WIHS Metabolic Working Group Summary

Transcription

1 Metabolics

2 Prepared by: Phyllis Tien on behalf of the Metabolic Working Group WIHS Metabolic Working Group Summary Aims of the Metabolic Working Group: To examine the role of chronic inflammation, coinfections, and the menopausal transition in the progression of fat and metabolic changes (including disorders of glucose and lipid metabolism, bone and kidney injury) in HIV infected and HIV uninfected women. Accomplishments of the Metabolic Working Group The Metabolic Working Group was originally founded to examine issues related to lipodystrophy, diabetes, and dyslipidemia in women. In response to changes in the HIV field during WIHS IV, the Metabolic Working Group expanded to include studies of chronic kidney disease, prioritized investigation of bone disorders in the WIHS, and forged collaborations across WIHS working groups to study the relationship of adipose tissue and metabolic changes with vascular and liver injury, neurocognitive disorders, and behavioral changes. The Metabolic Working Group has published over 25 peer reviewed scientific publications and has been extremely successful in obtaining external NIH funds to support the WIHS IV metabolic agenda with over 10 NIH funded grants leveraging the wealth of metabolic data collected and specimens banked in WIHS. Another important function of the Metabolic Working Group has been the development of standardized criteria to define metabolic outcomes in WIHS that are consistent with national guidelines and to ensure that high quality data are collected. For example, diabetes is defined by confirmation of elevated fasting glucose, report of medication use and elevation of hemoglobin A1C consistent with the guidelines of the American Diabetes Association. Training, certification, and recertification every 2 years are required for the performance of anthropometry, and metabolic parameters are tested centrally. The following summarize some of the key metabolic papers published by the WIHS Metabolic Working Group. References are attached. Fat Distribution Changes in HIV-infected and HIV-uninfected women The WIHS published the first large longitudinal analysis of the incidence of lipoatrophy (or fat loss) and lipohypertrophy (or fat gain) in HIVinfected women from WIHS over a 30 month period from 1999 to HIV infected women had an increased risk of developing lipoatrophy when compared to HIV uninfected women, while the incidence of central lipohypertrophy (or fat gain) was the same in HIV infected and uninfected women. These findings were significant because they challenged previous notions that lipohypertrophy predominated in HIV infected women. Prior studies in women were small and did not include an HIV uninfected comparison group. Disorders in glucose metabolism in HIV infection The WIHS published the first large longitudinal study of the incidence of diabetes mellitus (DM) in women over a 5 year period from in the WIHS. We found that longer cumulative exposure to nucleoside reverse transcriptase inhibitors (NRTIs) was associated with an increased risk of DM in HIV infected women and that regular DM monitoring should be performed, because NRTIs comprised the backbone of effective antiretroviral therapy. Our publication was recognized by the Faculty of 1000, which identifies the most significant articles from biomedical research publications. Bone injury in WIHS The WIHS has one of the longest running studies to investigate changes in bone mineral density (BMD) in a subset of women who have undergone serial DXA scanning. These studies were initially supported by WIHS Core Science and supplemental funding from NIDA and NIAID, and most recently by an R01, which will provide up to 10 years of BMD data in some women to study changes over the menopausal transition. We found that HIV infected women that were mainly premenopausal had lower BMD than HIVuninfected women over a 5 year period in WIHS. Self reported fracture incidence was not increased in HIVinfected premenopausal women compared to HIV uninfected women. These studies support the need to

3 Prepared by: Phyllis Tien on behalf of the Metabolic Working Group investigate changes in bone mineral density and bone microarchitecture as women transition to menopause and how menopause affects fracture incidence in WIHS. HIV and Renal injury Using urine collected from the year 2000 visit of WIHS, we measured a kidney injury panel that included a traditional marker of kidney glomerular injury (albuminuria) and novel markers of kidney tubular injury (interleukin 18 and kidney injury molecule 1 (KIM 1)). We found that distinct risk factor profiles were associated with glomerular versus tubular injury patterns, with tubular injury a consequence of HIV related risk factors and glomerular injury a consequence of traditional kidney disease risk factors. Longitudinally, we found that higher levels of albuminuria, IL 18, and KIM 1 were all independently associated with faster rates of kidney function decline after adjustment for other known kidney disease risk factors. Future directions will be to evaluate whether this novel panel of kidney injury markers can detect early toxicity in HIV infected women, such as that related to antiretroviral use, and to understand the prognostic importance of changes in the injury markers. Grants and supplemental funds resulting from the Metabolic Working Group Funded: NIDA Supplement (Glesby) 2009 Metabolomics Approach to Biomarker Development of DM in HIV & HIV/HCV Coinfection K23 (Estrella) 2009 Burden of Chronic Kidney Disease in HIV infection (WIHS/MACS) R01 (Shlipak) 2010 The Aging Kidney in HIV Infection (WIHS/MACS) R01 (Tien) 2010 Visceral Obesity, HIV, HCV: Biologic Mediators of Hepatic Steatosis (SF WIHS/SFVA cohort) ARRA Supplement ( Tien) 2010 Fat, Metabolism & Liver Fibrosis (SF WIHS) ARRA Supplement (Gustafson) 2010 Adipose Tissue Hormones in HIV/AIDS and in HIV/AIDSrelated Cognitive Impairment (Brooklyn WIHS) R01 (Yin) 2010 HIV and the Menopausal Transition: Effects on Musculoskeletal Health NIDA Supplement (Tien) 2010 & 2011 Association of Opiate Use & Androgen Hormones with Disorders of Glucose Metabolism in HIV-infected & uninfected Women K23 (Sharma) 2012 HIV and Adiposity in women: Effects on Bone Metabolism R03 (Estrella) 2012 Association of Kidney Disease, Klotho and FGF23 with functional decline in HIV. R01 (Weiser) 2012 Pathways from Food Insecurity to Health and Treatment Outcomes in Women with HIV

4 CLINICAL SCIENCE Incidence of Lipoatrophy and Lipohypertrophy in the Women s Interagency HIV Study Phyllis C. Tien, MD, Stephen R. Cole, PhD, Carolyn Masters Williams, PhD, Rui Li, MS, Jessica E. Justman, MD, Mardge H. Cohen, MD, Mary Young, MD, Nancy Rubin, DO, Michael Augenbraun, MD, and Carl Grunfeld, MD, PhD Objective: To estimate the incidence of lipoatrophy and lipohypertrophy among HIV-infected and HIV-uninfected women from the Women s Interagency HIV Study. Design: Eight hundred fifteen women with semiannual data on selfreport of bidirectional change in body fat, anthropometric measurements, weight, and bioelectric impedance analysis were included in a 30-month incidence analysis. Methods: Lipoatrophy and lipohypertrophy in both peripheral (arms, legs, and buttocks) and central (waist, chest, and upper back) sites were defined by self-report of either a decrease or an increase in a body fat region over the previous 6 months that was confirmed by a corresponding change in anthropometric measurement. Results: Weight and total body fat increased in HIV-uninfected women but remained stable in HIV-infected women over 30 months. Among HIV-infected women, the incidence of peripheral (relative hazard, 2.1; 95% confidence interval [CI], ) and central (relative hazard, 1.9; 95% CI, ) lipoatrophy was about double that among HIV-uninfected women, after adjustment for age and race. Received for publication July 13, 2003; accepted September 23, From Department of Medicine, University of California at San Francisco, and San Francisco Veterans Affairs Medical Center, San Francisco, CA (Drs Tien, Grunfeld); Department of Epidemiology, Johns Hopkins University, Baltimore, MD (Drs Cole, Li); Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD (Dr Williams); Department of Medicine, Bronx Lebanon Hospital Center, Bronx, NY (Dr Justman); Department of Medicine, Cook County Hospital, Chicago, IL (Dr Cohen); Department of Medicine, Georgetown University Medical Center, Washington, DC (Dr Young); Department of Medicine, University of Southern California, Los Angeles, CA (Dr Rubin); and Department of Medicine, State University New York Downstate Medical Center, Brooklyn, NY (Dr Augenbraun). The WIHS is funded by the National Institute of Allergy and Infectious Diseases, with supplemental funding from the National Cancer Institute, the National Institute of Child Health & Human Development, the National Institute on Drug Abuse, the National Institute of Dental and Craniofacial Research, the Agency for Health Care Policy and Research, and the Centers for Disease Control and Prevention (U01-AI-35004, U01-AI-31834, U01-AI-34994, U01-AI-34989, U01-HD [NICHD], U01-AI , U01-AI-42590, M01-RR00071, and M01-RR00083). Reprints: Phyllis Tien, VAMC, Metabolism Section, 111F, 4150 Clement Street, San Francisco, CA ( ptien@medicine.ucsf.edu). Copyright 2003 by Lippincott Williams & Wilkins The incidence of peripheral lipohypertrophy appeared lower among HIV-infected women than among HIV-uninfected women (relative hazard, 0.8; 95% CI, ), while the incidence of central lipohypertrophy did not differ by HIV status. Of HIV-infected women with 2 of 4 lipodystrophy outcomes, most (81%) had combined peripheral and central lipoatrophy or combined peripheral and central lipohypertrophy. Only 14% of these women had both peripheral lipoatrophy and central lipohypertrophy Conclusions: These prospective data suggest that lipoatrophy, affecting both peripheral and central sites, predominates in HIVinfected women. The simultaneous occurrence of peripheral lipoatrophy and central lipohypertrophy was uncommon. Key Words: fat distribution, HIV, lipodystrophy, women (J Acquir Immune Defic Syndr 2003;34: ) There is increasing concern regarding changes in fat distribution associated with HIV infection, including the development of a buffalo hump, increased breast size, and abdominal girth as well as fat loss in the face, limbs, and buttocks These changes are often referred to as lipodystrophy syndrome. 9 Previous studies have used different definitions of the syndrome including a combination of peripheral lipoatrophy and central lipohypertrophy, isolated peripheral lipoatrophy, and isolated central lipohypertrophy. The lack of a consensus definition may partly explain the varying prevalence of lipodystrophy syndrome, between 2% and 84% in previous reports. 15 Studies on fat redistribution in HIV-infected women are few and imprecise due to small samples. A case definition of lipodystrophy has recently been proposed, but only 15% of the cohort studied were women. 16 Increases in breast size and abdominal girth and decreases in lower limb fat are most commonly reported in case series and studies of HIV-infected women. 8,12,13,17,18 Two studies of HIV-infected women have described the prevalence of lipodystrophy syndrome as 18% 12 and 11%. 13 In the first study, the occurrence of a body habitus change was defined as any report of an increase in abdominal girth, breast size, or peripheral wasting. 12 In the second study, the syndrome was defined as self-report of an increase in abdominal girth or breast size accompanied by wasting of the J Acquir Immune Defic Syndr Volume 34, Number 5, December

5 Tien et al J Acquir Immune Defic Syndr Volume 34, Number 5, December buttocks, legs, and thighs, confirmed by physical examination. 13 Because central lipohypertrophy and peripheral lipoatrophy have been seen in HIV-infected individuals, it has been proposed that the 2 conditions are associated with HIV infection. However, some of these changes, particularly an increase in abdominal girth, are also commonly seen in healthy aging adults. Examination of the findings of small studies of HIVinfected women shows that women reporting body changes tend to be older than women not reporting such changes. 12,19 For example, Engelson et al 19 found that women infected with HIV who reported truncal enlargement were on average 4 years older than those without truncal enlargement (44 vs. 40 years, respectively). The definition of lipodystrophy syndrome must be explored in a large cohort including HIV-uninfected controls and, specifically, women must be studied independently before contributing factors including antiretroviral therapy can be examined. We report findings of the prevalence and incidence of lipoatrophy and lipohypertrophy separately, using self-report of bidirectional changes in body fat confirmed by anthropometric measurements over a 30-month period between April 1999 and March 2002 among HIV-infected and -uninfected women from a large longitudinal cohort study. METHODS Women s Interagency HIV Study The Women s Interagency HIV Study (WIHS) is a prospective multisite study (New York, Chicago, Los Angeles, San Francisco, and Washington, D.C.) of the history of HIV infection in women with or at risk for HIV infection that has been conducted since At each semiannual WIHS visit, participants complete an extensive interviewer-administered questionnaire covering sexual practices, HIV-related symptoms, and demographic and psychosocial characteristics. In addition, participants are asked about changes in body fat in the face, neck, upper back, chest, arms, legs, buttocks, and waist since their last visit; if a change is reported, they are asked whether the change was a gain or a loss in fat. Blood specimens are obtained for determination of CD4 cell count, plasma HIV-1 RNA level, and other laboratory parameters, and physical examination is also performed. Anthropometric measurements, including circumferences of the triceps, thigh, hip, chest, and waist region and skinfolds of the triceps, thigh, subscapular and suprailiac regions, were performed by trained clinicians beginning at the 10th semiannual WIHS visit (April to October 1999). Anthropometric measurements were performed using the techniques described by the Third National Health and Examination Survey procedures. 21 Examiners completed an intense standardized training and certification program before collection of anthropometric data. A single exercise physiologist conducted training and certification of all examiners at each site. Skinfolds were measured using Harpenden calipers. The percent total body fat was determined semiannually by bioelectric impedance analysis ( Quantum [BIA-101Q]; RJL Systems, Inc., Detroit, MI). Appropriate informed consent was obtained from all participants. Guidelines for human experimentation in accordance with the US Department of Health and Human Services and the institutional review board of each participating institution were followed in the conduct of the study. Definitions of Lipoatrophy and Lipohypertrophy Incident body fat changes were classified as peripheral (ie, arms, legs, and buttocks) or central (ie, abdomen, chest, and upper back) and as either lipoatrophy (fat loss) or lipohypertrophy (fat gain). The body fat regions that comprise the classification of central and peripheral fat are consistent with those used in previous studies of lipodystrophy. 9,15,22 A change in body fat was defined as lipoatrophy or lipohypertrophy when self-report of any change in body fat in the arms, legs, buttocks, abdomen, or chest was confirmed by a measured change in circumference measurements of 0.7 cm in the same direction in the corresponding body region (ie, triceps, thigh, hip, waist, or chest) over the prior 6 months. We selected the criterion of 0.7 cm because this is the limit for accepted differences between repeated circumference measurements used in the WIHS. This limit is based on the standardized values for reproducibility of circumference measurements used in the Third National Health and Examination Survey. 21 Circumference measurements rather than skinfold measurements were used when both circumference and skinfold measurements were available for a given body site, because in our data we found that circumference measurements were more strongly correlated with self-reported changes in body fat for the corresponding body site at each visit than skinfold measurements. Circumference measurements were rarely missing; however, subscapular skinfold measurements were used for the upper back because no circumference measurements corresponded to that body site. Results were similar when an alternative definition of body fat changes was used based primarily upon skinfold measurements rather than circumference (data not shown). Self-report of a change in body fat in the upper back was confirmed by a measured change of 0.2 cm in the subscapular skinfold in the same direction. We used the criterion of 0.2 cm because this is the standard cutoff value for reproducibility of the subscapular skinfold measurement. 21 Because self-report of bidirectional changes that is, increase, decrease, or no change in body fat at each of the body regions was obtained, both lipoatrophy and lipohypertrophy could be separately assessed. A report of any decrease in a Lippincott Williams & Wilkins

6 J Acquir Immune Defic Syndr Volume 34, Number 5, December Lipoatrophy and Lipohypertrophy in WIHS peripheral site confirmed by anthropometric measurement was classified as peripheral lipoatrophy. A report of any peripheral increase confirmed by examination was classified as peripheral lipohypertrophy. Likewise, a confirmed report of any increase at a central site was classified as central lipohypertrophy, and a confirmed report of any central decrease was classified as central lipoatrophy. Therefore, a participant may have completed the 30-month follow-up without any body fat changes or could have had confirmed reports of 1 to all 4 of the possible changes. Statistical Methods Of the 2628 women enrolled in the WIHS, 1656 were seen at the 10th semiannual WIHS visit; 43 were excluded due to pregnancy in the prior 2 years, leaving a sample of 1613 women, of whom 1266 were HIV positive and 347 were HIV negative. Analyses of incident lipoatrophy and lipohypertrophy were restricted to the 815 women (605 HIV-infected and 210 HIV-uninfected women) who remained after exclusion of 760 (643 HIV-infected and 117 HIV-uninfected women) who reported any prevalent body fat change during the 6 months before the baseline (ie, 10th semiannual) WIHS visit, 37 (18 HIV-infected and 19 HIV-uninfected women) without followup, and 1 who seroconverted at visit 11. Seventeen women became pregnant during follow-up and were censored at the visit before the reported pregnancy. For the 605 HIV-infected and 210 HIV-uninfected women included in the incidence analyses, changes in body weight and percent total body fat (over the 6 semiannual study visits [visits 10 15]) were analyzed by HIV status using a generalized linear model with robust variance estimates to account for repeated measurements. 23 Differences in characteristics measured at the baseline visit by HIV status were compared using Wilcoxon rank sum test statistics for continuously distributed characteristics or Fisher exact tests for discrete characteristics (ie, race group). Times to first body fat changes were measured from the date of the baseline visit to the first of either the date of the visit at which the self-reported body fat change was confirmed or the date of censoring (final visit). Differences in times to lipohypertrophy or lipoatrophy by HIV status were compared using Cox proportional hazards regression models 24 adjusting for age and race group. Adjustment for age was made using restricted cubic splines with knots at the 5th, 50th, and 95th percentiles, thereby creating a smoothly joined piecewise polynomial that allows for a flexible association with the outcome variable. Fifty-nine (7%) of the 815 women were excluded from the multivariable models due to missing data. Cumulative incidence plots were based on the complement of the Nelson Aalen estimator 25 and were adjusted for age and race using inverse probability weights. 26 Analyses were conducted using SAS version 8 (SAS Institute Inc., Cary, NC). RESULTS Of the 1613 women seen at the 10th semiannual WIHS visit, 18% (230/1266) of HIV-infected women and 20% (68/347) of HIV-uninfected women (Fisher exact test, P = 0.53) reported a decrease in peripheral body fat in the prior 6 months. Seventeen percent (214/1266) of HIV-infected women versus 11% (38/347) of HIV-uninfected women (Fisher exact test, P < 0.01) reported a decrease in central body fat, 35% (442/1266) versus 24% (83/347) (Fisher exact test, P < 0.01) reported an increase in central body fat, and 22% (276/1266) versus 7% (24/347) (Fisher exact test, P < 0.01) reported an increase in peripheral body fat, respectively. A change in body fat during the 6 months before the baseline visit was reported by 760 women (643 HIV-infected and 117 HIVnegative women); these women were excluded from the incident analyses. Of the 815 women included in the incidence analyses of clinical lipoatrophy and lipohypertrophy over a 30-month period, which corresponded with the 10th through 15th WIHS semiannual visits, 605 were HIV positive and 210 were HIV negative. The HIV-infected and HIV-uninfected women had similar distributions of age, race, and height (Table 1). At the baseline visit (visit 10), the 605 HIV-infected women had markedly lower median body weight and mean body weight (68 vs. 75 kg [Wilcoxon rank sum test, P < 0.01] and 73 vs. 78 TABLE 1. Characteristics of 815 Women Included in Incidence Analyses at the Baseline Visit, between April and October 1999 Characteristic at Index Visit* HIV-Infected n = 605 HIV-Uninfected n = 210 P Race, n (%) 0.01 Black 363 (60) 132 (63) Hispanic 142 (23) 60 (29) White 100 (17) 18 (9) Age, y (range) 41 (35 46) 40 (34 45) 0.03 Height, cm (range) 163 ( ) 163 ( ) 0.15 Weight, kg (range) 68 (59 82) 75 (61 90) <0.01 Percent body fat (range) 26 (19 36) 31 (20 40) <0.01 CD4 cell count, / mm 3 (range) 366 ( ) 984 ( ) <0.01 Clinical AIDS, n (%) 266 (44) NA HIV-1 RNA level, log 10 copies/ml (range) 3.3 ( ) NA *Median (interquartile range), unless noted otherwise. Nineteen women reported other racial groups combined with black. Number of missing data for HIV-infected (height, 20; weight, 18; CD4 cell count, 8; HIV RNA level, 8, body fat, 155) and HIV-uninfected (height, 12; weight, 9; CD4 cell count, 7; body fat, 54) women. From Wilcoxon rank sum or Fisher exact test, as appropriate. NA indicates not available Lippincott Williams & Wilkins 463

7 Tien et al J Acquir Immune Defic Syndr Volume 34, Number 5, December FIGURE 1. Average body weight (left) and average percent total body fat (right) for women included in incidence analyses over 6 semiannual study visits. FIGURE 2. Cumulative incidences of lipoatrophy (left top and bottom) and lipohypertrophy (right top and bottom) adjusted for age and race group. kg [t test, P < 0.01], respectively) and percent total body fat (26% vs. 31%, respectively; Wilcoxon rank sum test, P < 0.01) compared with the 210 HIV-uninfected women. Of the 605 HIV-positive women, 266 (44%) had reported clinical AIDS. The median CD4 cell count was 366/mm 3 (interquartile range, /mm 3 ), and the median HIV-1 RNA level was 3.3 log 10 copies/ml. Over the 6 WIHS study visits, the mean weight for the HIV-infected women remained relatively stable at 73 kg (Fig. 1). In contrast, the mean weight for the HIV-uninfected women increased at a rate averaging 0.6 kg per visit (robust 95% confidence interval [CI], ). This same pattern was observed for the mean percent total body fat, with HIV-infected women relatively stable at 29% and HIV-uninfected women with an increased rate averaging 0.9% per visit (robust 95% CI, ). Two hundred fifty-one of the 605 HIV-infected women and 47 of 210 HIV-uninfected women reported loss of fat in the periphery over the 6 study visits, and the proportion of these women with a confirmed change of 0.7 cm in the circumference measurement between visits was similar (53% [134/251] vs. 51% [24/47], respectively; 2 test, P = 0.87). Two hundred twenty HIV-infected and 45 HIV-uninfected women reported loss of fat in central body sites, and the proportion of women with a confirmed change was similar (54% [118/220] vs. 56% [25/45], respectively; 2 test, P = 0.87). Two hundred two HIV-infected and 78 HIV-uninfected women reported gain of fat in the periphery, and the proportion of women with a confirmed change was similar (46% [93/202] vs. 53% [41/78], respectively; 2 test, P = 0.35). Three hundred twenty-one HIV-infected and 91 HIV-uninfected women reported gain of fat in central sites, and the proportion of women with a confirmed change was lower among the HIVinfected women than among the HIV-uninfected women (38% [121/321] vs. 55% [50/91], respectively; 2 test, P < 0.01). The median change in anthropometric measurements between visits for both HIV-infected and HIV-uninfected women exceeded the standardized cutoff values of 0.7 cm for circumference measurements and 0.2 cm for the subscapular skinfold measurement by >2-fold for every site except the arm, where the average median circumference change was 1.0 cm (data not shown). The 30-month cumulative incidences of peripheral lipoatrophy were 27% (162/605) among HIV-infected women and 13% (27/210) among HIV-uninfected women (Fig. 2). The relative hazard for HIV-infected women compared with HIVnegative women was 2.1 (95% CI, ) after adjustment for age and race group. The cumulative incidences of central lipoatrophy were 23% (139/605) among HIV-infected women and 13% (27/210) among HIV-uninfected women. The adjusted relative hazard was 1.8 (95% CI, ) when HIVinfected women were compared with HIV-uninfected women. The cumulative incidences of central lipohypertrophy were 28% (169/605) among HIV-infected women and 31% (65/210) among HIV-uninfected women. The adjusted relative hazard was 1.0 (95% CI, ) when HIV-infected women were compared with HIV-uninfected women. Finally, the cumulative incidences of peripheral lipohypertrophy were 18% (109/605) among HIV-infected women and 25% (53/210) among HIV-negative women. The adjusted relative hazard was 0.8 (95% CI, ) when HIV-infected women were compared with HIV-uninfected women. All relative hazards were essentially the same in analyses further adjusting for height, weight, and percent body fat. Results were similar when an alternative definition of body fat changes was used based primarily upon skinfold measurements rather than circumference. The risk of lipoatrophy or lipohypertrophy was Lippincott Williams & Wilkins

8 J Acquir Immune Defic Syndr Volume 34, Number 5, December Lipoatrophy and Lipohypertrophy in WIHS not substantially increased between women with 10-unit increases in body weight or total percent body fat at baseline when stratified by HIV status (data not shown). Of the 605 HIV-infected women, 311 ended the 30- month follow-up without any incident lipoatrophy or lipohypertrophy. Of the 294 women who did develop changes, 99 developed peripheral lipoatrophy alone, central lipoatrophy alone, central lipohypertrophy alone, or peripheral lipohypertrophy alone. One hundred forty-six women had 2 of the 4 outcomes (81% had either combined peripheral and central lipoatrophy or combined peripheral and central lipohypertrophy, whereas only 14% had peripheral lipoatrophy and central lipohypertrophy). Thirty-two HIV-infected women had 3 of the 4 outcomes, and 17 had all 4 outcomes. Of the 210 HIV-uninfected women, 121 ended the 30- month follow-up without any incident lipoatrophy or lipohypertrophy. Of the 89 women who did develop changes, 32 developed peripheral lipoatrophy alone, central lipoatrophy alone, central lipohypertrophy alone, or peripheral lipohypertrophy alone. Forty-eight women had 2 of the 4 outcomes (94% had either combined peripheral and central lipoatrophy or combined peripheral and central lipohypertrophy, and 4% had peripheral lipoatrophy and central lipohypertrophy). Four HIV-uninfected women had 3 of the 4 outcomes, and 5 had all 4 outcomes. DISCUSSION HIV-infected women had a lower mean body weight and a lower percent total body fat than did HIV-uninfected women. The mean body weight and percent total body fat did not change for HIV-infected women over the 30-month follow-up period. HIV-uninfected women, on the other hand, had increases in both mean body weight and total percent body fat. The findings for HIV-uninfected women are expected with aging, especially among this middle-aged cohort. However, this same increase in mean weight and percent total body fat was not found for HIV-infected women, which suggests an effect of HIV infection on weight and total body fat. The differences in weight and percent total body fat between HIV-infected and HIV-negative women may be explained by our findings of the incident analysis of lipoatrophy and lipohypertrophy. Over the 30-month follow-up period, a marked increase in the incidence of both peripheral and central lipoatrophy among HIV-infected women compared with HIV-uninfected women in the WIHS was observed, suggesting that subcutaneous lipoatrophy may predominate in HIV-infected women. There was no difference between HIV-infected and HIVnegative women in the risk of central lipohypertrophy, and the risk of peripheral lipohypertrophy appeared lower for HIVpositive women than for HIV-uninfected women. To our knowledge, these findings have not previously been reported. Most previous studies 10,12,13,22 presumed an association between peripheral lipoatrophy and central lipohypertrophy associated with HIV infection and only investigated lipoatrophy in peripheral sites and lipohypertrophy in central sites. In addition, most studies did not include an HIV-uninfected comparison group. The simultaneous presence of peripheral lipoatrophy and central lipohypertrophy in HIV-infected women was uncommon in our study. Using the classification of having peripheral lipoatrophy, central lipoatrophy, central lipohypertrophy, and/or peripheral lipohypertrophy, most women who could be classified into 2 of the 4 groups had either both peripheral and central lipoatrophy or both peripheral and central lipohypertrophy. This suggests that lipoatrophy and lipohypertrophy should be assessed separately when describing the prevalence or incidence, or exploring the etiology of lipodystrophy syndrome. In addition, both fat loss and fat gain should be assessed in both peripheral and central sites. Interestingly, when the 1613 women seen at the 10th semiannual WIHS visit were studied in a cross-sectional analysis using self-reported data of body fat change since the prior WIHS visit 6 months earlier, we did not find that HIVpositive women reported more peripheral fat loss in the prior 6 months than HIV-uninfected women. This clearly demonstrates the difficulty in interpreting data on clinical lipoatrophy and lipohypertrophy across studies of women given differences in study design, varying definitions of lipoatrophy and lipohypertrophy, and the lack of a HIV-uninfected comparison group in most other previous studies. The use of subjective measures, especially self-report of body fat change alone, may not be adequate in studies of lipoatrophy and lipohypertrophy, especially cross-sectional studies. Our use of a more rigorous definition in the incidence analyses, not previously done in other studies, of a report of body fat change confirmed by an anthropometric measurement in the same direction for a given body site, rather than confirmation by subjective examination, may provide a more accurate reflection of the incidence of lipoatrophy and lipohypertrophy among women. The comparison of HIV-infected and HIV-uninfected women is also important to understanding what changes are pathologic. There are a number of limitations to our study. First, measurements of regional fat by dual-energy x-ray absorptiometry, computed tomography, or magnetic resonance imaging were not performed. Therefore, it is unclear whether there was a preferential increase in visceral fat in HIV-infected women compared with HIV-uninfected women, as was suggested in a case series of HIV-infected men 3. However, the description of the clinical syndrome of lipoatrophy and lipohypertrophy was based upon self-report and examination changes in abdomen size and not on measured visceral fat when assessing central body regions. Second, as demonstrated, regardless of the presence or absence of lipoatrophy or lipohypertrophy, HIV-infected women had lower mean weight and percent total body fat at the index visit, which on average did not change in the 30-month follow-up period. Therefore, 2003 Lippincott Williams & Wilkins 465

9 Tien et al J Acquir Immune Defic Syndr Volume 34, Number 5, December using a dichotomous clinical classification of having either lipoatrophy or lipohypertrophy may be limited; continuous measurements of regional fat using dual-energy x-ray absorptiometry or other imaging modalities are needed. Third, women who may have had lipoatrophy or lipohypertrophy at some more distant time in the past were not captured in the analysis using our definition of a report of body fat change confirmed by a measured change in fat in the previous 6 months beginning in Our findings suggest that studies of women need to use a more rigorous definition of clinical lipoatrophy and lipohypertrophy. The direct measurement of fat should be considered. Analyses of determinants of lipoatrophy and lipohypertrophy, particularly antiretroviral therapies, can be misleading in the absence of the use of a more rigorous definition of lipoatrophy and lipohypertrophy and the use of a HIV-uninfected comparison group. Therefore, we did not include data on the determinants of antiretroviral therapy on lipoatrophy and lipohypertrophy in this analysis to emphasize the importance of defining the outcomes of lipoatrophy and lipohypertrophy using confirmation with objective measures. In conclusion, our findings suggest that an HIVassociated lipoatrophy syndrome affecting both peripheral and central sites may predominate in women. The presence of peripheral lipoatrophy in combination with central lipohypertrophy was uncommon in these women; therefore, lipoatrophy and lipohypertrophy should be assessed separately. These findings suggest that future research efforts should be focused on the contributions of antiretroviral therapy and other factors to the development of lipoatrophy in HIV-infected women. ACKNOWLEDGMENTS Data in this report were collected by the WIHS Collaborative Study Group with centers (principal investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); and Data Coordinating Center (Alvaro Muñoz). REFERENCES 1. Lo JC, Mulligan K, Tai VW, et al. Buffalo hump in men with HIV-1 infection. Lancet. 1998;351: Miller KK, Daly PA, Sentochnik D, et al. Pseudo-Cushing s syndrome in human immunodeficiency virus-infected patients. Clin Infect Dis. 1998; 27: Miller KD, Jones E, Yanovski JA, et al. Visceral abdominal-fat accumulation associated with use of indinavir. Lancet. 1998;351: Saint-Marc T, Touraine JL. Buffalo hump in HIV-1 infection. Lancet. 1998;352: Hengel RL, Watts NB, Lennox JL. Benign symmetric lipomatosis associated with protease inhibitors. Lancet. 1997;350: Stricker RB, Goldberg B. Fat accumulation and HIV-1 protease inhibitors. Lancet. 1998;352: Striker R, Conlin D, Marx M, et al. Localized adipose tissue hypertrophy in patients receiving human immunodeficiency virus protease inhibitors. Clin Infect Dis. 1998;27: Madge S, Kinloch-de-Loes S, Mercey D, et al. Lipodystrophy in patients naive to HIV protease inhibitors. AIDS. 1999;13: Carr A, Samaras K, Burton S, et al. A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors. AIDS. 1998;12:F51 F Carr A, Samaras K, Thorisdottir A, et al. Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor- associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: a cohort study. Lancet. 1999;353: 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: 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 (HAART). J Acquir Immune Defic Syndr. 1999;21: Gervasoni C, Ridolfo AL, Trifiro G, et al. Redistribution of body fat in HIV-infected women undergoing combined antiretroviral therapy. AIDS. 1999;13: Saint-Marc T, Partisani M, Poizot-Martin I, et al. Fat distribution evaluated by computed tomography and metabolic abnormalities in patients undergoing antiretroviral therapy: preliminary results of the LIPOCO study. AIDS. 2000;14: Safrin S, Grunfeld C. Fat distribution and metabolic changes in patients with HIV infection. AIDS. 1999;13: Carr A, Emery S, Law M, et al. An objective case definition of lipodystrophy in HIV-infected adults: a case-control study. Lancet. 2003;361: Herry I, Bernard L, de Truchis P, et al. Hypertrophy of the breasts in a patient treated with indinavir. Clin Infect Dis. 1997;25: Ridolfo AL, Gervasoni C, Bini T, et al. Body habitus alterations in HIVinfected women treated with combined antiretroviral therapy. Aids Patient Care STDS. 2000;14: Engelson ES, Kotler DP, Tan Y, et al. Fat distribution in HIV-infected patients reporting truncal enlargement quantified by whole-body magnetic resonance imaging. Am J Clin Nutr. 1999;69: Barkan SE, Melnick SL, Preston-Martin S, et al. The Women s Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology. 1998;9: National Center for Health Statistics. The Third National Health and Examination Survey Reference Manuals and Reports p. CD-ROM. 22. Martinez E, Mocroft A, Garcia-Viejo MA, et al. Risk of lipodystrophy in HIV-1-infected patients treated with protease inhibitors: a prospective cohort study. Lancet. 2001;357: Zeger SL, Liang KY. Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42: Cox D. Regression models and life tables. J R Statist Soc. 1972;34: Aalen O. Nonparametric inference for a family of counting processes. Ann Stat. 1978;6: Cole S, Hernan M. Adjusted survival curves using inverse probability-oftreatment weights (abstract). Am J Epidemiol. 2002;155:S Lippincott Williams & Wilkins

10 Antiretroviral therapy exposure and incidence of diabetes mellitus in the Women s Interagency HIV Study Phyllis C. Tien a, Michael F. Schneider b, Stephen R. Cole b, Alexandra M. Levine c, Mardge Cohen d, Jack DeHovitz e, Mary Young f and Jessica E. Justman g Objective: To determine the incidence of diabetes mellitus (DM) in a nationally representative cohort of HIV-infected women and a comparison group of HIV-uninfected women. Design: A prospective study between October 2000 and March 2006 of 2088 participants from the Women s Interagency HIV Study who did not have evidence of DM at enrollment (1524 HIV infected and 564 HIV uninfected). Methods: Incident DM was defined as either having fasting glucose 1.26 g/l, reporting antidiabetic medication, or reporting DM diagnosis (with subsequent confirmation by fasting glucose 1.26 g/l or reported antidiabetic medication); all were assessed at semi-annual study visits. Results: DM developed in 116 HIV-infected and 36 HIV-uninfected women over 6802 person-years. HIV-infected women reporting no recent antiretroviral therapy had a DM incidence rate of 1.53/100 person-years; those reporting HAART containing a protease inhibitor (PI) had a rate of 2.50/100 person-years and those reporting non-picontaining HAART a rate of 2.89/100 person-years. None of these rates differed from the HIV-uninfected women (1.96/100 person-years) substantially or beyond levels expected by chance. Among HIV-infected women, longer cumulative exposure to nucleoside reverse transcriptase inhibitors (NRTI) was associated with an increased risk of DM incidence compared with no NRTI exposure: relative hazard (RH) 1.81 [95% confidence interval (CI), ] for > 0 to 3 years exposure and RH 2.64 (95% CI, ) for > 3 years exposure. Conclusion: Longer cumulative exposure to NRTI was associated with increased DM incidence in HIV-infected women. Regular DM monitoring is advisable because NRTI form the backbone of effective antiretroviral therapy. AIDS, 21: Keywords: antiretroviral therapy, diabetes mellitus, fasting glucose, HIV, nucleoside reverse transcriptase inhibitor From the a Department of Medicine, University of California, San Francisco, and San Francisco Veterans Affairs Medical Center, San Francisco, California, the b Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, the c Department of Medicine, University of Southern California, Los Angeles, California, the d CORE Center/Stroger Hospital of Cook County, Chicago, Illinois, the e Department of Medicine, State University of New York-Downstate Medical Center, Brooklyn, New York, the f Department of Medicine, Georgetown University Medical Center, Washington, DC, and the g Departments of Epidemiology and Medicine, Columbia University, New York, USA. Correspondence to Dr. P. Tien, VAMC, Infectious Disease Section, University of California San Francisco, 111W, 4150 Clement St, San Francisco, CA 94121, USA. ptien@ucsf.edu Received: 28 March 2007; revised: 30 April 2007; accepted: 15 May ISSN Q Lippincott Williams & Wilkins 1739 Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

11 1740 AIDS 2007, Vol 21 No 13 Introduction Diabetes mellitus (DM) has been of increased concern in HIV-infected individuals since an association between protease inhibitors (PI) and hyperglycemia, impaired glucose tolerance, and insulin resistance was first reported [1 4]. Recent studies [5,6] suggesting an increased risk of premature cardiovascular disease in HIV-infected individuals have heightened the need to understand the relation of HIV infection with risk of DM, a primary risk factor for cardiovascular disease. Prior publications on the risk of DM to date have been limited by small numbers of incident DM cases [7], reliance on self-report of incident DM [8], or lack of an HIV-negative comparison group [9]. This study uses a large well-characterized prospective cohort of HIV-infected and HIV-uninfected women followed over 5 1 / 2 years (October 2000 to March 2006) to examine the association of both the type and duration of antiretroviral therapy (ART) exposure on DM incidence determined, in part, using fasting glucose measurements. Methods Study population The Women s Interagency HIV Study (WIHS) is a multicenter prospective cohort study that was established in 1994 to investigate the progression of HIV in women with and at risk for HIV. A total of 3766 women (2791 HIV-infected and 975 HIV uninfected) were enrolled in either (n ¼ 2623) or (n ¼ 1143) from six United States cities (Bronx, Brooklyn, Chicago, Los Angeles, San Francisco, and Washington DC). Baseline sociodemographic characteristics and HIV risk factors were similar between HIV-infected and HIV-uninfected women [10,11]. Participating institutions institutional review boards approved study protocols and consent forms, and each study participant gave written informed consent. Every 6 months, participants complete a comprehensive physical examination, provide biological specimens for CD4 cell count and HIV RNA viral load determination, and complete an interviewer-administered questionnaire, which collected information on demographics, disease characteristics, and specific ART use. Beginning in October 2000, fasting glucose was measured after participants had fasted for at least 8 h. Of the 2859 women with a study visit between October 2000 and March 2006, 2554 had at least one fasting glucose measurement; the first study visit with fasting glucose data available will be referred to as the index study visit. Of the 2554 women with at least one fasting glucose measurement, 67 were excluded because of either a positive (n ¼ 60) or a missing (n ¼ 7) report of pregnancy at the index visit. Of the remaining 2487 women, 280 (202 of the 1808 HIV-infected and 78 of the 679 HIVuninfected women) were excluded owing to prevalent DM [fasting glucose 1.26 mg/l at index (n ¼ 119), a history of self-report DM prior to or at index (n ¼ 235), or antidiabetic medication use reported prior to or at index (n ¼ 126)]. Of the remaining 2207 women, 110 had no follow-up visits with data to determine incident DM after the index visit, resulting in a total of 2097 women. Nine women were excluded because they were HIV uninfected at study entry and seroconverted during follow-up. The final study population consisted of 2088 women (1524 HIV-infected and 564 HIV-uninfected women). The 2088 women contributed a total of study visits of followup (including the index visit). Fasting glucose data were available at 9537 (63%) of these study visits. The median number of study visits with glucose data for the 2088 women was 4 [interquartile range (IQR), 3 6]. Ascertainment of diabetes mellitus Incident DM was considered to have occurred at the first follow-up visit after the index visit at which fasting glucose was 1.26 mg/l, antidiabetic medication use was reported, or DM was reported with subsequent confirmation by either a fasting glucose 1.26 mg/l or report of antidiabetic medication use. Assessment of antiretroviral therapy use At each semiannual study visit, participants were shown photo-medication cards and were asked the names of specific ART medications used since their prior study visit. The WIHS uses a standard definition of HAART [12] adapted from the Department of Health and Human Services/Kaiser Panel guidelines [13]. All non-haart combination therapy regimens were classified as combination therapy; reports of a single nucleoside reverse transcriptase inhibitor (NRTI), PI, or nonnucleoside reverse transcriptase inhibitor (NNRTI) were classified as monotherapy. In analyses, ART regimens were classified as no therapy, monotherapy or combination therapy, PI-based HAART, and non-pi-based HAART. At each study visit from index through the end of followup, the cumulative drug-years of exposure to NRTI, PI, and NNRTI were determined. The time of exposure (in years) to each class of ART (i.e., NRTI, PI, NNRTI) at each study visit was defined as the product of the number of drugs which were reported in a given class since the last study visit and 0.5, since semiannual visits occurred approximately half a year apart. Drugs within a class were considered exchangeable and additive. Finally, the cumulative drug-years of exposure to the four most commonly used NRTI (i.e., zidovudine, abacavir, stavudine, and lamivudine) were determined in our cohort from index to the end of follow-up. For analyses, categories of cumulative drug exposure were defined as no exposure (reference category), exposed and less than Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

12 ART and diabetes in women Tien et al or equal to the median, and exposed and greater than the median. In adjusted cumulative drug-years analyses, adjustment was also made for cumulative amount of ART use reported prior to the index visit. Statistical analyses Comparisons of continuous and categorical characteristics among HIV-infected and HIV-uninfected women at the index visit were made by Wilcoxon rank sum test or Pearson x 2 test, respectively. Time at risk for incident DM was calculated from the date of the index visit through either the date of DM (for those with incident DM) or the date of the last study visit (for those without incident DM). The crude incidence rate of DM for each ART exposure group was calculated by dividing the number of incident DM cases by the total person-time at risk for DM. Cox proportional hazards models [14] were approximated using pooled logistic regression models [15]. Relative hazards (RH) were used as a measure of association and 95% confidence intervals (CI) as a measure of precision. Regression models were used to adjust for the potentially confounding effects of variables measured at the index visit. Specifically, adjustment was made for age, race, body mass index, smoking status (current versus noncurrent), hepatitis C virus antibody status, family history of DM, menopause, CD4 cell count (set to zero for HIVuninfected women so that the association between CD4 and incident DM was only assessed in HIV-infected women), history of ART use prior to the index visit, and enrollment cohort ( versus ). To explore changes in body size as a possible pathway through which ART may cause alterations in the rates of incident DM, separate analyses further adjusted for time-varying change in hip size. The proportional hazards assumption was explored by estimating the interaction between ART exposure and time. There was no strong evidence of nonproportional hazards among the different exposure groups (P ¼ 0.652); however, owing to the relatively few events, there was only modest statistical power to detect nonproportional hazards. All analyses were conducted using SAS version 9 (SAS Institute, Cary, North Carolina, USA). Results At the index visit, HIV-infected women were older, more likely to be postmenopausal, and more likely to be positive for hepatitis C virus antibodies than HIVuninfected women (Table 1). The racial distribution and the proportion with a family history of DM were similar. Body mass index and hip circumference were lower in HIV-infected women, and waist circumference was similar. The 2088 women at risk for DM contributed a total of 6802 person-years (4962 HIV-infected, 1840-uninfected) of follow-up from index until either incident DM or the Table 1. Characteristics of 2088 participants of the Women s Interagency HIV Study at index visit by HIV status. Characteristic HIV-infected women HIV-uninfected women P value a Total No Median age [years (IQR)] 39.2 ( ) 34.3 ( ) < Race [No. (%)] African American 858 (56) 329 (58) Hispanic 425 (28) 155 (27) Caucasian b 241 (16) 80 (14) Current smoker [No. (%)] c 717 (47) 292 (52) Median body mass index [kg/m 2 (IQR)] c 26.8 ( ) 28.0 ( ) Median waist size [cm (IQR)] c 87.1 ( ) 86.5 ( ) Median hip size [cm (IQR)] c 98.8 ( ) ( ) < Menopausal [No. (%)] c 260 (17) 50 (9) < Hepatitis C virus antibody positive [No. (%)] c 447 (30) 89 (16) < Family history of diabetes [No. (%)] c 439 (30) 149 (27) Clinical AIDS [No. (%)] 512 (34) NA NA Median HIV RNA [log 10 copies/ml (IQR)] c 2.92 ( ) NA NA Median CD4 count [cells/ml (IQR)] c 428 ( ) NA NA Median nadir CD4 count [cells/ml (IQR)] c 260 ( ) NA NA Antiretroviral therapy naive [No. (%)] 241 (16) NA NA IQR, interquartile range; NA, not applicable. a P value from Pearson x 2 test of overall association when percentages are compared, P value from Wilcoxon rank sum test when medians are compared. b Includes 49 (3%) and 25 (4%) Asian, Pacific Islander, Native American, Alaskan, other among HIV-infected and HIV-uninfected women, respectively. c Missing data: current smoking, 2 HIV-infected women; body mass index, 57 HIV-infected women and 14 HIV-uninfected women; waist size, 318 HIV-infected women and 115 HIV-uninfected women; hip size, 325 HIV-infected women and 116 HIV-uninfected women; menopausal status, 18 HIV-infected women and 4 HIV-uninfected women; hepatitis C infection, 31 HIV-infected women and 7 HIV-uninfected women; family history of diabetes, 46 HIV-infected women and 18 HIV-uninfected women; HIV RNA, 12 HIV-infected women; CD4 cell count, 27 HIV-infected women; nadir CD4 cell count, 90 HIV-infected women. Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

13 1742 AIDS 2007, Vol 21 No 13 Table 2. Risk of diabetes mellitus among 2088 women a by HIV infection status and antiretroviral therapy reported. Incident DM cases Person-years Incidence rate (per 100 person-years) Unadjusted relative hazard (95%CI) Adjusted relative hazard (95% CI) b HIV-uninfected women HIV-infected women: No therapy ( ) 0.66 ( ) Mono/combination therapy ( ) 1.68 ( ) HAART with PI ( ) 1.15 ( ) HAART, no PI ( ) 1.29 ( ) DM, diabetes mellitus; CI, confidence interval; PI, protease inhibitor. a Included 1524 HIV-infected and 564 HIV-uninfected women, who contributed a total of person-visits (9506 among HIV-infected and 3560 among HIV-uninfected women) following index visit; 172 HIV-infected women contributing 883 person-visits and 41 HIV-uninfected women contributing 215 person-visits were excluded from multivariable analysis because of missing data. b Adjusted for race, baseline hepatitis C antibody status, family history of DM, and index values of age, smoking status, body mass index, CD4 cell count (set to zero for HIV-uninfected women), and menopause. last study visit. The median fasting glucose at index was 830 mg/l (IQR, ) and was the same for HIVinfected and HIV-uninfected women. Over the followup period, 152 (116 HIV-infected, 36 uninfected) women developed DM, 121 died (114 HIV-infected, 7 uninfected) without developing DM, and 105 (81 HIVinfected, 24 uninfected) were lost to follow up prior to April 2005 free of DM. The remaining 1710 (1213 HIVinfected, 497 uninfected) women completed follow-up between April 2005 and March 2006 without DM. Of the 152 incident DM cases, 99 (72 HIV-infected and 27 uninfected) were defined as having a fasting glucose concentration 1.26 mg/l; 46 (39 HIV-infected and 7 uninfected) were based on reported antidiabetic medication use, and 7 (5 HIV-infected and 2 uninfected) were based on reported DM diagnosis subsequently confirmed by either a fasting glucose concentration 1.26 mg/l or reported antidiabetic medication use. Of the 46 classified by report of antidiabetic medications, 29 (26 HIVinfected and 3 HIV-uninfected) women reported antidiabetic medication specifically to lower blood sugar. For the remaining 17 (13 HIV-infected and 4 HIVuninfected) women, it was not possible to determine whether the antidiabetic medications were used to treat DM or for other reasons such as to treat pre-dm or fat distribution changes associated with HIV. Among the 116 HIV-infected women with incident DM: 25 reported using no ART immediately prior to the incident visit; 9 reported using ART (but not HAART) immediately prior to the incident visit; 41 reported a PIcontaining HAART regimen; and 41 reported a non-pi containing HAART regimen. Of the nine women on a non-haart regimen, three reported NRTI use only; four reported a single NRTI with at least one PI; one reported one NNRTI; and one reported an NNRTI and PI combination. Of the 41 women reporting a PIcontaining HAART regimen, the PI used included nelfinavir (16; one also with ritonavir), lopinavir (14), indinavir (6; two also with ritonavir). Of the 41 women reporting a non-pi containing HAART regimen, 14 reported nevirapine, 13 reported efavirenz, and 14 reported either abacavir or tenofovir as a third NRTI without an NNRTI. Compared with the DM incidence rate in HIVuninfected women (1.96/100 person-years): HIVinfected women reporting no recent ART had a DM incidence rate of 1.53/100 person-years; those reporting monotherapy or combination therapy at the visit prior to the incident event had a DM incidence rate of 3.40/100 person-years; those reporting PI-containing HAART had a rate of 2.50/100 person-years; and those reporting non-pi containing HAART a rate of 2.89/ 100 person-years. In adjusted analyses (Table 2), none of the associations between treatment group and DM incidence was substantial nor were any associations precise enough to rule out chance. Among HIV-infected women, after adjustment for potentially confounding factors measured at index (Table 3), longer cumulative exposure to NRTI was associated with an increased DM incidence compared with no NRTI exposure [RH, 1.81 (95% CI, ) for 0 to 3 years NRTI exposure; RH, 2.64 (95% CI, ) for > 3 years NRTI exposure]. Neither cumulative exposure to PI nor cumulative exposure to NNRTI was associated with DM incidence in adjusted analyses. Contrary to our mediation hypothesis, further adjustment for time-varying changes in hip size did not decrease the association between cumulative exposure to NRTI and incident DM (comparison of > 3 drug-years versus 0 drug-years: adjusted RH, 3.25; 95% CI, ). Because NRTI appeared most associated with increased incident DM, the rate of DM among HIV-infected women reporting the four most commonly used NRTI (zidovudine, abacavir, stavudine, and lamivudine) was also examined. No association between cumulative exposure to zidovudine, abacavir or stavudine and incident DM was observed in unadjusted or adjusted analyses (Table 4). However, cumulative exposure of > 1 year to lamivudine was associated with a nearly three-fold Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

14 ART and diabetes in women Tien et al Table 3. Risk of diabetes mellitus among 1524 HIV-infected women a by cumulative amount of each class of antiretroviral therapy reported at each visit after index. Incident DM cases Person-years Incidence rate (per 100 person-years) Unadjusted relative hazard (95% CI) Adjusted relative hazard (95% CI) b,c Cumulative number of NRTI-years reported after index > 0 to ( ) 1.81 ( ) > ( ) 2.64 ( ) Cumulative number of PI-years reported after index > 0 to ( ) 1.11 ( ) > ( ) 0.82 ( ) Cumulative number of NNRTI-years reported after index > 0 to ( ) 1.32 ( ) > ( ) 0.79 ( ) DM, diabetes mellitus; CI, confidence interval; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor. a Contributed a total of 9506 person-visits with antiretroviral therapy data; 172 HIV-infected women contributing 883 person-visits were excluded from multivariable analysis because of missing data. b Adjusted for cumulative number of NRTI, PI, or NNRTI reported from baseline through study visit prior to index, WIHS cohort status (1994 5or ), race, baseline hepatitis C antibody status, family history of DM, and index values of age, smoking status, body mass index, CD4 cell count, and menopause. c Each of the six pairs of two-way interactions were nonsignificant (P 0.05) when placed one pair at a time into the adjusted model; therefore no interaction terms were included in adjusted model. increase in the rate of DM incidence after adjustment for covariates including cumulative use of PI and NNRTI. Discussion In the largest prospective study to date among HIVinfected women of the risk of DM using fasting glucose measurements, we found that longer cumulative exposure to NRTI was associated with increased DM incidence compared with no NRTI exposure. No association between cumulative exposure to PI or NNRTI and incident DM was observed. Our findings of an association between longer cumulative NRTI exposure and DM incidence in HIV-infected women are consistent with those from a large study of Table 4. Risk of diabetes mellitus among 1524 HIV-infected women a by cumulative amount of zidovudine, abacavir, stavudine, and lamivudine reported at each visit after index. Incident DM cases Person-years Incidence rate (per 100 person-years) Unadjusted relative hazard (95% CI) Adjusted relative hazard (95% CI) b Cumulative number of person-years reporting zidovudine after index > 0 to ( ) 0.76 ( ) > ( ) 0.61 ( ) Cumulative number of person-years reporting abacavir after index > 0 to ( ) 0.90 ( ) > ( ) 1.17 ( ) Cumulative number of person-years reporting stavudine after index > 0 to ( ) 0.92 ( ) > ( ) 0.90 ( ) Cumulative number of person-years reporting lamivudine after index > 0 to ( ) 1.42 ( ) > ( ) 2.81 ( ) DM, diabetes mellitus; CI, confidence interval; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; NNRTI, nonnucleoside reverse transcriptase inhibitor. a Contributed a total of 9506 person-visits with antiretroviral therapy data; 172 HIV-infected women contributing 883 person-visits were excluded from multivariable analysis because of missing data. b Adjusted for cumulative number of nucleoside reverse transcriptase inhibitor, protease inhibitor, and nonnucleoside reverse transcriptase inhibitor drugs reported from WIHS baseline through index, cumulative number of protease inhibitors and nonnucleoside reverse transcriptase inhibitors reported from index to 6 months prior to each study visit, WIHS cohort status ( or ), race, baseline hepatitis C infection status, family history of DM, and index values of age, smoking status, body mass index, CD4 cell count, and menopause. Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

15 1744 AIDS 2007, Vol 21 No 13 HIV-infected men [16], where an association between cumulative NRTI exposure and insulin resistance was observed. In that study, the association between NRTI and incident DM was not studied owing to the small number of incident DM events. Recent findings of another large collaborative cohort study, which showed an increased risk of DM per year of NRTI use, also agree with our findings [9]. When the individual NRTI drugs were studied in these cohorts, however, a variety of NRTI were implicated. In one study, stavudine and lamivudine were independently associated with insulin resistance [16]. In another study, stavudine, zidovudine, and didanosine were associated with DM [9]. Among the NRTI, we found an association between increased DM incidence and the cumulative exposure to lamivudine. The mechanism by which NRTI may cause disorders in glucose metabolism is unclear. Some have postulated a role of NRTI-induced mitochondrial dysfunction and insulin resistance [17]. NRTI drugs (especially stavudine) have been implicated as a cause of lipoatrophy [18,19], and lipoatrophy has been associated with insulin resistance [20]. However, we observed that the association between NRTI and incident DM was not attenuated by further adjustment for changes in hip circumference (which we have previously demonstrated to be the most affected body site in this cohort [21]). We did not find a notable association between PI use and incident DM despite prior studies reporting an association between this group and disorders of glucose metabolism. This may be partially explained by the changing patterns in PI use. Indinavir, commonly used early in the HAART era, has been associated with hyperglycemia and decreased insulin sensitivity in HIVinfected individuals [3,22] and in healthy HIV-uninfected volunteers [23,24]. An early prospective study of selfreported DM incidence from the WIHS cohort (median follow-up of 2.9 years) demonstrated a three-fold rate of DM incidence among HIV-infected women reporting PI use compared with HIV-uninfected women [8]. Indinavir was the most frequently used PI during that time period (1994 8). A recent study in HIV-negative volunteers given a single dose of lopinavir/ritonavir, a commonly used PI in our study population, demonstrated an acute decrease in insulin sensitivity [25] but found no change in insulin sensitivity from baseline after 4 weeks on therapy [26]. In contrast, insulin sensitivity did not normalize after 4 weeks of indinavir use [23,24]. Therefore, the lack of an association between PI use and DM in our study may reflect the wider use of lopinavir/ritonavir, which may have less of an impact on insulin sensitivity than indinavir. Prior observations indicated that recent PI exposure was associated with disorders in glucose metabolism; consequently, we initially examined the risk of recent exposure to different types of ART regimen on DM incidence. We found that any recent use of ART, regardless of being monotherapy, combination therapy, or PI-containing or non-pi containing HAART regimen, appeared to be associated with increased DM incidence when compared with HIV-uninfected women, although the associations were not statistically significant. The magnitude of the association was not as large as previously reported. One prospective study of men over a median follow up of 2.3 years from 1999 to 2003 found a fourfold increase in DM incidence (24 cases in 506 personyears) among those reporting HAART compared with 361 HIV-uninfected men (10 cases in 709 person-years) [7]. Unlike the present study, that study did not adjust for key risk factors associated with DM, including family history of DM and hepatitis C infection, which have been shown to be independently associated with DM in several large epidemiological studies [27,28]. Among the limitations of our study is the definition of DM using a single fasting glucose measurement rather than confirmation on a subsequent day as recommended by the American Diabetes Association. In some cases, we also were not able to distinguish whether antidiabetic medications were used to treat DM or for other reasons such as to treat pre-dm or fat distribution changes associated with HIV. While we had many more incident DM cases than prior studies, the limited number of DM cases restricted our ability to explore possible synergistic effects of individual ART or interactions across ART classes. As with all observational studies, our findings are subject to possible unmeasured confounding. Finally, the design of this cohort study, which examines participants every 6 months, only allows us to coarsely definite cumulative exposure to specific ART, and, therefore, our results with respect to the association between specific ART drugs and DM should be interpreted with caution. In summary, we conclude that cumulative exposure to NRTI, but not PI or NNRTI, was associated with increased DM incidence, using the definition stated in our methods. NRTI remain the backbone of effective ART, and so regular monitoring of fasting glucose levels in HIV-infected patients is warranted. Study of the biological mechanisms by which NRTI might induce disorders in glucose metabolism is a priority. Acknowledgements Data in this manuscript were collected by the WIHS Collaborative Study Group with centers (principal investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, New York (Howard Minkoff); Washington DC Metropolitan Consortium (Mary Young); the Connie Wofsy Study Consortium of Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

16 ART and diabetes in women Tien et al Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); Data Coordinating Center (Stephen J. Gange). Sponsorship: The WIHS is funded by the National Institute of Allergy and Infectious Diseases, with additional supplemental funding from the National Cancer Institute, the National Institute of Child Health and Human Development, the National Institute on Drug Abuse, the Agency for Healthcare Policy and Research, the National Center for Research Resources, and the Centers for Disease Control and Prevention. U01-AI-35004, U01-AI-31834, U01-AI-34994, U01- AI-34994, U01-AI-34989, U01-HD (NICHD), U01-AI-34993, U01-AI-42590, M01-RR00079, and M01-RR Dr Tien is supported by the National Institute of Allergy and Infectious Diseases through K23 AI and currently has a research grant from Gilead. The funding sources had no role in the design and conduct of the study; collection, management, analysis and interpretation of the data; and preparation and review of the manuscript. Dr Tien had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. References 1. Carr A, Samaras K, Burton S, Law M, Freund J, Chisholm DJ, et al. A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors. AIDS 1998; 12:F51 F Behrens G, Dejam A, Schmidt H, Balks H, Brabant G, Körner T, et al. Impaired glucose tolerance, beta cell function and lipid metabolism in HIV patients under treatment with protease inhibitors. AIDS 1999; 13:F63 F Dube MP, Johnson DL, Currier JS, Leedom JM. Protease inhibitor-associated hyperglycaemia. Lancet 1997; 350: Walli R, Goebel FD, Demant T. Impaired glucose tolerance and protease inhibitors. Ann Intern Med 1998; 129: Mary-Krause M, Cotte L, Simon A, Partisani M, Costagliola D. Increased risk of myocardial infarction with duration of protease inhibitor therapy in HIV-infected men. AIDS 2003; 17: Friis-Moller N, Sabin CA, Weber R, d Arminio Monforte A, El-Sadr WM, Reiss P, et al. Combination antiretroviral therapy and the risk of myocardial infarction. N Engl J Med 2003; 349: Brown TT, Cole SR, Li X, Kingsley LA, Palella FJ, Riddler SA, et al. Antiretroviral therapy and the prevalence and incidence of diabetes mellitus in the Multicenter AIDS Cohort Study. Arch Intern Med 2005; 165: Justman JE, Benning L, Danoff A, Minkoff H, Levine A, Greenblatt RM, et al. Protease inhibitor use and the incidence of diabetes mellitus in a large cohort of HIV-infected women. JAcquir Immune Defic Syndr 2003; 32: de Wit S, Sabin C, Weber R, Worm SW, Reiss P, Thiebaut R, et al. for the DAD Study Group. d4t and lipodystrophy raise diabetes risk in DAD cohort; nevirapine protects. 8th International Congress on Drug Therapy in HIV infection. November Glasgow, Bacon MC, von Wyl V, Alden C, Sharp E, Robison N, Hessol S, et al. The Women s Interagency HIV Study: an observational cohort brings clinical sciences to the bench. Clin Diagn Lab Immunol 2005; 12: Barkan SE, Melnick SL, Preston-Martin S, Weber K, Kalish LA, Miotti P, et al. The Women s Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology 1998; 9: Cole SR, Hernan MA, Robins JM, Anastos K, Chmiel J, Detels R, et al. Effect of highly active antiretroviral therapy on time to acquired immunodeficiency syndrome or death using marginal structural models. Am J Epidemiol 2003; 158: Panel on Clinic Practices for Treatment of HIV infection from the US Department of Health and Human Services and Henry J. Kaiser Family Foundation. Guidelines for the Use of Antiretroviral Agents in HIV-1 Infected Adults and Adolescents. Bethesda, MD: National Institute of Health; Cox DH, Hinkley DV. Theoretical Statistics. London: Chapman & Hall; D Agostino RB, Lee ML, Belanger AJ, Cupples LA, Anderson K, Kannel WB. Relation of pooled logistic regression to time dependent Cox regression analysis: the Framingham Heart Study. Stat Med 1990; 9: Brown TT, Li X, Cole SR, Kingsley LA, Palella FJ, Riddler SA, et al. Cumulative exposure to nucleoside analogue reverse transcriptase inhibitors is associated with insulin resistance markers in the Multicenter AIDS Cohort Study. AIDS 2005; 19: Shikuma CM, Day LJ, Gerschenson M. Insulin resistance in the HIV-infected population: the potential role of mitochondrial dysfunction. Curr Drug Targets Infect Disord 2005; 5: FRAM Study Investigators. Fat distribution in men with HIV infection. J Acquir Immune Defic Syndr 2005; 40: FRAM Study Investigators. Fat distribution in women with HIV infection. J Acquir Immune Defic Syndr 2006, 42: Mynarcik DC, McNurlan MA, Steigbigel RT, Fuhrer J, Gelato MC. Association of severe insulin resistance with both loss of limb fat and elevated serum tumor necrosis factor receptor levels in HIV lipodystrophy. J Acquir Immune Defic Syndr 2000; 25: Tien PC, Schneider MF, Cole SR, Justman JE, French AL, Young M, et al. Relation of stavudine discontinuation to anthropometric changes among HIV-infected women. J Acquir Immune Defic Syndr 2007; 44: Dube MP, Edmondson-Melancon H, Qian D, Aqeel R, Johnson D, Buchanan TA. Prospective evaluation of the effect of initiating indinavir-based therapy on insulin sensitivity and B-cell function in HIV-infected patients. J Acquir Immune Defic Syndr 2001; 27: Noor MA, Lo JC, Mulligan K, Schwarz J-M, Halvorsen RA, Schambelan M, et al. Metabolic effects of indinavir in healthy HIV-seronegative men. AIDS 2001; 15:F11 F Noor MA, Seneviratne T, Aweeka FT, Lo JC, Schwarz JM, Mulligan K, et al. Indinavir acutely inhibits insulin-stimulated glucose disposal in humans: a randomized, placebo-controlled study. AIDS 2002; 16:F1 F Lee GA, Lo JC, Aweeka F, Schwarz J-M, Mulligan K, Schambelan M, et al. Single-dose lopinavir-ritonavir acutely inhibits insulinmediated glucose disposal in healthy volunteers. Clin Infect Dis 2006; 43: Lee GA, Seneviratne T, Noor MA, Lo JC, Schwarz JM, Aweeka FT, et al. The metabolic effects of lopinavir/ritonavir in HIV-negative men. AIDS 2004; 18: Mehta SH, Brancati FL, Sulkowski MS, Strathdee SA, Szklo M, Thomas DL. Prevalence of type 2 diabetes mellitus among persons with hepatitis C virus infection in the United States. Ann Intern Med 2000; 133: Mehta SH, Moore RD, Thomas DL, Chaisson RE, Sulkowski MS. The effect of HAART and HCV infection on the development of hyperglycemia among HIV-infected persons. J Acquir Immune Defic Syndr 2003; 33: Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

17 NIH Public Access Author Manuscript AIDS. Author manuscript; available in PMC 2011 November 13. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Published in final edited form as: AIDS November 13; 24(17): doi: /qad.0b013e32833f6294. Fracture incidence in HIV-infected women: results from the Women s Interagency HIV Study Michael T. Yin a, Qiuhu Shi b, Donald R. Hoover c, Kathryn Anastos d, Anjali Sharma e, Mary Young f, Alexandra Levine g, Mardge H. Cohen h, Elizabeth Shane a, Elizabeth T. Golub i, and Phyllis C. Tien j a Columbia University Medical Center, New York b New York Medical College, Valhalla, New York c Rutgers University, Piscataway, New Jersey d Montefiore Medical Center, Bronx, New York e State University of New York, Downstate, Brooklyn, New York f Georgetown School of Medicine, Washington DC g Keck School of Medicine, University of Southern California, Los Angeles, California h Departments of Medicine, Stroger (formerly Cook County) Hospital and Rush University, Chicago, Illinois i Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland j San Francisco Veterans Affairs Medical Center, University of California at San Francisco, California, USA Abstract Background The clinical importance of the association of HIV infection and antiretroviral therapy (ART) with low bone mineral density (BMD) in premenopausal women is uncertain because BMD stabilizes on established ART and fracture data are limited. Methods We measured time to first new fracture at any site with median follow-up of 5.4 years in 2391 (1728 HIV-infected, 663 HIV-uninfected) participants in the Women s Interagency HIV Study (WIHS). Self-report of fracture was recorded at semiannual visits. Proportional hazard models assessed predictors of incident fracture. Results At baseline, HIV-infected women were older (40 ± 9 vs. 36 ± 10 years, P <0.0001), more likely to report postmenopausal status and be hepatitis C virus-infected, and weighed less than HIV-uninfected women. Among HIV-infected women, mean CD4 + cell count was 482 cells/ μl; 66% were taking ART. Unadjusted incidence of fracture did not differ between HIV-infected and uninfected women (1.8 vs. 1.4/100 person-years, respectively, P = 0.18). In multivariate models, white (vs. African-American) race, hepatitis C virus infection, and higher serum creatinine, but not HIV serostatus, were statistically significant predictors of incident fracture. Among HIV-infected women, older age, white race, current cigarette use, and history of AIDSdefining illness were associated with incidence of new fracture. Conclusion Among predominantly premenopausal women, there was little difference in fracture incidence rates by HIV status, rather traditional risk factors were important predictors. Further research is necessary to characterize fracture risk in HIV-infected women during and after the menopausal transition. Keywords fracture; fragility fracture; HIV-infected women; premenopausal 2010 Wolters Kluwer Health Lippincott Williams & Wilkins Correspondence to Michael T. Yin, MD, MS, Division of Infectious Diseases, Columbia University Medical Center, 630 w168th Street, PH8-876, New York, NY 10032, USA. Tel: ; fax: ; Mty4@columbia.edu.

18 Yin et al. Page 2 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Introduction Methods Study population Low bone mineral density (BMD) is a recognized metabolic complication of HIV infection and its treatment [1]. Recent studies suggest that antiretroviral therapy (ART) initiation is associated with significant short-term bone loss in the range of 2 6% over 1 2 years, irrespective of the type of ART [2 7]. In contrast, longitudinal studies show that BMD is either stable or increases slightly over 2 years of follow-up in younger men and women on established ART [8 10]. It remains uncertain whether lower BMD associated with HIV infection or short-term bone loss associated with ART initiation will translate to increased fractures in younger HIVinfected individuals. There are several case series of fragility fractures in HIV-infected patients on ART [11 13], but only a few controlled studies evaluating fracture prevalence [14,15] and incidence [10,16 19]. Utilizing a large clinical database from the Partners HealthCare System with 8525 HIV-infected and HIV-uninfected patients, Triant et al. [15] found that prevalence of spine, hip, and wrist fracture as determined by ICD9 codes was higher in HIV-infected compared to HIV-uninfected men and women in all age groups; however, that study only controlled for age, sex, and race/ethnicity. Smaller studies with patient-level data have found that differences in fracture rates were attenuated after controlling for known risk factors for fracture [16,20]. We compared incidence rates of fracture and determinants of fracture among HIV-infected and uninfected women from the Women s Interagency HIV Study (WIHS), a large multicenter cohort of HIV-infected and uninfected women in the United States. The WIHS, a multicenter prospective study of the natural and treated history of HIV infection in women, enrolled 3766 women (2791 HIV-infected and 975 HIV-uninfected) in (n =2623) and (n =1143) from six sites (Bronx/Manhattan, New York; Brooklyn, New York; Chicago, Illinois; Los Angeles, California, San Francisco, California; and Washington, DC). WIHS methods and baseline cohort characteristics have been described previously [21,22]. HIV-uninfected women in the WIHS were recruited because they were at risk for HIV infection, and were comparable for a wide array of variables, including drug use, history of chronic illness, perceived health status, reproductive history, and income [21]. At semiannual visits, participants complete physical examinations and provide biological specimens; they also undergo extensive questionnaires assessing clinical and demographic information. Informed consent was obtained using procedures approved by committees on human research at each of the collaborating institutions. From April to September 2003 (visit 18), all WIHS participants were asked about personal history of fracture of the hip, wrist, and spine, both ever and within the prior 6 months. In those with a fracture history, fracture type(s) was determined, that is, fragility (resulting from fall from standing height or less) and nonfragility. In all subsequent study visits, participants were asked whether they had fractures of the hip, wrist, spine, and/or other site since the last visit. After exclusion of those who seroconverted during follow-up (n =17), 2391 (1728 HIV-infected, 663 HIV-uninfected) participants with at least one additional core visit after the index visit (visit 17, October 2002 March 2003) were included in the analysis. Visit 17 was designated as the index visit as incident fractures were only recorded after visit 17. We used fragility fractures, nonfragility fractures, and all fractures (fragility and nonfragility combined) at any body site as study outcomes given recent data that hightrauma fractures are associated with lower BMD and are similar to fragility fractures in AIDS. Author manuscript; available in PMC 2011 November 13.

19 Yin et al. Page 3 predicting future fractures [23]. Our analysis included biannual observations from October 2002 (visit 17) to September 2008 (visit 28). NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Statistical methods Demographics, medical history, HIV treatment history, and laboratory data were extracted from the WIHS database at index visit and subsequent visits. Presence of prognostic factors for fracture [24] at the index visit were quantified from the database: age, menopause (no menses for greater than 1 year), personal history of fracture, parental history of hip fracture, weight, race/ethnicity, current use of cigarettes, greater than two drinks/day of alcohol, glucocorticoid use (ever), and history of rheumatoid arthritis, or other cause of secondary osteoporosis. In addition, we obtained history of diabetes, hepatitis C RNA, estimated glomerular filtration rate using the Modification of Diet in Renal Disease (MDRD) calculation, reported current calcium use or multivitamin supplementation, and any use of injection drugs (IDU), opiates or cocaine, statins, hormonal contraception, or hormone replacement therapy (HRT). We also considered HIV-specific variables: current and nadir CD4 + lymphocyte counts, history of AIDS-defining illness (ADI), type of ART at index visit, and cumulative ART exposure by type of ART (total years of exposure to nucleoside (NRTI) and nonnucleoside (NNRTI) reverse transcriptase inhibitors, protease inhibitors, and tenofovir from initiation, whether before or after enrollment into WIHS, to either last visit or fracture event). Means, medians, standard deviations, interquartile ranges, and proportions were derived to summarize study variables, depending on whether they were continuous or categorical. The primary outcome was time to self-reported fracture event, defined as a fragility fracture at any body site or any fracture (fragility or nonfragility) at any site. Incidence rate was defined as the number of participants with a new fracture divided by the total time (in person-years) at risk for new fracture during the study period. Visit 17 was considered the start time; the end of follow-up was either the visit, at which the first fracture event was observed, or, if no fracture was observed, the last visit for each subject. Thirty-nine participants (22 HIVinfected, 17 HIV-uninfected) reported more than one fracture during the study period; only the first fracture was included in the analysis. Proportional hazard models with time-dependent covariates were constructed to determine predictors of incidence of new fracture. All covariates were assessed at the index visit, except for cumulative ART exposure, which was calculated from time of ART initiation to time of fracture event or last visit. Bivariate analyses comparing each predictor to the outcome were considered for HIV-infected and uninfected participants together, and for HIV-infected participants separately. Bivariate and multivariate proportional hazard models were performed for HIV-infected only and HIV-infected plus uninfected participants together. All variables were entered into multivariate models and a backward selection strategy with a stay P value 0.1 or less was used to determine which variables remained in the final model. HIV status was the only variable forced into the model. A P value less than 0.05 was considered statistically significant. Proportional hazard curves were constructed without and with adjustment for covariates using the Kalbfleisch Prentice approach to estimate baseline hazards and were fit twice with all variables except HIV set to their mean values and HIV set to negative and positive [25]. All analyses were performed using SAS software (Version 9.1.3; SAS Institute Inc., Cary, North Carolina, USA). AIDS. Author manuscript; available in PMC 2011 November 13.

20 Yin et al. Page 4 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Results Participant characteristics at index visit Fractures Table 1 presents participant characteristics at the index visit. HIV-infected women were older, of lower weight, and more likely to have detectable hepatitis C virus (HCV) RNA, to report postmenopausal status, and to use HRT, calcium and vitamin D supplementation. HIV-infected women were also less likely to report moderate and heavy alcohol consumption or current smoking than uninfected women. Among the 1728 HIV-infected women at the index visit, 39% reported a prior ADI and 66% reported taking ART (30% used a protease inhibitor-based regimen, 30% a NNRTI-based regimen, and 6% a NRTIonly regimen). Mean duration of ART exposure at the index visit (including self-reported ART use prior to enrollment in WIHS) was 4.31 ± 3.4 years. Among the 34% not receiving ART at the index visit, half were ART-naive and the other half reported prior ART use. Thirty-nine percent of participants on NNRTI-based regimens and 52% of those on protease inhibitor-based regimens at the index visit remained on the same class of ART throughout the duration of study. At the index visit, prevalence of self-reported history of fracture did not differ between HIVinfected and HIV-uninfected groups. Most fractures prior to the index visit occurred at the wrists in both groups. The median follow-up for the study was 5.4 (IQR 5.3, 5.5) years. During the study period, 217 new fractures occurred in 195 of 2391 participants (8.6% of HIV-infected and 7.1% of HIV-uninfected women; Table 2). In both groups, fractures were more common at sites other than the spine, hip, and wrist. Approximately one-third of the new fractures were classified as fragility fractures: 49/147 (33%) in HIV-infected and 18/47 (38%) in HIV-uninfected. Unadjusted incidence rates of fracture at any site were similar between HIV-infected and uninfected women (1.8 vs. 1.4/100 person-years, P = 0.13) as were rates of fragility fractures or rates of fracture limited to the hip, spine, or wrist (Table 2). Distribution of fractures at the spine, hip, and wrist did not differ by HIV serostatus. Correlates of incident fracture In bivariate analysis of all participants, traditional factors that were statistically associated with incidence of new fracture were older age, white race, self-report of menopause, history of fracture prior to index at any site, HCV infection, higher diastolic blood pressure at index, cigarette use at index, history of IDU or opiate use (Table 3). Use of HRT or oral contraceptives was associated with new fracture, but use of progestin-only formulations of hormonal contraception was not. HIV infection was not a significant predictor of incidence of new fracture [hazard ratio (HR) = 1.28; 95% confidence interval (CI): ]. In multivariate analysis, only older age, white race, HCV infection, and higher serum creatinine were statistically significant predictors of incidence of new fracture. Proportional hazard curves depicting proportion without new fracture in HIV-infected and uninfected women are presented without adjustment (Fig. 1a). There is no statistically significant difference between groups (P = 0.13), although the curves appear to diverge with longer follow-up. After adjustment for significant covariates (age, white race, HCV infection, and serum creatinine) using the Kalbfleisch-Prentice approach [25] (Fig. 1b), the curves remain similar throughout the study period (P = 0.54) In bivariate analyses restricted to HIV-infected participants older age, self-report of menopause, history of fracture prior to index at any site, HCV infection, cigarette use at index, history of IDU, and history of ADI were associated with incidence of new fracture (Table 4). In contrast, CD4 + cell count (at index visit or nadir), ART class treated as a AIDS. Author manuscript; available in PMC 2011 November 13.

21 Yin et al. Page 5 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Discussion categorical variable, cumulative protease inhibitor (HR = 1.03; 95% CI:, 0.98, 1.08, P = 0.29), NNRTI (HR = 0.95; 95% CI: 0.88, 1.02, P = 0.18), and tenofovir exposure (HR = 1.03, 95% CI: 0.98, 1.09, P = 0.26) were not associated with incidence of new fracture in bivariate analysis. In multivariate analysis, older age, white race, current smoking, and ADI were associated incidence of new fracture (Table 4). Cumulative NNRTI use was associated with a modest but statistically significant protective effect (HR = 0.92, 95% CI: 0.85, 0.99, P = ) (Table 4). In our cohort of predominantly premenopausal women, we did not find a statistically significant increase in incident fractures in HIV-infected women, even though the HIVinfected women were on average 4 years older than the HIV-uninfected women. Our findings are consistent with a prior analysis in a subset of pre-menopausal WIHS women, which found little difference in rates of bone loss or self-reported incident fracture over 2 years [10]. Furthermore, rates of fracture observed at the hip and wrist in our HIV-infected women over a median of 5.4 years (0.2/100 and 0.3/100 person-years, respectively) are consistent with large epidemiological studies of healthy premenopausal women in the United Kingdom, where rates of hip and wrist fracture were 0.3/100 [26] and 0.1/100 person-years [27], respectively. Finally, we demonstrated that traditional risk factors including older age, white race, and renal insufficiency were statistically significant predictors of fracture, as was HCV infection. Among HIV-infected women, history of an ADI was also an independent predictor of fracture. Our fracture incidence rates are somewhat higher than those reported in two other studies of HIV-infected patients of comparable age. A French cohort of 1281 ART-treated, mainly HIV-infected men with a median age of 36 years and follow-up of 7.1 years reported an incidence density rate of 0.33/100 person-years [17]. This study also found that the observed rates within their HIV-infected cohort were similar to that observed in the general population [17]. In the HIV Outpatient Study (HOPS), Dao et al. [19] reported a fracture incidence rate of approximately 0.8/100 person-years in HIV-infected patients (median age of 37, 81% men, 56% white), which was higher than rates from the National Hospital Discharge Survey. Estimates from our study (1.8/100 person-years for all fractures, 0.6/100 person-years for fragility fractures) are likely higher than previous studies because of more complete ascertainment resulting from use of a dedicated fracture questionnaire at semiannual visits, and possibly because all participants were women. Another critical difference between our study and the aforementioned studies is that our HIV-uninfected controls were prospectively enrolled and evaluated with the same questionnaire as our HIVinfected participants. The French study also defined fracture as having a fracture leading to severe or complete limitation of activity [17]. Two studies have also examined fracture rates in older HIV-infected men. In a study of 559 men 49 years of age or older, rates of self-reported incident fractures were similar for HIVinfected and uninfected groups (3.1/100 versus 2.6/100 person-years, P = 0.69) during a 2- year follow-up [16]. In contrast, the much larger Veterans Aging Cohort Study observed that HIV-infected veterans have higher rates of fragility fracture at the spine, hip, and wrist than uninfected veterans after the age of 50 [18]. There are no published data on fracture risk in HIV-infected postmenopausal women. We found that in addition to white race and renal insufficiency, HCV infection but not HIV infection was associated with fracture. Several studies in HIV-uninfected individuals have described the association between noncirrhotic HCV infection and low bone mass [28] or increased fracture [29,30]. Putative mechanisms are likely multifactorial and may be AIDS. Author manuscript; available in PMC 2011 November 13.

22 Yin et al. Page 6 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Acknowledgments associated with increased alcohol use in patients infected with HCV, alterations in calciotropic and gonadotropic hormone levels, weight loss, and increased fall risk [28 30]. In the setting of HIV infection, Lo Re et al. [31] found that HIV/HCV-coinfected women had lower BMD than HIV-monoinfected women, but did not observe a similar difference in men. Collin et al. [17] found an association of HCV coinfection and higher alcohol consumption with increased fracture risk. In our cohort, HIV/HCV-coinfected and HIV monoinfected women were similar in age, weight, and alcohol consumption; we were unable to assess for group differences in calciotropic or gonadotropic hormones. Among HIV-infected women in our cohort, history of ADI was associated with incidence of new fracture. This may reflect more advanced HIV disease stage, or the impact of hormonal, nutritional or neurological alterations associated with AIDS, opportunistic infections or their treatment on fracture risk. AIDS wasting syndrome has also been associated with lower BMD [32] and loss of muscle mass and functional status [33] that may lead to increased falls. Interestingly, we found an association of cumulative exposure to NNRTIs with decreased fracture risk in multivariate analyses. It is unclear if these results represent a direct effect of NNRTI on fracture risk. First, there was only a weak, nonsignificant association in the unadjusted models; therefore, this may be a spurious finding. Second, women on NNRTI-based regimens may be less sick than those on protease inhibitor-based regimens; protease inhibitors may more likely be used in those with advanced disease. Studies comparing bone loss with initiation of protease inhibitor versus NNRTI-based regimens have yielded inconclusive results [34 36]. Moreover, a prior analysis in a subset of WIHS women suggested that BMD may be slightly higher in those on an NNRTI-based regimen [14], but no statistically significant difference in change in BMD occurred over 2 years between ART groups [10]. We found no association between tenofovir use at index and new fractures. There are several strengths to our study. WIHS is one of the only cohorts that has collected detailed data semiannually regarding fractures in HIV-infected women and a well matched comparison group of HIV-uninfected women. Traditional risk factors for fracture have also been collected at regular intervals, and participant retention is excellent. Limitations include the lack of characterization of fracture sites other than spine, hip, and wrist and of radiographic confirmation of fractures. We also did not assess muscle strength or fall risk, which are other potential determinants of fracture. As the majority of women in WIHS are African-American and overweight, our results may not be generalizable to all HIV-infected women (although our enrolled cohort is representative of the HIV epidemic among US women). Finally, our sample size may be insufficient for comparisons of fracture rates by ART exposure. In conclusion, after 5.4 years of median follow-up in predominantly premenopausal women of color, fracture rates were not significantly increased in HIV-infected as compared to uninfected women. Traditional risk factors were important predictors of fracture, whereas HIV status was not. Among HIV-infected women, history of ADI was more predictive of fracture than ART exposure. Our data provide some reassurance that fracture risk is modest in predominantly premenopausal HIV-infected women. However, further research is necessary to assess fracture risk as these women transition through menopause and to clarify whether fracture risk differs among antiretroviral regimens. All authors played a role in editing the article and approved the text as submitted to AIDS. M.T.Y. designed the study and wrote the manuscript. Q.S. and D.R.H. performed the data analysis and assisted in the interpretation of statistical data. K.A., A.S., M.Y., A.L., M.H.C., E.S., E.T.G., and P.C.T. reviewed and edited the manuscript. AIDS. Author manuscript; available in PMC 2011 November 13.

23 Yin et al. Page 7 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript References Data in this manuscript were collected by the Women s Interagency HIV Study (WIHS) Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (K.A.); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Chicago Consortium (M.H.C.); Data Coordinating Center (Stephen Gange). The WIHS is funded by the National Institute of Allergy and Infectious Diseases (UO1-AI-35004, UO1-AI-31834, UO1-AI-34994, UO1-AI-34989, UO1-AI-34993, and UO1-AI-42590) and by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (UO1-HD-32632). The study is co-funded by the National Cancer Institute, the National Institute on Drug Abuse, and the National Institute on Deafness and Other Communication Disorders. Funding is provided by the National Center for Research Resources (UCSF-CTSI Grant Number UL1 RR024131). Drs. Yin and Tien are supported by the National Institute of Allergy and Infectious Diseases through K23 AI and K23 AI , respectively. The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health. 1. Brown TT, Qaqish RB. Antiretroviral therapy and the prevalence of osteopenia and osteoporosis: a meta-analytic review. AIDS. 2006; 20: [PubMed: ] 2. Brown TT, McComsey GA, King MS, Qaqish RB, Bernstein BM, da Silva BA. Loss of bone mineral density after antiretroviral therapy initiation, independent of antiretroviral regimen. J Acquir Immune Defic Syndr. 2009; 51: [PubMed: ] 3. Cassetti I, Madruga JV, Suleiman JM, Etzel A, Zhong L, Cheng AK, Enejosa J. The safety and efficacy of tenofovir DF in combination with lamivudine and efavirenz through 6 years in antiretroviral-naive HIV-1-infected patients. HIV Clin Trials. 2007; 8: [PubMed: ] 4. Duviver, C.; Kolta, S.; Assoumou, L.; Gohosn, J.; Rozenberg, S.; Murphy, R., et al. First-line PIcontaining regimens enhance decreased bone mineral density greater than NNRTI-containing regimen in HIV-1 infected patients: A substudy of the HIPPOCAMPE-ANRS 121 Trial. 15th Conference on Retroviruses and Opportunistic Infections; Boston, MA Grund B, Peng G, Gibert CL, Hoy JF, Isaksson RL, Shlay JC, et al. Continuous antiretroviral therapy decreases bone mineral density. AIDS. 2009; 23: [PubMed: ] 6. Rivas P, Gorgolas M, Garcia-Delgado R, Diaz-Curiel M, Goyenechea A, Fernandez-Guerrero ML. Evolution of bone mineral density in AIDS patients on treatment with zidovudine/lamivudine plus abacavir or lopinavir/ritonavir. HIV Med. 2008; 9: [PubMed: ] 7. Tebas, P.; Umbleja, T.; dube, MP.; Parker, RA.; Mulligan, K.; Roubenoff, R., et al. Initiation of ART is associated with bone loss independent of the three specific ART regimens: Results of ACTG A5005s. 14th Conference on Retroviruses and Opportunistic Infections; Los Angeles, US Dolan SE, Kanter JR, Grinspoon S. Longitudinal analysis of bone density in human immunodeficiency virus-infected women. J Clin Endocrinol Metab. 2006; 91: [PubMed: ] 9. Mondy K, Yarasheski K, Powderly WG, Whyte M, Claxton S, DeMarco D, et al. Longitudinal evolution of bone mineral density and bone markers in human immunodeficiency virus-infected individuals. Clin Infect Dis. 2003; 36: [PubMed: ] 10. Yin MT, Lu D, Cremers S, Tien PC, Cohen MH, Shi Q, et al. Short-term bone loss in HIV-infected premenopausal women. J Acquir Immune Defic Syndr. 2010; 53: [PubMed: ] 11. Guaraldi G, Ventura P, Albuzza M, Orlando G, Bedini A, Amorico G, Esposito R. Pathological fractures in AIDS patients with osteopenia and osteoporosis induced by antiretroviral therapy. AIDS. 2001; 15: [PubMed: ] 12. Stephens EA, Das R, Madge S, Barter J, Johnson MA. Symptomatic osteoporosis in two young HIV-positive African women. AIDS. 1999; 13: [PubMed: ] 13. McComsey GA, Huang JS, Woolley IJ, Young B, Sax PE, Gerber M, et al. Fragility fractures in HIV-infected patients: need for better understanding of diagnosis and management. J Int Assoc Physicians AIDS Care (Chic Ill). 2004; 3: AIDS. Author manuscript; available in PMC 2011 November 13.

24 Yin et al. Page 8 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript 14. Anastos K, Lu D, Shi O, Mulligan K, Tien PC, Freeman R, et al. The association of bone mineral density with HIV infection and antiretroviral treatment in women. Antivir Ther. 2007; 12: [PubMed: ] 15. Triant VA, Brown TT, Lee H, Grinspoon SK. Fracture prevalence among human immunodeficiency virus (HIV)-infected versus non-hiv-infected patients in a large U.S. healthcare system. J Clin Endocrinol Metab. 2008; 93: [PubMed: ] 16. Arnsten JH, Freeman R, Howard AA, Floris-Moore M, Lo Y, Klein RS. Decreased bone mineral density and increased fracture risk in aging men with or at risk for HIV infection. AIDS. 2007; 21: [PubMed: ] 17. Collin F, Duval X, Le Moing V, Piroth L, Al Kaied F, Massip P, et al. Ten-year incidence and risk factors of bone fractures in a cohort of treated HIV1-infected adults. AIDS. 2009; 23: [PubMed: ] 18. Womack, J.; Goulet, J.; Gilbert, C.; Brandt, C.; Mattocks, K.; Rimland, D., et al. HIV-infection and fragility fracture among male veterans. 17th Conference on Retroviruses and Opportunistic Infections; San Francisco February 2010; 19. Dao, C.; Young, B.; Buchacz, K.; Baker, R.; Brooks, J. Higher and increasing rates of fracture among HIV-infected persons in the HIV Outpatient Study compared to the general US population, 1994 to th Conference on Retroviruses and Opportunistic Infections; San Francisco, CA February 2010; 20. Prior J, Burdge D, Maan E, Milner R, Hankins C, Klein M, Walmsley S. Fragility fractures and bone mineral density in HIV positive women: a case-control population-based study. Osteoporos Int. 2007; 18: [PubMed: ] 21. Barkan SE, Melnick SL, Preston-Martin S, Weber K, Kalish LA, Miotti P, et al. The Women s Interagency HIV Study. WIHS Collaborative Study Group. Epidemiology. 1998; 9: [PubMed: ] 22. Hessol NA, Schneider M, Greenblatt RM, Bacon M, Barranday Y, Holman S, et al. Retention of women enrolled in a prospective study of human immunodeficiency virus infection: impact of race, unstable housing, and use of human immunodeficiency virus therapy. Am J Epidemiol. 2001; 154: [PubMed: ] 23. Mackey DC, Lui LY, Cawthon PM, Bauer DC, Nevitt MC, Cauley JA, et al. High-trauma fractures and low bone mineral density in older women and men. JAMA. 2007; 298: [PubMed: ] 24. Kanis JA, McCloskey EV, Johansson H, Oden A, Strom O, Borgstrom F. Development and use of FRAX in osteoporosis. Osteoporosis Int. 2010; 21 (Suppl 2):S407 S Kalbfleisch, JD.; Prentice, RL. The statistical analysis of failure time data. 2. Hoboken: Wiley; Banks E, Reeves GK, Beral V, Balkwill A, Liu B, Roddam A. Hip fracture incidence in relation to age, menopausal status, and age at menopause: prospective analysis. PLoS Med. 2009; 6:e [PubMed: ] 27. Thompson PW, Taylor J, Dawson A. The annual incidence and seasonal variation of fractures of the distal radius in men and women over 25 years in Dorset, UK. Injury. 2004; 35: [PubMed: ] 28. Schiefke I, Fach A, Wiedmann M, Aretin AV, Schenker E, Borte G, et al. Reduced bone mineral density and altered bone turnover markers in patients with noncirrhotic chronic hepatitis B or C infection. World J Gastroenterol. 2005; 11: [PubMed: ] 29. Collier J. Bone disorders in chronic liver disease. Hepatology. 2007; 46: [PubMed: ] 30. Nanda KS, Ryan EJ, Murray BF, Brady JJ, McKenna MJ, Nolan N, et al. Effect of chronic hepatitis C virus infection on bone disease in postmenopausal women. Clin Gastroenterol Hepatol. 2009; 7: [PubMed: ] 31. Lo Re V 3rd, Guaraldi G, Leonard MB, Localio AR, Lin J, Orlando G, et al. Viral hepatitis is associated with reduced bone mineral density in HIV-infected women but not men. AIDS. 2009; 23: [PubMed: ] AIDS. Author manuscript; available in PMC 2011 November 13.

25 Yin et al. Page 9 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript 32. Huang JS, Wilkie SJ, Sullivan MP, Grinspoon S. Reduced bone density in androgen-deficient women with acquired immune deficiency syndrome wasting. J Clin Endocrinol Metab. 2001; 86: [PubMed: ] 33. Grinspoon S, Corcoran C, Rosenthal D, Stanley T, Parlman K, Costello M, et al. Quantitative assessment of cross-sectional muscle area, functional status, and muscle strength in men with the acquired immunodeficiency syndrome wasting syndrome. J Clin Endocrinol Metab. 1999; 84: [PubMed: ] 34. Brown TT, McComsey GA. Association between initiation of antiretroviral therapy with efavirenz and decreases in 25 hydroxyvitamin D. Antiviral Ther. 2010; 15: Duvivier C, Kolta S, Assoumou L, Ghosn J, Rozenberg S, Murphy RL, et al. Greater decrease in bone mineral density with protease inhibitor regimens compared with nonnucleoside reverse transcriptase inhibitor regimens in HIV-1 infected naive patients. AIDS. 2009; 27: [PubMed: ] 36. McComsey, G.; Kitch, D.; Daar, E.; Tierney, C.; Jahed, N.; Tebas, P., et al. Bone and limb fat outcomes of ACTG A5224 s, a substudy of A5202: A prospective, randomized, partially blinded phase III trial of ABC/3TC or TDF/FTC with EFV or ATV/r for initial treatment of HIV-1 infection. 17th Conference on Retroviruses and Opportunistic Infections; San Francisco, CA AIDS. Author manuscript; available in PMC 2011 November 13.

26 Yin et al. Page 10 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Fig. 1. (a) Incidence of new fracture (fragility or nonfragility) at any site, unadjusted in HIVinfected (dashed line) and uninfected (solid line) women (P = 0.13). (b). Incidence of new fracture (fragility or nonfragility) at any site adjusted for age, race, hepatitis C virus infection, and serum creatinine in HIV-infected (dashed line) and uninfected (solid line) women (P = 0.54). AIDS. Author manuscript; available in PMC 2011 November 13.

27 Yin et al. Page 11 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Demographic and clinical characteristics. Table 1 Demographics HIV-infected (N =1728) HIV-uninfected (N =663) P Age 40.4 ± ± 9.9 < Race 0.30 White 230 (13.3%) 71 (10.7%) Black 972 (56.3%) 387 (58.4%) Latina 470 (27.2%) 179 (27.0%) Other 56 (3.2%) 26 (3.9%) Weight (kg) (mean ± SD) 74.5 ± ± 22.9 < BMI (kg/m 2 ) (mean ± SD) 28.5 ± ± 8.2 < Smoking at index visit 783 (45.3%) 337 (50.8%) IDU use 21 (3.2%) 49 (2.8%) 0.68 Ever opiate use 62 (3.6%) 33 (5.0%) 0.13 Ever cocaine use 50 (2.9%) 27 (4.1%) 0.15 Alcohol use < Abstainer 918 (53.1%) 265 (40.0%) Light (<3 drinks/week) 618 (35.8%) 259 (39.1%) Moderate (3 13 drinks/week) 155 (9.0%) 113 (17.0%) Heavy ( 14 drinks/week) 37 (2.1%) 26 (3.9%) Menstrual history Self-reported menopause 338 (19.6%) 74 (11.2%) < Hormone replacement therapy ever 124 (7.2%) 26 (3.9%) Medical history Previous fracture 74 (4.5%) 35 (5.6%) 0.37 Diabetes 332 (19.2%) 127 (19.2%) 1.00 HCV RNA positive 438 (25.4%) 96 (14.5%) < Diastolic blood pressure (mmhg) 75.6 ± ± Serum creatinine (mg/dl) (mean ± SD) 1.0 ± ± Estimated GFR < 60 (MDRD) 139 (8.5%) 36 (5.9%) Calcium supplementation 95 (5.5%) 21 (3.2%) Vitamin D supplementation 184 (41.9%) 724 (27.8%) < Statin use 88 (5.1%) 13 (2.0%) HIV ADI ever 671 (38.8%) CD4 + cell count, at index visit (cells/μl) 482 ± 320 CD4 + cell count, nadir (cells/μl) 233 ± 183 ART at index visita 1134 (65.6%) Protease inhibitor-art at index visit 523 (30.4%) NNRTI-ART at index visit 510 (29.6%) ADI, AIDS-defining illness; ART, antiretroviral therapy; GFR, glomerular filtration rate; IDU, injection drugs; MDRD, Modification of Diet in Renal Disease; NNRTI, nonnucleoside reverse transcriptase inhibitor. AIDS. Author manuscript; available in PMC 2011 November 13.

28 Yin et al. Page 12 a Including NRTI-only and combination ART. NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript AIDS. Author manuscript; available in PMC 2011 November 13.

29 Yin et al. Page 13 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Table 2 Incidence rates of new fractures in HIV-infected and uninfected women. HIV-infected (N =1728) HIV-uninfected (N =663) Fractures: number and site distribution Fractures N (%) Fractures/100 person-year Fractures N (%) Fractures/100 person-years P Subjects with (any site): New fracture 148 (8.6%) (7.1%) New fragility fracture 49 (2.8%) (2.7%) Fracture sites a Spine 15 (9.4%) (12.3%) Hip 15 (9.4%) (7.0%) Wrist 25 (15.6%) (19.3%) Other 105 (65.6%) (61.4%) Includes only one fracture per person. a Includes multiple fractures per person. There were a total of 160 fractures in 148 HIV-infected women and a total of 57 fractures in 47 HIV-uninfected women. (%) Percentage of total fractures occurring at body site. AIDS. Author manuscript; available in PMC 2011 November 13.

30 Yin et al. Page 14 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Table 3 Bivariate and multivariate models for incidence of new fractures in HIV-infected and uninfected women. Variables Bivariate model HR (95% CI) P Multivariate model a HR (95% CI) P HIV status 1.28 (0.93, 1.78) (0.78, 1.60) 0.54 Age (10 years) 1.37 (1.18, 1.58) < (1.02, 1.43) Race Black (Ref) White 1.50 (1.03, 2.17) (1.08, 2.39) Hispanic 0.70 (0.49, 1.01) (0.55, 1.18) 0.26 Other 0.83 (0.37, 1.89) (0.39, 2.38) 0.94 BMI (kg/m 2 ) 1.00 (1.00, 1.01) 0.66 Postmenopausal 1.71 (1.23, 2.36) Fracture before index 2.04 (1.24, 3.37) (0.94, 2.73) HCV infection 2.0 (1.49, 2.70) < (1.11, 2.20) Serum creatinine (per mg/dl) 1.09 (1.00, 1.19) (1.01, 1.21) Diastolic blood pressure (per mmhg) 1.01 (1.00, 1.03) Cigarette use (index) 1.62 (1.22, 2.15) <0.001 Alcohol use (index) 1.09 (0.91, 1.30) 0.34 Cocaine use (ever) 1.78 (0.94, 3.36) Injected drugs (ever) 2.21 (1.17, 4.18) Opiate use (ever) 1.86 (1.04, 3.33) Calcium use (index) 1.61 (0.95, 2.73) Vitamin D use (index) 0.92 (0.68, 1.23) 0.55 HRT (ever) 1.67 (1.05, 2.66) Oral contraceptive use (ever) 1.75 (1.12, 2.71) Statin Use (ever) 1.43 (0.80, 2.56) CI, confidence interval; HCV, hepatitis C virus; HR, hazard ratio; HRT, hormone replacement therapy. a All variables listed in table were considered in the multivariate model and eliminated if they did not meet backward selection criteria of P 0.1. AIDS. Author manuscript; available in PMC 2011 November 13.

31 Yin et al. Page 15 NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript Table 4 Bivariate and multivariate models for incidence of new fractures in HIV-infected women. Variables Bivariate model HR (95% CI) P Multivariate model a HR (95% CI) P Age (10 years) 1.36 (1.14, 1.62) < (1.02, 1.51) Race Black (Ref) White 1.49 (0.98, 2.27) (1.02, 2.39) Hispanic 0.74 (0.49, 1.11) (0.55, 1.27) 0.40 Other 0.78 (0.29, 2.14) (0.35, 2.62) 0.93 BMI (per kg/m 2 ) 1.01 (0.99, 1.03) (1.00, 1.04) Postmenopausal 1.54 (1.06, 2.22) Fracture before index 1.88 (1.02, 3.48) HCV coinfection 1.86 (1.33, 2.61) <0.001 Cigarette use (index) 1.66 (1.20, 2.30) (1.09, 2.12) IDU use (ever) 2.17 (1.01, 4.62) CD4 cell count (per 100 cells/μl) 0.97 (0.91, 1.03) 0.29 Nadir CD4 (per 100 cells/μl) 0.96 (0.87, 1.06) 0.41 ADI 2.10 (1.52, 2.91) < (1.34, 2.64) <0.001 NRTI use (index) 0.76 (0.54, 1.07) 0.12 TDF use (index) 1.25 (0.90, 1.74) 0.19 Protease inhibitor use (index) 1.14 (0.82, 1.58) 0.43 NNRTI use (index) 0.79 (0.54, 1.16) 0.23 Cumulative HAART (years) 1.01 (0.97, 1.05) 0.72 Cumulative NRTI (years) 1.01 (0.97, 1.05) 0.78 Cumulative protease inhibitor (years) 1.03 (0.98, 1.08) 0.29 Cumulative NNRTI (years) 0.95 (0.88, 1.02) (0.85, 0.99) ADI, AIDS-defining illness; CI, confidence interval; HCV, hepatitis C virus; HR, hazard ratio; IDU, injection drugs; NNRTI, nonnucleoside reverse transcriptase inhibitor. a Backward selection with stay criteria P 0.1. AIDS. Author manuscript; available in PMC 2011 November 13.

32 CLINICAL SCIENCE Urinary Markers of Kidney Injury and Kidney Function Decline in HIV-Infected Women Michael G. Shlipak, MD,* Rebecca Scherzer, PhD,* Alison Abraham, PhD, Phyllis C. Tien, MD,* Carl Grunfeld, MD,* Carmen A. Peralta, MD,* Prasad Devarajan, MD, Michael Bennett, PhD, Anthony W. Butch, PhD,k Kathryn Anastos, MD, # Mardge H. Cohen, MD,** Marek Nowicki, Anjali Sharma, MD, Mary A. Young, MD, Mark J. Sarnak, MD,kk and Chirag R. Parikh, MD ## Objective: HIV-infected persons have substantially higher risk of kidney failure than persons without HIV, but serum creatinine levels are insensitive for detecting declining kidney function. We hypothesized that urine markers of kidney injury would be associated with declining kidney function among HIV-infected women. Methods: In the Women s Interagency HIV Study, we measured concentrations of albumin-to-creatinine ratio, interleukin-18 (IL-18), kidney injury marker-1 (KIM-1), and neutrophil gelatinase-associated lipocalin from stored urine among 908 HIV-infected and 289 HIV-uninfected participants. Primary analyses used cystatin C-based estimated glomerular filtration rate (CKD-EPI egfrcys) as the outcome, measured at baseline and 2 follow-up visits over 8 years; secondary analyses used creatinine (CKD-EPI egfrcr). Each urine biomarker was categorized into tertiles, and kidney decline was modeled with both continuous and dichotomized outcomes. Results: Compared with the lowest tertiles, the highest tertiles of albumin-to-creatinine ratio (20.15 ml/min per 1.73 m 2, P, ), IL-18 (20.09 ml/min per 1.73 m 2, P, ) and KIM-1 (20.06 ml/min per 1.73 m 2, P, 0.001) were independently associated with faster egfrcys decline after multivariate adjustment Received for publication April 26, 2012; accepted September 7, From the *Department of Medicine, San Francisco VA Medical Center, San Francisco, CA; Department of Epidemiology and Biostatistics, University of California, San Francisco, CA; Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; Division of Nephrology and Hypertension, Cincinnati Children s Hospital Medical Center, Cincinnati, OH; kdepartment of Pathology and Laboratory Medicine, Clinical Immunology Research Laboratory, University of California, Los Angeles, CA; Departments of Medicine; and #Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, NY; **Departments of Medicine, Stroger Hospital and Rush University, Chicago, IL; Department of Medicine, University of Southern California, Los Angeles, CA; Division of Infectious Diseases, Department of Medicine, SUNY Downstate Medical Center, Brooklyn, NY; Division of Infectious Disease, Georgetown University Medical Center, Washington, DC; kkdivision of Nephrology, Tufts Medical Center, Boston, MA; Section of Nephrology, Department of Medicine, Yale University, New Haven, CT; and ##Program of Applied Translational Research, Yale University, New Haven, CT. Supported by the Women s Interagency HIV Study (WIHS) Kidney Aging Study Grant 1R01AG A2 (principal investigator, M. G. Shlipak), which was administered by the Northern California Institute for Research and Education, and with resources of the Veterans Affairs Medical Center, San Francisco, CA. Data in this manuscript were collected by the WIHS Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (K. Anastos); Brooklyn, NY (H. Minkoff); Washington DC Metropolitan Consortium (M. A. Young); The Connie Wofsy Study Consortium of Northern California (R. Greenblatt); Los Angeles County/Southern California Consortium (A. Levine); Chicago Consortium (M. H. Cohen); Data Coordinating Center (S. Gange). The WIHS is funded by the National Institute of Allergy and Infectious Diseases Grants UO1-AI-35004, UO1- AI-31834, UO1-AI-34994, UO1-AI-34989, UO1-AI-34993, and UO1-AI and by the Eunice Kennedy Shriver National Institute of Child Health and Human Development Grant UO1-HD The study is cofunded by the National Cancer Institute, the National Institute on Drug Abuse, and the National Institute on Deafness and Other Communication Disorders. Funding is also provided by the National Center for Research Resources (UCSF-CTSI Grant UL1 RR024131). The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of the National Institutes of Health. Authors contributions: M. G. Shlipak: study concept and design, analysis and interpretation of data, drafting of the manuscript; R. Scherzer: analysis and interpretation of data, drafting of the manuscript; A. Abraham: manuscript revision, cohort development; P. C. Tien: manuscript revision and cohort development; C. Grunfeld: analysis and interpretation of data and manuscript revision; C. A. Peralta: manuscript revision; P. Devarajan: measurement of data and manuscript revision; M. Bennett: measurement of data and manuscript revision; A. W. Butch: measurement of data and manuscript revision; K. Anastos: manuscript revision and cohort development; M. H. Cohen: manuscript revision and cohort development; M. Nowicki: cohort development; A. Sharma: cohort development; M. Young: cohort development; M. J. Sarnak: manuscript revision; C. R. Parikh: analysis and interpretation of data and manuscript revision. Dr M. G. Shlipak had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr C. R. Parikh is a coinventor on IL-18 patent issues to University of Colorado. Dr P. Devarajan is a coinventor on patents related to the use of NGAL as a marker of acute and chronic kidney injury. The other authors have no conflicts of interest to disclose. Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal s Web site ( Correspondence to: Michael G. Shlipak, MD, MPH, Chief of General Internal Medicine, San Francisco Veterans Affairs Medical Center, 4150 Clement St, Box 111A1, San Francisco, CA ( michael.shlipak@ucsf.edu). Copyright 2012 by Lippincott Williams & Wilkins J Acquir Immune Defic Syndr Volume 61, Number 5, December 15,

33 Shlipak et al J Acquir Immune Defic Syndr Volume 61, Number 5, December 15, 2012 including all 3 biomarkers among HIV-infected women. Among these biomarkers, only IL-18 was associated with each dichotomized egfrcys outcome: $3% (relative risk = 1.40; 95% confidence interval: 1.04 to 1.89); $5% (1.88; 1.30 to 2.71); and $10% (2.16; 1.20 to 3.88) for the highest versus lowest tertile. In alternative models using egfrcr, the high tertile of KIM-1 had independent associations with 5% (1.71; 1.25 to 2.33) and 10% (1.78; 1.07 to 2.96) decline, and the high IL-18 tertile with 10% decline (1.97; 1.00 to 3.87). Conclusions: Among HIV-infected women in the Women s Interagency HIV Study cohort, novel urine markers of kidney injury detect risk for subsequent declines in kidney function. Key Words: HIV, KIM-1, NGAL, IL-18, albumin-to-creatinine ratio, cystatin C, kidney injury (J Acquir Immune Defic Syndr 2012;61: ) INTRODUCTION Kidney disease is a major complication of HIV infection. Persons with HIV have substantially elevated risk of end-stage renal disease compared with HIV-uninfected individuals. 1,2 Elevated creatinine, a standard measure of kidney function, is associated with adverse health outcomes in HIV infection, including mortality, cardiovascular disease, and heart failure. 3 6 However, creatinine is an imperfect marker of kidney function, due to bias by age, sex, race, muscle mass, and hydration status. 7,8 In HIV-infected persons, early reductions in kidney function expressed as estimated glomerular filtration rate (egfr) may not be detected by serum creatinine levels but seem to be captured by cystatin C. 9,10 In the Study of Fat Redistribution and Metabolic Change in HIV Infection, reduced kidney function (egfr,60 ml/min per 1.73 m 2 ) was twice as prevalent when defined by cystatin C as by creatinine and had a much stronger association with mortality risk. 11 Although cystatin C may detect early reductions in GFR, substantial kidney injury may occur in HIV-infected persons before any discernible effect on kidney filtration. In some patients, kidney injury can be detected before GFR is decreased by measuring the urine albumin-to-creatinine ratio (ACR). One study found 5-fold odds of albuminuria in HIVinfected individuals compared with HIV-uninfected persons, most of whom had egfr in the normal range. 12 Both albuminuria and total urinary protein levels have been associated with mortality risk in HIV-infected persons, 5,6 and albuminuria and cystatin C have complementary associations with mortality. 11 However, urine albumin is a marker of glomerular injury and thus may not capture renal injury at other sites of the nephron. This is important because both HIV and its treatments may have nonglomerular effects on the kidney. Kidney biopsy studies suggest that the proximal and distal tubules are early targets of HIV Several tubular injury markers have been developed for early detection of acute kidney injury (AKI); among these, kidney injury marker-1 (KIM-1), interleukin-18 (IL-18), and neutrophil gelatinaseassociated lipocalin (NGAL) have been extensively studied in humans. 17 To our knowledge, these markers have not been evaluated in longitudinal studies of HIV-infected persons. We designed this study within the Women s Interagency HIV Study (WIHS) to test the hypothesis that markers of tubular injury could forecast the rate of subsequent egfr decline among women at risk for kidney disease. Our primary hypotheses were that ambulatory women with and without HIV infection would have detectable damage to their kidney tubules, measurable by IL-18, KIM-1, and NGAL levels, and that the level of the injury markers would be associated independently with the rate of subsequent kidney function decline. METHODS Design The WIHS is the largest, long-term, observational study of the progression of HIV in US women. The WIHS study design and methods have been described previously. 18,19 In brief, 3766 women (2791 HIV-infected and 975 HIV-uninfected) were enrolled in either (n = 2623) or (n = 1143) from 6 US sites (Bronx, Brooklyn, Chicago, Los Angeles, San Francisco, and Washington, DC). HIV-infected women were recruited to be representative of HIV-infected women in each community. HIV-uninfected women were recruited from similar venues. Participants are interviewed and examined every 6 months. Serum specimens were stored in a 280 C freezer until biomarker measurement. The WIHS HIV Kidney Aging study was designed as a nested cohort study to investigate the onset of kidney disease in the setting of HIV, using stored urine and serum specimens. Baseline measures were collected between October 1999 and March 2000, as this was the most recent visit that had collected and stored urine in WIHS. This study included all participants with available serum specimens during this time interval. A total of 1403 women (1032 HIV-infected and 371 HIVuninfected) had cystatin C measured at baseline; 1197 (908 HIV-infected and 289 HIV-uninfected) had stored urine available and at least one follow-up cystatin C measure; these participants were included in the longitudinal analyses of this manuscript. WIHS was approved by the relevant institutional review boards at all study sites. This study of kidney injury was also approved by the University of California, San Francisco; San Francisco Veterans Affairs Medical Center; and Yale committees on human research. Predictors Primary predictors in this study were urine ACR, IL-18, KIM-1, and NGAL. All urinary kidney injury biomarkers were measured at the Cincinnati Children s Hospital Medical Center Biomarker Laboratory. Urine albumin and creatinine were measured by immunoturbidimetry and colorimetric enzyme assay, respectively, using a Siemens Dimension Xpand plus HM clinical analyzer (Siemens, Munich, Germany). Urine IL-18 was measured using a commercially available ELISA kit (Medical & Biological Laboratories Co, Nagoya, Japan). The urine KIM- 1 ELISA was constructed using commercially available reagents (R&D Systems, Inc, Minneapolis, MN). 20 Urine NGAL was assayed using a human-specific commercially available ELISA (AntibodyShop, Grusbakken, Denmark). 21 All urine specimens Ó 2012 Lippincott Williams & Wilkins

34 J Acquir Immune Defic Syndr Volume 61, Number 5, December 15, 2012 Biomarkers and Kidney Decline were in continuous storage without prior freeze-thaw. Laboratory personnel were blinded to clinical information about the participants, including HIV status, and specimens were evaluated in random order. Coefficients of variation for the urine measures were: albumin, 5.9%; creatinine, 4.1%; IL-18, 7.2%; KIM-1, 5.2%; and NGAL, 5.4%. Outcomes Primary outcomes of this study were derived from cystatin C measures from the 3 WIHS visits, which were conducted concurrently at the UCLA Clinical Immunology Research Laboratory. Cystatin C was chosen as the outcome because it is less correlated with muscle mass or health status than creatinine and may thus be less biased in the setting of HIV infection. 9 Cystatin C was measured by a particle-enhanced immunoturbidimetric assay (Gentian, Moss, Norway), which has been calibrated against the new World Standard Reference material ERM-DA471/IFCC. 22 Intra-assay coefficients of variation, based on 10 replicates, were,2% at serum concentrations of 0.7 and 1.1 mg/l. Inter-assay coefficients of variation were 4.4% and 3.9% at serum concentrations of 0.8 and 2.2 mg/l, respectively. We estimated GFR using the CKD-EPI equation for cystatin C (egfrcys). 22,23 Alternative analyses were conducted using serum creatinine estimates of GFR by the CKD- EPI equation. Creatinine measures were conducted at the clinical laboratories of each WIHS site at the time of collection. As in our prior work, we truncated egfr at 120 ml/min per 1.73 m 2 because the equations have not been validated in persons with very high GFR, and higher GFR estimates are unlikely to be accurate or precise. 10 We analyzed egfr decline over the 8 years of follow-up as both continuous and dichotomous outcomes. The linear outcome was expressed as annual change in egfr in milliliters per minute per 1.73 square meters over the entire 8-year follow-up. We dichotomized egfr to distinguish persons with clinically meaningful rates of declining kidney function: mild ($3% annual egfr decline), moderate ($5% decline), or severe ($10% decline). We defined each dichotomized outcome by calculating the relative change in egfr from baseline to each follow-up visit for each participant. We defined the cystatin C analyses as primary a priori because the measures were conducted concurrently and were less likely to be biased by health status. Secondary analyses were implemented with the clinical creatinine measures and are presented in the Supplemental Digital Content (see Tables and Figures, Adjustment Variables Other covariates of interest included demographic characteristics, kidney disease risk factors, and HIV-specific risk factors obtained as part of the WIHS semiannual assessment. Specifically, the following characteristics were tested as candidate covariates in all multivariate models: age and race/ ethnicity; menopause status, antihypertensive use, diabetes (fasting glucose $126 mg/dl, self-reported diabetes, selfreported medication, or HbA1c $6.5), cigarette smoking (current, former, never); systolic and diastolic blood pressure, LDL and HDL cholesterol, triglycerides, body mass index, and waist circumference. We also tested the following HIV-related characteristics: hepatitis C virus (HCV) infection (defined by detectable HCV RNA), current heroin use, current CD4 cell count, nadir CD4 cell count, history of AIDS diagnosis (including low CD4), current HIV viral load, current highly active antiretroviral therapy use, current nucleoside reverse transcriptase inhibitor use, current nonnucleoside reverse transcriptase inhibitor use, and current protease inhibitor use. There was minimal use of tenofovir at the baseline of this study. In a sensitivity analysis, we included tenofovir use during follow-up. Factors forced in the full model included age, race/ethnicity, hypertension and diabetes, current HIV viral load, current CD4 cell count, and HCV infection. All time-varying covariates in the model were time updated, except for the urine biomarkers that were collected only at baseline. After forcing the above variables, we used a stepwise backward selection with a significance level of a = 0.05 to remove candidate covariates that were not associated with the outcome. Multiple imputation with the Markov chain Monte Carlo method was used to impute missing covariates, with 5 imputations to yield ;95% relative efficiency. 24 The percentage of observations with missing covariates ranged from less than 1% to 15%. Statistical Analysis We began our analyses by comparing the baseline characteristics of HIV-infected and HIV-uninfected participants. We first analyzed the urine injury biomarkers as continuous (log-transformed) predictors of kidney decline, but the assumption of linearity did not hold for all measures. We therefore present results with the biomarkers categorized into tertiles, with the HIV-infected and uninfected participants tertiled separately. We analyzed tertiles of ACR to facilitate comparisons of the effect sizes of ACR with the novel urine markers; however, we also dichotomized ACR at 30 mg/g, the standard clinical cut point. The association of each injury marker with annual mean change in egfr was assessed using linear mixed models with random intercepts and slopes. 25 Interaction terms between each injury marker and time were used to determine the rate of change in egfr. We used generalized estimating equations using a Poisson working model to account for clustering from repeated events to determine relative risks for the association of each covariate with the dichotomized kidney outcomes. To determine whether injury markers were independently associated with kidney outcomes, multivariable models were sequentially adjusted for (1) demographics and (2) traditional kidney disease risk factors, HIV-specific risk factors, and ACR. We included ACR because it is an established marker in widespread clinical use. Therefore, any novel marker should be demonstrated to have associations independent of ACR. In our final model, we adjusted for all 4 injury markers concurrently, allowing them to compete as predictors of each kidney outcome. We also conducted a sensitivity analysis in which we forced in tenofovir use, as a time-dependent covariate, into the final model. To account for informative lost to follow-up, we also adjusted estimates using an inverse probability weighting Ó 2012 Lippincott Williams & Wilkins 567

35 Shlipak et al J Acquir Immune Defic Syndr Volume 61, Number 5, December 15, 2012 TABLE 1. Baseline Characteristics of HIV-Infected and HIV- Uninfected Women in WIHS Cystatin C Study Characteristic HIV-Infected (N = 1032) HIV-Uninfected (N = 371) P Baseline age, yrs 41 (36 45) 38 (32 44),0.0001,30 65 (6%) 62 (17%) (40%) 145 (39%) (44%) 130 (35%) (10%) 34 (9%) Race African American 610 (59%) 228 (61%) 0.09 White 198 (19%) 53 (14%) Other 224 (22%) 90 (24%) History of coronary 202 (20%) 41 (11%), artery disease Cigarette smoking At baseline 542 (53%) 221 (60%) Past 246 (24%) 73 (20%) Never 244 (24%) 77 (21%) Diabetes mellitus 102 (10%) 39 (11%) 0.76 Systolic blood 118 ( ) 119 ( ) pressure, mm Hg Diastolic blood 72 (66 80) 72 (68 80) 0.68 pressure, mm Hg Hypertension 254 (25%) 92 (25%) 0.94 Antihypertensive use 115 (11%) 38 (10%) 0.70 LDL, mg/dl 101 (78 129) 104 (84 129) 0.13 HDL, mg/dl 44 (35 56) 51 (42 62), Triglycerides, mg/dl 131 (91 192) 97 (69 142), Body mass index, 27 (23 31) 29 (24 36), kg/m 2 Waist 88 (80 99) 92 (80 104) circumference, cm Hepatitis C 315 (31%) 67 (18%), Baseline heroin use 48 (5%) 28 (8%) Baseline HAART use 596 (58%) Baseline NRTI use 671 (65%) Baseline NNRTI use 276 (27%) Baseline PI use 411 (40%) Baseline CD4 398 ( ) Nadir CD4 214 ( ) History of AIDS* 483 (47%) Plasma HIV RNA # (31%) (23%) (16%).10, (31%) egfr cystatin, ml/ min/1.73 m 2 89 (75 105) 105 (91 119), Data are presented as median (interquartile range) or numbers (percent). *Defined by CD4 count or opportunistic infection. HAART, Highly Active Antiretroviral Therapy; NNRTI, nonnucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor. approach by modeling the participant s probability of having a nonmissing outcome at each visit. 26 The inverse of this probability was then used as a weight applied to persons with known outcome in the multivariable regression analyses of kidney decline. All analyses were conducted using SAS version 9.2 (SAS Institute Inc, Cary, NC). RESULTS In our study, 1032 HIV-infected women and 371 HIVuninfected women with cystatin C measured at baseline were included. More than half of women were African American or reported current smoking; one-fourth had hypertension (Table 1). HCV infection was more common in HIV-infected than HIV-uninfected women. Baseline egfrcys was significantly lower in HIV-infected compared with uninfected women (median 89 vs. 105 ml/min per 1.73 m 2, P, ). During the approximately 8-year follow-up period, rates of annual egfrcys decline were (95% CI: to 21.06) in HIV-infected women and (21.16 to 20.79) ml/min per 1.73 m 2 in HIV-uninfected women (P = 0.06 for difference). Among the HIV-infected women, egfrcys decline was faster among those coinfected with HCV (21.46; to 21.14) compared with monoinfected women (20.92; to 20.93; P, ). We initially examined the association of tertiles of each biomarker with the linear outcome of rate of change in egfrcys in HIV-infected women (Fig. 1). All 4 markers were independently associated with faster egfrcys decline in both demographic and multivariate-adjusted models. When we controlled for all 4 biomarkers simultaneously, the highest tertiles of ACR, IL-18, and KIM-1 remained independently associated with kidney function decline, whereas the association of NGAL weakened substantially (Fig. 1). The magnitude of association for the highest ACR tertile was larger than that of IL-18 or KIM-1. When we repeated analyses with egfrcr as the outcome, the biomarker associations with egfrcr decline were approximately half the magnitude of the egfrcys associations (see Figure S1, Supplemental Digital Content 1, The high tertiles of ACR, KIM-1, and NGAL were independently associated with faster egfrcr decline in single marker models, but only ACR and KIM-1 remained associated in the combined model. Adjustment for tenofovir use during follow-up had no material impact on the associations between the urine biomarkers and kidney function decline. We next compared biomarker associations with the dichotomized egfrcys outcomes in HIV-infected women, using similar staged adjusted models (Table 2). In fully adjusted analysis, the highest tertile of ACR was not significantly associated with any of the dichotomized outcomes relative to the lowest tertile. Dichotomized at ACR.30 mg/g, fully adjusted associations were 1.17 (0.93 to 1.47), 1.49 (1.14 to 1.95), and 1.43 (0.83 to 2.44) for 3%, 5%, and 10% decline, respectively. The highest tertile of IL-18 was progressively associated with increased risk for each endpoint. The middle tertile of IL- 18 was also significantly associated with the 3% and 5% thresholds of kidney function decline. The highest tertile of KIM-1 was significantly associated with a doubling in risk for the 10% annual egfr decline outcome, but this finding was attenuated by adjustment for IL-18. NGAL showed little association with any of these outcomes; relative risks Ó 2012 Lippincott Williams & Wilkins

36 J Acquir Immune Defic Syndr Volume 61, Number 5, December 15, 2012 Biomarkers and Kidney Decline FIGURE 1. Multivariable-adjusted associations of urine biomarkers with annual change in egfrcys (milliliters per minute per 1.73 square meters) in 908 WIHS HIV-infected participants. Estimates are highest versus lowest biomarker tertile. Tertile 1 is reference category and represents those with lowest amount of biomarker; Tertile 3 results shown. Cutoff points for biomarker tertiles areasfollows:acrt1:,7.1 mg/g, T2: mg/g, T3:.15.8 mg/g; IL-18 T1:,81 pg/ml, T2: pg/ml, T3:.196 pg/ml; KIM-1 T1:,319 pg/ml, T2: pg/ml, T3:.723 pg/ml; NGAL T1:,22 ng/ml, T2: ng/ml, T3:.57 ng/ml. Results reported as estimated annual change in egfr from baseline in milliliters per minute per 1.73 square meters (95% confidence interval). Demographic-adjusted model includes single biomarker, age, and ethnicity. Single injury marker multivariable model includes age, ethnicity, traditional kidney risk factors, HIV-related risk factors, and ACR. Additional adjustment for other injury biomarker model includes single injury biomarker multivariable model and all 4 markers. were 1.02 (0.79 to 1.31), 0.90 (0.65 to 1.24), and 0.85 (0.44 to 1.63) for 3%, 5%, and 10% decline, respectively. In analyses based on egfrcr (see Table S1, Supplemental Digital Content 1, significant associations were observed for the high tertile of KIM-1 with the 5% and 10% decline outcomes, and the high tertile of IL-18 was associated with the 10% decline outcome. We also evaluated associations of biomarker tertiles with kidney function decline in HIV-uninfected women (Fig. 2). In the linear analyses, the highest tertiles of IL-18, KIM-1, and NGAL were each individually associated with more rapidly declining kidney function in demographic and individual multivariateadjusted analyses, whereas ACR was not associated with decline. When we controlled for all 4 biomarkers simultaneously, only TABLE 2. Multivariable-Adjusted Associations of Urine Biomarkers With Rapid Decline by egfrcys in HIV-Infected WIHS Participants ACR IL-18 KIM-1 Tertile 2 Tertile 3 Tertile 2 Tertile 3 Tertile 2 Tertile 3 Cutoff Points mg/g.15.7 mg/g pg/ml.196 pg/ml pg/ml.721 pg/ml 3% (N = 553)k Model (0.84 to 1.51) 1.20 (0.89 to 1.62) 1.26 (0.96 to 1.64) 1.38 (1.07 to 1.76)* 0.92 (0.70 to 1.21) 1.02 (0.80 to 1.29) Model 2# 1.09 (0.82 to 1.47) 1.13 (0.84 to 1.52) 1.23 (0.97 to 1.56) 1.32 (1.01 to 1.72)* 0.92 (0.71 to 1.18) 0.97 (0.75 to 1.26) Model 3** 1.09 (0.82 to 1.45) 1.15 (0.86 to 1.53) 1.27 (1.00 to 1.63) 1.40 (1.04 to 1.89)* 0.85 (0.66 to 1.10) 0.84 (0.64 to 1.11) 5% (N = 381)k Model (0.66 to 1.50) 1.34 (0.90 to 1.99) 1.53 (1.09 to 2.14)* 1.88 (1.36 to 2.59) 0.96 (0.65 to 1.41) 1.26 (0.89 to 1.78) Model 2# 1.06 (0.73 to 1.53) 1.32 (0.91 to 1.91) 1.40 (1.03 to 1.91)* 1.74 (1.25 to 2.43) 0.88 (0.61 to 1.27) 1.10 (0.77 to 1.57) Model 3** 1.04 (0.73 to 1.49) 1.39 (0.99 to 1.97) 1.47 (1.06 to 2.03)* 1.88 (1.30 to 2.71) 0.80 (0.56 to 1.13) 0.89 (0.62 to 1.29) 10% (N = 138)k Model (0.45 to 2.07) 1.36 (0.65 to 2.86) 1.58 (0.88 to 2.81) 2.49 (1.34 to 4.62) 1.48 (0.85 to 2.57) 2.36 (1.35 to 4.10) Model 2# 1.11 (0.52 to 2.34) 1.50 (0.74 to 3.03) 1.42 (0.78 to 2.58) 2.46 (1.32 to 4.57) 1.31 (0.76 to 2.27) 2.05 (1.17 to 3.58)* Model 3** 1.11 (0.60 to 2.04) 1.58 (0.80 to 3.14) 1.38 (0.73 to 2.64) 2.16 (1.20 to 3.88) 1.23 (0.69 to 2.19) 1.62 (0.99 to 2.65) Estimates are calculated from generalized estimating equations relative risk models to account for multiple episodes of rapid decline, with inverse probability weighting to account for dropout, and multiple imputation for missing covariates. Tertile 1 is reference category, and represents those with lowest amount of biomarker; cutoff points for Tertile 1 of each biomarker are as follows: ACR,7.1 mg/g; IL-18,81 pg/ ml; KIM-1,318 pg/ml. *P, P, P, P, denotes statistical significance of estimates above. Results reported as risk of egfr decline of $3%, $5%, or $10% per year (95% confidence interval). kn = number of participants with at least one outcome. Model 1: Demographic-adjusted model controls for age, ethnicity, and single biomarker. #Model 2: Multivariable-adjusted full model controls for model 1 plus traditional kidney risk factors, HIV-related risk factors, and ACR. **Model 3: Multivariable-adjusted full model controls for model 2 plus all 4 biomarkers (ACR, IL-18, KIM-1, and NGAL). Ó 2012 Lippincott Williams & Wilkins 569