The Women’s Interagency HIV Study (WIHS) has been described. [6] HIV-infected and high-risk seronegative women were enrolled from six United States sites. Social and medical history, HIV-RNA, CD4, hemoglobin, serum albumin, and urinalysis were ascertained at enrollment and every six months. Hepatitis serologies and serum creatinine were performed at enrollment, and creatinine was repeated annually. Glomerular filtration rate (GFR) was estimated using the 4-variable Modification of Diet in Renal Disease (MDRD) equation.[7] The institutional review board of each site approved the WIHS protocol, and participants provided written informed consent.

Urine specimens were collected every 6 months for the first 3 years and stored at −70°C in a central repository. Specimens were excluded if there was evidence of cystitis, as previously described. [5] Semi-quantitative microalbuminuria testing (Clinitek Microalbumin Reagent Strips and Clinitek 50 Analyzer, Siemens Healthcare Diagnostics, Deerfield, IL) was performed in banked specimens from the first two eligible visits for each participant. Quality assurance testing was performed using standardized controls (Bio-Rad Laboratories, Hercules, CA). Microalbuminuria was defined as an albumin: creatinine ratio > 30 mg/g (3.4 mg/mmol). Women with urinalysis protein ≥1+ at both visits were considered to have “confirmed proteinuria.” Women with microalbuminuria at both visits or with microalbuminuria at one visit and proteinuria at the other were considered to have “confirmed microalbuminuria.” Women with proteinuria or microalbuminuria at only one visit were considered to have “unconfirmed albuminuria.” This analysis excluded women who initiated HAART prior to the second urine visit. Among included participants, the median time from the second urine visit to HAART initiation was 728 days.

Data on vital status and date of death were collected from medical records, providers, personal contacts, and the National Death Index. Cause of death was ascertained from the death certificate, medical records, providers, or personal contacts, in that order, and categorized as AIDS or non-AIDS using previously described methods. [8]

HIV-infected women enrolled in 1994–1995 who subsequently initiated HAART were included in this analysis. Sustained HAART use was defined as self-reported use for ≥ 3 consecutive visits prior to March 2009. The date of HAART initiation was defined as the mid-point between the last pre-HAART visit and the first visit at which HAART use was reported. Women with > 2 missed visits between the last pre-HAART visit and the first HAART visit were excluded. Continuous and categorical variables were compared by level of albuminuria using Kruskal-Wallis and chi-square tests, respectively. A 2-sided p-value of <0.05 was considered significant. Kaplan-Meier and proportional hazards models evaluated the association of albuminuria with time from HAART initiation to death. Multivariate models adjusted for demographics and laboratory results at the last pre-HAART visit (age, race-ethnicity, log transformed HIV-RNA, CD4, hemoglobin, serum albumin, and GFR), as well as for relevant clinical conditions (hepatitis C co-infection, diabetes, and blood pressure). Prior AIDS-defining illness (ADI) was ascertained at the first HAART visit, in order to include events that prompted initiation. When data were missing, the value was obtained from the most recent prior visit. All analyses were performed using SAS (version 9.1.3; SAS).

A total of 2,059 HIV-infected women were enrolled in WIHS during 1994–1995. Among the 1,390 participants who subsequently initiated HAART during the study period, 1,073 (77%) were included in this analysis. Women were excluded if they initiated HAART prior to the second urine visit (n=54), had < 2 eligible urine specimens (n=206), or had > 2 missed visits between the last pre-HAART visit and the first HAART visit (n=57). Excluded women were similar with respect to age and race-ethnicity (data not shown).

Among included participants, 107 (10%) had unconfirmed albuminuria, 42 (4%) had confirmed microalbuminuria, and 33 (3%) had confirmed proteinuria. Albuminuria was associated with black race, as well as with GFR<60ml/min/1.73m², higher HIV-RNA, higher blood pressure, and lower hemoglobin and serum albumin at the last pre-HAART visit (Table 1).

There were 301 deaths during the study period, including 130 AIDS deaths and 171 non-AIDS deaths (Table 1). Figure 1 shows the time to all-cause, AIDS, and non-AIDS death following HAART initiation. Compared to the reference group of women with no albuminuria, women with confirmed microalbuminuria had the highest risk for all-cause mortality (HR 3.0; 95% CI 2.0–4.6), followed by those with confirmed proteinuria (HR 2.3; 95% CI 1.4–3.9). Only confirmed microalbuminuria was significantly associated with AIDS death when considered separately (Figure 1b), while women with either confirmed microalbuminuria or confirmed proteinuria were at increased risk for non-AIDS death (Figure 1c). Women with a single episode of albuminuria were at modestly increased risk for all-cause mortality (HR 1.6; 95% CI 1.1–2.2), but unconfirmed albuminuria was not statistically associated with AIDS or non-AIDS death.

In multivariate analysis (Table 2), confirmed microalbuminuria remained significantly associated with all-cause (HR 1.9; 95% CI 1.2–2.9) and AIDS mortality (HR 2.3; 95% CI 1.3–4.3), but the association with non-AIDS death was no longer significant. Only the association of confirmed proteinuria with non-AIDS death remained significant (HR 2.4; 95% CI 1.2–4.6). In addition to albuminuria, the only other factor that was independently associated with time to both AIDS and non-AIDS death was GFR<60ml/min/1.73m². Results were similar when the CKD-EPI equation [9] was used to estimate GFR (data not shown). Other factors independently associated with AIDS death included CD4, HIV-RNA, and hemoglobin prior to HAART initiation (Table 2). Factors associated with non-AIDS death included older age, hepatitis C co-infection, current smoking, and lower serum albumin. Although CD4, HIV-RNA, and time to HAART initiation did not predict non-AIDS death, history of ADI remained marginally associated with non-AIDS mortality.

We have previously demonstrated an increased risk of all-cause and AIDS-related mortality among HIV-infected women with microalbuminuria prior to the introduction of HAART. [5] In the current analysis of 1,073 HIV-infected women who subsequently initiated HAART, pre-existing microalbuminuria was associated with a similar 2-fold increase in the risk of all-cause and AIDS death. These observations suggest that screening for albuminuria may identify HIV-infected individuals at increased risk for mortality even after HAART initiation.

Chronic kidney disease (CKD), as indicated by overt proteinuria or decreased GFR, has been associated with increased mortality in the general population and in HIV-infected women. [10][11] [12] [13] In an earlier analysis of WIHS censored in 2002, proteinuria detected by routine urinalysis was associated with a more than 2-fold increased risk of all-cause mortality among women who initiated HAART, although deaths were not classified by relationship to AIDS. [11] More recently, studies in predominantly male populations have demonstrated associations between overt proteinuria and cardiovascular events in HIV-infected individuals. [14] [15] While detailed data on cause of death were not available for the current analysis, cardiovascular disease was a common cause of non-AIDS mortality in an earlier analysis of WIHS. [16] Future studies should consider whether HIV-infected individuals with overt proteinuria would benefit from more intensive management of cardiovascular risk factors and other comorbidities that may contribute to non-AIDS mortality.

Microalbuminuria has also been associated with mortality and cardiovascular disease in HIV-negative patient populations. [4] Although microalbuminuria may indicate early CKD, it has also been hypothesized to reflect systemic inflammation [3] and generalized endothelial dysfunction. [4] In a prior analysis of 162 WIHS participants with serial albuminuria measurements, albuminuria stabilized in women who initiated HAART, but continued to increase in untreated controls. [17] Microalbuminuria was hypothesized to reflect inflammation and endothelial dysfunction resulting from active viral replication. The results of the current analysis suggest that pre-existing microalbuminuria may continue to predict AIDS-related outcomes even after the initiation of HAART.

Future studies are needed to determine whether microalbuminuria predicts mortality in HIV-infected men or from specific causes. Other important limitations are the possible underestimation of albuminuria in women taking angiotensin converting enzyme inhibitors or angiotensin receptor blockers, and the misclassification of women who developed new albuminuria in the interval between the confirmatory urine test and HAART initiation. Misclassification may also have occurred due to the limited sensitivity of dipstick assays to detect lower levels of proteinuria [18] and microalbuminuria, or due to decreased sensitivity of microalbuminuria testing in banked urine.[19] These types of misclassification, however, would be expected to bias our results towards the null. Finally, we had limited power to exclude an association between overt proteinuria and AIDS mortality or to evaluate the effect of individual ART regimens.

The current study suggests that albuminuria testing before initiation of HAART may identify individuals at increased risk for mortality despite HAART. Screening for CKD, including assessment for overt proteinuria, is currently standard of care in HIV-infected individuals.[20] Future studies should investigate whether additional microalbuminuria testing is a cost-effective approach to identify high-risk patients, who may benefit from more aggressive management of comorbidities and behavioral factors known to contribute to mortality.