Peripheral and central lipoatrophy affecting subcutaneous fat is often associated with HIV infection 91231. Although central fat accumulation was noted in early reports 5716, appreciation of the clinical significance of HIV-associated fat accumulation has come slowly in studies of HIV-associated lipodystrophy.

There may be a number of reasons why fat accumulation has received less attention than peripheral lipoatrophy in the literature. Firstly, there appears to be a perception that fat loss is more common than fat accumulation 932. Secondly, patients may report fat loss more readily to their physicians because of the undesirable cosmetic effects of fat loss from the face, buttocks and extremities 33. Thirdly, the presence of obesity may confound the detection of HIV-associated accumulation of fat 34. Fourthly, accumulation of VAT is not readily amenable to objective measurement in the clinical setting 3536, and sophisticated imaging equipment is needed to visualize and specifically quantify HIV-associated accumulation of VAT 36.

The 1998 Advisory Committee of the International Association of Physicians in AIDS Care (IAPAC) estimated that approximately 1/3 of all treated HIV-infected patients show evidence of intra-abdominal (visceral) fat accumulation along with, or independent of, generalized obesity 8. This observation coincided with the introduction of one of the first protease inhibitor (PI) drugs (indinavir) that may have contributed significantly to lipodystrophy with VAT accumulation 8. In summary, HARS is a subset of HIV-associated lipodystrophy in which there is abnormal accumulation of trunk fat, including VAT, appearing with or without concurrent lipoatrophy of subcutaneous adipose tissue (SAT) in the limbs and face, or both. Between 5% and 10% of HARS patients exhibit dorsocervical fat accumulation 24.

To date, it has not been possible to bring the diverse observations of altered fat distribution and metabolism under the rubric of a single syndrome 32, and case definition remains a work in progress 37. Without a clear definition of what constitutes a "case" of HIV lipodystrophy, the field will continue to be plagued by a wide variation in prevalence estimates, uncertainty about risk factors and indecision about proper course of treatment 38.

Methodologic difficulties have plagued investigations of HIV lipodystrophy. Differing case definitions, reliance on detection methods with various levels of sensitivity and specificity, and enrollment of different patient populations have made it difficult to compare results across studies. A low level of agreement between clinical and patient assessments of fat redistribution has also contributed to the problem of clarifying the case definition 39. Finally, the cross-sectional design of many studies is appropriate for generating hypotheses about the possible association of factors in lipodystrophy, but such studies cannot be used to identify causal relationships.

To resolve these ambiguities, the HIV Lipodystrophy Case Definition Study Group developed a statistical model for the diagnosis of lipodystrophy (including age, sex, duration of HIV infection, HIV disease stage, waist-to-hip ratio [WHR], anion gap, serum high-density lipoprotein-cholesterol [HDL-C] concentration, trunk-to-peripheral fat ratio, percentage leg fat, and intra- and extra-abdominal fat ratio as variables) with a quantitative scale for detecting cases 4041. The model efficiently captured cases of HIV lipodystrophy (70% sensitivity, 80% specificity), but this approach failed to develop a model specifically for peripheral fat loss or central fat gain.

In fact, this case-definition exercise deliberately excluded patients with moderate or severe "abdominal obesity" without lipoatrophy on the assumption that it was "age-related adiposity", and was thus biased against those with central fat accumulation or HARS. Nevertheless, regional fat accumulation was a prominent feature of the model with high scores for fat accumulation on the dorsocervical spine and abdominal region. There was no difference in body mass index between the cases and the comparison subjects (24.1 ± 3.7 vs 23.8 ± 3.8, P = 0.454). However, in cases of lipodystrophy, WHR was greater (0.95 ± 0.09 vs 0.91 ± 0.07, P < 0.001), intra-abdominal fat was greater (138 ± 88 cm² versus 97 ± 71 cm², P < 0.001), and the ratio of intra-to-extra-abdominal fat was dramatically greater (3.21 ± 9.87 vs 0.90 ± 1.16, P < 0.001) compared with controls.

In the Aquitaine Cohort (1999), in which 61% of the 581 patients were treated with PI, the overall incidence of fat maldistribution was 38% 42. Some form of peripheral fat loss was observed in 90% of the patients, but 21% had increased abdominal girth and 10% of the patients had both peripheral fat loss and central fat accumulation. At least one metabolic abnormality was seen in 54% of patients; 51% of men and 45% of women had some form of lipid or glucose abnormality. These results led the authors of the Aquitaine study to suggest that there may be 3 separate presentations of lipodystrophy – fat atrophy, fat accumulation, and a mixed syndrome of both fat atrophy and accumulation 42.

A number of studies have consistently reported a significant proportion of patients with central fat accumulation: the Australian Prevalence Survey of Lipodystrophy Syndrome43^{the HIV Outpatient Study (HOPS) 44, the LIPOCO Study 45, the Norwegian patient survey 46, the Swiss cohort study 47, and the Australia cohort survey 434849. Safrin and Grunfeld (1999) surveyed reports of localized fat accumulation and loss in HIV-infected patients, and offered prevalence estimates of 1% to 56% for fat accumulation, 1% to 24% for regional loss of fat, and 2% to 83% for the pooled lipodystrophy syndromes 10.}

Lichtenstein et al. reviewed 15 published surveys with >100 patients with descriptions suggestive of HARS 50. They determined that prevalence estimates for HARS vary widely, from 9% 51 to 48% 52 (mean 32%), depending on the criteria used for assessment: increased abdominal girth, abdominal lipohypertrophy, abdominal enlargement, central or abdominal fat accumulation, truncal obesity, central fat gain, increased abdominal wall thickness, or pseudo-obesity. From a review of published, large cross-sectional and longitudinal studies, Tien and Grunfeld determined similar estimates 12; for any lipodystrophy prevalence ranged from 30% to 62%, peripheral lipoatrophy from 22% to 38%, and any central fat accumulation from 18% to 45%. The highest estimates in the ranges reported by Tien and Grunfeld, however, all came from one study, relying on physical examination only and not on anthropometric or other objective measurements 53.

The Fat Redistribution and Metabolic Changes in HIV Infection Study (FRAM) was a large cross-sectional analysis of 425 HIV-infected and 152 HIV-negative men, age 33 to 45, in which lipoatrophy or lipohypertrophy were defined as concordance between subject report of the direction of fat change and results of the physical examination 931. The objective of this study was to compare regional fat distribution (determined by self-report, physical examination and magnetic resonance imaging [MRI]) in HIV-positive men and population-based comparison subjects. Whereas previous studies evaluated only the presence of suspected changes, FRAM used bidirectional instruments (increase or decrease, as well as scales) to grade the magnitude of reported change as mild, moderate or severe.

The FRAM study found no correlation between changes in central fat and peripheral fat in HIV-infected men, measured either by self-report or physical examination 931. In fact, the amount of central accumulation of fat was equivalent in HIV-infected men and comparison men without HIV, and peripheral fat loss did not necessarily correlate with central fat loss. The study authors concluded that central fat accumulation and peripheral fat loss are not reciprocal processes; there is probably no "relocation" of fat between these two depots.

The FRAM study results were somewhat controversial, however. The researchers did not control for body mass index and the HIV-negative control patients were significantly heavier than the HIV-infected patients. This may have contributed to the observation that increases in VAT were more common in the control patients 54. Furthermore, FRAM was a cross-sectional study; longitudinal studies, such as ACTG 5005s (a substudy of ACTG 384), have demonstrated an increase in truncal adiposity with prolonged antiretroviral therapy 55. In ACTG 5005s, the prevalence of a high WHR increased from 35% at baseline to 47% after 64 weeks of treatment with antiretrovirals. This moderate relative change underscores the importance of knowing baseline values. However, unlike FRAM, the ACTG studies did not include an HIV-negative control group, so the extent to which the WHR compares with age-matched HIV-negative controls is not clear.

The cross-sectional design of FRAM also does not shed light on how quickly VAT might have accumulated after a patient became infected with HIV or after initiating treatment with HAART. Follow-up data from FRAM are awaited with interest. With further study, perhaps the rate of VAT accumulation may help distinguish HARS from VAT accumulation related to general obesity.

Physical symptoms associated with HARS primarily include bodily discomfort due to local accumulation of a large mass of VAT, with abdominal distension, respiratory difficulty (similar to Pickwickian syndrome), umbilical herniation, gastro-esophageal reflux, difficulty swallowing, abdominal cramping, peptic ulcer, and constipation or diarrhea 33. It remains unknown if or how much the inflammation associated with HIV adds to or enhances the effects of increased VAT.

VAT is an independent risk factor for insulin resistance 60. When Goodpaster et al. (1999) examined the effects of weight loss on regional fat distribution and insulin sensitivity in the general population with obesity, they found that the reduction in VAT was the only adipose tissue parameter that predicted the improvement in insulin sensitivity 61.

An estimated 30% to 90% of patients receiving PI agents are insulin resistant, although the incidence of diabetes mellitus is less than 10% 62. Lipodystrophy (peripheral lipoatrophy and/or lipohypertrophy) is associated with hyperinsulinemia and insulin resistance 136364. The dorsocervical fat pad or "buffalo hump" has also been associated with insulin resistance in an analysis of two separate studies involving 1765 HIV-infected patients 65. In addition, impaired glucose tolerance is part of the spectrum of metabolic changes seen in patients taking antiretroviral therapy 29. The direct effects of PI and nucleoside reverse transcriptase inhibitor agents, and the indirect effects of visceral adiposity and peripheral fat loss, with fatty infiltration of muscle and liver, account in part for the insulin resistance seen in the HIV-infected population 66.

Cohort studies suggest that patients with HIV already face an increased risk of acute myocardial infarction (MI) compared with matched non-HIV-infected controls 596768. In the subset of patients with HARS, the metabolic consequences of fat deposition may further increase cardiovascular risk 596869, although HIV patients tend also to have high rates of other cardiovascular risk factors relative to age-matched HIV-negative individuals, including diabetes, hypertension and dyslipidemia 67. In addition, protease inhibitor use is a strong predictor of MI in HIV-infected patients 70.

Patient-reported health-related quality-of-life issues associated with abnormal fat distribution in HIV-positive patients came to the attention of investigators early 86. Changes in body shape stigmatize patients 87 and may interfere with treatment adherence 88. In addition, patients' anxieties regarding long-term skeletal and cardiovascular risks can interfere with treatment decisions 89. Although it is difficult to isolate the effect of body habitus changes alone on quality of life in HIV patients 90, significant body image distress is associated with adipose tissue maldistribution 30.

Patients with HARS tend to exhibit excess visceral adiposity without increased total body fat, and have more VAT than non-obese HIV-negative people with similar height-adjusted weight 91. Patients exhibiting signs of HARS also may have a higher WHR, trunk-to-limb fat ratio, and VAT/SAT ratio 36 than would be expected from their body mass index 92. HIV-infected patients with truncal enlargement had 2.5–4 times more VAT on MRI scan 4 and 4 times more VAT on single-slice CT scan than healthy control subjects of similar age, sex and height. Engelson and colleagues found that VAT volumes could be quite large in HARS patients, and ranged from 2.1–9.8 L in HIV-positive men with truncal obesity compared with 0.3–1.7 L (P < 0.001) in HIV-positive comparison patients without truncal enlargement 4. Men with truncal enlargement had 6.5 times more VAT than men without truncal enlargement 4.

Body measurement of patients with fat redistribution syndrome is fraught with difficulties 96979899. Studies of HIV lipodystrophy have relied on several different means of detecting and measuring changes in body shape: self-report by means of questionnaires, scales for tabulating clinicians' observations on physical examination, anthropometric formulas, and radiographic techniques. Self-report and questionnaires are open to subjective bias. In general, levels of agreement (tested by kappa statistic) between subjective measures have been poor 39.

Because HARS is marked by depots of VAT rather than SAT, the condition may not always be evident on inspection or physical examination. In patients with mixed lipodystrophy, the decrease in truncal SAT (lipoatrophy) may contribute to a smaller than expected increase in waist circumference or WHR and thus lead to underestimation of the amount of VAT 100. The 4-year Multicenter AIDS Cohort Study (MACS) suggested that, in at least some HAART recipients, an increase in WHR could be attributed to a relatively lower increase in hip circumference rather than an increased rate of change in waist circumference 101. Visual inspection on physical exam in some HARS patients may reveal abdominal distension as VAT pushes the abdominal musculature forward 102. In these, there may be a lack of flabbiness or pinchable subcutaneous fat stretched over a taut abdomen. However, other HARS patients may present with large bellies that are difficult to differentiate from general obesity.

Standard anthropometric tests have the advantage of being readily available to clinicians 103104. Anthropometric criteria, based on waist circumference and/or WHR, have been used to estimate the relative mass of visceral fat 24105106. Some studies have applied a composite criterion of waist circumference >88 cm and WHR ≥ 0.95 for men, plus waist circumference >75 and WHR ≥ 0.90 for women to effectively detect HIV-infected patients with high VAT content 2435107108. These cutoffs for WHR were based on published anthropometric criteria considered to define visceral adiposity and increased cardiovascular risk in adults 106109. As mentioned earlier, WHR data may be misleading, especially if a patient has a normal waist but massive loss of fat and muscle from the hips and buttocks. Thus, in order to exclude patients with apparently high WHR due to such dramatically reduced hip circumference, a minimum waist circumference >88.2 cm was set for males and >75.3 cm for females 24. While these criteria should be subjected to independent validation, it has been suggested that they may be conservative 108.

The most sensitive and specific methods for the detection of VAT are CT scan and MRI because they provide quantifiable data on the location and mass of visceral fat and SAT 110. DXA is sensitive and produces results within acceptable variability for measuring the subcutaneous fat of the appendages (SAT), but is more difficult for estimation of VAT since changes in SAT and VAT independently affect the mass of truncal fat 111. Limitations of these methods include: expense and limited availability; radiation exposure associated with CT scan, limit to single-slice and not whole-body studies; few data exist on normal distribution of VAT and SAT, so it is difficult to make comparisons; subcompartments of SAT and VAT are not homogeneous, so whole-body estimates may be simplistic.

Useful as these radiologic techniques are, CT, MRI and DXA instruments are too expensive to be practical for the identification of HARS patients in routine clinical practice. Routine measurement of waist circumference and/or WHR may be helpful and are recommended for all patients as it enhances diagnostic sensitivity for cardiovascular risk factors 112.