Reports of fat redistribution, insulin resistance, and pro-atherogenic hyperlipidemia attributed to protease inhibitor (PI) therapy in HIV-infected patients have resulted in increasing concern among patients and their physicians. The first of the reported body composition changes involved increases (lipohypertrophy) of fat in the posterior neck (‚buffalo hump‚), neck, chest, abdomen and other localized areas[1-20]. Subsequently, a syndrome of ‚lipodystrophy‚ was reported, which was predominantly loss (lipoatrophy) of fat in the face, arms or legs, often accompanied by abdominal distension[11,15,21-29]. However, ‚lipodystrophy‚ is increasingly being used to represent any change in fat distribution in an HIV-infected patient. Due to varying definitions of lipodystrophy, there are substantial problems in delineating and understanding the syndrome(s). Reports of the prevalence of lipodystrophy have been as low as 2% and as high as 84% (see Table 1), which may be due to variations in definition, methodology (e.g., self-report versus measurement versus chart review), treatment regimen or patient population. Moreover, both lipohypertrophy and lipoatrophy have been reported in patients receiving non-PI-containing antiretroviral regimens[1,2,7,15,16,20,30-33], raising the question as to whether fat redistribution is associated with factors such as highly active antiretroviral therapy (HAART) rather than PI per se. In addition to the use of antiretroviral medications, many potential contributing factors to lipodystrophy have been proposed, such as age, time on PI or antiretroviral therapy, viral load, duration of HIV infection, and change in body weight.
Metabolic changes, including hypertriglyceridemia, hypercholesterolemia, hyperglycemia, and insulin resistance have also been reported in patients taking PI ([8,21,34-36] K. Mulligan, C. Grunfeld, V. W. Tai, H. Algren, M. Pang, D.N. Chernoff, J.C. Lo, M. Schambelan; unpublished data). However, the relationship of the metabolic changes to fat redistribution has been debated. There is growing concern that these symptoms resemble Syndrome X, also known as Multiple Metabolic Syndrome (central/visceral obesity, hyperlipidemia, insulin resistance, hypertension and hyperuricemia), which confers an increased risk of cardiovascular disease.
In this paper, we review the key studies in the field, with attention to methodology, nature of the patient population, and entry criteria. We compare the objective measurements of body composition with the descriptions of the syndrome(s). We assess what is known about the relation of the changes to type and effectiveness of antiretroviral therapy, as well as to other potential cofactors. Finally, we discuss the implications for designing future studies.
Reports of fat redistribution (lipohypertrophy)
Early case reports and small study series of fat redistribution included descriptions of the striking appearance of an increase in the dorsocervical fat pad known as ‚buffalo hump‚[1-12], which were frequently attributed to PI use. Due to the association of buffalo hump with Cushing‚s syndrome, these investigations were careful to rule out the presence of hypercortisolism[1-3]. Various surveys have noted buffalo hump in two out of 95 (2%), four out of 116 (3%), and three out of 63 (5%) patients taking PI [9,11,15] (see Table 1). Thus, although the development of buffalo hump is feared in the HIV-infected community, the prevalence may be low (i.e., <5%). In two studies evaluating the presence of buffalo hump in HIV-infected patients not on PI, the prevalence was none of 32 patients and none of 42 patients[9,15]. While these data suggest an association with PI, the differences are not significant, as the prevalence is low; larger numbers are needed to determine an association; other cases of buffalo hump have been reported in patients who had not received PI therapy (see below). The specific factors associated with the development of buffalo hump remain unclear.
Other localized deposits of fat, particularly in the upper torso and breasts, are also described in case reports[4,8,11-16]. Patients with buffalo hump may have diffuse neck fat that resembles Madelung‚s syndrome or multiple symmetric lipomatosis[6,10]. Self-reported increase in breast size was noted in 6%, 13% and 37% of HIV-infected women taking PI[9,11,15]. Another study found that 1% (three out of 272) of subjects on PI reported breast enlargement; one of the three patients was a man. Prevalence estimates of breast enlargement are clearly dependent on whether men or women are studied together or separately. Also, however, is the issue of whether a non-PI comparison group was examined. In one of the above studies, none of 42 (0%) women not taking PI reported changes. In contrast, another study found no difference in prevalence when comparing women who were and were not taking PI [35 out of 95 (37%) versus 10 out of 32 (31%), respectively]. The failure to find a difference when the prevalence is in the optimal range (one-third) is highly suggestive: it is difficult to demonstrate a substantial difference between two groups when the prevalence is low. Although some of the studies described a change in bra size, none directly measured change in breast size. Thus, the true prevalence of increased breast size in HIV-infected patients remains unclear, and the association with use of PI is variable.
Surveys of self-reported increase in abdominal girth have noted frequencies of 1%, 16%, 21%, and 56% in patients on PI [9,11,] see Table 1). One report utilizing physical examination found a frequency of 7%. Two studies compared patients receiving and not receiving PI: one found a greater frequency of increased abdominal girth in those taking PI [13 out of 63 (21%) versus 0 out of 42 (0%) patients], while another found no difference [35 out of 95 (56%) versus 13 out of 32 (41%) patients]. It should be noted that most reports have not documented that the increased girth is adipose tissue. Furthermore, an increase in weight, abdominal girth and visceral fat is a normal component of aging in the general population. Therefore, in AIDS patients with increased abdominal girth, it is important to differentiate between three possible scenarios: increased fat due to aging in a population with restored health due to effective antiretroviral regimens, abnormally increased fat deposition due to PI (or other) therapy, or increase in non-adipose (e.g., gas or lean) abdominal tissue.
Reports of fat redistribution (lipoatrophy)
Lipoatrophy of the face, arms and/or legs has been reported in patients receiving PI therapy[11,16,21-29]. In prevalence surveys, facial atrophy in patients taking PI occurred in 1%, 3%, 22%, and 24% of patients on PI [9,12,] see Table 1]. The only one of these studies that compared patients who were and were not taking PI found no clear differences in frequency, perhaps due to low rates: two out of 63 (3%) versus 0 out of 42 (0%) patients.
Wasting of fat in the arms or legs has been reported with a frequency of 8%, 11% and 13% in patients on PI[9,11,22]. In one study in which lipodystrophy was reported in 64% of AIDS patients on PI, a subset of patients was confirmed to have lower levels of arm and leg fat than patients not taking PI and than a healthy control group, using dual energy X-ray absorptiometry (DEXA). Improved understanding of the large variation in reported prevalence of facial or peripheral lipoatrophy, ranging from 1 to 24%, will require larger studies with standardized definitions which incorporate objective as well as subjective measures.
Although peripheral lipoatrophy is not often viewed as a common component of the normal aging process, it is well recognized that waist-to-hip ratio (WHR) increases with aging, particularly with weight gain. In fact, it has been found that WHR increases with age even in those who maintain stable weight, suggesting that waist will increase while peripheral fat decreases. In some subsets of patients, such as smokers, WHR can increase despite weight loss. Similarly, in patients with Type 2 diabetes mellitus, there is an increase in waist circumference and a decrease in hip circumference. Thus, peripheral lipoatrophy accompanying central obesity may occur in non-HIV-infected populations as well as in HIV-infected patients, further complicating our understanding of these newly reported syndromes.
Prevalence of multiple changes in fat distribution
Some studies have reported the overall prevalence of changes in fat distribution (lipodystrophy) by combining the syndromes described above. The prevalence estimates of the combined syndromes vary widely, ranging from 2% to 84% (see Table 1) [9,11,12,16, 21-23,26-28]. In a key primary report, Carr et al. found lipodystrophy syndrome in 74 out of 116 (64%) subjects taking PI; a long-term follow-up study reported a prevalence of 84% (95 out of 113 patients). Comparison with patients not on PI reported frequencies of one out of 32 (3%) and one out of 28 (4%) patients in the initial and follow-up studies, respectively (Table 1)[21,28]. Two other studies which evaluated patients on non-PI-containing regimens also found low prevalence rates: 0 out of 16 patients and 0 out of 42 patients, respectively (see Table 1)[9,26]. In two studies of HIV-infected women taking variable antiretroviral regimens, rates of lipodystrophy were found to be 18% and 11% (32 out of 306 women)[11,16].
Central obesity versus sparing of loss of central visceral fat
Many reports of PI-associated lipodystrophy suggest that peripheral lipoatrophy is often accompanied by an increase in central obesity or adiposity (lipohypertrophy). However, it is well recognized that with thinning of the arms and legs, the abdomen may appear to be more protuberant (pseudotruncal obesity), and with increased waist circumference, arms and legs may appear to be smaller. Thus, to understand these changes, it is important to measure regional body composition to determine fat content precisely. Fat content can be measured by several techniques. Bio-electrical impedance analysis is inexpensive and easy to perform with a hand-held machine that uses undetectable electrical currents. However, as bio-electrical impedance analysis measures only total body fat, the results are not very helpful for understanding recently described changes in HIV-infected patients. DEXA uses the differential absorption of two X-rays generated at different energies. Originally used for bone density, it quantifies bone, muscle and fat by their varying densities. DEXA can measure regional fat content (arms, legs and trunk routinely; other areas can be analyzed separately by the user). However, the image is two-dimensional and does not allow separate quantification of visceral fat. Body composition by DEXA is increasingly available, but experimental and moderately expensive (≊$125). Three-dimensional analysis by total body scanning with computed tomography (CT) or magnetic resonance imaging (MRI) can quantify visceral fat as well as regional fat. These techniques are expensive ($500-1000) and the analysis of results is not well- standardized between groups.
It is of note that in the few series of patients reported with peripheral lipoatrophy in which regional body composition has been measured, the actual measurements do not substantiate an increase in true central obesity. In one study, a subset of patients with peripheral lipoatrophy in whom body composition was measured by DEXA had less total truncal fat with no increase in ‚abdominal‚ fat (narrow DEXA window) than did either healthy controls or HIV-infected patients not taking PI  (see Table 2). Similarly, Kotler et al. using DEXA and anthropometric procedures for measuring regional fat, found that patients on PI had a decrease in peripheral fat without an absolute increase in truncal or estimated visceral adiposity when compared to patients not receiving PI and to healthy controls (see Table 3). Thus these data support pseudotruncal obesity, rather than increased truncal fat.
In the first study cited in the previous paragraph that utilized DEXA for measurement of regional fat, the PI cohort had 4kg less (fat plus fat-free) mass than the non-PI HIV-positive or the HIV-negative controls (73.3 versus 77.4 versus 77.5kg, respectively), as well as 5.5kg less total fat (see Table 2). Similarly, patients taking PI in the second study discussed  had lower weight and total body fat than did HIV-negative controls (Table 3). In another series of patients with peripheral fat loss, weight loss averaging 5.7kg was found in seven subjects. Thus, loss of weight and/or fat may be a contributing factor to lipoatrophy. In these studies, the reduction in total adipose stores was predominantly from subcutaneous depots, whereas truncal fat was relatively spared[21,30]. These data should be contrasted with those in obesity, where there is preferential loss of visceral fat with weight loss[39-41].
The data do not support increased central obesity in the majority of subjects with peripheral lipoatrophy. Rather, they support the concept that during fat loss, visceral fat is relatively spared, giving the impression of increased abdominal girth, hence pseudotruncal obesity. Yet many surveys accept self- or physician-report of increased abdominal girth as evidence of central adiposity or lipodystrophy, without quantification of abdominal or peripheral fat.
Studies in normal subjects (lean and obese) have shown that increased WHR correlates with increased central adiposity and increased risk of coronary artery disease[42-46]. While WHR was originally found to be an excellent predictor of visceral obesity, more recent data raise the possibility that waist circumference may be better ([47-51] C.E. Lewis, S. Sidney, J.O.Hill, unpublished data). Thus, a correlation between increased WHR and visceral adiposity cannot be presumed in HIV infection: actual measurements are needed. The situation in HIV-infected patients is complicated further by the frequent presence of drug-induced abdominal bloating or fullness. Because even waist circumference is a relatively weak predictor of visceral obesity, direct measurement by MRI or CT is required to more accurately establish true truncal versus pseudotruncal obesity[42,52,53].
One study evaluated 10 patients with complaints of increased abdominal girth and symptoms of abdominal fullness, distension or bloating after starting indinavir therapy. Abdominal CT in these 10 patients documented an increase in visceral adipose tissue, as compared with 10 HIV-infected patients taking indinavir who did not have abdominal complaints and with 10 HIV-infected patients not taking a PI agent. Increased visceral adipose tissue was particularly marked in those patients with increased girth. Thus, the study documented visceral obesity in some patients complaining of increased girth. However, confirmation of these findings, as well as estimation of frequency, will require a larger sample of randomly selected patients to ensure lack of ascertainment bias. It is of note that the subjects in this study had unusually high mean CD4 cell counts (797 and 796×106 cells/l) and that the comparison group had CT performed for medical indications. There were trends for patients with complaints of increased girth toward a longer duration of PI therapy (14.1±6.9 versus 9.7±3.7 months), and for those on PI to have higher CD4 cell counts and lower viral load compared with those not taking PI. Whereas weight change on therapy was not statistically different between the groups, PI-taking subjects with complaints of increased girth had gained 1.2±2.5kg in the 6months before the study, whereas those without complaints lost 0.6±3.8kg. These data raise the possibility that weight gain is accompanied by visceral adiposity, whereas those of other studies [21,] see above] suggest that loss of fat and/or weight may result in a syndrome of peripheral lipoatrophy.
Another report compared regional fat quantified by DEXA and MRI in 26 HIV-infected patients (15 men, 11 women) with that in age- and gender-matched healthy controls. In HIV-infected men, peripheral fat and body weight were lower, and there was a trend towards less truncal fat also. Although there were no significant differences in regional fat between HIV-infected women and controls, women with HIV infection had slightly less lean body mass on DEXA and less muscle on MRI. In a subset of 12 HIV-infected subjects (seven men, five women) reporting truncal enlargement, men had higher total and truncal fat with no difference in peripheral fat as compared with eight HIV-infected men not reporting such changes. However, age was greater and body mass indexes (BMI) were higher. Similar trends in fat distribution and age were found in HIV-infected women compared to controls, but these did not reach statistical significance. These data suggest a role for aging, and in men, weight gain in the induction of increased visceral fat in HIV-infected patients. Larger studies of randomly selected subjects are needed to determine the extent to which the reported changes represent restoration to health due to effective antiretroviral therapy (e.g., weight gain with central adiposity) versus abnormal distribution of fat.
Resolving the contradiction of lipoatrophy and lipohypertrophy
The papers described above describe conflicting phenotypes. The reports of lipoatrophy present facial lipoatrophy as the most striking finding, yet many of the subjects with upper body lipohypertrophy have increased neck and facial fat. While some patients with symptoms of truncal enlargement have documented increases in truncal fat[17,20], those with predominantly lipoatrophy often do not when measured objectively[21,30]. If energy balance were neutral, the loss of peripheral fat (e.g., by apoptosis) should be accompanied by an increase in visceral fat or by fatty liver (see Fig. 1b). The failure to see this compensation (see Fig. 1c) has implications for mechanism.
Rather than invoking a single mechanism as cause for both lipoatrophy and lipohypertrophy in HIV-infected patients, we propose a different conceptual approach that considers multiple contributing factors. The major drive behind fat redistribution could be a cellular metabolic program that promotes visceral fat formation relative to subcutaneous fat, or inhibits subcutaneous fat formation relative to visceral fat. However, the net outcome in terms of phenotype would depend on what happens to absolute amounts of fat. Those with higher total body fat will present with proliferation of truncal obesity including visceral, chest and upper back (see Fig. 1e), whereas those with lower total body fat will present with decreased arm, leg and/or facial fat (Fig. 1c). Such a hypothesis is consistent with the published data, and may also explain some of the wide discrepancies in published prevalence estimates of lipoatrophy and lipohypertrophy. An alternative possibility is that there is not one, but several discrete syndromes affecting HIV-infected patients, such as separate syndromes leading to lipoatrophy versus lipohypertrophy (see below).
Does fat redistribution occur with non-PI-containing regimens
Many components of the abnormal fat distribution syndrome have been described in patients not taking PI (Table 4). In an early report, Lo et al. found that four out of eight patients with ‚buffalo hump‚ were not taking PI; subsequently, when PI usage was more common, they reported six more cases, one of which occurred without PI exposure. Another report found that one out of four subjects with a buffalo hump was not on PI  Another case of buffalo hump has been reported in a patient not taking PI. Similarly, increased breast and waist size have also been reported in women and men not taking PI[15,16,20,33]. One study found no difference in visceral or subcutaneous adipose tissue or waist circumference when comparing HIV-infected subjects on PI or not taking PI. Another study found truncal adiposity by DEXA in women to be associated with HIV infection, but independent of PI use; only four out of 13 patients who had abnormal fat distribution were taking PI. Lipoatrophy has also been reported in patients taking non-PI-containing regimens[15,16,32,33].
A recent study found that 32 out of 106 (11%) women had fat redistribution characterized by wasting of glutei, thighs and calves, accompanied by increases in breasts and abdominal girth. The body composition of these 32 women was compared to an equal number of BMI-matched control HIV-infected subjects using DEXA (see Table 5). As expected by matching BMI, total fat was similar, but those with fat redistribution had more abdominal and less limb fat. Of note, 12 out of the 32 women with fat redistribution were not on PI-containing regimens. More importantly, all 32 women with fat redistribution were taking lamivudine. Univariate analysis linked lamivudine and stavudine, but not PI, to fat redistribution. Multivariate analysis found that duration of therapy and high viral load at the beginning of the last change in antiretroviral therapy were the only independent predictors of fat redistribution. Another series reported nine patients with similar changes in fat distribution who had never received PI; all nine had taken lamivudine and four had taken stavudine. Thus, data is accumulating to suggest that PI therapy per se may not be the primary or sole determinant of changes in fat distribution. As lamivudine and stavudine are frequently used in combination with PI discernment of their respective roles in fat redistribution syndromes must be specifically undertaken.
The effect of PI therapy on body weight
In light of the above, it is important to assess the extent to which PI therapy affects body weight. In a longitudinal study of 38 subjects, weight gain averaged 1.5kg at 12.2 months after PI therapy was initiated. In another study of patients with wasting (<95% usual body weight), weight gain was 1.5kg in 186 patients evaluated 49 (range, 21-100) days after PI initiation and 4kg in 160 patients evaluated 176 (range, 109-232) days after starting therapy. There was, however, significant dropout rate due to complications that might have caused weight loss, such as opportunistic infections. Thus we do not know accurately how much weight change is induced by PI, as data are limited and the mean changes reported are small. Furthermore, in the few studies where body composition was performed, the subjects gained predominantly fat[54-56].
Lipid and glucose metabolism in HIV infection
Metabolic disturbances are a classic host response to infection, including HIV (Table 6). In men, high density lipoprotein (HDL) cholesterol decreases early in HIV infection, followed by decreases in low density lipoprotein (LDL) cholesterol (Table 6)[58,59]. At the time of transition to AIDS, triglycerides rise due to increased very (V)LDL[58,59]. There is both a decreased clearance and an increased production of triglyceride-rich particles[59,60]. The changes in triglycerides are related to interferon-agr; levels, the host response to virally-infected cells[59-63].
Studies of HIV-infected patients receiving PI have typically found increased triglyceride and cholesterol levels ([21,35,36] K. Mulligan, C. Grunfeld, V. W. Tai, H. Algren, M. Pang, D.N. Chernoff, J.C. Lo, M. Schambelan, unpublished data). In one study, fasting triglycerides were 106% higher in HIV-infected patients on PI therapy compared with those not on PI therapy, which were in turn 33% higher than in HIV-negative controls. Cholesterol levels were 31% higher in those on PI, and HDL levels were unchanged. In another study, initiation of PI therapy was associated with an increase in fasting triglycerides by 47%, in cholesterol levels by 23%, in calculated LDL cholesterol by 20%, and in directly measured LDL by 27%; HDL cholesterol levels were unchanged (K. Mulligan, C. Grunfeld, V. W. Tai, H. Algren, M. Pang, D.N. Chernoff, J.C. Lo, M. Schambelan, unpublished data). In comparison, a group of HIV-infected patients not receiving PI showed no such changes. Higher LDL levels have also been found in patients taking PI therapy in other studies[17,35]. One study of women reported increases in total and HDL cholesterol levels after PI therapy. Another study found elevated serum triglyceride levels in 75 HIV-infected women when compared with 30 healthy controls, without relation to either PI use or body weight. However, this study did find a strong correlation between duration of PI use and triglyceride levels. Additionally, triglyceride levels correlated positively with viral load, as might be predicted from studies in the pre-PI era.
A previous study of initiating zidovudine in antiretroviral-naïve patients showed decreased triglyceride levels; therefore, an increase in triglycerides in patients on PI is likely to be a direct effect of the PI, rather than a result of improvement in health. Yet, given that LDL is low in HIV infection[58,59], the data on increases in LDL levels ( K. Mulligan, C. Grunfeld, V. W. Tai et al. unpublished data) cannot distinguish between restoration to health by PI-containing regimens with return-to-‚normal‚ LDL levels and increased formation of LDL from VLDL.
In the pre-PI era, HIV-infected subjects were generally found to have normal or decreased glucose levels, without insulin resistance[63,65-68]. While several reports found hyperglycemia in patients on PI therapy[8,29,69-74], most did not[21,34,36,75-77]. However, many of these studies involved chart review of random glucose values. When fasting glucose levels are compared, there is little difference in hyperglycemia between patients on PI therapy and those who are PI-naïve ([21,35] K. Mulligan, C. Grunfeld, V. W. Tai, H. Algren, M. Pang, D.N. Chernoff, J.C. Lo, M. Schambelan,unpublished data).
In contrast, insulin resistance, documented by high fasting insulin-to-glucose ratios, has been more consistently reported in association with PI use ([21,34-36] K. Mulligan, C. Grunfeld, V. W. Tai et al. unpublished data); such studies have predominantly analyzed male populations. Insulin resistance associated with PI therapy has been documented by provocative testing in two cohort studies. Oral glucose tolerance testing showed impaired glucose tolerance accompanied by higher stimulated insulin or C-peptide levels ( K. Mulligan, C. Grunfeld, V. W. Tai et al. unpublished data). Furthermore, a study using insulin tolerance testing also demonstrated insulin resistance in patients taking PI. However, a recent report comparing 75 HIV-infected women with 30 healthy weight-matched controls found significant hyperinsulinemia in association with HIV infection that was independent of PI use. In contrast, one study found that initiation of PI therapy in PI-naïve subjects induces a small increase in glucose levels (11%) and a larger increase in insulin levels (95%). Because their study compared values before and after PI therapy in the same subjects, the results are not confounded by other potential differences that may be found in cohort studies (, K. Mulligan, C. Grunfeld, V. W. Tai, H. Algren, M. Pang, D.N. Chernoff, J.C. Lo, M. Schambelan, unpublished data). In all likelihood, therefore, the use of PI is associated with insulin resistance. The prevalence and severity of this condition requires further evaluation.
What is the relation between changes in metabolism and fat redistribution?
Varying results have also been obtained when different groups have analyzed the relationships between fat redistribution syndromes and metabolic changes. These differences may be the product of differences in syndrome description, cohort composition, drug use and other unknown differences between sites.
In their initial description of the lipodystrophy syndrome, Carr et al. found that higher levels of triglyceride, insulin, and C-peptide in patients on PI therapy with lipodystrophy, as well as more severe insulin resistance than those on PI therapy without lipodystrophy. Sixty four percent of their subjects on PI therapy reported at least mild lipodystrophy. In follow-up, at which time 83% of subjects reported lipodystrophy, there was no further worsening of triglyceride, insulin and C-peptide levels, or insulin resistance.
In contrast, other investigators found no differences in lipid or glucose metabolism in 32 HIV-infected women with fat redistribution when compared with 74 women without body changes. However, PI use was present in both groups. A study of 10 patients on PI therapy who presented with complaints of increased abdominal girth also found limited relation between fat distribution and metabolic changes. An increase in the ratio of visceral adipose tissue to total adipose tissue correlated with triglyceride levels but not with cholesterol levels. A more recent paper by the same investigators found no correlation of triglyceride levels with BMI in 12 patients with lipodystrophy. Two other reports found that triglycerides were higher [78,79] in those subjects who had more lipohypertrophy. Another found that total cholesterol, LDL cholesterol and triglycerides were not higher in women who presented with fat redistribution compared to a group of women without fat redistribution who were on similar therapy and had similar demographics. Thus, the data on the association between body composition changes and metabolic alterations in cohorts of HIV-infected subjects are in conflict.
A potential resolution is found in one recent report, in which subjects were studied before and after (3.5±0.5 months) the initiation of PI therapy (K. Mulligan, C. Grunfeld, V. W. Tai, H. Algan, M. Pang, D.N. Chernoff, J.C. Lo, M. Schambelan, unpublished data). Increases in triglyceride, cholesterol, LDL cholesterol, glucose and insulin levels were documented, despite the absence of changes in fat distribution (either lipoatrophic or lipohypertrophic) as measured by DEXA. Furthermore, two studies had subsets in which some lipid measurements were available before and after initiation of PI therapy. In one study evaluating visceral obesity, initiation of PI therapy was independently found to be associated with an increase in triglycerides and cholesterol. In another study of fat redistribution in women, initiation of PI therapy was found to be associated with an increase in total and HDL cholesterol. These data support the concept that PI-induced changes in metabolism precede any changes in fat distribution, and are consistent with that of other investigators[16,21,28]. Also, the data are consistent with the weak correlation of metabolic with body composition changes in some studies[11,17,64]. Thus, a working hypothesis is that metabolic changes are induced by PI therapy independent of body composition changes. It is then important to determine the individual influence of both lipoatrophic and lipohypertrophic changes on metabolic parameters, as well as whether metabolic changes predict changes in fat distribution.
Syndrome or syndromes?
The initial description of ‚a syndrome of peripheral lipodystrophy, hyperlipidemia and insulin resistance in patients receiving HIV protease inhibitors‚ assumed that each of those components was part of a single syndrome directly induced by PI[21,80]. The data presented above strongly suggest that there are either multiple syndromes or a variety of factors inducing different changes that influence the ultimate phenotype.
For example, each fat distribution change has been reported on non-PI-containing regimens, suggesting that a specific, single drug- or drug class-induced mechanism cannot explain all of the changes. Several possible hypotheses may explain these data, including: (i) the mechanism for changes in fat redistribution is unrelated to PI therapy per se (e.g., any form of HAART may contribute); (ii) the changes in fat distribution caused by non-PI-containing regimens represent a separate syndrome with a different mechanism than that induced by PI; or (iii) each drug regimen interacts by a different pathway with a common contributor.
Certain described changes are bidirectional. Many patients with ‚buffalo hump‚ and increased upper body fat have increased deposition of fat in the neck and face, yet facial and neck lipoatrophy are a striking characteristic of the ‚peripheral lipodystrophy‚ syndrome. Likewise, whereas there are reports of central obesity that include an increase in subcutaneous fat, the ‚peripheral lipodystrophy‚ syndrome is often accompanied by loss of subcutaneous fat in the same area. It is difficult to envisage a single mechanism that could lead to either lipohypertrophy or lipoatrophy in the same area. It is therefore likely that the lipohypertrophic syndrome(s) are distinct from the lipoatrophic syndrome(s). Alternatively, it is possible that the primary metabolic disturbance is a preferential deposition of fat in visceral adipose stores, and that the phenotypic presentation is dependent on a secondary factors such as energy balance. Those patients in negative energy balance present with dominant lipoatrophy with retention of central fat, whereas those in positive energy balance present with true central obesity and increased deposition in related areas such as neck and face. It may also be necessary to invoke genetic factors that would contribute to which depots are affected at a given energy balance in an individual subject to explain other variations. Thus multiple different mechanisms may contribute to the ultimate phenotype.
The presence of metabolic changes seem more strongly associated with PI therapy than with fat redistribution. However, there have been inadequate studies of the relative contribution of individual changes in fat distribution to metabolic changes. For example, large numbers of patients with documented visceral lipohypertrophy need to be studied to determine the extent to which that depot contributes to metabolic changes. Similarly, a separate study of subjects with peripheral lipoatrophy and sparing of central fat is needed to determine the contribution of lipoatrophy per se to metabolic changes. Furthermore, it is important to determine whether the adverse metabolic consequences observed are influenced by the absolute amount of visceral fat or the relative amount of visceral fat (visceral/total or visceral/subcutaneous), as the ratio is perturbed in lipoatrophy independent of visceral obesity.
Multiple factors have been reported to contribute to fat distribution or metabolic changes. These include time on antiretroviral therapy, time on HAART regimens, change in viral load, prior treatment regimens, age, weight change, BMI and WHR [16,19,21,23,27,28, 32,33,36,37,74,77-79]. The presentation may vary by gender, as results from studies on women and men often differ. The potential existence of multiple contributing factors also supports the concept that the phenotypic presentations of these changes do not represent a single entity with a single cause.
This review has not dealt with other changes that have been ascribed to PI therapy such as dry skin, cracked lips, brittle hair, hair loss, ingrown toe nails, paronychia and avascular necrosis among others. Many of these changes remain at case report level and probably represent distinct syndromes.
What must be done next?
It is obvious that more research is needed to understand the changes that are occurring in HIV-infected individuals in the era of HAART, which has been accompanied by dramatic reductions in death and complications. HIV infection is now a chronic disease and the approach to the new changes must take into account both the therapies and the time course of events. While a common thread in past studies of complications has been the consequences of immunosuppression, a single mechanism has not clearly been found that accounts for the changes in fat distribution and metabolism. Indeed, a case has been made above that there is likely to be more than one syndrome and/or more than one underlying mechanism at work. To make progress in understanding these changes, both in terms of what changes are associated with each other and what factors contribute to those changes, it is necessary to perform future research in a manner that takes these possibilities into account. Based on review of the data to date, several recommendations can be made: (i) wherever possible, future studies should have regional body composition studies performed and the results treated as a continuous, quantitative parameter. It will be desirable to determine empirically whether a given value represents a reliable cut off to define a syndrome; (ii) for studies relying on self-report and physical examination, responses should be bidirectional (i.e., increased or decreased). The development of common instruments would be helpful, but the dependence of self-report on language and cultural norms will limit this possibility; (iii) certain changes in fat distribution should be analyzed separately. In particular, it is important to separate atrophy from hypertrophy in most depots, but particularly in the face, neck, chest and abdomen. The association of changes in the different depots should then be determined; (iv) the contribution of each change in fat distribution to metabolic parameters should be analyzed in terms of absolute amounts of regional fat as well as ratios, because ratios may be similarly influenced by opposite changes in two different depots; (v) bloods must be drawn in the fasting state for all metabolic studies; (vi) complete medication histories are clearly needed, given the controversies over the respective roles of PI, lamivudine, stavudine and combination antiretroviral regimens; (vii) comparison of the treatment in cohorts to the pattern of the clinic at large must be made; (viii) a careful analysis of the many potential differences between cohorts pre-selected by presenting symptoms or by use of specific drug regimens must be performed. They should include potential contributing factors such as drug exposure, time on antiretroviral therapy, time on HAART, change in viral load, duration of HIV infection, age, weight change, and gender among others; (ix) the factors contributing to changes in each depot as well as metabolic changes should be analyzed separately and used to understand how the changes are associated with each other; (x) the results from these analyses can then be used to define syndromes more precisely.
The data reviewed above suggest that HIV infection is a strong underlying contributor to both changes in body composition and metabolic alterations. Distinctions must be made between restoration to normal from abnormal and the appearance of new pathology. The implication of age as a confounding factor implies that careful comparisons to age-induced changes are needed. Some changes might represent an accelerated return to normal, similar to the ‚catch-up‚ growth seen in children who are treated successfully. Other changes being reported are clearly pathological. As a consequence however, it would be appropriate to study as ‚control‚ groups not only age-matched HIV-negative healthy subjects (to determine the extent to which the changes in HIV infection exceed the norm), but also those with other chronic infections (to determine which are unique to HIV and its therapies).
Given the large number of potential changes and contributing factors, coupled with the differences between groups reported to date, several types of future studies are warranted. These include: (i) randomized, controlled trials of different therapies in which body composition and metabolic studies are included. While such trials may directly test hypotheses about drug causality, careful attention needs to be made to how subjects are allocated to the nucleoside reverse transcriptase inhibitors that have recently been implicated in fat distribution changes and not merely to PI versus non-nucleoside reverse transcriptase inhibitors as part of HAART. Furthermore, it must be recognized that studies of newly diagnosed antiretroviral naïve patients may underestimate changes, if duration of HIV infection, drug exposure, time on prior antiretroviral therapy and age are major contributors; (ii) intensive body composition and metabolic studies before and after changes in therapy; (iii) large multicenter studies using common survey and body composition methods with analysis of metabolic changes using standardized techniques and a central laboratory. Such studies are needed to determine whether differences in reports to date are due to differing definitions, methodology, prescribing patterns or patient populations.
Finally, the long-term consequences of these changes need study. It will be important to monitor the incidence (and not merely prevalence) of diabetes mellitus, hypertension and cardiovascular disease in future studies and to relate them to fat distribution and metabolic changes. The predicted consequences may take years to be manifest after the changes start. Given the currently reported prevalence of cardiovascular disease[80-83], large cohorts will be needed to attribute changes to a therapy or its consequences. Enough has been learned to date to suggest that clinical care of HIV-infected patients now needs to consider the long-term effects of HIV infection and its therapies. The side-effects of therapies may influence when and how they should be used. However, the side-effects of therapies must then be balanced against their obvious efficacy in decreasing complications and death. That progress in preventing HIV-related morbidity and mortality should not be ignored as we enter the era of HIV as a chronic disease.
1. Lo JC, Mulligan K, Tai VW, Algren H, Schambelan M. Buffalo hump in men with HIV-1 infection. Lancet
2. Miller K, Daly P, Sentochnik D, et al.Pseudo-Cushing‚s syndrome in human immunodeficiency virus-infected patients.Clin Infect Dis
3. Roth V, Kravcik S, Angel J. Development of cervical fat pads following therapy with human immmunodeficiency virus type 1 protease inhibitors. Clin Infect Dis
4. Hengel R, Watts N, Lennox J. Benign symmetric lipomatosis associated with protease inhibitors [letter].Lancet
5. Wurtz R. Abnormal fat distribution and use of protease inhibitors. Lancet
6. Stricker R, Goldberg B, Martinez E. Fat accumulation and HIV-1 protease inhibitors.Lancet
7. Lo JC, Mulligan K, Tai VW, Algren H, Schambelan M. Body shape changes in HIV-infected patients.J Acquir Immune Def Syndr
8. Striker R. Conlin D, Marx M, Wiviott L. Localized adipose tissue hypertrophy in patients receiving human immunodeficiency virus protease inhibitors. Clin Infect Dis
9. Reboli AC, Marshall S. Changes in body habitus in HIV-infected patients treated with protease inhibitors.
IDSA 36th Annual Meeting. Denver, November 1998 [abstract 476 Sa].
10. Hengel RL, Geary JAM, Vuchetich MA, et al.Multiple symmetrical lipomatosis associated with protease inhibitor therapy. Fifth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1998 [abstract 407].
11. Dong K, Bausserman LL, Flynn MM, et al.Changes in body habits and serum lipid abnormalities in HIV-positive women on highly active antiretroviral therapy (HAART).J Acquir Immune Def Syndr Hum Retrovirol
12. Boix V, Reus S, Priego M, et al.Expanding the spectrum of protease inhibitors-induced lipodystrophy.12th World AIDS Conference.
Geneva, June 1998 [abstract 12398].
13. Lui A, Karter D, Turrett G. Another case of breast hypertrophy in a patient treated with indinavir. Clin Infect Dis
14. Mann M., Piazza-Hepp T, Koller E, Gibert C. Abnormal fat distribution in AIDS patients following protease inhibitor therapy: FDA summary. Fifth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1998 [abstract 412].
15. Sutinen J, Mathur-Wagh U. Changes in body shape during PI therapy in HIV+ women. 6th Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 662].
16. Gervasoni C, Ridolfo AL, Trifiro G, et al.Redistribution of body fat in HIV-infected women undergoing combined antiretroviral therapy. AIDS
17. Miller K, Jones E, Yanovski J, Shankar R, Feuerstien I, Falloon J. Visceral abdominal-fat accumulation associated with use of indinavir.Lancet
18. Lipsky J. Abnormal fat accumulation in patients with HIV-1 infection. Lancet
19. Rosenberg HE, Mulder J, Sepkowitz KA, Giordano MF. ‚Protease-paunch‚ in HIV+ persons receiving protease inhibitor therapy: incidence, risks and endocrinologic evaluation.Fifth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1998 [abstract 408].
20. Engelson ES, Kotler DP, Tan YX, et al.Fat distribution in HIV-infected patients reporting truncal enlargement quantified by whole-body magnetic resonance imaging. Am J Clin Nutr
21. Carr A, Samaras K, Burton S, Freund I, Chisholm DJ, Cooper DA. A syndrome of peripheral lipodystrophy, hyperlipidemia and insulin resistance due to HIV protease inhibitors. AIDS
22. Viraben R, Aquuilina C. Indinavir-associated lipodystrophy. AIDS
23. Shaw A, McLean K, Evans B. Disorders of fat distribution in HIV infection.Int J STD & AIDS
24. Bonnet E, Cuzin L, Sailler L, et al.Associated lipodystrophy metabolic disorders due to protease inhibitor containing regimens. 12th World AIDS Conference.
Geneva, June 1998 [abstract 12299].
25. Wong KH, Ho Ty, Chan CW, Lee SS. Indinavir-associated facial atrophy in HIV positive patients.12th World AIDS Conference.
Geneva, June 1998 [abstract 12391].
26. Bonjoch A, Veny A, Johnston S, et al.Body composition measurements by anthropometry and bioelectrical impedance to evaluate redistribution of body fat in HIV-Positive patients on antiretroviral treatment.38th Interscience Conference on Antimicrobial Agents and Chemotherapy.
San Diego, September 1998 [abstract I-66].
27. Veny A, Bonjoch J, Romeu M, et al.Cumulative risk for developing protease inhibitor-associated lipodystrophy (PI-AL) in HIV-infected patients. 38th Interscience Conference on Antimicrobial Agents and Chemotherapy.
San Diego, September 1998 [abstract I-92].
28. Carr A, Samaras K, Thorisdottir A, Kauffman GR, Chisholm DJ, Cooper DA. Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor-associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: a cohort study. Lancet
29. Gharakhanian, Salhi Y, Nguyen TH, et al.Frequency of lipodystrophy and factors associated with glucose/lipid abnormalities in a cohort of 650 patients treated by protease inhibitors. Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 642].
30. Kotler DP, Rosenbaum KB, Wang J, Pierson RN. Studies of body composition and fat distribution in HIV-infected and control subjects. J Acquir Immune Def Syndr Human Retrovirol
31. Saint-Marc T, Touraine, IL. Buffalo hump in HIV-1 infection. Lancet
32. Saint-Marc T, Poizot-Martin I, Partisani M, Fabre J, Touraine JL. A syndrome of lipodystrophy in patients receiving a stable nucleoside-analogue therapy. Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 653].
33. Madge S, Kinloch-DeLoes S, Tyrer M, Johnson MA. Lipodystrophy in patients naïve to protease inhibitors. AIDS
34. Walli R, Herfort O, Michl GM, et al.Treatment with protease inhibitors associated with peripheral insulin resistance and impaired oral glucose tolerance in HIV-1-infected patients. AIDS
35. Behrens GMN, Dejam A, Schmidt H, et al.Impaired glucose tolerance, beta cell function and lipid metabolism in HIV patients under treatment with protease inhibitors.AIDS
36. Yanovski JA, Miller KD, Kino T, et al.Endocrine and metabolic evaluation of human immunodeficiency virus-infected patients with evidence of protease inhibitor-associated lipodystrophy. J Clin Endocrinol Metab
37. Caan B, Armstrong MA, Selby JV, et al.Changes in measurements of body fat distribution accompanying weight change. Int J Obes Relat Metab Disord
38. Seidell JC, Han TS, Feskens EJ, Lean ME. Narrow hips and broad waist circumferences independently contribute to increased risk of non-insulin-dependent diabetes mellitus. J Intern Med
39. Ross R, Rissanen J. Mobilization of visceral and subcutaneous adipose tissue in response to energy restriction and exercise. Am J Clin Nutr
40. Chowdhury B, Kvist H, Andersson B, Bjorntorp, P, Sjostrom L. CT-determined changes in adipose tissue distribution during a small weight reduction in obese males.Int J Obes Relat Metab Disord
41. Van der Kooy K, Leenen R, Seidell JC, Deurenberg P, Droop A, Bakker CJ. Waist-hip ratio is a poor predictor of changes in visceral fat. Am J Clin Nutr
42. Kissebah AH, Vydelingum N, Murray R, et al.Relation of body fat distribution to metabolic complications of obesity. J Clin Endocrinol Metab
43. Lapidus L, Bengtsson C, Larsson B, Pennert K, Rybo E, Sjostrom L. Distribution of adipose tissue and risk of cardiovascular disease and death. I. 12-year follow-up of participants in the population study of women in Goethenburg, Sweden. BMJ
44. Larsson B, Svardsudd K, Welin L, Wilhelmsen L, Bjorntorp P, Tibblin G. Abdominal adipose tissue distribution, obesity and risk of cardiovascular disease and death: a 13-year follow-up of participants in the study of men born in 1913. BMJ
45. Despres JP, Marette A. Relation of components of insulin resistance syndrome to coronary disease risk. Curr Opin Lipidol
46. Bjorntorp P. Body fat distribution, insulin resistance and metabolic diseases. Nutrition
47. Bouchard C, Depres J-P, Mauriege P. Genetic and nongenetic determinants of regional fat distribution.Endocrinol Rev
48. Despres J-P, Prud‚Homme D, Pouliot M-C, Tremblay A, Bouchard C. Estimation of deep abdominal adipose-tissue accumulation from simple anthropometric measurements in men. Am J Clin Nutr
49. Hill JO, Sidney S, Lewis CE, Tolan K, Scherzinger AL, Stamm ER. Racial differences in amounts of visceral adipose tissue in young adults: the CARDIA (Coronary Artery Risk Development in Young Adults) Study.Am J Clin Nutr
50. Sidney S, Lewis CE, Hill JO, et al.Association of total and central adiposity measures with fasting insulin in a biracial population of young adults with normal glucose tolerance: the CARDIA study. Obes Res
51. Han TS, McNeill G, Seidell JC, Lean ME. Predicting intra-abdominal fatness from anthropometric measures: the influence of stature. Int J Obes Relat Metab Disord
52. Van der Kooy, Leenen R, Seidell JC, Deurenberg P, Visser M. Abdominal diameters as indicators of visceral fat: comparison between magnetic resonance imaging and anthropometry.Br J Nutr
53. Van der Kooy K, Seidell JC. Techniques for the measurement of visceral fat: a practical guide.Int J Obes Relat Metab Disord
54. Silva M, Skolnik P, Grobach S, et al.The effect of protease inhibitors on weight and body composition in HIV-infected patients. AIDS
55. Carbonnel F, Maslo C, Beaugerie L, et al.Effect of indinavir on HIV-related wasting. AIDS
56. Schwenk A, Beijenherz A, Kermer G, et al.Impact of protease inhibitor treatment on body composition and prevalence of malnutrition in HIV-positive outpatients. 12th World AIDS Conference.
Geneva, June 1998 [abstract 32177].
57. Grunfeld C, Feingold KR. Seminars in medicine of the Beth Israel Hospital, Boston. Metabolic disturbances and wasting in the acquired immunodeficiency syndrome.N Engl J Med
58. Grunfeld C, Kotler DP, Hamadeh R, Tierney A, Wang J, Pierson Jr RN. Hypertriglyceridemia in the acquired immunodeficiency syndrome.Am J Med
59. Grunfeld C, Pang M, Doerrler W, Shigenaga JK, Jensen P, Feingold KR. Lipids, lipoproteins, triglyceride clearance and cytokines in human immunodeficiency virus infection and the acquired immunodeficiency syndrome.J Clin Endocrinol Metab
60. Hellerstein MK, Grunfeld C, Wu, K, et al.Increased de novo hepatic lipogenesis in human immunodeficiency virus-infected humans. J Clin Endocrinol Metab
61. Grunfeld C, Kotler DP, Shigenaga JK, et al.Circulating interferon alpha levels and hypertriglyceridemia in the acquired immunodeficiency syndrome. Am J Med
62. Mildvan D, Machado SG, Wilets I, Grossberg SE. Endogenous interferon and triglyceride concentrations to assess response to zidovudine in AIDS and advanced AIDS-related complex. Lancet
63. Stein TP, Nutinsky C, Condoluci D, Schluter MD, Leskiw MJ. Protein and energy substrate metabolism in AIDS patients. Metabolism
64. Hadigan C, Miller K, Corcoran C, Anderson E, Basgoz N, Grinspoon S. Fasting hyperinsulinemia and changes in regional body composition in human immunodeficiency virus-infected women. J Clin Endocrinol Metab
65. Hommes MJT, Romijn JA, Endert E, Eeftinck-Schattenkerk JKM, Sauerwein HP. Basal fuel homeostasis in symptomatic human immunodeficiency virus infection. Clin Sci
66. Hommes MJT, Romijn JA, Endert E, Eeftinck-Schattenkerk JKM, Sauerwein HP. Insulin sensitivity and insulin clearance in human immunodeficiency virus-infected men. Metabolism
67. Mulligan K, Grunfeld C, Hellerstein MK, Neese RA, Schambelan M. Anabolic effects of recombinant human growth hormone in patients with wasting associated with human immunodeficiency virus infection. J Clin Endocrinol Metab
68. Heijligenberg R, Romijn JA, Hommes MJT, Endert E, Eeftinck-Schattenkerk JKM, Sauerwein HP. Non-insulin-mediated glucose uptake in human immunodeficiency virus-infected men. Clin Sci
69. Dube MP, Johnson DL, Currier JS, Ledom JM. Protease inhibitor-associated hyperglycaemia. Lancet
70. Eastone JA, Decker CF. New-onset diabetes mellitus associated with use of protease inhibitor.Ann Intern Med
71. Visnegarwara F, Krause KL, Musher DM. Severe diabetes associated with protease inhibitor therapy. Ann Intern Med
72. Caldwell R, Rothstein A, Allmon C, Montoya JG, Zolopa A. Evaluation of protease inhibitor-associated hyperglycemia in a university-based HIV clinic.12th World AIDS Conference.
Geneva, June 1998 [abstract 12395].
73. Dong BJ, Gruta CI, Legg JJ. Diabetes and use of protease inhibitors. 12th World AIDS Conference.
Geneva, June 1998 [abstract 12327].
74. Tsiodras S, Mantzoros C, Hammer SM, Samore M. Effects of protease inhibitor use on hyperglycemia and hyperlipidemia: a five year analysis. Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 643].
75. Chang E, Deleo M, Liu YT, et al.The effects of antiretroviral protease inhibitors (PIs) on serum lipids and glucose in HIV-infected patients. 12th World AIDS Conference.
Geneva, June 1998 [abstract 12381].
76. Diaz L, Wiewora R. Diabetes, HIV infection and use of protease inhibitors 12th World AIDS Conference.
Geneva, June 1998 [abstract 60148].
77. Lee D, Matthews WC. Prevalence and risk factors for hyperglycemia, dyslipidemia, and coronary disease among HIV-infected patients on initial protease inhibitor therapy. Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 644].
78. Falutz J. Lipid abnormalities and body composition alterations in progressive stages of the HIV/HAART associated lipodystrophy (HAL) syndrome. Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 646].
79. Bausserman L, Flynn M, Dickenson R, Mileno M, Carpenter C. Relationship between body fat distribution, lipoprotein levels and fasting insulin in HIV+ women treated with protease inhibitors (PI). Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 659].
80. Carr A, Samaras K, Chisholm DJ, Cooper DA.Pathogenesis of HIV-1-protease inhibitor-associated peripheral lipodystrophy, hyperlipidaemia, and insulin resistance.Lancet
1998, 351: 1881-3.
81. Juette A, Salzberger B, Franzen C, Roemer K, Diehl V, Faetkenheuer G. Increased morbidity from severe coronary heart disease in HIV-patients receiving protease inhibitors.Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 656].
82. Klein D, Sidney S, Hurley L, Iribaren C. Do protease inhibitors increase the risk for coronary heart disease among HIV positive patients? Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 657].
83. Coplan P, Nikas A, Saah A, et al.No association observed between indinavir therapy for HIV/AIDS and myocardial infarction in 4 clinical trials with 2,825 subjects.
Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January 1999 [abstract 658].