Evaluation of the Prognostic Value of Impaired Renal Function on Clinical Progression in a Large Cohort of HIV-Infected People Seen for Care in Italy

Whilst renal dysfunction, especially mild impairment (60<eGFR<90 ml/min), has been often described in HIV-infected population, its potential contribution to HIV evolution and risk of cerebro-cardiovascular disease (CCVD) has not been clarified. Data from HIV-1 infected patients enrolled in the Italian Cohort of Antiretroviral-Naïve (Icona) Foundation Study collected between January 2000 and February 2014 with at least two creatinine values available. eGFR (CKD-epi) and renal dysfunction defined using a priori cut-offs of 60 (severely impaired) and 90 ml/min/1.73m2 (mildly impaired). Characteristics of patients were described after stratification in these groups and compared using chi-square test (categorical variables) or Kruskal Wallis test comparing median values. Follow-up accrued from baseline up to the date of the CCVD or AIDS related events or death or last available visit. Kaplan Meier curves were used to estimate the cumulative probability of occurrence of the events over time. Adjusted analysis was performed using a proportional hazards Cox regression model. We included 7,385 patients, observed for a median follow-up of 43 months (inter-quartile range [IQR]: 21-93 months). Over this time, 130 cerebro-cardiovascular events (including 11 deaths due to CCVD) and 311 AIDS-related events (including 45 deaths) were observed. The rate of CCVD events among patients with eGFR >90, 60-89, <60 ml/min, was 2.91 (95% CI 2.30-3.67), 4.63 (95% CI 3.51-6.11) and 11.9 (95% CI 6.19-22.85) per 1,000 PYFU respectively, with an unadjusted hazard ratio (HR) of 4.14 (95%CI 2.07-8.29) for patients with eGFR <60 ml/min and 1.58 (95%CI 1.10-2.27) for eGFR 60-89 compared to those with eGFR ≥90. Of note, these estimates are adjusted for traditional cardio-vascular risk factors (e.g. smoking, diabetes, hypertension, dyslipidemia). Incidence of AIDS-related events was 9.51 (95%CI 8.35-10.83), 6.04 (95%CI 4.74-7.71) and 25.0 (95%CI 15.96-39.22) per 1,000 PYFU, among patients with eGFR >90, 60-89, <60 ml/min, respectively, with an unadjusted HR of 2.49 (95%CI 1.56-3.97) for patients with eGFR <60 ml/min and 0.68 (95%CI 0.52-0.90) for eGFR 60-89. The risk of AIDS events was significantly lower in mild renal dysfunction group even after adjustment for HIV-related characteristics. Our data confirm that impaired renal function is an important risk marker for CCVD events in the HIV-population; importantly, even those with mild renal impairment (90<eGFR<60) seem to be at increased risk of cerebro-cardiovascular morbidity and mortality.


Introduction
HIV infected individuals, because of the availability of effective and well tolerated antiretroviral treatments (ART), show considerably prolonged survival and are now facing a new set of health challenges. Several studies have addressed non-HIV related diseases in HIV-positive population, identifying aging, antiretroviral treatments and HIV itself as leading causes of premature atherosclerosis vascular disease leading to increased cardiovascular disease mortality in these patients [1]. Accumulated evidence in the general population suggests that reduced kidney function exposes patients to an increased rate of cerebro-cardiovascular disease (CCVD) both in populations at high risk for CCVD and even in the absence of traditional risk factors [2][3].
Proportion of patients with HIV infection and renal impairment of any grade are variable in published studies, ranging between 5% and up to 32% in some cases and mostly represented by mild renal dysfunction (90<eGFR<60 ml/min) [4][5][6][7]. Mild renal dysfunction is also frequent in HIV-infected population seen for care in Italy with a prevalence of 25% in ART-naïve patients [8].
Previous analysis that investigated the association between kidney disease and the risk of CCVD in HIV-population showed that both an eGFR less than 60 ml/min and albuminuria were associated with higher risk of CCVD and heart failure [9]. Moreover, severe renal dysfunction has been associated with increased risk of HIV disease progression and even death in several studies [10][11][12][13]. However, most of these studies evaluated patients not receiving ART or they have been conducted in the early highly active antiretroviral therapy era [9], thus concerning settings with a greater incidence of AIDS events. These studies often involve populations of patients living in settings where antiretroviral treatments are not largely available or there is limited access to healthcare [11][12], such as sub-Saharian Africa, exposing people to a greater risk of HIV progression. In addition, none of these studies focused on mild renal dysfunction and its predictive role on clinical outcome has not been assessed.
We, therefore, aimed to use the data of patients enrolled in the Icona Foundation Study to evaluate the prognostic value of reduced eGFR (both <60 ml/min and 60-90 ml/min) in predicting the occurrence of CCVD, AIDS defining diseases and consequent deaths over prospective follow-up.

Patient population
This analysis includes data from HIV-1 infected patients enrolled in the Italian Cohort of Antiretroviral-Naïve Patients (Icona) Foundation Study and collected after January 2000 (date of routine collection of creatinine for all patients). All information recorded in the database up to February 1st 2014. According to protocol inclusion criteria, patients had to be > 18 years old and start ART when they were naïve to antiretrovirals. In addition, for inclusion in this analysis, they had to have at least two creatinine values available after January 2000.
The ICONA Foundation Study is an Italian multi-centric prospective observational cohort study of HIV-1-positive persons enrolled since 1997. CD4+ T cell count, HIV-1-RNA, HCV antibody, HBV surface antigen and antibody and creatinine are systematically recorded every 6 months. Creatinine is measured using a method traceable to IDMS (upper limit of normal 1.3 mg/dL) at the local laboratories of the various sites. Clinical and laboratory data and data regarding any drug taken by the patient, either antiretroviral or else, are collected for all participants and recorded using an electronic data collection form (www.iconafoundation.it). All data are updated at the occurrence of any clinical event and, in the absence of such an event, at least every 6 months. AIDS-defining diseases and CCVD events captured by medical record abstraction are recorded in the database at the date that this diagnosis is confirmed.
The eGFR was used to identify patients in the cohort with potential renal dysfunction. To better estimate renal function, eGFR was calculated using the CKD-epi formula:GFR = 141 X min(Scr/κ,1) α X max(Scr/κ,1) -1.209 X 0.993 Age X 1.018 [if female] X 1.159 [if black]. Where Scr is serum creatinine (mg/dL), κ is 0.7 for females and 0.9 for males, α is −0.329 for females and −0.411 for males, min indicates the minimum of Scr/κ or 1, and max indicates the maximum of Scr/κ or 1.
This method of calculation recently emerged as a useful filtration marker even in HIV positive population, because of reducing bias for all traditional cardiovascular risk factors [14][15][16].
All individuals signed an informed consent prior to enrollment and the study was approved by the Ethics Committee of each participating institution that are listed in the Acknowledgments.

Outcome
The outcomes of interest were the occurrence of newly diagnosed cerebro-cardiovascular events or AIDS-related events. Only the first of these diagnoses were counted as events and relapses have been ignored in the analysis. The cerebro-cardiovascular endpoint included the following events: angioplasty, coronary artery bypass, cardiac arrhythmia, cerebral haemorrhage, carotid endarterectomy, cerebral stroke, dilated cardiomyopathy, other cardiac surgical procedures, death due to cardiovascular events. HIV-related events were defined as the occurrence of a new AIDS-related opportunistic infections or neoplasms (as defined by the Center for Disease Control and Prevention 1993 classification) and occurrence of AIDS-related deaths.

Statistical analysis
Baseline was defined as the date of the first of the two consecutive creatinine values measured after January 2000. Patients were grouped according to eGFR values calculated at baseline using the a priori defined cut-offs of 60 (severely impaired, if at least one value < 60 ml/min) and 90 ml/min/1.73m 2 (mildly impaired, if at least two values between 60-89 ml/min). Characteristics of patients were described after stratification in these groups and compared using chisquare test (categorical variables) or Kruskal-Wallis test comparing median values (quantitative measures).
Follow-up accrued from baseline up to the date of the event of interests or last available clinical follow-up visit. Rates of events have been calculated as number of new diagnoses occurring after baseline divided by number of person years of follow-up (PYFU). Confidence intervals for these estimates were calculated using a Poisson approximation. Survival analysis has also been employed. Kaplan Meier curves were used to estimate the cumulative probability of occurrence of the events by longer time from baseline. Log-rank test was used to compare survival of patients according to different eGFR strata. A multivariable Cox regression analysis was performed to control for confounding. A sequential manual adjustment was used so that estimates were first adjusted for demographic factors (i.e. age, gender, mode of HIV transmission, nationality and calendar year at baseline). Then they were further adjusted for other set of covariates such as traditional risk factors for CCVD disease and other clinical parameters (i.e. smoking, diabetes, use of blood pressure and lipids lowering drugs, total cholesterol, blood glucose and high density lipoprotein cholesterol [HDL]). Finally a further adjustment was made by adding virus-host-and treatment-related factors such as use of ART, baseline CD4, CD8 count and nadir CD4 count, HIV RNA and co-infection with HBC/HCV into the model. Unless noted, only baseline values of time-varying covariates were included in the model. Relative hazards of cardiovascular disease according to defined eGFR strata have been estimated and tabulated together with 95% confidence intervals. eGFR was also examined in the log10 scale because the distribution was skewed in the raw scale. All statistics were 2-sided and a p-value 0.05 was considered as significant. SAS version 9.3 was used for all analyses.

Study population
Seven thousand three hundred eighty-five patients were included in the study and observed over a median follow-up of 43 months (inter-quartile range [IQR]: 21-93 months).
Patients were followed-up for a total of 35,968 years for an overall incidence of cerebrocardiovascular events of 3.6 (95% CI: 3.0-4.3) events per 1,000 PYFU. During study follow-up, 311 AIDS-events occurred, of whom 45 (14.5%) were AIDS-related deaths. In this analysis a total follow-up of 35,393 years were accumulated for an overall incidence of AIDS-related events of 8.8 (95% CI:7.8-9.8) events per 1,000 PYFU. A description of the type of cardiovascular and HIV-related events is reported in Table 2.

Impaired renal function and risk of AIDS-related events
An increased incidence of AIDS-related events was found in patients with deeply reduced renal function (eGFR<60 ml/min) as reported in Table 5 eGFR 90, 60-89, <60 ml/min, respectively (Fig 2). Again, in the unadjusted analysis the incidence rates among patients with reduced renal function (eGFR<60 ml/min) was significantly higher than among HIV-infected subjects with normal renal function (eGFR90 ml/min), with an unadjusted hazard ratio of 2.49 (95%CI 1.56-3.97). In contrast, HIV-infected subjects with mild renal impairment (60eGFR<90 ml/min) showed decreased incidence rates of HIV-related events as compared to patients with normal renal function (eGFR90 ml/min), with an unadjusted hazard ratio of 0.68 (95% CI 0.52-0.90). Reduced renal function (eGFR<60 ml/min) was independently associated with a higher risk of developing

Discussion
In a large cohort of antiretroviral treatment-naïve patients seen for care in Italy and starting ART, our primary goal was to evaluate the role of renal function measurement in predicting clinical events, especially cerebro-cardiovascular events, AIDS-related outcomes and even death from both causes.
In the great majority of previously published reports [9][10][11][12][13], kidney disease has been defined by classifying patients in two categories: those with eGFR<60 ml/min and those with eGFR>60 ml/min; in contrast, our study additionally focuses on mild renal impairment, defined as a 60>eGFR<90 ml/min, as before its predictive value has not been extensively investigated despite its high prevalence in HIV-positive populations.
Indeed, while mild renal dysfunction (60<eGFR<90) is a relatively frequent condition in our cohort of HIV-infected patients (26%), we found a low prevalence of severe renal impairment (3% of population had eGFR<60ml/min) at baseline. This was anticipated as people are a relatively healthy population enrolled when they are antiretroviral-naïve by study design.
Our study highlights the existence of a strong link between kidney function and CCVD risk in HIV-infected individuals. Indeed, we observed a marked gradient in the association between eGFR levels and the risk of cerebro-cardiovascular disease, with an intermediate excess of risk of CCVD in patients with mild renal impairment (eGFR 60-89 ml/min) and a greater risk in patients with severe chronic disease (eGFR< 60 ml/min). Importantly, we found that lower eGFR was associated with a significantly increased risk of CCVD even after adjustment for baseline demographic characteristics of patients, HIV-associated factors and traditional cardiovascular risk factors (smoking, diabetes, hypertension, use of lipid lowering drugs, total and HDL cholesterol values, blood glucose levels). This held true both in the analysis using the categorical variable comparing different levels of renal impairment and in the analysis using eGFR as continuous in the log scale. Thus, our results confirm that reduced impaired renal function is an important risk factor for cerebro-cardiovascular events in the HIV-population. As outlined above, one of the key finding is that, compared to other studies investigating the association between kidney function and the risk of CCVD 9 , this is the first to identify HIV-infected patients with mild renal impairment (90<eGFR<60) as a population at increased risk of cerebro-cardiovascular morbidity and mortality. This result is however consistent with that of multiple studies conducted in general population, which showed that kidney disease is associated with dyslipidemia, anemia, left ventricular hypertrophy, arterial stiffness, inflammation and endothelial dysfunction [17].
Conversely, in our analysis, HIV-associated parameters do not seem to play a substantial role in the development of CCVD events; indeed in our population only age and ethnicity were independent markers of CCVD risk, while poor immunological condition, represented by CD4 cells count and CD4 cells nadir, did not predict cerebro-cardiovascular events. These finding differ from other studies which found that lower CD4+ count, together with traditional CCVD risk factors and duration of ART are associated with increased incidence of CCVD events [18][19]. But recent data from D.A.D. And COHERE show the same thing i.e. That CD4 count does not predict risk of MI and CVD events in general. This difference might be explained by the characteristics of the patients in our cohort (CD4 count were high on average at the date in which the first eGFR was measured) and the analysis takes into account only baseline CD4 count and not the current value.
Our study highlights the importance of estimation of filtrate through creatinine measurement as a clinical tool for identifying HIV-infected individuals at elevated risk for cerebro-cardiovascular events, and strengthens the fact that managing traditional risk factors in HIV population with renal impairment should be a priority.
Kidney disease is also feared as a predisposing condition for progression toward AIDS and death. In our analysis, initial serious renal impairment (eGFR<60) seems to be associated with an increase in number of AIDS related illnesses. However, this link was not confirmed after adjustment for the presence or history of antiretroviral treatment, CD4+ T cells nadir, baseline CD4+ and CD8+ T cells count, and baseline HIV viral load together with chronic hepatitis co-infections. This is consistent with the results of Sczech et al who found that dipstick proteinuria but not inverse creatinine was significantly associated with the development of AIDSdefining illness [10]. Unfortunately dipstick proteinuria was not available for our patients. This finding is suggestive for the many sided clinical picture of patients who usually develops AIDS defining illnesses in which HIV-related parameters, as a mirror of immunological condition, play a pivotal role in determining collapse toward AIDS development. On the other hand, advanced HIV disease is also associated with renal dysfunction and it is therefore a confounder for the association between renal disease and risk of AIDS and AIDS-related death. All the parameters in our analysis have been measured concomitantly at baseline and therefore it is difficult to understand dynamic mechanisms that might have led to disease progression. Even in a situation in which prospective measures of both creatinine and CD4 count are considered is often difficult to establish links of causality. Nevertheless, it is conceivable that renal impairment is likely to be a consequence and not the cause of a worse immunological situation. Of interest, participants with mild dysfunction were at lower risk of HIV progression and the association remained significant after controlling for a number of potential confounding factors. This might be due to the combined effect of low impact of renal disease on long term HIV disease progression and the fact that these patients were more able to tolerate and respond well to antiretroviral treatment. Some of this question might be answered by including timedependent confounding but more sophisticated methods than standard survival analysis are needed to disentangle the complex confounding issue of time-dependent markers affected by prior use of ART [20] Our results, however, differ from those of other studies assessing and justifying the development of renal abnormalities as predictors for AIDS progression [10][11]. Gupta et al, found that lower renal function defined as eGFR < 60 ml/min was independently associated with an increased risk of HIV disease progression [12]. Reasons for the discrepancy are unclear but might be attributable to the different endpoint used (time to developing either a CD4 count <200 cells/mm3 or WHO stage 3 or 4 disease) clinical setting, patients ethnicity or a combination of these.
In a recent analysis Lucas G et al evaluated the association of eGFR, estimated by three different equations (Cockcroft/Gault, MDRD and CKD-Epi with and withous the inclusion of cystatin C besides creatinine), with clinical events in HIV-infected patients in the SMART study and found that eGFR estimates using plasma cystatin C had stronger associations with mortality, cardiovascular events and opportunistic diseases than eGFR based on plasma creatinine [21]. Indeed, one limit of our study is the unavailability of other markers of kidney damage, expecially in patients with eGFR above 60 ml/min, for whom CKD-EPI, like other existing GFR-estimating equations have increasing bias. Unfortunately, plasma cystatin C was not available in our patients and its predictive value in patients receiving ART should be further investigated.
In conclusion, in this large cohort of HIV-infected patients seen for care in Italy, we observed a strong association between levels of eGFR and cerebro-cardiovascular diseases, while severe renal impairment did not predict HIV-related events. A major strength of our study is the analysis of the predictive role of mild renal impairment. It is important to increase our understanding of the mechanisms leading to severe non-AIDS morbidity anf mortality in the context of the emerging clinical challenge of treating an ageing HIV-infected population. These results may improve clinical management of HIV-infected patients by helping clinicians in identifyng people at higher risk of CCVD and provide a rationale for the evaluation of therapeutic strategies aimed to ameliorate patients' CCVD risk profile.