HIV & Kidney Disease: ART Interruption In SMART
"Renal glomerular and tubular epithelial cells are susceptible to infection by HIV, and infected cells have been detected in the kidney, even in persons receiving ART with undetectable plasma HIV-RNA levels [37,38]. HIV infection of human renal tubular epithelial cells induces apoptosis and production of inflammatory mediators [39-42]. Studies [38,43,44] have demonstrated that ART can induce marked improvement in functional and histological abnormalities in some patients with HIVAN after a reduction in viral load. Participants in the drug conservation arm of the SMART trial who were previously taking ART had a marked increase in plasma HIV-RNA levels after discontinuation or deferral of ART following randomization. Moreover, participants in the drug conservation arm were more likely to have an increase in cystatin C if they had an HIV-RNA level 400 copies/ml or less at randomization before ART discontinuation. It is therefore plausible that withdrawal of ART resulted in increased HIV replication in renal epithelial cells, resulting in local inflammation and renal injury."
Interruption of antiretroviral therapy is associated with increased plasma cystatin C
AIDS:Volume 23(1)2 January 2009p 71-82
Mocroft, Amandaa; Wyatt, Christinab; Szczech, Lyndac; Neuhaus, Jacquied; El-Sadr, Wafaae; Tracy, Russellf; Kuller, Lewisg; Shlipak, Michaelh; Angus, Briani,j; Klinker, Hartingk; Ross, Michaelb; for the INSIGHT SMART study group
aUniversity College London Medical School, London, UK
bMount Sinai School of Medicine, New York, USA
cDuke University Medical Center, Durham, USA
dUniversity of Minnesota School of Public Health, Minnesota, USA
eHarlem Hospital Center and Columbia University, New York, USA
fUniversity of Vermont College of Medicine, Vermont, USA
gUniversity of Pittsburgh, Pittsburgh, USA
hSan Francisco Veterans Affairs Medical Center and University of California, San Francisco, USA
iNuffield Department of Medicine, University of Oxford, Oxford, UK
jMedical Research Council, HIV Clinical trials Unit, London, UK
kKlinikum der Universitt, Wrzburg, Germany.
Background: Cystatin C has been proposed as an alternative marker of renal function. We sought to determine whether participants randomized to episodic use of antiretroviral therapy guided by CD4+ cell count (drug conservation) had altered cystatin C levels compared with those randomized to continuous antiretroviral therapy (viral suppression) in the Strategies for Management of Antiretroviral Therapy trial, and to identify factors associated with increased cystatin C.
Methods: Cystatin C was measured in plasma collected at randomization, 1, 2, 4, 8 and 12 months after randomization in a random sample of 249 and 250 participants in the drug conservation and viral suppression groups, respectively. Logistic regression was used to model the odds of at least 0.15 mg/dl increase in cystatin C (1 SD) in the first month after randomization, adjusting for demographic and clinical characteristics.
Results: At randomization, mean (SD) cystatin C level was 0.99 (0.26 mg/dl) and 1.01 (0.28 mg/dl) in the drug conservation and viral suppression arms, respectively (P = 0.29). In the first month after randomization, 21.8 and 10.6% had at least 0.15 mg/dl increase in cystatin C in the drug conservation and viral suppression arms, respectively (P = 0.0008). The difference in cystatin C between the treatment arms was maintained through 1 year after randomization. After adjustment, participants in the viral suppression arm had significantly reduced odds of at least 0.15 mg/dl increase in cystatin C in the first month (odds ratio 0.42; 95% confidence interval 0.23-0.74, P = 0.0023).
Conclusion: These results demonstrate that interruption of antiretroviral therapy is associated with an increase in cystatin C, which may reflect worsened renal function.
The increased risk of kidney disease demonstrated in the drug conservation arm of the SMART trial  motivated the current study of changes in plasma cystatin C, a proposed early marker of subclinical kidney damage. In the subset of SMART trial participants evaluated in this study, those who discontinued or deferred ART (drug conservation arm) had a rapid small, but significant, increase in plasma cystatin C compared with participants who were treated with continuous ART (viral suppression arm), and were significantly more likely to have cystatin C levels higher than the level at randomization. The difference in cystatin C levels between the drug conservation and viral suppression arms of the SMART trial were maintained through 1 year of follow-up.
Plasma cystatin C levels are determined by the net balance of secretion of cystatin C into plasma and clearance by glomerular filtration . The increased cystatin C levels found in the drug conservation arm compared with the viral suppression arm must therefore reflect either a reduction in GFR in the drug conservation group that was not detected by changes in serum creatinine or estimated GFR, increased secretion of cystatin C from cells into the plasma or both. The latter explanation seems unlikely, as research has shown that cystatin C levels do not increase in the setting of acute stress . Systemic levels of cystatin C have been shown to be independent of several characteristics, including age , sex [29,30] and muscle mass . There are conflicting data in the literature regarding whether systemic inflammation can influence cystatin C levels independent of renal function. Knight et al.  reported that cystatin C levels were independently associated with C reactive protein (CRP) levels in a cohort of healthy individuals, nearly all of whom were whites with normal renal function. However, another study by Singh et al. , which included a more racially diverse population with a higher prevalence of chronic kidney disease, found that the association of cystatin C with inflammatory markers was not independent of GFR. The latter study suggested that impaired renal function is associated with inflammation and that levels of cystatin C are reflective of decreased GFR even in the presence of systemic inflammation. In this study, there was a strong correlation between randomization levels of cystatin C and serum creatinine in the upper quintile of cystatin C at randomization. One explanation for this observation is that at higher levels, cystatin C and creatinine are well correlated and reflect impaired renal function, whereas at lower levels, cystatin C, creatinine or both are more prone to the influence of other factors such as inflammation and muscle mass, respectively. Although we identified weak correlations between levels of inflammatory markers and cystatin C, there was no independent association between hsCRP or IL-6 and change in cystatin C after adjustment for other variables. We therefore hypothesize that the increase in cystatin C observed in the drug conservation arm of the SMART trial upon interruption of ART-reflected systemic inflammation that was concurrent with and may have contributed to renal injury.
We also found that an increase in cystatin C from randomization to month 1 correlated with the value of D-dimers and HDL levels at 1 month after randomization and with the change in these markers from randomization to month 1. Increased D-dimers and decreased HDL are associated with increased risk of cardiovascular disease in the general population and can be induced by systemic inflammation . D-dimers and HDL are also more likely to be elevated and decreased, respectively, in participants with chronic kidney disease, possibly contributing to the increased risk of cardiovascular mortality in that population [35,36].
There are several mechanisms by which interruption or deferral of ART may have resulted in an increase in cystatin C. Although it is possible that the increase in cystatin C could reflect systemic inflammation rather than a decrease in renal function, we hypothesize that the inflammation was concurrent with and may have contributed to renal injury. Renal glomerular and tubular epithelial cells are susceptible to infection by HIV, and infected cells have been detected in the kidney, even in persons receiving ART with undetectable plasma HIV-RNA levels [37,38]. HIV infection of human renal tubular epithelial cells induces apoptosis and production of inflammatory mediators [39-42]. Studies [38,43,44] have demonstrated that ART can induce marked improvement in functional and histological abnormalities in some patients with HIVAN after a reduction in viral load. Participants in the drug conservation arm of the SMART trial who were previously taking ART had a marked increase in plasma HIV-RNA levels after discontinuation or deferral of ART following randomization. Moreover, participants in the drug conservation arm were more likely to have an increase in cystatin C if they had an HIV-RNA level 400 copies/ml or less at randomization before ART discontinuation. It is therefore plausible that withdrawal of ART resulted in increased HIV replication in renal epithelial cells, resulting in local inflammation and renal injury.
Our study has several limitations. Although there was a statistically significant difference in cystatin C levels between the drug conservation and viral suppression arms at month 1 after randomization, the magnitude of the difference was small, and the clinical significance and relationship of this difference with renal disease are unclear. As serum creatinine measurements were only available on a subset of participants, we cannot exclude the possibility that the lack of difference in creatinine and creatinine-based estimates of GFR between drug conservation and viral suppression arms was a result of ascertainment bias. Finally, as we did not use a 'gold standard' to measure GFR (such as iothalamate or inulin clearance), we cannot be certain that the rise in cystatin C reflects a decline in GFR. However, cystatin C has been shown to be a better marker of GFR than serum creatinine in several disease settings, and creatinine itself has never been validated as a marker of GFR in HIV-infected patients using a 'gold standard'. Prospective, well designed studies are therefore urgently needed to explore the accuracy of cystatin C and creatinine as markers of GFR in patients with HIV infection.
In summary, in this randomized comparison, an increase in cystatin C levels was demonstrated after stopping ART in the drug conservation arm of the SMART trial. Cystatin C levels in the drug conservation arm were consistently higher than the levels seen in the viral suppression arm, and these increases were associated with increased production of inflammatory and coagulation-related biomarkers. These data demonstrate that discontinuation of ART was associated with an increase in cystatin C, which may be attributable to activation of inflammatory mediators. The reasons for this increase and the association with other biomarkers of inflammation and kidney function and with clinical renal disease require further investigation.
The introduction of effective antiretroviral therapy (ART) for the treatment of HIV-1 infection has led to a dramatic decline in mortality and morbidity [1,2]. As patients survive longer, renal disease has become an important contributing factor to morbidity and mortality . Although ART has been demonstrated to reduce the incidence and severity of HIV-associated nephropathy (HIVAN) [4-6], specific antiretrovirals may be associated with nephrotoxicity or with increased rates of hypertension and diabetes, which in turn may increase the risk of kidney disease [7,8]. Several studies [9-12] have considered the impact of ART on serum creatinine, proteinuria, estimated glomerular filtration rate (GFR) and progression of kidney disease. Cystatin C, a nonglycosylated basic protein secreted by all nucleated cells and freely filtered by the glomerulus, has been considered as a prognostic marker for mortality and morbidity  and also proposed as an alternative marker of GFR . Serum cystatin C levels correlate well with GFR in patients without HIV [15-17], and may be more sensitive than serum creatinine in the detection of early renal insufficiency [18,19]. Serum cystatin C levels are not affected by muscle mass, which may be decreased in patients with advanced HIV disease . Further, serum cystatin C levels have been correlated with renal function in HIV-infected patients and have been noted to be higher in HIV-1-infected patients compared with uninfected individuals [21,22]. To our knowledge, changes in cystatin C associated with starting or stopping ART have not been previously described.
The Strategies for Management of Antiretroviral Therapy (SMART) trial was a randomized, controlled clinical trial, which compared the episodic use of ART (drug conservation arm) guided by CD4+ cell counts with continuous therapy aimed at virologic suppression (viral suppression arm) . The study  demonstrated that the drug conservation strategy was associated with inferior outcomes with regards to development of opportunistic disease or death and the composite endpoint of cardiovascular, hepatic and renal disease; an increased risk of fatal or nonfatal renal disease was described in the drug conservation arm compared with the viral suppression arm. In another study  from the SMART trial, plasma biomarkers of inflammation [interleukin 6 (IL-6), high sensitivity C-reactive protein (hsCRP), amyloid A and amyloid P], coagulation (D-dimer and prothrombin fragments 1 + 2) and lipids [total cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL) and triglycerides] were measured to determine whether they were associated with all-cause mortality and cardiovascular disease.
The aims of the current study were to describe changes in plasma cystatin C in SMART trial participants included in the previous biomarker substudy, and to investigate the factors associated with increase in cystatin C levels after randomization. As many of the previously measured biomarkers are also associated with renal disease, we sought to determine whether these biomarkers were associated with cystatin C.
Strategies for Management of Antiretroviral Therapy study participants and treatment protocol
Between January 2002 and January 2006, 5472 HIV-infected participants were enrolled in the SMART trial by 318 sites in 33 countries. Participants were eligible if they had a CD4+ cell count more than 350 cells/μl and were willing to initiate, modify or stop ART as per study guidelines . SMART participants were randomized to one of two ART strategies. For the viral suppression group, ART was used in an uninterrupted manner with the goal of maximal suppression of HIV replication. The experimental drug conservation strategy entailed episodic use of ART for periods defined by CD4+ cell count thresholds. ART was stopped (or deferred) until the CD4+ cell count dropped to less than 250 cells/μl, at which time ART was to be (re)initiated and continued until the CD4+ cell count rose to more than 350 cells/μl. Upon confirmation that the CD4+ cell count was more than 350 cells/μl, ART was to be stopped and resumed again when the CD4+ cell count was less than 250 cells/μl. During periods of ART use, the goal was to achieve maximal viral suppression. As previously reported, on 11 January 2006, investigators and participants were notified of a safety risk in the drug conservation group, enrollment was stopped, and participants in the drug conservation group were advised to restart ART .
Specimens collected for the biomarker substudy
In the SMART trial, follow-up study visits occurred at months 1 and 2, then every 2 months for the first year, and every 4 months thereafter. Five thousand one hundred fifty-one participants (94%) had blood stored for future use. Plasma specimens were collected using EDTA (lavender top) tubes, aliquoted, and shipped frozen to a central repository. Only specimens obtained prior to 11 January 2006 were used for the biomarker analyses described in this paper. Plasma specimens collected at randomization and 1 month after randomization were identified for participants who consented to specimen storage for future research. A random sample of 250 participants without a preexisting history of cardiovascular disease from each treatment group was chosen from 1286 participants with available samples. One participant in the drug conservation group who did not have an adequate specimen at randomization was subsequently excluded.
Measurement of Cystatin C and inflammatory markers
Cystatin C was measured on plasma samples that were stored at -70ºC. A BNII nephelometer (Dade Behring Inc., Deerfield, Illinois, USA) that utilized a particle-enhanced immunonephelometric assay (N Latex Cystatin C) was used. The assay range was 0.195-7.330 mg/dl. The reference range for young, healthy individuals was 0.53-0.92 mg/dl. The inflammatory markers were measured by the Laboratory for Clinical Biochemistry Research at the University of Vermont. Estimated glomerular filtration rate (GFR) was calculated using the modification of diet in renal disease (MDRD) formula  and the Cockcroft-Gault formula .
Descriptive statistics were used to demonstrate the change in cystatin C in the first year after randomization in the drug conservation versus viral suppression arm. Pearson correlation coefficients were used to determine the correlation between values of cystatin C and markers of inflammation, coagulation and lipids at randomization, marker values at randomization and change in cystatin C between randomization and month 1, marker values and cystatin C at month 1 and changes in marker and cystatin C values between randomization to month 1. Logistic regression, using forward selection with P value more than 0.2 inclusion criteria, was used to model the odds of at least 1 SD increase in cystatin C, 0.15 mg/dl, in the first month after randomization. Variables included in multivariate models were sex, age, race, mode of HIV infection, hepatitis B and C status, prior AIDS, HIV-RNA 400 copies/ml or less at randomization, current and prior smoking status, diabetes, treatment with blood pressure (BP) or lipid-lowering therapy, previous antiretroviral experience, ART status at enrolment, and exposure to protease inhibitors, nonnucleoside reverse transcriptase inhibitors (NNRTIs) or nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs). Randomization values for age, weight, BMI, CD4+ cell count, HIV-RNA levels, total cholesterol, HDL, LDL and triglycerides, as well as nadir CD4+ cell count were also included. All analyses were performed using SAS version 9.1 (SAS Institute Inc., Cary, North Carolina USA). All tests of significance were two sided.
Characteristics at randomization of participants assigned to the drug conservation and viral suppression arms in this substudy of the SMART trial were well balanced with no significant differences (P > 0.1 for all comparisons) (Tables 1 and 2). At randomization, 73.1 and 75.2% of drug conservation and viral suppression participants, respectively, were on ART. At randomization, mean (SD) cystatin C level was 0.99 (0.26 mg/dl) and 1.01 (0.28 mg/dl) in the drug conservation and viral suppression arms, respectively (P = 0.29). Ninety participants in the drug conservation arm (36.3%) had a cystatin C level above 1.0 mg/dl as compared with 101 participants in the viral suppression arm (41.2%, P = 0.26, chi-squared test).
The difference in cystatin C levels between the drug conservation and viral suppression arms together with the proportion of participants having a at least 0.15 mg/dl rise in cystatin C in the drug conservation and viral suppression arms is shown in Fig. 1. Six participants with cystatin C levels above 5 mg/dl at randomization were excluded from this analysis (one drug conservation and five viral suppression); inclusion of these participants did not alter our findings (data not shown). In the first month after randomization, there was a small rise in cystatin C (mean 0.05, SD 0.18 mg/dl) in the drug conservation arm and a small decrease in cystatin C in the viral suppression arm (mean -0.02, SD 0.17 mg/dl; P < 0.0001 for difference). The proportion of participants who experienced at least 0.15 mg/dl increase in cystatin C was significantly higher in the drug conservation arm (21.8%) compared with the viral suppression arm (10.6%; P = 0.0008, chi-square test). Cystatin C remained higher in the drug conservation arm than the viral suppression arm from month 1 through month 12 (Fig. 1). Highly consistent results were seen censoring the analysis at the date of restarting ART in the drug conservation arm (data not shown). There were no significant differences between the treatment arms in the proportion of participants with a cystatin C level of above 1.0 mg/dl at months 1, 4, 8 and 12 (P > 0.1, chi-squared test).
Other markers of kidney function were measured in a subset of participants at randomization, month 4 and month 12 of follow-up. There was no significant difference in serum creatinine or MDRD estimated GFR between participants in the drug conservation and viral suppression arms at any time point (Table 3) or when comparing participants from different races within the drug conservation and viral suppression arms (data not shown). In addition, there was no change in weight or albumin from randomization to month 4 or month 12 or from month 4 to month 12 in either treatment arm (P > 0.25, all comparisons). Among all participants, there was a significant correlation between cystatin C and MDRD estimated GFR at months 0, 4 and 12 (correlation coefficients -0.41, -0.22 and -0.35; P < 0.0001, P = 0.092, P < 0.0001, respectively) and between cystatin C and serum creatinine at months 0, 4 and 12 (correlation coefficients 0.234, 0.308 and 0.665; P < 0.0001 for all). Consistent results were seen between the drug conservation and viral suppression arms when GFR was estimated using the Cockcroft-Gault equation (data not shown).
Figure 2 shows the unadjusted odds ratio (OR) of more than 1 SD increase in cystatin C comparing the viral suppression arm with the drug conservation arm within subgroups of patients. For example, in patients with a CD4 cell count of 500/μl or less at randomization, 5.2% of participants in the viral suppression arm had at least 0.15 mg/dl increase in cystatin C compared with 22.2% in the drug conservation arm [OR 0.19; 95% confidence interval (CI) 0.07-0.53, P = 0.0015]. The corresponding proportions in participants with a CD4 cell count more than 500 cells/μl at randomization were 14.1 and 21.5% (OR 0.60, 95% CI 0.33-1.10, P = 0.098, P value for interaction = 0.059). The OR of more than 1 SD increase in cystatin C comparing the viral suppression arm with the drug conservation arm was similar in patients who were on or off antiretrovirals at baseline and in patients with undetectable plasma HIV-RNA (<400 copies/ml) compared with those with plasma HIV-RNA at least 400 copies/ml.
In a multivariate model adjusted for treatment arm, HIV exposure route, exposure to BP-lowering medication and previous treatment with protease inhibitors, participants in the viral suppression arm had less than half the odds of at least 0.15 mg/dl increase in cystatin C in the first month after randomization to the SMART trial compared with participants in the drug conservation arm (adjusted OR 0.42; 95% CI 0.23-0.74, P = 0.0023). There was a significantly higher odd of increased cystatin C in participants treated at randomization with BP-lowering medications (adjusted OR 1.91; 95% CI 1.06-3.44, P = 0.038) and those who had been treated previously with protease inhibitors (adjusted OR 2.17; 95%CI 1.07-4.40, P = 0.030). Participants infected with HIV through intravenous drug use had marginally significantly increased odds of an increase in cystatin C (adjusted OR 2.07; 95% CI 0.91-4.71, P = 0.077). There was no significant association between race and increase in cystatin C in univariate (P = 0.57) or multivariate analyses (P = 0.43). In addition, there was no relationship between exposure to specific NRTIs and change in cystatin C, although the power for this analysis was limited. These analyses were repeated, modeling the change in cystatin C rather than the proportion with at least 0.15 mg/dl increase with highly consistent results (data not shown). Multivariate analyses were also performed separately for the drug conservation and viral suppression arms. There was no qualitative difference in the association of variables with change in cystatin C at month 1 between the treatment groups (data not shown). In the subset of patients with cystatin C levels measured at 12 months after randomization (n = 283; 150 drug conservation versus 133 viral suppression), participants in the viral suppression arm had lower odds of at least 0.15 mg/dl increase in cystatin C at 12 months after adjustment for the same factors as shown in Table 4 (OR 0.38, 95% CI 0.13-1.10, P = 0.075), although this was of marginal significance.
The lowest and highest quintiles of cystatin C at randomization were 0.81 and 1.14, respectively; the correlation between cystatin C and serum creatinine at randomization in the lowest quintile was 0.27 (n = 57 observations, P = 0.045) and in the highest quintile was 0.71 (n = 57 observations, P < 0.0001). Additional correlations between cystatin C, inflammatory, coagulation and lipid biomarkers are shown in Table 4. The increase in cystatin C at 1 month after randomization was correlated with an increase in hsCRP, IL-6, amyloid A and D-dimer, and a decrease in HDL and total cholesterol between randomization and month 1. Biomarker values at randomization were added to the logistic regression model described above. After adjustment, none of the biomarkers were independently associated with at least 0.15 mg/dl increase in cystatin C at 1 month after randomization. Further analyses detected no correlation between randomization values of cystatin C and HIV-RNA in the drug conservation arm (Table 4, column 2; correlation coefficient 0.037, P = 0.57), but there was a weak correlation in the viral suppression arm (correlation coefficient 0.195, P = 0.0021). The correlation coefficients were similar in the drug conservation and viral suppression arms when considering the correlation between HIV-RNA at randomization and change in cystatin C over the first month (Table 4, column 3), and change in HIV-RNA versus change in cystatin C over the first month (Table 4, column 5). In addition, there was a weak correlation between HIV-RNA at 1 month after randomization and change in cystatin C in the drug conservation arm (correlation coefficient -0.141, P = 0.027), which was not apparent in the viral suppression arm (correlation coefficient 0.061, P = 0.34).