Statins and Renin Angiotensin System Inhibitors Dose-Dependently Protect Hypertensive Patients against Dialysis Risk

Background Taiwan has the highest renal disease incidence and prevalence in the world. We evaluated the association of statin and renin–angiotensin system inhibitor (RASI) use with dialysis risk in hypertensive patients. Methods Of 248,797 patients who received a hypertension diagnosis in Taiwan during 2001–2012, our cohort contained 110,829 hypertensive patients: 44,764 who used RASIs alone; 7,606 who used statins alone; 27,836 who used both RASIs and statins; and 33,716 who used neither RASIs or statins. We adjusted for the following factors to reduce selection bias by using propensity scores (PSs): age; sex; comorbidities; urbanization level; monthly income; and use of nonstatin lipid-lowering drugs, metformin, aspirin, antihypertensives, diuretics, and beta and calcium channel blockers. The statin and RASI use index dates were considered the hypertension confirmation dates. To examine the dose–response relationship, we categorized only statin or RASI use into four groups in each cohort: <28 (nonusers), 28–90, 91–365, and >365 cumulative defined daily doses (cDDDs). Results In the main model, PS-adjusted hazard ratios (aHRs; 95% confidence intervals [CIs]) for dialysis risk were 0.57 (0.50–0.65), 0.72 (0.53–0.98), and 0.47 (0.41–0.54) in the only RASI, only statin, and RASI + statin users, respectively. RASIs dose-dependently reduced dialysis risk in most subgroups and in the main model. RASI use significantly reduced dialysis risk in most subgroups, regardless of comorbidities or other drug use (P < 0.001). Statins at >365 cDDDs protected hypertensive patients against dialysis risk in the main model (aHR = 0.62, 95% CI: 0.54–0.71), regardless of whether a high cDDD of RASIs, metformin, or aspirin was used. Conclusion Statins and RASIs independently have a significant dose-dependent protective effect against dialysis risk in hypertensive patients. The combination of statins and RASIs can additively protect hypertensive patients against dialysis risk.


Introduction
In Taiwan, 92.4% of patients with renal diseases undergo hemodialysis; this percentage is 91.7% in the United States and 18.7% in Hong Kong [1]. The mean total lifetime treatment cost for dialysis patients is NT$6,112,755 ± NT$317,559 [2]. Furthermore, Taiwan has the highest incidence and prevalence of renal diseases and dialysis use worldwide [3]. The costeffect problem in the Taiwanese National Health Insurance (NHI) system for dialysis use has emerged as a public health burden. Therefore, introducing an optimal therapy to avoid dialysis use among susceptible patients may aid in reducing national expenditure in the NHI program.
Hypertension, a major cause of renal diseases [3], is frequently seen in patients with acute and chronic renal diseases, particularly glomerular and vascular disorders [4]. Hypertension may primarily be caused by fluid overload, as indicated by a suppressed renin-angiotensinaldosterone system and enhanced atrial natriuretic peptide release [5]. Hypertension is presented by 80%-85% of patients with chronic kidney disease (CKD) [6]. In patients with CKD, hypertension likely occurs because of a combination of factors including sodium retention, increased renin-angiotensin system activity, and enhanced sympathetic nervous system activity [7]. Hypertension is also common in acute vascular diseases, such as vasculitis and scleroderma renal crisis. In these settings, blood pressure increases because of ischemia-induced renin-angiotensin system activation, rather than volume expansion [8]. Renin-angiotensin system inhibitors (RASIs), including angiotensin-converting enzyme inhibitors (ACEIs), angiotensin II receptor blockers (ARBs), and direct renin inhibitors, are commonly used in hypertension treatment. Furthermore, inhibiting angiotensin II formation with an ACEI is effective in patients with vasculitis or scleroderma renal crisis [9]. In patients with proteinuric CKD, an ACEI or ARB is recommended in the first-line hypertension therapy [10][11][12][13]. However, no clear evidence indicating that early RASI use reduces dialysis risk in hypertensive patients without CKD has been reported.
Indirect evidence has indicated the beneficial effects of statins on vessel stiffening and endothelial function in patients with CKD [14,15]. After renal injury, dyslipidemia may accelerate and perpetuate the yearly decline in the glomerular filtration rate (GFR) [16][17][18]; however, this effect has been confirmed through post hoc analyses, which can be limited by unmeasured confounders closely correlated with dyslipidemia [18,19]. If present, the aforementioned effect is extremely uncertain and may require many trials to obtain conclusive results [20]. Two metaanalyses of small-scale randomized trials have demonstrated that statin therapy significantly alleviates albuminuria [21,22]. However, the patients included in these trials were not uniformly using RASIs. By contrast, two large-scale randomized trials have revealed that statins do not affect albumin excretion in patients receiving optimal RASI therapy to reduce CKD progression and achieve satisfactory blood pressure control [23,24]. Thus, conflicting data concerning the effect of statins on renal disease progression have been reported [11,[25][26][27]. Most data derived from large-scale intervention studies, with hard clinical endpoints, have suggested that statins do not prevent renal function loss [28][29][30]. All trials evaluating the effects of statin therapy on renal disease progression have used subset analyses of trials designed to evaluate the efficacy of statin therapy in treating cardiovascular disease in patients with CKD [31,32]. However, experimental evidence has indicated that reducing lipid levels by using a drug such as lovastatin reduces renal injury progression [33][34][35].
Currently, statins and RASIs are not recommended for renal protection alone in hypertensive patients without CKD [36,37]. In this study, we clarified the potential protective effects of statins and RASIs against dialysis risk in hypertensive patients without CKD.

Materials and Methods
In Taiwan, the NHI program, established in 1995, currently provides comprehensive health insurance coverage to 98% of the population of more than 23 million people. We used data from the National Health Insurance Research Database (NHIRD). Distributions of age, sex, and health care costs in the NHIRD and among NHI enrollees do not differ significantly. Data that can be used to identify patients or care providers, including the names of medical institutions and physicians, are encrypted before being sent to the National Health Research Institutes (NHRI) for inclusion in the NHIRD. The NHRI further encrypts the data before releasing the database to researchers. Theoretically, the NHIRD data alone are insufficient to identify any individual. All researchers using the NHIRD and its data subsets must sign a written agreement declaring that they have no intention of attempting to obtain information that could potentially violate the privacy of patients or care providers [38][39][40].
Our study cohort comprised all patients who received a hypertension diagnosis (according to International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes) at all health care facilities in Taiwan (n = 248,791) between January 1, 2001, and December 31, 2012. We excluded patients without subsequent outpatient visits, subsequent antihypertension medications, and emergency department visits or inpatient hospitalizations for hypertension within 12 months of first presentation (n = 37,881), because these patients were considered to not have hypertension (Fig 1). In Taiwan, most dialysis patients are >40 years old [41] [2], and <40-year-old patients rarely receive a diagnosis of hypertension [42]. Thus, we focused only on patients aged >40 years. Consequently, we excluded 64,693 patients aged <40 years (n = 125,849), those with any inpatient or outpatient diagnosis associated with CKD before the date of cohort entry (n = 3,484), those with any inpatient or outpatient diagnosis associated with dialysis before the date of cohort entry (n = 39), those with any inpatient or outpatient diagnosis associated with renal transplantation before the date of cohort entry (n = 7), those having a RASI prescription before the date of cohort entry (n = 7,596), and those who had a statin prescription before the date of cohort entry (n = 3,894).
Our final study cohort comprised 110,829 patients with hypertension; of them, 44,764 used RASIs alone, 7,606 used statins alone, 27,836 used both RASIs and statins, and 33,716 used neither RASIs nor statins (Table 1). After literature review [43][44][45][46][47], we selected covariates on the basis of a logistic regression model. Each patient was followed to assess dialysis risk and protective factors. We evaluated the following demographic characteristics by using propensity scores (PSs): age; sex; diabetes; dyslipidemia; cerebrovascular disease; cardiovascular disease; hepatitis B and C virus infection; liver cirrhosis; moderate and severe liver disease; asthma; use of antihypertensives, diuretics, beta and calcium channel blockers, nonstatin lipid-lowering drugs, metformin, and aspirin; urbanization level; and monthly income (S1 Table). The index date of statin and RASI use was considered the confirmation date of hypertension. To evaluate the  protective effects of statins and RASIs against dialysis risk in hypertensive patients, dialysis risk was considered the primary endpoint, whereas the secondary endpoints were the differential effects of various doses of and additive effect of RASIs and statins. The defined daily dose (DDD)-recommended by the World Health Organization-is a measure of the prescribed drug amount. The DDD is the assumed average maintenance dose per day of a drug consumed for its main indication in adults [40]. To examine the dose-response relationship, we categorized statin use into four groups in each cohort (<28, 28-90, 91-365, and >365 cumulative DDDs [cDDDs]) because the duration of the refill card was 3 months. Patients receiving <28 cDDDs were defined as nonusers (Tables 2-4) [48]. Furthermore, to examine the additive effect of RASI and statin use, we used sensitivity analysis of adjusted hazard ratios (aHRs) of RASIs and statins in reducing dialysis risk (Tables 2-4). PSs were derived using a logistic regression model to estimate the effect of RASIs and statins by accounting for the covariates predicting intervention (statins and RASIs) receipt. All potential confounders were included in the list of regressors (C statistic: 0.684). This method is used in observational studies to reduce selection bias [49]. The following covariates in the main model were adjusted according to the PS: age; sex; diabetes; dyslipidemia; cerebrovascular disease; cardiovascular disease; hepatitis B and C virus infection; liver cirrhosis; moderate and severe liver disease; asthma; use of antihypertensives, diuretics, beta and calcium channel blockers, nonstatin lipid-lowering drugs, metformin, and aspirin; urbanization level; and monthly income (NT$0, NT$1-NT$21,000, NT$21,000-NT$33,300; and !NT$33,301; Table 2). The endpoint for users of RASIs alone, statins alone, and RASIs + statins and nonusers was the recommendation of dialysis (ICD-9-CM V45.11 or V45.12), with a subsequent outpatient visit, emergency department visit, or inpatient hospitalization for any dialysis treatment within 12 months of diagnosis; nonusers were treated as the reference arm.
A time-dependent Cox proportional hazard model was used to calculate the HRs of dialysis risk in the users of RASIs alone, statins alone, and RASI + statin and nonusers. In the multivariate analysis, the HRs were adjusted for the aforementioned covariates. All analyses were conducted using SAS version 9.3 (SAS, Cary, NC, USA); two-tailed test results with P < 0.05 were considered significant. In sensitivity analyses, external adjustments are used to improve the understanding of the effects of drugs and other covariates in epidemiological database studies [50]. Hence, in our sensitivity analysis, data were adjusted in different models to estimate the association of dialysis incidence with age; sex; diabetes; dyslipidemia; cerebrovascular disease; cardiovascular disease; hepatitis B and C virus infection; liver cirrhosis; moderate and severe liver disease; asthma; and use of antihypertensives, diuretics, beta and calcium channel blockers, nonstatin lipid-lowering drugs, statins, RASIs, metformin, and aspirin.

Results and Discussion
Compared with the users of other drugs, only statin users exhibited a higher prevalence of preexisting medical comorbidities including cerebrovascular disease, cardiovascular disease, and dyslipidemia (all P < 0.001). In addition, significant differences were observed among the four groups in the distributions of age; sex; monthly income; urbanization level; and use of nonstatin lipid-lowering drugs, aspirin, RASIs, and metformin (Table 1). A higher proportion of nonusers used nonstatin lipid-lowering drugs, metformin, and aspirin at <28 cDDDs; however, most RASI or statin users used these drugs at >365 cDDDs. A lower proportion of statin nonusers had a monthly income of !NT$33,301 or resided in urban areas.
In the sensitivity analysis, PS adjustments were made to estimate the associations of age; sex; diabetes; dyslipidemia; cerebrovascular disease; cardiovascular disease; hepatitis B and C virus infection; liver cirrhosis; moderate and severe liver disease; asthma; and use of antihypertensives, diuretics, beta and calcium channel blockers, nonstatin lipid-lowering drugs, metformin, and aspirin with the incidence of dialysis in different models. Table 2 shows that the effects of the use of RASIs alone, statins alone, or RASIs + statins remained significant in the different groups when the main model was PS adjusted. A stratified sensitivity analysis was performed to evaluate the dialysis risk among the users of different drugs. After PS adjustments for the main model, aHRs (95% confidence intervals [CIs]) of dialysis were 0.57 (0.50-0.65) for those using RASIs alone, 0.72 (0.53-0.98) for those using statins alone, and 0.47 (0.41-0.54) for those using RASIs + statins (Table 2). Table 2 also shows that the effects of RASI + statin use remained significant in the subgroups of various covariates namely age; sex; diabetes; cerebrovascular disease; cardiovascular disease; asthma; and use of antihypertensives, diuretics, beta and calcium channel blockers, and metformin. The combined use of RASIs and statins might have the highest potential for reducing dialysis risk, as indicated by the RASI + statin group having the lowest aHR among all groups. The effects were nonsignificant in users of RASIs alone, statins alone, and RASIs + statins when the cDDDs of nonstatin lipid-lowering drugs were moderate to high (>28) or when those of aspirin were high (>365). When the dose of metformin was moderate to high, the effects were significant only for the RASI + statin group.
RASIs dose-dependently reduced dialysis risk in most subgroups and the main model ( Table 3). All aHRs indicated that RASI use caused significant reductions in dialysis risk in most subgroups, regardless of comorbidities or drug use (P < 0.001). Our data revealed that RASI use, with a dose-dependent effect frequency, has a protective effect against dialysis risk, which was particularly predominant in female patients using RASIs at >365 cDDDs (aHR = 0.42, 95% CI: 0.34-0.54) and in patients without diabetes who used RASIs at >365 cDDDs (aHR = 0.38, 95% CI: 0.32-0.45), those without cerebrovascular disease who used RASIs at >365 cDDDs (aHR = 0.45, 95% CI: 0.39-0.51), and those not using antihypertensives,  diuretics, beta and calcium channel blockers, statins, nonstatin lipid-lowering drugs, metformin, or aspirin (P < 0.001). If the patients used high cDDDs of statins, nonstatin lipid-lowering drugs, or aspirin, no protective effect against dialysis risk was observed, even for high RASI cDDDs. If the patients used high cDDDs of nonstatin lipid-lowering drugs, no protective effect against dialysis risk was noted, even when RASIs were used; in addition, the aHRs significantly increased with the dose of nonstatin lipid-lowering drugs, but not with the dose of RASIs. As presented in Table 4, we performed a sensitivity analysis with PS adjustments in the main model for age; sex; diabetes; dyslipidemia; cerebrovascular disease; cardiovascular disease; hepatitis B and C virus infection; liver cirrhosis; moderate and severe liver disease; asthma; urbanization level; monthly income; and use of antihypertensives, diuretics, beta and calcium channel blockers, RASIs, nonstatin lipid-lowering drugs, metformin, and aspirin. Statins at >365 cDDDs conferred a protective effect against dialysis risk in hypertensive patients in the main model (aHR = 0.62, 95% CI: 0.54-0.71). This protective effect was more predominant in patients with cerebrovascular disease (aHR = 0.59, 95% CI: 0.37, 0.59), those with asthma (aHR = 0.56, 95% CI: 0.32, 0.96), and those not using diuretics, beta and calcium channel blockers, RASIs, nonstatin lipid-lowering drugs, metformin, or aspirin (all P < 0.001). If patients used high cDDDs of nonstatin lipid-lowering drugs, no protective effect against dialysis risk was observed, even for high cDDD statin use; furthermore, aHRs increased nonsignificantly with the dose of nonstatin lipid-lowering drugs. Therefore, regardless of whether high cDDDs of RASIs, metformin, or aspirin were used, high cDDDs of statins had a protective effect against dialysis risk in the statin alone group. RASIs are used in first-line antihypertensive therapy in all patients with heart failure or asymptomatic left ventricle dysfunction, those with anterior myocardial infarction and diabetes or systolic dysfunction, and those with proteinuric CKD [51,52]. RASIs have a cardioprotective effect, independent of blood pressure lowering noted in patients at a high risk of cardiovascular events. Hypertension can be a causative or contributory factor in kidney disease development [53]. No data are available for estimating the protective effect of RASIs in reducing dialysis risk in hypertensive patients without CKD. According to our research, our study is the first to report a dose-dependent effect of RASIs in reducing dialysis risk in hypertensive patients without CKD. We observed that in addition to their cardioprotective effect, RASIs have a protective effect against dialysis risk in hypertensive patients. The use of RASIs dosedependently reflected their protective effect against dialysis risk in hypertensive patients without CKD (Table 3). Hyperlipidemia is common in patients with renal diseases, particularly nephrotic syndrome [54]. In addition to accelerating systemic atherosclerosis development, experimental studies have suggested that high lipid levels also may promote renal disease progression [55,56]. The major experimental evidence supporting this hypothesis in animals is that loading cholesterol increases glomerular injury and that reducing lipid levels by using statins slows injury progression [33][34][35]. Furthermore, the beneficial lipid-lowering effect may be supplementary to that of blood pressure lowering, as observed in some renal disease models [34]. However, the factors responsible for the lipid-lowering effects remain unclear. In various animal models, high cholesterol intake can be deleterious, causing an increase in intraglomerular pressure [35]; by contrast, lipid-lowering drugs, which do not affect glomerular hemodynamics, can be more beneficial [56]. These contradictory observations suggest that in addition to intraglomerular pressure, other mechanisms may contribute to blood pressure lowering. Moreover, statins may act independent of plasma lipid levels by directly inhibiting mesangial cell proliferation and monocyte chemoattractant production [57,58]. The applicability of these findings to human diseases is uncertain; hence, our present human data are valuable. Numerous secondary analyses of data from lipid-related trials have suggested that high lipid levels accelerate renal disease progression, whereas statins delay this progression. In the current study, statins at >365 cDDDs conferred an independent protective effect against dialysis risk to hypertensive patients (aHR = 0.62, 95% CI: 0.54-0.71) in the PS-adjusted main model ( Table 4). The combined use of RASIs and statins may have the highest potential in reducing dialysis risk with the smallest aHR (0.47, 95% CI: 0.41-0.54) compared with that of the use of RASIs alone or statins alone ( Table 2). The current data regarding the additive effect of dialysis risk reduction in humans corroborate those of a preclinical study [34]. The novelty of our study is the establishment of clinical data demonstrating that statins confer an independent protective effect against dialysis risk to hypertensive patients, and this can be further enhanced through combined use of statins and RASIs.
The effects of RASIs alone, statins alone, or RASIs + statins were nonsignificant when the cDDDs of nonstatin lipid-lowering drugs were moderate to high (>28) or when those of aspirin were high (>365). If a moderate-to-high dose of metformin was used, the effect was significant only when RASIs and statins were used in combination. A history of diabetes, hypertension, cerebrovascular disease, or cardiovascular disease can confer the highest CKD risk to patients [59]. In this study, the protective effect was more predominant in female patients, those without diabetes, and those without cerebrovascular disease (Table 2). These outcomes correspond to those of a previous study [59]. Our findings imply that early RASI and statin use by hypertensive patients without diabetes or cerebrovascular disease might strengthen its protective effect against dialysis risk. Higher cDDDs of aspirin for cerebrovascular disease, metformin for diabetes, and nonstatin lipid-lowering drugs for hyperlipidemia might correspond to disease severity and duration, both of which cannot be PS adjusted. Thus, the severity and duration of comorbidities might mask the protective effect of RASIs or statins (Tables 2-4). The following was most predominantly observed in nonstatin lipid-lowering drug users: no protective effect against dialysis risk, even with RASI or statin use, and aHRs increased with an increase in the cDDDs of nonstatin lipid-lowering drug use (Tables 2-4); this is because poor hyperlipidemia control is a major risk factor for renal disease progression, according to preclinical studies [55,56]. In addition, the protective effect against the dialysis risk of statins might be superior to that of nonstatin lipid-lowering drugs, according to the highest aHR (2.29) being observed in hypertensive patients using >365 cDDDs of nonstatin lipid-lowering drugs and 28-90 cDDDs of statins (Table 4). This is the first study with human clinical data demonstrating that statins have a stronger protective effect against dialysis risk than nonstatin lipid-lowering drugs do in hypertensive patients without CKD, potentially because statins directly inhibit mesangial cell proliferation and monocyte chemoattractant production [57,58].
Regardless of whether high doses of RASIs, metformin, or aspirin are used, the protective effect against dialysis risk was observed with the use of high cDDDs of statins in the statin alone group (Table 4). These findings suggest that the pharmacological mechanism underlying the protective effect of statins is independent from that of RASIs. Although diabetes, hypertension, and cardiovascular disease are risk factors for CKD [59], statins have a significant protective effect, even when >365 cDDDs of metformin, aspirin, or RASIs are used (Table 4). By contrast, no protective effect is observed when >365 cDDDs of RASIs are used with >365 cDDDs of statins or aspirin (Table 3). Therefore, statins have independent protective and additive effects along with RASIs against dialysis risk in hypertensive patients. Two randomized trials have demonstrated that statins combined with angiotensin blockers do not slow CKD progression but lead to favorable blood pressure control [23,24]. This may be because the doses and duration of statin use in these trials were insufficient. In the future, randomized trials considering higher doses and longer durations of statin use should be conducted.
This study has six limitations. First, different statin and RASI types were considered but not analyzed separately; thus, the potential effects of a specific statin or RASI remain unknown. Second, evidence from observational studies has suggested that lifestyle factors, particularly social, mental, and physical activities, are inversely associated with dialysis risk. However, methodological concerns may obscure the precise relationship between these factors and dialysis risk. In our study, we used PSs to match age, sex, comorbidities, urbanization level, and monthly income. Third, urbanization level and monthly income were used as unvalidated alternatives to lifestyle factors. To obtain such information regarding the actual factors, a largescale randomized trial should be conducted along with a suitable regimen and appropriately selected patients to compare standard approaches. Fourth, in the present study, dialysis recommendation and comorbidity diagnoses were completely dependent on the ICD codes. Nevertheless, the NHI Administration randomly reviews medical records and interviews patients to validate diagnoses. Hospitals with outlier diagnoses and practices may be audited and subsequently heavily penalized if malpractice or discrepancies are discovered. Fifth, the NHIRD contains no information on several unmeasured confounders including body mass index, laboratory data, compliance with drug use, smoking status, alcohol intake, and use of other dialysis-associated over-the-counter drugs. However, if patient compliance is poor, the drug effects are underestimated, causing bias toward the null hypothesis [60]. Thus, in cases of poor patient compliance, the true effects of statins or RASIs may have been underestimated. Considering the magnitude and significance of the observed effects, it is unlikely that this limitation compromised the results. Finally, our study was not prospective, randomized, or blinded; hence, a cause-effect relationship could not be established. The findings of this study suggest that statins or RASIs independently exert a significant protective effect against dialysis risk in hypertensive patients in a dose-dependent manner. The combined use of statins and RASIs has an additive effect against dialysis risk in hypertensive patients. Additional randomized studies are warranted to verify our findings.

Conclusions
Statins and RASIs independently exert a significant dose-dependent protective effect against dialysis risk in hypertensive patients without CKD. Statins in combination with RASIs can additively protect hypertensive patients against dialysis risk.