![](images/null.gif) |
![](images/null.gif) |
![](images/null.gif) |
|
Predictors for Change in Estimated Glomerular Filtration Rate in HIV-infected Individuals with or without cART: The Swiss HIV Cohort Study
|
|
|
Reported by Jules Levin
CROI 2010 Feb 16-19 SF
J Schaefer1, C Fux1, E Bernasconi2, M Cavassini3, R Weber4, P Vernazza5, B Hirschel6, M Battegay7, Heiner Bucher*7, and Swiss HIV Cohort Study
1Univ Hosp and Univ of Bern, Switzerland; 2Ospedale Regionale di Lugano, Switzerland; 3Univ Hosp Lausanne, Switzerland; 4Univ Hosp Zurich, Switzerland; 5Cantonal Hosp, St Gallen, Switzerland; 6Univ Hosp Geneva, Switzerland; and 7Univ Hosp Basel, Switzerland
Interpretation
Our results suggest that the null hypothesis of no cART treatment group differences (TDF-based versus TDF-sparing) in the changes of eGFR can be rejected (p < 0:001). The estimated mean change during the first year under a TDF-sparing cART regimen is -1.586 mL/min/1.73m2, comparing with a more modest decline of only -0.442 mL/min/1.73m2 under a TDF sparing regimen. However, the estimated variances of the random effects (data not shown) indicate that there is substantial variation from patient to patient. For example, approximately 95% of treatment-naive patients (no intravenous drug use, no prior history of AIDS, no diabetes, no hypertension) are expected to have annual changes in eGFR between -9.13 and 11.76.
ABSTRACT
Background: HIV may accelerate the loss of renal function. Combination antiretroviral therapy (cART) may protect the kidney through its anti-HIV effect, but some anti-retroviral drugs are nephrotoxic. Past studies have not always modelled risk factors and cART components known to be related to renal function.
Methods: We estimated glomerular filtration rate (eGFR) with the MDRD study equation. To determine how changes in eGFR over time are related to various covariates, we fitted a multivariable linear mixed effects model with fixed baseline covariates (gender, ethnicity, transmission mode (intravenous drug use), CD4 cell count, log viral load, years with HIV infection, and years with diabetes) and time-varying covariates (age, hypertension, years of exposure to ART, years of exposure to tenofovir (TDF)). We allowed the intercept to vary randomly from one patient to another and assumed an autoregressive correlation structure of order 1 to model the dependence among repeated observations on the same patient.
Results: Since 2002, 2253 patients in the SHCS who were antiretroviral-naïve until the date of the first creatinine measurement were followed over a median of 2.5 years (IQR: 0.8 to 4.9).
Cumulative exposure to ART was associated with an increase in eGFR (0.68 mL/min/1.73m2 (standard error (SE) 0.20); P <0.001) per year.
Exposure to TDF was associated with an average decrease in eGFR of -2.40 mL/min/1.73m2 (SE 0.33); P<0.001) per year.
eGFR showed an approximately linear relationship with age over time (-0.56 mL/min/1.73m2 (SE 0.41) per year P<0.001).
Female gender, black ethnicity and intravenous drug use as the likely mode of transmission were the only factors associated with a statistically significant effect on eGFR (- 2.40 mL/min/1.73m2 (SE 0.94); P =0.01, 21.59 mL/min/1.73m2 (SE 1.26); P <0.001 and 6.93 mL/min/1.73 m2 (SE 1.23); P<0.001), respectively).
Conclusions: In this well characterized cohort of cART naïve patients, time on ART was associated with an increase in eGFR, whereas exposure to TDF and age were related with a discrete decrease in eGFR. The overall explained variance of individual eGFR variability was modest.
Background
Human immunodeficiency virus (HIV)-infection is hypothesized to accelerate the loss of renal function, whereas combination antiretroviral therapy (cART) may protect the kidney through its anti-HIV effect. However, some antiretroviral drugs are presumably nephrotoxic. The focus of this presentation is on assessing the effect of cART and the antiretroviral agent tenofovir (TDF) on renal impairment in an increasingly aging population of HIV-infected individuals.
Methods
Since January 2002, plasma creatinine, which is a commonly used indicator of kidney function, is collected in the Swiss HIV Cohort study (SHCS), a nationwide study including over half of the HIV-infected patients in Switzerland. As of 17 December 2009, 2694 patients were included who were either on their first-line therapy (that is, boosted PI, non-boosted PI, NNRTI or NRTI) at the date of the first creatinine measurement or were treatment-naive and started cART during follow-up. These patients were followed for a median of 3.77 years (interquartile range (IQR): 1.63-6.33) with a median of 11 creatinine measurements (IQR: 6-17) per patient. The median follow-up time for the 1703 patients with an initial TDF-sparing drug combination was 4.95 years (IQR: 2.31-6.78) with a median of 13
measurements (IQR: 7-21) per subject, whereas the 991 patients with an initial TDF-containing drug combination were followed only for a median of 2.52 years (IQR: 1.04-4.30) with a median of 8 measurements (IQR: 4-12.5) per subject. Demographic and selected patient characteristics at the time of antiretroviral treatment initiation are shown in Table 1. Glomerular filtration rate (GFR) is estimated using the Modification of Diet in Renal Disease (MDRD) equation [Levey et al., 2006]
![](../images/030610/030610-5/image002.gif)
for creatinine in mg/dl.
To investigate changes in eGFR before and after cART initiation and factors that may influence change, we fit a homoscedastic linear mixed-effects model [Laird and Ware, 1982] by restricted maximum likelihood (REML), assuming independent within-patient errors. The random-effects structure represents the variability in subject-specific intercepts and slopes that is not explained by the included covariates. Of note, data were analysed by intention-to-continue-treatment, that is, ignoring treatment changes and interruptions.
Table 1: Demographic and selected patient characteristics at the time of antiretroviral treatment initiation.
![](../images/030610/030610-5/image004.gif)
Results
Figure 1 assesses the trend in mean eGFR using lowess smoothed curves. It reveals a steeper decline in mean eGFR after cART initiation for the TDF-based first-line therapy group. In addition, the time plot indicates greater variability of measurements in the post-treatment start period. Next we consider the hypothesis that eGFR decreases linearly with age, but with different slopes before and after treatment initiation. In addition, the mean rate of change under cART can differ depending on whether or not the chosen drug combination includes TDF. Table 2 shows the REML estimates for the fixed effects in the corresponding model.
Figure 1: Estimated GFR against time, relative to the individual initiation of cART (in years), with lowess smoothed curves.
![](../images/030610/030610-5/image006.gif)
Table 2: Estimates, standard errors and p-values for all fixed effects in the fitted model, with time expressed in years since initiation of cART.
![](../images/030610/030610-5/image008.gif)
References
Laird, N. M. and Ware, J. H. (1982).
Random-effects models for longitudinal data.
Biometrics, 38(4):963{974.
Levey, A. S., Coresh, J., Greene, T., Stevens, L. A., Zhang, Y. L., Hendriksen, S., Kusek, J. W., and Lente, F. V. (2006).
Using standardized serum creatinine values in the modi_cation of diet in renal disease study equation for estimating glomerular filtration rate.
Ann Intern Med, 145(4):247{254.
|
|
|
![](images/null.gif) |
![](images/null.gif) |
|
|