HIV/AIDS Kenneth H. Mayer, Section Editor MAJOR ARTICLE Modifiable Dietary Habits and Their Relation to Metabolic Abnormalities in Men and Women with Human Immunodeficiency Virus Infection and Fat Redistribution Colleen Hadigan, 1 Shafali Jeste, 1 Ellen J. Anderson, 2 Rita Tsay, 3 Helen Cyr, 3 and Steven Grinspoon 1 1 Neuroendocrine Unit, Department of Medicine, and 2 General Clinical Research Center, Massachusetts General Hospital, Harvard Medical School; and 3 Clinical Research Center, Massachusetts Institute of Technology, Boston We assessed the relationship between dietary intake, body composition, and metabolic parameters in 85 consecutive human immunodeficiency virus (HIV) infected patients with fat redistribution. Dietary history and values for fasting glucose, insulin, lipids, and oral glucose tolerance were obtained for 62 men and 23 women with HIV infection and fat redistribution (mean age standard error of the mean [SEM], 43.5 0.9 years; mean body mass index [BMI] SEM, 26.3 0.5 kg/m 2 ). A multivariate regression analysis was used to predict insulin area under the curve (AUC) following the oral glucose tolerance test; this included age, sex, BMI, waist-to-hip ratio, kilocalories, duration of protease inhibitor (PI) use, fat redistribution pattern, alcohol intake, dietary fiber intake, and polyunsaturated-to-saturated (P:S) fat ratio. Only age ( P p.004), PI use duration ( P p.02), and P:S fat ratio ( P p.003) were positively associated with insulin AUC. Dietary fiber intake was inversely associated with the insulin AUC ( P p.001). In a similar analysis, alcohol consumption was a significant positive predictor of low-density lipoprotein cholesterol. Polyunsaturated fats, fiber, and alcohol are strongly associated with insulin resistance and hyperlipidemia in this population and may be important targets for dietary modification. In recent years, abnormal fat distribution and metabolic abnormalities have been increasingly recognized among individuals with HIV infection, particularly those receiving combination antiretroviral therapy [1 5]. Patients experience fat redistribution that may include facial and peripheral fat atrophy, as well as increased adiposity of the neck and abdomen [6 9]. In addition to clinically evident fat redistribution, significant metabolic disturbances, such as hyperinsulinemia, dys- Received 3 January 2001; revised 9 March 2001; electronically published 30 July 2001. All subjects gave prior written consent, and the study was approved by the Human Research Committee of the Massachusetts General Hospital and/or the Committee on Research of Human Subjects at the Massachusetts Institute of Technology. Financial support: National Institutes of Health (grant numbers R01-DK59535, T32-DK07703, K23-DK02844, M01-RR01066, and M01-RR300088) and Serono Labs. Reprints or correspondences: Dr. Colleen Hadigan, Massachusetts General Hospital, 55 Fruit St., BUL457-B, Boston, MA 02114 (chadigan@partners.org). Alternate correspondent: Dr. Steven Grinspoon, Massachusetts General Hospital, 55 Fruit St., BUL457-B, Boston, MA 02114 (sgrinspoon@partners.org). Clinical Infectious Diseases 2001; 33:710 7 2001 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2001/3305-0017$03.00 lipidemia, and impaired glucose tolerance, have been identified in these patients [5, 10, 11]. Exposure to protease inhibitors (PIs) and, more recently, nucleoside reverse transcriptase inhibitors has been implicated in the etiology of this syndrome [2, 3, 12], but the exact mechanism of fat redistribution and metabolic abnormalities has not been identified. Using food-frequency questionnaires, Batterham et al. [13] investigated energy, fat, and saturated fat intake of HIV-infected patients with fat redistribution versus patients without fat redistribution. No relationship was noted between saturated or total fat intake and the metabolic and body composition abnormalities associated with lipodystrophy. However, this study did not assess such dietary factors as intake of polyunsaturated fats [14] and dietary fiber [15], which are known to modify insulin resistance and hyperlipidemia. Furthermore, no study to date has investigated the potential dietary contribution to the various manifestations of fat redistribution (i.e., lipoatrophy vs. lipohypertrophy or a combination of both). Therefore, we evaluated the dietary intake and body fat distribution of 85 HIV-infected men 710 CID 2001:33 (1 September) HIV/AIDS
and women with fat redistribution in order to assess the relationships among dietary variables, metabolic abnormalities, and fat redistribution associated with this syndrome. METHODS Subjects. Eighty-five HIV-infected men and women with recent complaints of fat redistribution were recruited from 1998 through 1999 from the multidisciplinary HIV practice at the Massachusetts General Hospital, Boston, from communitybased practices, and by newspaper advertisement. Subjects were eligible for enrollment on the basis of the following inclusion criteria: documented HIV infection, age of 18 60 years, receipt of stable antiviral medication regimen for at least 6 weeks, and evidence of fat redistribution. Metabolic and anthropometric data from a subset of these patients have been published elsewhere [5]. Fat redistribution was scored by a single investigator as present or absent on the basis of evidence of fat accumulation in the trunk, chest, or neck and/or loss of fat in the face or extremities on physical examination. All subjects had evidence of fat redistribution in 1 or more areas. Subjects were then subcategorized according to the pattern of fat redistribution. Patients with fat atrophy of the face and/or extremities in the absence of any fat accumulation were coded as patients with lipoatrophy. Patients with increased abdominal and/or neck fat, in the absence of any fat atrophy, were coded as patients with lipohypertrophy. Finally, patients who demonstrated both characteristics of regional fat atrophy and hypertrophy were coded as patients with mixed lipodystrophy. Subjects were excluded from the study if they had a history of diabetes mellitus; were receiving concurrent therapy with insulin, antidiabetic agents, glucocorticoids, testosterone, growth hormone, estrogen, or anabolic steroids; were current substance abusers; or had a major opportunistic infection within the previous 6 weeks. Similarly, 17 men with HIV infection without any evidence of fat redistribution were recruited for participation in a study of metabolic effects of HIV infection. Inclusion criteria were identical for both patient groups, but nonlipodystrophic patients were required to have no evidence of fat redistribution (according to self-reports, with confirmation by physical examination). Thirty-five age- and body mass index (BMI) matched healthy men were recruited through hospital and local advertisements to serve as controls. In order to be included, subjects were required to be free of any significant medical condition, including diabetes, coronary artery disease, and renal disease, and to be receiving no medications. Protocol. Subjects were studied after an overnight fast. Each subject had a complete medical history obtained and a physical examination performed, which included measurement of height by stadiometer and weight by digital scale, with use of standard procedures [16]. In addition, a metal spring-tension measuring tape was used to determine circumference of the hip and the waist at the level of the umbilicus. The mean of three measurements was recorded for each parameter, and the results for waist and hip circumference were used to calculate the waist-to-hip ratio (WHR). Fasting serum samples were obtained to determine insulin, glucose, cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), and triglyceride concentrations. HIV-infected subjects with fat redistribution also received a standard 75-g oral glucose challenge, with glucose and insulin determinations at 30, 60, 90, and 120 min. Subjects also underwent total-body dual-energy x-ray absorptiometry (DEXA) with use of the Hologic QDR-4500A scanner (Hologic). This technique estimates body composition from the attenuation of x-rays pulsed in synchrony between 100 and 140 kv with the line frequency for each pixel of the scanned image. As an external standard for analysis of body composition, a step phantom, with 6 fields of acrylic and aluminum of varying thickness and known absorptive properties, was scanned periodically for calibration. The DEXA scan was used to determine total body fat as well as regional body fat measurements, such as percentage of trunk fat (trunk fat/total trunk mass) and percentage of extremity fat (extremity fat/total extremity mass). Dietary data. Dietary intake information was obtained from each subject by nutrition staff members trained in the modified Burke diet history technique [17, 18]. Subjects were asked both open-ended and closed-ended questions to determine their food intake within the past month. Food models, cups, and measuring tools were used to aid in visualization and quantification of portion sizes. Nutrient calculations were performed with the Minnesota Nutrition Data System, version 2.93, software developed by the Nutrition Coordinating Center of the University of Minnesota, in Minneapolis (Food Database, version 14A; Nutrient Database 29). The reported dietary intakes were analyzed for energy intake and macronutrient consumption. Biochemical and immunologic assays. Insulin levels were measured by radioimmunoassay (Diagnostic Product), with intra-assay and interassay coefficients of variation ranging from 5.0% to 10.0% and cross-reactivity with proinsulin at midcurve of at least 40%. HIV load was determined with Amplicor HIV- 1 Monitor testing (Roche Molecular Systems), with a lower limit of detection of 400 copies/ml. Concentrations of glucose, cholesterol, HDL, LDL, and triglyceride were determined by methods described elsewhere [5]. Direct quantitative determination of LDL cholesterol was done for patients with HIV infection and fat redistribution, with use of N-geneous LDL cholesterol reagent (Genzyme Diagnostics), with interassay co- HIV/AIDS CID 2001:33 (1 September) 711
efficients of variation between 0.62 and 0.73 and intra-assay coefficients of variation between 1.73 and 2.27. Data analysis. All data analyses were performed with use of SAS JMP software, version 3.1 (SAS Institute). Data are expressed as mean values SEM. Student s t tests were used to determine statistically significant differences between groups. Categorical variables were compared between groups with a x 2 statistic. Macronutrients are expressed as percentages of total daily energy intake. A multivariate regression model was used to evaluate the relationship between dietary intake and metabolic parameters, with adjustment for age, sex, BMI, WHR, energy intake, duration of PI use (coded as a continuous variable), and pattern of fat redistribution (coded as lipoatrophy, lipohypertrophy, or mixed lipodystrophy ). A subanalysis was performed to assess dietary patterns in 55 male patients with fat redistribution who were age- and BMI-matched to the 17 HIV-infected men without fat redistribution and 35 healthy male control subjects. RESULTS Sixty-two men and 23 women with HIV infection and fat redistribution were evaluated. All patients except 1 were receiving treatment with an antiretroviral regimen that contained a nucleoside reverse transcriptase inhibitor, and 59 (69%) of 85 patients were receiving a PI at the time of evaluation. Of those patients who were currently receiving a PI, the mean duration of PI exposure was 27.3 1.4 months. Eleven patients were currently using a lipid-lowering medication. Clinical characteristics of women and men were similar except for age and PI use. Women had a mean age 5 years younger than that of men ( P p.01), and men were more likely to be receiving a PIincluding regimen ( P p.01) (table 1). Dietary intake and metabolic parameters. Dietary intake and metabolic parameters for all patients are presented in table 1. A multivariate regression analysis was used to assess the contribution of dietary fiber intake, alcohol intake, and the polyunsaturated-to-saturated (P:S) fat ratio on metabolic abnormalities, independent of age, sex, BMI, WHR, pattern of fat redistribution, energy intake, and duration of PI use. Age ( P p.004), duration of PI use ( P p.02), and P:S fat ratio ( P p.003) were strong positive predictors of insulin area under the curve (AUC), whereas dietary fiber was inversely associated with insulin AUC ( P p.001) (table 2). A 5-g increase in daily dietary fiber intake was associated with a 14% reduction in insulin AUC. With use of the same model to determine independent predictors of LDL cholesterol, alcohol consumption ( P p.003) was associated with elevated LDL cholesterol levels. When lipid-lowering therapy (coded as current therapy: yes or no ) was added to the model, there was no change in these findings. Alcohol consumption was also positively associated with HDL levels ( P p.01), as was BMI ( P p.01) and female sex ( P p.007). WHR was inversely related to HDL concentration ( P p.005). There were no significant clinical or dietary variables identified as predictors of triglyceride levels with use of the same regression model. Fifty-three percent of all patients consumed!20 g/d of dietary fiber, and 54% of the patients reported cholesterol intake of 1300 mg/d. Body composition and pattern of fat redistribution. Fifteen patients had a lipoatrophy pattern of fat redistribution, 14 had primarily lipohypertrophy, and 56 demonstrated a pattern of mixed lipodystrophy (table 3). Differences in fat distribution between these subclassifications were verified by DEXA scan, which showed significantly lower extremity fat in patients with lipoatrophy and mixed lipodystrophy and significantly greater truncal fat in patients with lipohypertrophy and mixed lipodystrophy. Subjects with lipoatrophy had the lowest total percentage of body fat, compared with patients with lipohypertrophy or mixed lipodystrophy ( P p.001 for each group comparison). When dietary intake was analyzed by pattern of fat redistribution, only kilocalories per kilogram of body weight differed among patient groups. Patients with lipoatrophy ate significantly more kilocalories/kg than did patients with lipohypertrophy or mixed lipodystrophy. Fiber, alcohol, and cholesterol intake and P:S fat ratio did not differ by fat redistribution pattern. Patients with mixed lipodystrophy were found to have significantly greater fasting hyperinsulinemia than did patients with either lipoatrophy or lipohypertrophy (table 2). Similarly, insulin AUC was significantly increased and more than doubled in patients with mixed lipodystrophy, compared with patients with lipoatrophy ( P p.007). Lipoatrophy patients had lower HDL cholesterol than did patients with lipohypertrophy and mixed lipodystrophy, but this difference was significant only for the patients with mixed lipodystrophy ( P p.04). There were no other significant differences in lipid profiles among the patients with different patterns of fat redistribution. Dietary, metabolic, and body composition parameters, and sex. Women consumed significantly less fiber ( P p.003) and cholesterol ( P p.03) than did men (table 1). Energy per kilogram of body weight, macronutrient intake, alcohol consumption, and the P:S fat ratio were similar regardless of sex. Women were more likely than men to have a diet low in fiber (!20 g/d; 78% vs. 44%; P p.004). Women had significantly lower fasting glucose concentrations ( P p.03) and higher HDL cholesterol levels ( P p.01) than did men. There was no significant difference between men and women with fat redistribution in terms of response to the oral glucose tolerance test or other lipid measurements. There were no significant differences in type of fat redistribution pattern, BMI, or WHR between men and women. However, as a group, regardless of fat redistribution pattern, women had 712 CID 2001:33 (1 September) HIV/AIDS
Table 1. Patient characteristics and sex-related differences. Characteristics All (n p 85) Men (n p 62) Women (n p 23) Age, years 43.5 0.9 44.8 1.0 a 39.8 1.7 a Duration HIV Infection, years 7.2 0.4 7.4 0.5 6.6 0.8 HIV load, copies/ml 7448 2879 8818 3439 599 100 Current PI use, % 69 77 a 48 a Duration of NRTI exposure, years 4.8 0.3 5.0 0.4 4.1 0.5 Body composition BMI, kg/m 2 26.3 0.5 26.3 0.5 26.4 0.9 WHR 0.97 0.006 0.97 0.006 0.96 0.01 Body fat, % 22.7 0.9 19.7 0.7 b 30.9 1.5 b Trunk fat, % 25.7 0.9 22.9 0.8 b 33.3 1.5 b Extremity fat, % 18.9 1.1 15.1 0.8 b 28.9 2.1 b Glucose, insulin, lipids Fasting glucose, mg/dl 92.4 1.4 94.3 1.7 a 87.3 2.4 a Fasting insulin, mu/ml 19.6 1.8 20.9 2.4 16.0 1.5 Glucose AUC, mg/dl 120 min 17,494 426 18,012 513 16,058 676 Insulin AUC, mu/ml 120 min 10,342 847 10,217 999 10,698 1630 Cholesterol, mg/dl 227 6 231 8 219 7 LDL, mg/dl 134 5 134 6 132 7 HDL, mg/dl 41 1 39 1 a 45 3 a Triglyceride, mg/dl 338 34 369 44 255 39 Daily dietary intake Kilocalories 2490 86 2675 102 b 1992 106 b Kilocalories/kg 32.6 1.1 33.6 1.3 29.9 2.0 Fat, % 33.2 0.8 33.4 1.0 32.8 1.5 Carbohydrate, % 50.2 0.9 49.8 1.1 51.3 1.9 Protein, % 17.2 0.4 17.3 0.4 16.9 0.7 Alcohol, g 3.8 0.7 4.6 0.9 1.5 0.7 Dietary fiber, g 21.3 1.1 23.3 1.3 b 16.1 1.3 b Polysaturated:saturated fat ratio 0.66 0.04 0.68 0.05 0.61 0.06 Cholesterol, mg 351 19 376 24 a 283 22 a NOTE. Data are mean values SEM, unless otherwise indicated. AUC, area under the curve; BMI, body mass index; HDL, high-density lipoprotein; LDL, low-density lipoprotein; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; Poly, polyunsaturated; WHR, waist-to-hip ratio. a P!.05 for comparisons between men and women. b P!.01 for comparisons between men and women. greater total body, trunk, and extremity percentages of fat than did men ( P p.0001, for all comparisons). HIV-infected men with fat redistribution versus HIV-infected men without fat redistribution and healthy control subjects. There was no statistically significant difference between the 3 groups in terms of age or BMI (table 4). The mean duration of HIV infection was 7.2 0.6 and 5.7 0.7 years in patients with and without fat redistribution, respectively. Sixty-three percent of patients with fat redistribution and 41% of patients without fat redistribution had an HIV load!400 copies/ml. HIV-infected subjects, with or without fat redistribution, had similar dietary habits and did not differ from one another in total energy intake or macronutrient composition. Both groups of HIV-infected subjects reported a higher percentage of energy from fat ( P!.01) and a lower percent of energy from carbohydrates ( P!.01), compared with control subjects. Total cholesterol intake was significantly higher among the 2 groups with HIV infection than it was among control subjects ( P!.01). HIV-infected patients with fat redistribution were more likely to have a cholesterol intake of 1300 mg/d than were controls (56% vs. 29%; P!.01), as were HIV-infected patients without fat redistribution (64% vs. 29%; P p.01). Cholesterol, LDL, and triglyceride levels were higher in subjects with fat redistribution than they were in those without fat redistribution ( P p.01) and in control subjects ( P p.001). HIV/AIDS CID 2001:33 (1 September) 713
Table 2. Relationship of dietary intake of polyunsaturated fat, fiber, and alcohol to insulin and lipids. Variable Estimate (95% CI) P Insulin AUC Age, years 286 (94 478).004 PI therapy duration, months 121 (24 219).02 P:S ratio 7087 (2581 11,592).003 Dietary fiber, g 289 ( 131 to 446).001 LDL cholesterol Alcohol, g 2.2 (0.8 3.7).003 HDL cholesterol Alcohol, g 0.5 (0.1 0.9).01 Female sex 8.3 (2.4 14.3).007 BMI 0.9 (0.2 1.5).01 WHR 72 ( 22 to 122).005 NOTE. Estimate, 95% CI, and P value represent statistically significant results for multivariate regression model to predict insulin AUC and LDL and HDL cholesterol. The model included age, sex, BMI, WHR, energy intake, duration of PI use, pattern of fat redistribution, dietary fiber and alcohol intake, and P:S fat ratio. AUC, area under the curve; BMI, body mass index; HDL, high-density lipoprotein; LDL, low-density lipoprotein; PI, protease inhibitor; P: S ratio, ratio of polyunsaturated to saturated fat ; WHR, waist-to-hip ratio. Patients with fat redistribution also had significantly lower HDL cholesterol levels than did control subjects ( P p.001). Fasting insulin levels were significantly higher in patients with fat redistribution than they were in control subjects (19.5 2.3 mu/ ml vs. 11.8 1.3 mu/ml; P p.01]. There were significant differences between patient groups and control subjects in measures of body fat distribution. HIVinfected patients with fat redistribution had significantly higher WHRs ( P!.0001) and lower percentages of extremity fat ( P!.05 vs. patients without fat redistribution; P!.0001 vs. control subjects) than did both control groups. In addition, patients with fat redistribution had higher percentages of trunk fat than did nonlipodystrophic patients ( P!.05). DISCUSSION HIV-associated fat redistribution is increasingly recognized among HIV-infected individuals and has been estimated to occur in more than half of patients receiving protease inhibitor therapy [4]. Fat redistribution is associated with metabolic disturbances, such as hyperinsulinemia and hyperlipidemia, which may increase the risk of coronary artery disease in affected patients [5]. Our data indicate that certain modifiable dietary components, such as polyunsaturated fats, fiber, and alcohol, are strongly associated with insulin resistance and hyperlipidemia among these patients, independent of age, sex, protease inhibitor use, and body fat distribution. Using food-frequency questionnaires, Batterham et al. [13] evaluated dietary intake among patients with HIV infection, both with and without fat redistribution. In this study, no relationship was detected between dietary total or saturated fat intake and measured lipid, glucose, or insulin levels. In contrast, we demonstrate that high P:S fat ratio and low dietary fiber intake were strong independent predictors of hyperinsulinemia among patients with HIV infection and fat redistribution. The use of detailed diet history and of oral glucose tolerance testing as a sensitive measure of insulin resistance may explain, in part, the discrepancies between the 2 studies. Although still investigational, differential effects of polyunsaturated, mono-unsaturated, and saturated fat content of the diet have been shown to influence insulin action [19, 20]. Larsson et al. [14] demonstrated that habitual intake of a diet rich in polyunsaturated fatty acids (PUFAs) was associated with higher fasting insulin levels, decreased insulin sensitivity, and impaired glucose tolerance in nonobese postmenopausal women. In the current study, increased PUFA intake relative to intake of saturated fats by the patients with fat redistribution was associated with higher insulin AUC. This is consistent with the previous observations among non-hiv-infected individuals that increased PUFA may adversely effect insulin dynamics, particularly in insulin-resistant states. Dietary fiber was evaluated to determine the influence of fiber intake on insulin and cholesterol in patients with HIV infection and fat redistribution. In this study, low dietary fiber intake was associated with higher insulin AUC but was not associated with lipid abnormalities. Enhanced intake of dietary fiber has been shown to improve insulin sensitivity in healthy individuals [21] and to improve both glycemic control and lipid concentrations in individuals with type 2 diabetes mellitus [15]. We found that 53% of patients consumed less than the 20 g of dietary fiber per day recommended by the American Diabetes Association [22]. Although there is no consensus about whether there is a single lipodystrophy syndrome with various phenotypic expressions or several distinct syndromes of fat redistribution among HIV-infected patients, our data demonstrate that there is a spectrum of metabolic disturbances among patients with fat redistribution and that patients with mixed lipodystrophy may be more severely affected with regard to insulin resistance. However, there was no clear association between dietary habits and pattern of fat distribution in this cohort. Use of PIs was associated with increased insulin AUC. This is consistent with prior findings of the direct effect of shortterm administration of PIs on decreasing insulin sensitivity in HIV-negative healthy volunteers [23]. Although the use of PIs had the anticipated effect on insulin, it is important to note that dietary factors such as P:S fat ratio, fiber intake, and alcohol consumption were significant predictors of insulin AUC and cholesterol, independent of the effect of PI use. There are important potential limitations to this study. Pro- 714 CID 2001:33 (1 September) HIV/AIDS
Table 3. Patient characteristics, as related to pattern of fat redistribution. Variable Lipoatrophy (n p 15) Lipohypertrophy (n p 14) Mixed lipodystrophy (n p 56) Age, years 43.9 2.3 38.7 l.6 a 44.6 1.1 a Sex, no. male/no. female 13/2 7/7 42/14 Duration HIV infection, years 7.0 0.9 6.8 1.1 7.3 0.6 HIV load, copies/ml 14,834 6175 11,077 5913 3275 3053 Current PI use, % 53 71 73 Duration NRTI exposure, years 5.1 0.8 4.9 0.8 4.7 0.4 Body composition BMI, kg/m 2 23.0 0.5 b,c 29.7 1.3 a,c 26.4 0.5 a,b WHR 0.93 0.01 b,c 0.99 0.01 c 0.98 0.007 b Body fat, % 16.2 1.4 b,c 32.5 2.1 a,c 22.0 0.8 a,b Trunk fat, % 17.4 0.9 b,c 35.2 1.9 a,c 25.6 0.9 a,b Extremity fat, % 13.9 2.3 c 30.4 2.7 a,c 17.3 1.0 a Glucose, insulin, lipids Fasting glucose, mg/dl 92 5 93 4 92 2 Fasting insulin, mu/ml 11.9 1.7 b 13.6 1.0 a 23.3 2.6 a,b Glucose AUC, mg/dl 120 min 17,426 1258 17,850 1009 17,429 500 Insulin AUC, mu/ml 120 min 6235 618 b 8229 989 12,625 1185 b Cholesterol, mg/dl 227 18 216 11 230 8 LDL, mg/dl 134 12 128 9 135 6 HDL, mg/dl 35 2 b 42 4 42 1 b Triglyceride, mg/dl 433 152 299 78 324 30 Daily dietary intake Kilocalories 2639 226 2610 303 2420 89 Kilocalories/kg 38.8 3.1 b,c 29.9 2.4 c 31.6 1.2 b Fat, % 31.6 2.2 34.7 1.9 33.3 1.0 Carbohydrate, % 53.2 2.1 48.2 2.7 50.0 1.1 Protein, % 16.4 0.7 17.8 1.1 17.2 0.4 Alcohol, g 2.9 1.5 2.4 1.3 4.4 1.0 Dietary fiber, g 22.5 2.6 20.6 3.2 21.2 1.3 Polysaturated:saturated fat ratio 0.63 0.07 0.58 0.08 0.69 0.05 Cholesterol, mg 355 54 392 59 339 20 NOTE. Data are mean values SEM, unless otherwise indicated. AUC, area under the curve; BMI, body mass index; HDL, high-density lipoprotein; LDL, low-density lipoprotein; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; Poly, polyunsaturated; WHR, waist-to-hip ratio. a P!.05 for lipohypertrophy vs. mixed lipodystrophy. b P!.05 for lipoatrophy vs. mixed lipodystrophy. c P!.05 for lipoatrophy vs. lipohypertrophy. spectively collected dietary food records may provide a more accurate assessment of dietary behaviors than do patients recollections of the preceding month. Furthermore, it is possible that some of the observed dietary behaviors reflect changes made by patients in response to changes in body composition (e.g., increased energy per kilogram of body weight in patients with fat atrophy). Dietary factors may not be the primary cause of the metabolic disturbances, and the associations between diet and metabolic parameters detected in this study may represent the normal physiological relationship between these factors observed in healthy individuals. Nonetheless, the identification of expected associations, such as fiber and insulin resistance, and their persistence after fat redistribution and PI exposure were controlled for suggest a possible role for dietary modification for HIV-infected patients. Our data indicate a number of potential targets for dietary modification among patients with HIV infection and fat redistribution. In this regard, prospective dietary intervention studies are necessary to determine whether alteration in polyunsaturated fat intake and increased dietary fiber will be of benefit in treating the insulin resistance associated with HIV-related fat redistribution, and whether reduction of alcohol and cholesterol intake HIV/AIDS CID 2001:33 (1 September) 715
Table 4. Clinical characteristics and body composition of HIV-infected subjects with and without fat redistribution and healthy control subjects. Characteristic Yes (n p 55) Fat redistribution No (n p 17) Controls (n p 35) Age, years 43 1 40 1 41 1 Duration HIV infection, years 7.2 0.6 5.7 0.7 HIV load of!400 copies/ml, % 63 41 Current PI therapy, % 78 a 41 a BMI, kg/m 2 26.6 0.6 24.7 0.6 25.8 0.7 Waist-to-hip ratio 0.97 0.01 a,b 0.91 0.01 a 0.89 0.01 b Trunk fat, % 23.1 0.9 a 18.9 1.1 a,c 23.2 1.5 c Extremity fat, % 15.3 0.9 a,b 19.0 1.2 a 22.8 1.2 b Insulin, mu/ml 19.5 2.3 b 14.0 4.1 11.8 1.3 b Glucose, mg/dl 94 2 95 4 92 1 Cholesterol, mg/dl 231 8 a,b 166 9 a 182 7 b LDL, mg/dl 136 6 a,b 96 10 a 107 5 b HDL, mg/dl 38 1 a,b 44 2 a 48 2 b Triglyceride, mg/dl 379 49 a,b 149 36 a 106 11 b Energy, kcal 2705 111 2838 319 2408 181 Protein, % 17.4 0.4 16.5 1.1 16.3 0.5 Carbohydrate, % 50.0 1.1 b 47.2 2.4 c 54.6 1.2 b,c Fat, % 33.3 1.1 b 36.1 1.9 c 28.8 1.0 b,c Alcohol, g 3.7 0.8 5.5 2.6 6.6 1.8 Dietary fiber, g 23.9 1.5 20.2 1.7 24.1 2.0 Polysaturated:saturated fat ratio 0.66 0.05 0.52 0.09 0.54 0.04 Cholesterol, mg 381 26 b 459 87 c 250 25 b,c NOTE. Data are mean values SEM, unless otherwise indicated. AUC, area under the curve; BMI, body mass index; HDL, high-density lipoprotein; LDL, low-density lipoprotein; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; Poly, polyunsaturated; WHR, waist-to-hip ratio. a Patients with vs. patients without fat redistribution, P!.05. b Patients with fat redistribution vs. controls, P!.05. c Patients without fat redistribution vs. controls, P!.05. will be beneficial. Further investigation is needed to assess the impact of dietary intake and its modification on metabolic risk factors in HIV-associated fat redistribution. Acknowledgments We thank the nursing and dietary staffs of the Massachusetts Institute of Technology and the Massachusetts General Hospital Clinical Research Centers, Boston, for their dedicated patient care, and Gregory Neubauer, for his assistance in performing the radioimmunoassays. References 1. Carr A, Samaras K, Burton S, et al. A syndrome of peripheral lipodystrophy, hyperlipidemia and insulin resistance in patients receiving protease inhibitor therapy. AIDS 1998; 12:F51 8. 2. Saint-Marc T, Partisani M, Poizot-Martin I, et al. A syndrome of peripheral fat wasting (lipodystrophy) in patients receiving long-term nucleoside analogue therapy. AIDS 1999; 13:1659 67. 3. Gervasoni C, Ridolfo AL, Trifiro G, et al. Redistribution of body fat in HIV-infected women undergoing combined antiretroviral treatment. AIDS 1999; 13:465 71. 4. Carr A, Samaras K, Thorisdottir A, et al. Diagnosis, prediction and natural course of HIV-1 protease inhibitor associated lipodystrophy, hyperlipidemia, and diabetes mellitus: a cohort study. Lancet 1999; 353:2093 9. 5. Hadigan C, Meigs JB, Corcoran C, et al. Metabolic abnormalities and cardiovascular disease risk factors in adults with human immunodeficiency virus infection and lipodystrophy syndrome. Clin Infect Dis 2001; 32:130 9. 6. Lo JC, Mulligan K, Tai VW, et al. Buffalo hump in men with HIV infection. Lancet 1998; 351:867 70. 7. Miller KD, Jones E, Yanovski JA, et al. Visceral abdominal fat accumulation associated with use of indinavir. Lancet 1998; 351:871 5. 8. Dong KL, Bausserman LL, Flynn MM, et al. Changes in body habitus and serum lipid abnormalities in HIV-positive women on highly active antiretroviral therapy. JAIDS 1999; 21:107 13. 9. Roth VR, Kravcik S, Angel JB. Development of cervical fat pads following therapy with human immunodeficiency virus type-1 protease inhibitors. Clin Infect Dis 1998; 27:65 7. 716 CID 2001:33 (1 September) HIV/AIDS
10. Walli RK, Herfort O, Michl GM, et al. Treatment with protease inhibitors associated with peripheral insulin resistance and impaired glucose tolerance in HIV-1 infected patients. AIDS 1998; 12:F167 73. 11. Vigouroux C, Gharakhanian S, Salhi Y, et al. Diabetes, insulin resistance and dyslipidemia in lipodystrophic HIV-infected patients on highly active antiretroviral therapy (HAART). Diabetes Metab (Paris) 1999; 25:225 32. 12. Carr A, Miller J, Law M, Cooper D. A syndrome of lipoatrophy, lactic acidaemia and liver dysfunction associated with HIV nucleoside analogue therapy: contribution to protease inhibitor related lipodystrophy syndrome. AIDS 2000; 14:F25 32. 13. Batterham MJ, Garsia R, Greenop PA. Dietary intake, serum lipids, insulin resistance and body composition in the era of highly active antiretroviral therapy. Diet FRS Study. AIDS 2000; 14:1839 43. 14. Larsson H, Elmståhl S, Berglund G, Ahrén B. Habitual dietary intake versus glucose tolerance, insulin sensitivity and insulin secretion in post-menopausal women. J Intern Med 1999; 245:581 91. 15. Chandalia M, Garg A, Lutjohann D, et al. Beneficial effects of high dietary fiber intake in patients with type 2 diabetes mellitus. N Engl J Med 2000; 342:1392 8. 16. Lohman TG, Roche AF, Martoell R, eds. Anthropometric standardization reference manual. Champaign, IL: Human Kinetics Books, 1988. 17. Burke BS. The diet history as a tool in research. J Am Diet Assoc 1947; 23:1041 6. 18. Van Staveren WA, de Boer JO, Burema J. Validity and reproducibility of a dietary history method estimating the usual food intake during one month. Am J Clin Nutr 1985; 42:554 9. 19. Berry EM. Dietary fatty acids in the management of diabetes mellitus. Am J Clin Nutr 1997; 66(Suppl):991S 7S. 20. Storlien LH, Kriketos AD, Jenkins AB, et al. Does dietary fat influence insulin action? Ann NY Acad Sci 1997; 827:287 301. 21. Liljeberg HG, Akerberg K, Bjorck IM. Effect of glycemic index and content of indigestible carbohydrates of cereal-based breakfast meals on glucose tolerance at lunch in healthy subjects. Am J Clin Nutr 1999; 69:647 55. 22. Clinical practice recommendations, 2000: nutrition recommendations and principles for people with diabetes mellitus. American Diabetes Association. Diabetes Care 2000; 23(Suppl1):S43 6. 23. Noor M, Lo J, Mulligan K, et al. Metabolic effects of indinavir in healthy HIV-seronegative subjects. AIDS 2001; 15:F11 8. HIV/AIDS CID 2001:33 (1 September) 717