio	0.84 (0.79–0.92)	0.84 (0.80–0.89)	0.358	0.83 (0.77–0.88)	0.103	0.83 (0.80–0.89)	0.501	0.84 (0.79–0.88)	0.571	0.86 (0.81–0.91)	0.022	0.143
Mid-upper arm (cm)	27.3 (24.0–27.0)	28.5 (25.5–31.5)	<0.001	29.5 (26.0–32.5)	0.001	29.4 (26.0–33.0)	0.280	29.5 (25.8–33.0)	0.086	29.5 (25.5–33.5)	0.533	<0.001
Mid-thigh (cm)	55.0 (50.0–60.5)	56.0 (51.0–61.0)	0.021	59.0 (52.0–64.0)	0.009	59.0 (53.0–64.25)	0.084	59.0 (52.0–64.0)	0.486	60.0 (52.0–65.0)	0.898	<0.001
Skinfold thicknesses
Biceps (mm)	7.0 (4.8–10.1)	7.4 (5.4–11.5)	0.133	8.4 (6.0–12.4)	0.013	8.7 (5.75–13.2)	0.574	8.6 (6.0–14.0)	0.073	9.0 (5.0–14.4)	0.279	<0.001
Triceps (mm)	17.2 (11.9–22.8)	19.2 (12.9–24.8)	<0.001	21.4 (15.6–25.8)	<0.001	22 (14.2–27.8)	0.684	21.2 (13.4–26.6)	0.449	21.1 (13.2–29.2)	0.831	<0.001
Abdomen (mm)	19.4 (13.9–31.4)	22.0 (15.8–32.0)	<0.001	22.9 (17.0–34.0)	0.169	26.2 (18.65–38)	0.003	27.2 (19.0–37.0)	0.041	29.1 (19.1–39.8)	0.179	<0.001
Thigh (mm)	27.6 (20.4–36.0)	32.2 (23.2–39.0)	<0.001	33.3 (23.2–41.4)	0.153	35.0 (24.1–44.2)	0.008	34.1 (25.0–42.3)	0.263	32.3 (22.0–41.2)	0.158	<0.001
Sub-scapular (mm)	15.3 (10.0–24.4)	17.2 (11.3–27.3)	0.203	19.4 (11.9–29.2)	0.002	22.2 (12.6–32.0)	0.054	20.0 (13.0–30.3)	0.053	23.8 (13.6–31.0)	0.069	<0.001
Supra-iliac (mm)	11.1 (7.4–21.4)	13.4 (8.6–23.3)	0.007	15.5 (9.0–27.0)	<0.002	18.5 (10.1–31.1)	<0.001	17.6 (11.8–28.2)	0.799	21.2 (9.8–32.1)	0.203	<0.001
Calf (mm)	17.3 (13.4–21.4)	18.2 (14.6–23)	<0.001	20.4 (15.0–25.5)	0.002	20.9 (16.2–27.3)	0.016	20.0 (15.0–26.2)	0.351	20.0 (14.4–26.2)	0.101	<0.001

BMI: body mass index; IQR: interquartile range.

^aPaired t-test examining change from baseline to three months.

^bPaired t-test examining change from three to six months.

^cPaired t-test examining change from six to 12 months.

^dPaired t-test examining change from 12 to 18 months.

^ePaired t-test examining change from 18 to 24 months.

^fJonckheere–Terpstra test for ordered variables.The bold values identify p-values≥0.05.

Table 3.

Comparison of anthropometric measures in men at baseline, three, six, 12, 18, and 24 months.

	Baseline Median (IQR) n = 55	Three months Median (IQR) n = 43	P-value^a (42 pairs)	Six months Median (IQR) n = 45	P-value^b (42 pairs)	12 months Median (IQR) n = 35	P-value^c (35 pairs)	18 months Median (IQR) n = 31	P-value^d (28 pairs)	24 months Median (IQR) n = 26	P-value^e (25 pairs)	P-value^f (increasing trend)
Height (m)	1.7 (1.7–1.8)	1.7 (1.7–1.7)	0.351	1.7 (1.7–1.8)	0.324	1.7 (1.7–1.8)	0.909	1.7 (1.7–1.7)	0.174	1.7 (1.7–1.7)	0.517	0.475
Weight (kg)	61.0 (55.0–67.0)	62.1 (55.8–68.2)	0.067	63 (56.2–69.0)	0.621	65.0 (60.6–72.0)	0.018	65.0 (60.0–74.0)	0.716	65.2 (59.0–70.5)	0.067	0.009
BMI	21.3 (19.7–23.1)	21.2 (19.7–23.6)	0.034	21.4 (20.0–23.3)	0.753	22.0 (21.0–24.5)	0.019	21.9 (21.2–24.6)	0.421	21.8 (21.3–23.9)	0.147	0.003
Sagittal abdominal diameter (cm)	17.0 (15.8–18.5)	17.0 (15.5–18.0)	0.631	16.8 (15.0–18.0)	0.121	17.5 (16.3–20.0)	0.086	17.5 (16.5–19.5)	0.452	17.6 (16.3–19.0)	1.0	0.017
Circumferences
Waist (cm)	78.3 (74.0–83.0)	78.8 (76.0–83.0)	0.911	78.0 (75.0–83.5)	0.703	80.5 (75.0–86.5)	0.004	80.0 (76.0–86.0)	0.305	79.6 (75.0–83.5)	0.073	0.092
Hip (cm)	88.4 (85.0–92.0)	88.0 (84.0–92.5)	0.366	90.0 (85.0–93.0)	0.442	90.0 (88.0–95.0)	0.481	91.0 (87.0–94.0)	0.726	89.0 (87.0–93.0)	0.323	0.071
Waist/hip ratio	0.89 (0.86–0.92)	0.89 (0.86–0.92)	0.400	0.87 (0.85–0.92)	0.324	0.89 (0.85–0.93)	0.022	0.89 (0.85–0.93)	0.510	0.89 (0.84–0.93)	0.244	0.593
Mid-upper arm (cm)	25.0 (24.0–27.0)	25.0 (24.0–27.0)	0.049	25.5 (24.0–27.0)	0.204	26.5 (25.0–28.5)	0.040	26.5 (25.0–28.0)	0.124	25.3 (24.5–27.0)	0.086	0.011
Mid-thigh (cm)	48.0 (45.0–52.0)	48.0 (45.5–51.5)	0.250	48.5 (45.5–52.0)	0.618	50.0 (48.0–52.5)	0.115	50.0 (48.0–54.5)	0.302	48.0 (46.0–51.0)	0.012	0.017
Skinfold thicknesses
Biceps (mm)	3.2 (2.8–4.0)	3.4 (2.8–3.7)	0.422	3.3 (3.0–3.8)	0.675	3.4 (3.0–4.3)	0.365	3.4 (3.0–4.3)	0.167	3.7 (3.0–4.6)	0.880	0.030
Triceps (mm)	5.8 (4.9–8.6)	6.3 (4.8–8.2)	0.394	6.0 (5.0–8.8)	0.665	7.0 (5.1–10.9)	0.062	7.0 (5.1–10.9)	0.478	6.8 (5.7–8.9)	0.579	0.035
Abdomen (mm)	10.9 (7.8–15.6)	10.4 (8.6–14.4)	0.566	10.4 (7.7–17.8)	0.587	13.0 (8.4–20.4)	0.474	13.0 (8.4–20.4)	0.828	13.3 (9.3–18.5)	0.796	0.033
Thigh (mm)	7.7 (5.7–12.0)	8.0 (6.0–13.5)	0.191	8.6 (6.0–12.4)	0.699	8.0 (5.9–12.4)	0.365	8.0 (5.9–12.4)	0.840	9.0 (6.8–10.0)	0.519	0.141
Sub-scapular (mm)	9.0 (6.5–9.7)	7.1 (6.4–9.3)	0.977	7.5 (6.6–10.2)	0.309	8.6 (6.6–11.4)	0.234	8.6 (6.6–11.4)	0.672	9.1 (6.8–11.9)	0.840	0.011
Supra-iliac (mm)	5.8 (4.8–7.6)	6.0 (4.8–7.5)	0.745	6.0 (4.8–8.2)	0.437	6.4 (5.0–10.4)	0.063	6.4 (5.0–10.4)	0.842	6.7 (4.6–9.0)	0.714	0.084
Calf (mm)	5.8 (4.6–8.2)	6.2 (5.0–8.4)	0.307	5.8 (5.0–9.1)	0.077	6.8 (4.4–8.3)	0.254	6.8 (4.4–8.3)	0.567	6.5 (5.2–8.2)	0.180	0.043

BMI: body mass index; IQR: interquartile range.

^aPaired t-test examining change from baseline to three months.

^bPaired t-test examining change from three to six months.

^cPaired t-test examining change from six to 12 months.

^dPaired t-test examining change from 12 to 18 months.

^ePaired t-test examining change from 18 to 24 months.

^fJonckheere–Terpstra test for ordered variables.The bold values identify p-values≥0.05.

Cumulative measures for lipoatrophy and lipohypertrophy in men and women are shown in Table 4. Few self-reported cases of lipoatrophy or lipohypertrophy were recorded. The incidence of lipoatrophy, defined by limb fat loss as measured by DXA, increased in both men and women over the 24-month period. A greater proportion of men (9 of 42; 21.4%) than women (11 of 114; 9.6%) developed lipoatrophy by 24 months on ART when defined as ≥20% limb fat loss. The proportion of participants who developed lipohypertrophy (defined as ≥20% trunk fat gain) increased in both men and women by 24 months on ART. A similar proportion of men (25 of 42; 60%) and women (65 of 114; 57%) developed lipohypertrophy.

Table 4.

Cumulative number of participants with lipodystrophy assessed by patient report.

	Three months	Six months	12 months	18 months	24 months
Women
Number at risk	114	101	94	83	75
Lost to observation	14	10	7	10	15
Lipoatrophy
Patient report	1	1	1	1	1
DXA ≥20% limb fat loss	1	1	6	8	11
DXA ≥10% limb fat loss	6	7	15	19	26
Lipohypertrophy
Patient report	6	15	16	17	17
DXA ≥20% trunk fat gain	17	31	53	63	65
Men
Number at risk	42	44	37	30	28
Lost to observation	6	1	4	8	4
Lipoatrophy
Patient report	1	1	1	1	1
DXA ≥20% limb fat loss	2	3	5	8	9
DXA ≥10% limb fat loss	8	13	15	17	19
Lipohypertrophy
Patient report	1	3	3	3	3
DXA ≥20% trunk fat gain	7	13	21	24	25

DXA: dual-energy X-ray absorptiometry.

The highest ROC AUC of anthropometric measures for lipoatrophy was observed in women at 18 months on ART and in men at 24 months on ART (Table 5). Anthropometric measures for lipohypertrophy performed poorly in women with no ROC AUCs ≥0.7 (Table 6). In men, for lipohypertrophy at 18 months, only SAD generated a ROC AUC ≥0.7 (AUC = 0.732). Optimum cut-points for lipoatrophy variables with ROC AUCs of ≥0.7, based on sensitivity, specificity, and % correctly classified, were selected for women (Table 7). The thigh skinfold cut-point (≤24 mm) was able to correctly classify the greatest number of women (80.3%), both with lipoatrophy (sensitivity = 62.5) and those without lipoatrophy (specificity = 82.35).

Table 5.

ROC AUC and 95% CI for anthropometric measures used to predict lipoatrophy defined as ≥20% limb fat loss in women and men at 12, 18, and 24 months on ART. ROC AUC values ≥0.7, which was our criterion for determining anthropometric cut-points, are in bold.

	12 months	18 months	24 months
	ROC AUC (95%CI)	ROC AUC (95%CI)	ROC AUC (95%CI)
Women	n=87	n=76	n=65
Waist/hip ratio	0.639 (0.534–0.744)	0.568 (0.447–0.679)	0.606 (0.471–0.720)
Mid-thigh circumference	0.694 (0.581–0.785)	0.843 (0.740–0.916)	0.700 (0.566–0.801)
Hip circumference	0.748 (0.643–0.834)	0.859 (0.756–0.925)	0.748 (0.631–0.852)
Mid-arm circumference	0.695 (0.594–0.795)	0.740 (0.623–0.831)	0.616 (0.486–0.733)
Biceps skinfold	0.711 (0.606–0.805)	0.605 (0.487–0.716)	0.625 (0.502–0.747)
Triceps skinfold	0.714 (0.606–0.805)	0.764 (0.652–0.853)	0.660 (0.534–0.774)
Calf skinfold	0.677 (0.569–0.774)	0.791 (0.681–0.875)	0.680 (0.559–0.798)
Thigh skinfold	0.725 (0.618–0.815)	0.814 (0.710–0.895)	0.730 (0.598–0.827)
Men	n=33	n=28	n=23
Waist/hip ratio	0.560 (0.364–0.719)	0.614 (0.406–0.785)	0.667 (0.427–0.836)
Mid-thigh circumference	0.565 (0.392–0.745)	0.621 (0.406–0.785)	0.672 (0.427–0.836)
Hip circumference	0.637 (0.451–0.796)	0.504 (0.306–0.694)	0.650 (0.427–0.836)
Mid-arm circumference	0.528 (0.335–0.692)	0.602 (0.406–0.785)	0.706 (0.471–0.868)
Biceps skinfold	0.607 (0.421–0.771)	0.546 (0.339–0.725)	0.667 (0.427–0.836)
Triceps skinfold	0.542 (0.364–0.719)	0.617 (0.406–0.785)	0.500 (0.306–0.732)
Calf skinfold	0.597 (0.421–0.771)	0.519 (0.339–0.725)	0.506 (0.306–0.732)
Thigh skinfold	0.583 (0.392–0.745)	0.553 (0.339–0.725)	0.511 (0.306–0.732)

AUC: area under the ROC curve; CI: confidence interval; ROC: receiver operating characteristic.

Table 6.

ROC AUC and 95% CI of anthropometric measures used to predict lipohypertrophy defined as ≥20% trunk fat gain in women and men at 12, 18, and 24 months on ART. ROC AUC values ≥0.7, which was our criterion for determining anthropometric cut-points, are in bold.

	12 months	18 months	24 months
	ROC AUC (95%CI)	ROC AUC (95%CI)	ROC AUC (95%CI)
Women	n=87	n=76	n=65
Sagittal abdominal diameter	0.518 (0.408–0.626)	0.559 (0.441–0.675)	0.540 (0.417–0.672)
Waist circumference	0.507 (0.396–0.615)	0.580 (0.460–0.691)	0.550 (0.425–0.677)
Hip circumference	0.543 (0.430–0.648)	0.594 (0.473–0.704)	0.592 (0.456–0.706)
Waist/hip ratio	0.520 (0.408–0.626)	0.521 (0.408–0.642)	0.531 (0.410–0.663)
Abdominal skinfold	0.519 (0.408–0.626)	0.577 (0.460–0.691)	0.502 (0.381–0.634)
Sub-scapular skinfold	0.540 (0.430–0.648)	0.530 (0.408–0.642)	0.527 (0.395–0.649)
Supra-iliac skinfold	0.526 (0.419–0.637)	0.606 (0.487–0.716)	0.510 (0.381–0.634)
Men	n=33	n=28	n=23
Sagittal abdominal diameter	0.664 (0.482–0.820)	0.732 (0.537–0.889)	0.659 (0.427–0.836)
Waist circumference	0.567 (0.392–0.745)	0.661 (0.476–0.841)	0.579 (0.345–0.768)
Hip circumference	0.584 (0.392–0.745)	0.631 (0.441–0.814)	0.702 (0.471–0.868)
Waist/hip ratio	0.592 (0.421–0.771)	0.655 (0.441–0.814)	0.501 (0.306–0.732)
Abdominal skinfold	0.546 (0.364–0.719)	0.608 (0.406–0.785)	0.643 (0.427–0.836)
Sub-scapular skinfold	0.523 (0.335–0.692)	0.576 (0.372–0.755)	0.532 (0.306–0.732)
Supra-iliac skinfold	0.564 (0.392–0.745)	0.544 (0.339–0.725)	0.575 (0.345–0.768)

ART: antiretroviral therapy; AUC: area under the ROC curve; CI: confidence interval; ROC: receiver operating characteristic.

Table 7.

Anthropometric cut-points to identify lipoatrophy in women at 18 months on ART and men at 24 months on ART. Anthropometric measures were selected for cut-point determination if their ROC AUC was ≥0.7.

Lipoatrophy variable	ROC AUC	Cut-point	Sensitivity (%)	Specificity (%)	Likelihood ratio positive test	Likelihood ratio negative test	Positive predictive value (%)	Negative predictive value (%)	Per cent correctly classified
Women (n=76)
≥20% limb fat loss
Hip circumference	0.859	≤96 cm	75.00	73.53	2.83	0.34	25.00	73.53	73.68
Mid-thigh circumference	0.843	≤55 cm	87.50	70.59	2.98	0.18	25.93	70.59	72.37
Mid-arm circumference	0.740	≤25 cm	23.53	93.22	3.47	0.82	50.00	93.22	77.63
Thigh skinfold	0.814	≤24 mm	62.50	82.35	3.54	0.46	29.41	82.35	80.26
Triceps skinfold	0.764	≤16 mm	25.00	96.15	6.50	0.78	75.00	96.15	73.68
Calf skinfold	0.791	≤16 mm	21.74	94.34	3.84	0.83	62.50	94.34	72.37

ART: antiretroviral therapy; AUC: area under the ROC curve; ROC: receiver operating characteristic.

Discussion

This is the first study to develop simple diagnostic measures for lipodystrophy using longitudinal changes in DXA-derived measures of body fat composition as the reference standard. We found that simple anthropometric measures performed well for defining lipoatrophy in women and lipohypertrophy in men. The best predictors of lipoatrophy were hip and mid-thigh circumference, and thigh and calf skinfold thickness. The proportion of men and women who developed DXA-defined lipoatrophy increased over the 24-month period, but it occurred more frequently in men, possibly due to their lower baseline BMI as well as having less body fat, especially in the limbs. The proportion of men and women with DXA-defined lipohypertrophy increased until 18 months on ART, before decreasing. The best predictor of lipohypertrophy in men was SAD and no anthropometric measure was useful for predicting lipohypertrophy in women. We found very poor agreement between DXA-defined lipodystrophy and patient report, which grossly underestimated lipodystrophy.

Patient reports, using standardised questionnaires,^6,21 are commonly used to assess lipodystrophy in low- and middle-income countries.^26–28 Results from studies comparing patient report to clinical examination have been contradictory.^14–17 One study reported a high level of agreement between patient- and physician-reported lipoatrophy and DXA-measured limb fat.¹⁸ We, on the other hand, found very poor agreement between self-report and DXA-defined lipodystrophy, with the proportion of participants with lipoatrophy and lipohypertrophy by self-report being far less than when defined by change in DXA measures in both men and women. These findings highlight the subjective nature of self-report and suggest that the incidence of lipodystrophy in African studies may have been underestimated when the diagnosis was based solely on self-report.

In low- and middle-income countries zidovudine continues to be used and stavudine was widely used until recently.²⁹ Lipoatrophy remains a common antiretroviral adverse drug reaction as both these drugs, especially stavudine, are associated with the development of lipoatrophy.^27,30–33 Accurate identification of lipoatrophy is important as it is independently associated with increased vascular risk.^2–5 Objective measures based on DXA-derived variables have been developed to define lipodystrophy in France,³⁴ Portugal,³⁵ India,^35,36 and Brazil.³⁷ These cross-sectional studies, which consisted mainly of men, proposed that fat mass ratio, defined as the ratio of the percentage of the trunk fat mass to the percentage of the lower limb fat mass measured by DXA, be used for the early diagnosis of lipodystrophy.^34–36 These measures may not be generalisable to sub-Saharan Africa, where black South African women have less visceral adipose tissue than white South African women, despite being more insulin resistant and showing a different metabolic phenotype in sub-Saharan Africa.¹⁹ Using an earlier cross-sectional study consisting of 550 participants on ART,¹² our group developed objective measures for defining lipoatrophy and lipohypertrophy based on anthropometry and DXA-derived variables. We found the anthropometric variables triceps and thigh skinfold thickness to be the best predictors of lipoatrophy in women. However, all measures used patient report as the reference standard and may therefore be subject to the bias evident in self-report.

To our knowledge no prior studies have used change from baseline in DXA-derived variables as the reference standard to define lipoatrophy and lipohypertrophy in order to develop cut-points for anthropometric measures. We showed that several anthropometric measures could be used to diagnose lipoatrophy in women (hip, mid-thigh, and mid-arm circumferences; and thigh and calf skinfold), but none were good enough to diagnose lipoatrophy in men, possibly due to the small sample of men and the relatively large loss to follow-up. Given that skinfold measurement requires equipment and training to ensure accuracy, we propose the use of either hip, mid-thigh, or mid-arm circumferences for defining lipoatrophy in women in African studies. While the percentage of women correctly classified were similar across the three measures (72.4–77.6%), mid-arm circumference has high specificity and is better at identifying women without lipoatrophy, while mid-thigh circumference had high sensitivity and is better at identifying women with lipoatrophy. If resources allow for only one measure of lipoatrophy, then hip circumference is recommended. SAD was the only anthropometric measure that defined lipohypertrophy in men.

We showed that in both men and women, weight, waist circumference, SAD (a predictor of visceral fat),³⁸ and abdominal skinfold thickness increased over time. After 18 months on ART more than half the women and almost half the men had developed lipohypertrophy as defined by ≥20% increase in trunk fat. Even though lipohypertrophy does not appear to be an antiretroviral adverse drug reaction, and is instead thought to be a consequence of treating the HIV infection,³³ detecting increased trunk fat is clinically relevant as it has been shown to be associated with an increased risk of cardiovascular disease^2–5 and death in HIV-infected populations.³⁹

There are some limitations to our study. There were no ART naïve or HIV-uninfected control participants. The study sample was relatively small, with few participants being exposed to protease inhibitors, so only inferences about first-line ART can be made. Finally, rates of loss to follow up were high. Despite these limitations, ours is the only study to use longitudinal changes from baseline in fat distribution measured by DXA to develop cut-points of simple anthropometric measures for diagnosing lipoatrophy and lipohypertrophy.

Conclusion

We developed anthropometric cut-points for defining lipodystrophy in South African HIV-infected people on ART. These measures could be used in sub-Saharan Africa to identify lipoatrophy in women and lipohypertrophy in men. In resource-limited settings such as South Africa, where health professionals are overburdened and need simple, inexpensive, and quick methods for diagnosing patients, these measures are of particular relevant. Further research in other African countries is needed to validate these cut-points.

Footnotes

Acknowledgements

We thank Linda Bewerunge for doing the DXA scans, Sasha West for anthropometric measurements, and Carmen Delport for co-ordinating the study.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: GM was supported in part by the National Research Foundation (NRF) of South Africa. The Grant holder acknowledges that opinions, findings, and conclusions or recommendations expressed in any publication generated by the NRF supported research are that of the author(s) and that the NRF accepts no liability whatsoever in this regard.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the NRF of South Africa (grant number 85810), the World Diabetes Foundation, and the South African Department of Health.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

References

Grinspoon

Carr

Cardiovascular risk and body-fat abnormalities in HIV-infected adults. N Engl J Med 2005; 352: 48–62.

Freitas

Carvalho

Souto

et al . Impact of lipodystrophy on the prevalence and components of metabolic syndrome in HIV-infected patients. BMC Infect Dis 2011; 11: 246.

Masia

Padilla

Garcia

et al . Endothelial function is impaired in HIV-infected patients with lipodystrophy. Antivir Ther 2010; 15: 101–110.

Lake

Wohl

Scherzer

et al . Regional fat deposition and cardiovascular risk in HIV infection: the FRAM study. AIDS Care 2011; 23: 929–938.

Finkelstein

Gala

Rochford

et al . HIV/AIDS and lipodystrophy: implications for clinical management in resource-limited settings. J Int AIDS Soc 2015; 18: 19033.

Carr

HIV Lipodystrophy Case Definition Study Group. An objective case definition of lipodystrophy in HIV-infected adults: a case-control study. Lancet 2003; 361: 726–735.

Casado

Iglesias

del Palacio

et al . Social isolation in HIV-infected patients according to subjective patient assessment and DEXA-confirmed severity of lipodystrophy. AIDS Care 2013; 25: 1599–1603.

Saves

Francois

Jacqueline

et al . Factors related to lipodystrophy and metabolic alterations in patients with human immunodeficiency virus infection receiving highly active antiretroviral therapy. Clin Infect Dis 2002; 34: 1396–1405.

McComsey

Kitch

Sax

et al . Peripheral and central fat changes in subjects randomized to abacavir-lamivudine or tenofovir-emtricitabine with atazanavir-ritonavir or efavirenz: ACTG Study A5224s. Clin Infect Dis 2011; 53: 185–196.

10.

Bacchetti

Heymsfield

Lewis

et al . Fat distribution in women with HIV infection. J Acquir Immune Defic Syndr 2006; 42: 562–571.

11.

Cavalcanti

Cheung

Raboud

et al . Reproducibility of DXA estimations of body fat in HIV lipodystrophy: implications for clinical research. J Clin Densitom 2005; 8: 293–297.

12.

Abrahams

Dave

Maartens

et al . The development of simple anthropometric measures to diagnose antiretroviral therapy-associated lipodystrophy in resource limited settings. AIDS Res Ther 2014; 66: 839–844.

13.

Wouters

Heunis

van Rensburg

et al . Patient satisfaction with antiretroviral services at primary health-care facilities in the Free State, South Africa – a two-year study using four waves of cross-sectional data. BMC Health Serv Res 2008; 8: 1

14.

Huang

Harrity

Lee

et al . Body image in women with HIV: a cross-sectional evaluation. AIDS Res Ther 2006; 3: 1–7.

15.

Guaraldi

Murri

Orlando

et al . Severity of lipodystrophy is associated with decreased health-related quality of life. AIDS Patient Care STDs 2008; 22: 577–585.

16.

Belloso

Quiros

Ivalo

et al . Agreement analysis of variables involved in lipodystrophy syndrome definition in HIV-infected patients. J Acquir Immune Defic Syndr 2003; 32: 104–111.

17.

Mercier

Gueye

Cournil

et al . Lipodystrophy and metabolic disorders in HIV-1-infected adults on 4- to 9-year antiretroviral therapy in Senegal: a case-control study. J Acquir Immune Defic Syndr 2009; 51: 224–230.

18.

Tungsiripat

O’Riordan

Storer

et al . Subjective clinical lipoatrophy assessment correlates with DEXA-measured limb fat. HIV Clin Trials 2009; 10: 314–319.

19.

Goedecke

Levitt

Lambert

et al . Differential effects of abdominal adipose tissue distribution on insulin sensitivity in black and white South African women. Obesity 2009; 17: 1506–1512.

20.

Department of Health. South African antiretroviral treatment guidelines, http://www.sahivsoc.org/upload/documents/2013%20ART%20Guidelines-Short%20Combined%20FINAL%20draft%20guidelines%2014%20March%202013.pdf (accessed 18 October 2013).

21.

Lichtenstein

Ward

Moorman

et al . Clinical assessment of HIV-associated lipodystrophy in an ambulatory population. AIDS 2001; 15: 1389–1398.

22.

Haubrich

Riddler

DiRienzo

et al . Metabolic outcomes in a randomized trial of nucleoside, nonnucleoside and protease inhibitor-sparing regimens for initial HIV treatment. AIDS 2009; 23: 1109–1118.

23.

Dube

Parker

Tebas

et al . Glucose metabolism, lipid, and body fat changes in antiretroviral-naive subjects randomized to nelfinavir or efavirenz plus dual nucleosides. AIDS 2005; 19: 1807–1818.

24.

Kolta

Flandre

Ngo Van

et al . Fat tissue distribution changes in HIV-infected patients treated with lopinavir/ritonavir. Results of the MONARK trial. Curr HIV Res 2011; 9: 31–39.

25.

Valantin

Flandre

Kolta

et al . Fat tissue distribution changes in HIV-infected patients with viral suppression treated with darunavir/ritonavir (DRV/r) monotherapy versus 2 NRTIs DRV/r in the MONOI-ANRS 136 randomized trial: results at 48 weeks. Conference on Retroviruses and Opportunistic Infections, 2010. San Francisco.

26.

Womack

HIV-related lipodystrophy in Africa and Asia. AIDS Read 2009; 19: 131–139.

27.

van Griensven

De Naeyer

Mushi

et al . High prevalence of lipoatrophy among patients on stavudine-containing first-line antiretroviral therapy regimens in Rwanda. Trans R Soc Trop Med Hyg 2007; 101: 793–798.

28.

Mutimura

Stewart

Rheeder

et al . Metabolic function and the prevalence of lipodystrophy in a population of HIV-infected African subjects receiving highly active antiretroviral therapy. J Acquir Immune Defic Syndr 2007; 46: 451–455.

29.

World Health Organisation. Phasing out stavudine: progress and challenges, http://www.who.int/hiv/pub/guidelines/arv2013/arv2013supplement_to_chapter09.pdf (accessed 20 July 2016).

30.

Van Vonderen

MGA

Van Agtmael

Hassink

EAM

et al . Zidovudine/lamivudine for HIV-1 infection contributes to limb fat loss. PLoS One 2009; 4: e5647.

31.

Cameron

Becker

King

et al . Exploratory study comparing the metabolic toxicities of a lopinavir/ritonavir plus saquinavir dual protease inhibitor regimen versus a lopinavir/ritonavir plus zidovudine/lamivudine nucleoside regimen. J Antimicrob Chemother 2007; 59: 957–963.

32.

Menezes

Maskew

Sanne

et al . A longitudinal study of stavudine-associated toxicities in a large cohort of South African HIV infected subjects. BMC Infect Dis 2011; 11: 1.

33.

de Waal

Cohen

Maartens

Systematic review of antiretroviral-associated lipodystrophy: lipoatrophy, but not central fat gain, is an antiretroviral adverse drug reaction. PLoS One 2013; 8: e63623.

34.

Bonnet

Delpierre

Sommet

et al . Total body composition by DXA of 241 HIV-negative men and 162 HIV-infected men: proposal of reference values for defining lipodystrophy. J Clin Densitom 2005; 8: 287–292.

35.

Freitas

Santos

Carvalho

et al . Fat mass ratio: an objective tool to define lipodystrophy in HIV-infected patients under antiretroviral therapy. J Clin Densitom 2010; 13: 197–203.

36.

Asha

Seshadri

Paul

et al . Human immunodeficiency virus-associated lipodystrophy: an objective definition based on dual-energy x-ray absorptiometry-derived regional fat ratios in a South Asian population. Endocr Pract 2012; 18: 158–169.

37.

Beraldo

Vassimon

Aragon

et al . Proposed ratios and cutoffs for the assessment of lipodystrophy in HIV-seropositive individuals. Eur J Clin Nutr 2015; 69: 274.

38.

Zamboni

Turcato

Armellini

et al . Sagittal abdominal diameter as a practical predictor of visceral fat. Int J Obes Relat Metab Disord 1998; 22: 655–660.

39.

Scherzer

Heymsfield

Lee

et al . Decreased limb muscle and increased central adiposity are associated with 5-year all-cause mortality in HIV infection. AIDS 2011; 25: 1405–1414.