Distinct impact of three different statins on arteriovenous fistula outcomes: a retrospective analysis

Abstract

Purpose:

Whether statins improve arteriovenous fistula (AVF) outcomes is still a matter of debate. Taking into consideration the existing physicochemical differences between individual drugs, this study evaluates the impact of three different statins (atorvastatin, rosuvastatin and simvastatin) on one-stage and two-stage AVF outcomes.

Methods:

Using a retrospective cohort of 535 patients, we analyzed the effects of each statin on primary failure and primary patency using multivariate logistic regressions and Cox proportional hazard models.

Results:

Out of the three statins analyzed, only atorvastatin improved the overall primary failure of AVF (odds ratio [OR] = 0.18, p = 0.005). Comparisons between the two AVF types demonstrated that this effect was due to a prominent reduction in primary failure of one-stage (OR = 0.03; p = 0.005), but not two-stage fistulas (OR = 0.43; p = 0.25). In contrast, primary patency of two-stage (hazards ratio [HR] = 0.51; p = 0.024), but not one-stage fistulas (HR = 0.98; p = 0.95), was improved by all statins as a group, but not by individual drugs.

Conclusions:

Our results suggest that the potential benefit of statins on AVF outcomes is a drug-specific and not a class effect, and that such effect is also influenced by the type of fistula.

Introduction

Arteriovenous fistulas (AVFs) are the preferred vascular access for hemodialysis. However, 30%−50% of AVFs fail to mature independently or require an intervention shortly after creation (1–3). This problem is known as AVF primary failure, which exposes patients to additional morbidities and surgical risks due to the need for salvage procedures, increased use of central venous catheters, or the creation of less favorable permanent vascular accesses. Most cases of primary failure are caused by insufficient dilation of the vein and/or occlusion due to stenosis in the juxtaposed venous segment (4). Unfortunately, there is no drug at the moment that is indicated to prevent primary failure.

Statins are effective lipid-lowering drugs with beneficial pleiotropic effects for a variety of vascular remodeling processes. They are known to improve mortality rates and procedural outcomes in cases of acute coronary syndrome (5), heart failure (6), percutaneous coronary intervention (7), coronary artery bypass grafting (8), peripheral vascular disease (9), and abdominal aortic aneurysms (10). The vasculoprotective and anti-inflammatory effects of statin therapy have been attributed to both lipid-reducing and non-lipid-related mechanisms (11–14). Through inhibition of HMG-CoA reductase, statins reduce the synthesis of cholesterol and of a number of isoprenoid intermediates involved in the post-translational modification and activation of cell cycle- and cytoskeleton-associated proteins (11, 12). This way, statins decrease smooth muscle cell (SMC) proliferation and migration (15, 16) and chemotaxis of immune cells (13, 14), all important players in vascular remodeling. The use of these drugs is also associated with improved endothelial cell (EC) function, via lower circulating levels of oxidized LDL, upregulation of endothelial nitric oxide synthase, inhibition of endothelin-1, and reduced adhesion of leukocytes to the endothelium (11, 13, 14).

Experiments in animal models have suggested a potential benefit of statin therapy in AVF vascular remodeling (17). However, retrospective clinical studies have produced controversial results (3, 18–20). One of the main limitations of these clinical studies is that they have assessed the effects of statins as a group, without investigating individual drugs. This approach overlooks the existing pharmacokinetic and physicochemical differences between individual statins, which determine their binding affinity for HMG-CoA reductase, duration of action, and penetrability into extrahepatic cell types (21, 22). Interestingly, in vitro studies disagree on the efficacy of different statins in reducing SMC proliferation and invasion in human veins (15, 23). A comparative analysis of six statins indicates that this anti-proliferative property is a class effect (15), while a separate study suggests that lipophilic statins perform better at this function (23).

This study analyzes the effects of three different statins (atorvastatin, rosuvastatin and simvastatin) on primary failure and primary patency of one-stage and two-stage AVFs. We demonstrate that atorvastatin, but not the other two drugs, significantly decreases the frequency of primary failure in one-stage, but not two-stage fistulas. In contrast, all statins as a group improve primary patency of two-stage AVF, but this effect is not statistically significant when evaluated for individual drugs. Our results suggest that the potential benefit of statins on AVF primary failure is a drug-specific and not a class effect, and that such effect varies depending on the type of fistula.

Methods

Study participants

This retrospective study included patients over 17 years of age who underwent either one-stage or two-stage AVF creation at Jackson Memorial Hospital or the University of Miami from January 2008 to January 2015. All AVF creation and transposition procedures were performed by Dr. Marwan Tabbara. The patients’ medical records were reviewed to collect information on demographics, comorbidities, drug treatment at the time of AVF creation, prior vascular access history, and fistula characteristics and outcomes. Patients were excluded if they were lost to follow-up or, in the case of two-stage AVF, if the first-stage surgery was performed by a different physician. All sections of the study were performed according to the ethical principles of the Declaration of Helsinki and regulatory requirements at both institutions. The ethics committee and Institutional Review Board at the University of Miami approved the study.

Definitions

Primary failure was defined as an AVF that failed to mature (diameter less than 6 mm or approximate blood flow less than 600 mL/min on ultrasound or inability to cannulate for hemodialysis), or that required a radiologic or surgical intervention within three months of creation (one-stage AVF) or transposition (two-stage AVF). Primary patency was defined as the time from creation (one-stage AVF) or transposition (two-stage AVF) until the first radiologic or surgical intervention. All AVFs that received an intervention within three months of creation or transposition required the conversion to a graft. Therefore, AVFs classified as primary failure were not included in the primary patency analyses.

Statistical analyses

Statistical analyses were performed using XLSTAT (New York, NY) and GraphPad Prism (La Jolla, CA). Results were considered significant when p<0.05. Normally distributed data were compared using the Student’s t-test and expressed as mean ± standard deviation, while non-normally distributed values were compared using the Mann-Whitney U-test and expressed as median and interquartile range (IQR). Comparisons between group frequencies were performed using Fisher’s exact test. Statistical associations between baseline covariates and primary failure were assessed using multivariate logistic regressions (Logit algorithm). This algorithm was also used to look for correlations between the statin daily dose and primary failure while adjusting for baseline covariates. Lastly, the effects of baseline covariates on primary patency were explored using multivariate Cox proportional hazards models.

Results

Baseline characteristics of the study population

The medical records of 574 patients who underwent AVF creation by a single surgeon at our institution were reviewed. From this total, 28 were excluded from further analyses due to loss of follow-up, while 11 two-stage AVF patients were removed from the study because the first surgery was performed by a different physician. The final study population of 535 patients was formed by 41% females, 59% African Americans, 38% Hispanics, and only 4% Caucasians (Tab. I). More than three-fourths (77%) initiated hemodialysis with a catheter and 19% had a history of previous AVF. Comorbidities and medication use are listed in Table I. Forty-four percent of patients (238) were on statin therapy in the overall population. From these, 126 were taking rosuvastatin (53%), 58 simvastatin (24%), 39 atorvastatin (16%), and 15 were using either lovastatin or pravastatin (6%, Supplementary Tab. S1, available online as Supplementary material at www.vascular-access.info).

TABLE I -.

Baseline characteristics of the study population

	All fistulas	One-stage	Two-stage	pa
No. of patients	535	230	305	-
Demographics
Age (y; mean ± SD)	52 ± 15	51 ± 15	53 ± 14	0.11
Female (%)	220 (41)	92 (40)	128 (42)	0.66
Black (%)	315 (59)	116 (50)	199 (65)	0.0007
Hispanic (%)	201 (38)	104 (45)	97 (32)	0.0016
White (%)	19 (4)	10 (4)	9 (3)	0.48
Comorbidities (%)
Hypertension	520 (97)	222 (97)	298 (98)	0.44
CAD	87 (16)	40 (17)	47 (15)	0.56
CHF	43 (8)	22 (10)	21 (7)	0.27
Diabetes	278 (52)	117 (51)	161 (53)	0.66
Drug therapy (%)
Antiplatelet agentsb	106 (20)	39 (17)	67 (22)	0.16
Statins	238 (44)	98 (43)	140 (46)	0.48
ACE-I/ARB	259 (48)	119 (52)	140 (46)	0.19
Fistula features (%)
Brachial-basilic	296 (55)	2 (1)	294 (96)	0.0001
Brachial-cephalic	177 (33)	172 (75)	5 (2)	0.0001
Radial-basilic	4 (1)	2 (1)	2 (1)	1.00
Radial-cephalic	58 (11)	54 (23)	4 (1)	0.0001
Previous fistula	100 (19)	32 (14)	68 (22)	0.014
Catheter use	412 (77)	181 (79)	231 (76)	0.47

SD = standard deviation; CAD = coronary artery disease; CHF = congestive heart failure; ACE-I/ARB = angiotensin converting enzyme inhibitor/angiotensin II receptor blocker.

^ap value for the comparison between one-stage and two-stage fistulas.

^bIncludes aspirin and clopidogrel.

In terms of AVF type, 230 and 305 patients received a one-stage and two-stage AVF, respectively. The majority of one-stage procedures resulted in either brachial-cephalic (75%) or radial-cephalic fistulas (23%), while 96% of two-stage AVFs used the brachial-basilic combination. The median time interval between surgeries for two-stage AVF was 70 days (IQR = 49–112). Functional outcomes were similar between one-stage and two-stage fistulas, with primary failure frequencies of 24% versus 25% (Supplementary Tab. S1), and 2-year primary patency of 83% versus 81%, respectively.

Effects of statin therapy on primary failure

Primary failure occurred in 132 of 535 patients (24.7%; Supplementary Tab. S1). Patients treated with atorvastatin or rosuvastatin, but not simvastatin, had lower frequencies of failure than non-statin-treated individuals (Fig. 1). The lowest frequency of primary failure was observed in patients on atorvastatin therapy and with one-stage AVF (6%) compared to 27% for the same AVF type and no statins (Fig. 1). The reduction in primary failure was statistically significant for atorvastatin and all fistulas (p = 0.019).

Fig. 1 -.

Frequency of primary failure in one-stage and two-stage arteriovenous fistulas (AVFs) per statin treatment group. *p value = 0.019 for all fistulas; † = 0.072 for one-stage AVF in the atorvastatin group when compared to the no statin group using Fisher’s exact test.

Adjusting for the statistical impact of 10 covariates (age, sex, ethnicity, coronary artery disease [CAD], congestive heart failure [CHF], diabetes, antiplatelet agents, angiotensin converting enzyme inhibitor/angiotensin II receptor blocker [ACE-I/ARB], catheter use, and history of previous AVF), statins in general decreased the risk of primary failure in all AVF (OR = 0.63, p = 0.046), and specifically in one-stage fistulas (OR = 0.45, p = 0.034; Tab. II). Out of the three individual statins analyzed, only atorvastatin decreased the frequency of primary failure in all AVF (OR = 0.18, p = 0.005) and one-stage fistulas (OR = 0.03, p = 0.005). There was no significant association between statin therapy and primary failure in two-stage fistulas.

TABLE II -.

Effects of different statins on primary failure of arteriovenous fistulas adjusting for clinical covariates^a

	All statinsb	Atorvastatin	Rosuvastatin	Simvastatin
All fistulas
No. on statins/no statins	238/297	39/297	126/297	58/297
Median statin dose (mg; IQR)	NA	20 (20–40)	10 (10–15)	20 (20–40)
Odds ratio	0.63	0.18	0.61	1.29
95% confidence interval	0.40–0.99	0.06–0.60	0.34–1.08	0.66–2.52
p value	0.046	0.005	0.087	0.45
One-stage fistulas
No. on statins/no statins	98/132	17/132	54/132	19/132
Median statin dose (mg; IQR)	NA	40 (20–40)c	10 (10–20)	20 (10–40)
Odds ratio	0.45	0.03	0.43	0.94
95% confidence interval	0.21–0.94	0.002–0.35	0.17–1.12	0.27–3.36
p value	0.034	0.005	0.085	0.93
Two-stage fistulas
No. on statins/no statins	140/165	22/165	72/165	39/165
Median statin dose (mg; IQR)	NA	20 (10–20)c	10 (5–10)	20 (20–40)
Odds ratio	0.79	0.43	0.79	1.59
95% confidence interval	0.44–1.43	0.10–1.81	0.37–1.68	0.70–3.61
p value	0.44	0.25	0.55	0.27

IQR = interquartile range.

^aPatients in the indicated treatment groups were compared to those on no statins using multivariate logistic regressions adjusted for age, sex, ethnicity, coronary artery disease, congestive heart failure, diabetes, antiplatelet agents, angiotensin converting enzyme inhibitor/angiotensin II receptor blocker, prior catheter use, and history of previous AVF.

^bThe All Statins group includes patients on atorvastatin, rosuvastatin, simvastatin, lovastatin, and pravastatin (see Supplementary Tab. S1, available online as Supplementary material at www.vascular-access.info).

^cThe difference in the atorvastatin daily dose between one-stage and two-stage patients is not statistically significant (Mann-Whitney U-test; p = 0.088).

To confirm that the differential effect of atorvastatin on both types of fistulas was not determined by dose, a dose response analysis was performed adjusting for the same 10 covariables (Fig. 2). The results indicate a significant decrease in the probability of primary failure of one-stage AVF with increasing atorvastatin dose (B regression coefficient = −0.162, p = 0.027), while no effect was observed in two-stage fistulas (B = 0.011, p = 0.59). Individuals who benefited the most in this model were those with a history of previous AVF (Fig. 2). This risk factor was consistently associated with primary failure in all multivariate logistic regression models for all statins grouped together as well as for individual drugs (Supplementary Tab. S2, available online as Supplementary material at www.vascular-access.info).

Fig. 2 -.

Probability of primary failure of arteriovenous fistulas (AVFs) by atorvastatin daily dose. Dose-response analysis for primary failure of one-stage (A) and two-stage AVF (B). Logistic regression curves are shown for patients with and without history of previous AVF. B coefficients (slope) and p values adjusted for 10 clinical covariates are indicated in each graph.

Effects of statin therapy on primary patency

The effects of statins on AVF patency were evaluated using Cox proportional hazards models adjusted for the above 10 covariables (Tab. III). Overall, patients on statin therapy and with two-stage AVF had improved primary patency with respect to those on no statins (HR = 0.51, p = 0.024), but this effect was not statistically significant when evaluated for individual drugs. Statin therapy was not associated with primary patency in one-stage AVF (Tab. III, Supplementary Tab. S3, available online as Supplementary material at www.vascular-access.info). This differential effect on both fistula populations is further illustrated in Figure 3.

TABLE III -.

Effects of different statins on primary patency of arteriovenous fistulas adjusting for clinical covariates^a

	All statinsb	Atorvastatin	Rosuvastatin	Simvastatin
All fistulas
No. on statins/no statins	187/216	35/216	101/216	38/216
Hazard ratio	0.72	0.39	1.03	0.54
95% confidence interval	0.46–1.12	0.15–1.06	0.63–1.70	0.22–1.32
p value	0.15	0.064	0.90	0.18
One-stage fistulas
No. on statins/no statins	79/96	16/96	44/96	13/96
Hazard ratio	0.98	0.32	1.64	1.03
95% confidence interval	0.50–1.92	0.06–1.57	0.76–3.55	0.24–4.35
p value	0.95	0.16	0.21	0.97
Two-stage fistulas
No. on statins/no statins	108/120	19/120	57/120	25/120
Hazard ratio	0.51	0.43	0.65	0.34
95% confidence interval	0.28–0.91	0.12–1.54	0.32–1.30	0.10–1.17
p value	0.024	0.20	0.22	0.087

^aPatients in the indicated treatment groups were compared to those on no statins using multivariate Cox proportional hazards models adjusted for age, gender, ethnicity, coronary artery disease, congestive heart failure, diabetes, antiplatelet agents, angiotensin converting enzyme inhibitor/angiotensin II receptor blocker, prior catheter use, and history of previous arteriovenous fistulas.

^bThe All Statins group includes patients on atorvastatin, rosuvastatin, simvastatin, lovastatin, and pravastatin (see Supplementary Tab. S1, available online as Supplementary material at www.vascular-access.info).

Fig. 3 -.

Primary patency analysis of arteriovenous fistulas (AVFs) by statin treatment group. Kaplan-Meier survival curves of one-stage (A) and two-stage AVF (B) as determined by the statin treatment status of patients. The All Statins group includes atorvastatin, rosuvastatin, simvastatin, lovastatin and pravastatin.

Discussion

Our work demonstrates that i) out of the three drugs analyzed, only atorvastatin improved primary failure in one-stage but not two-stage AVF; ii) higher doses of atorvastatin are associated with a lower risk of one-stage AVF primary failure; and iii) all statins as a group, but not when analyzed individually, extended primary patency in two-stage fistulas.

The superiority of atorvastatin in improving AVF outcomes may be explained by its combined physicochemical and pharmacokinetic properties. Unlike the hydrophilic rosuvastatin, which requires cell membrane transporters that increase its hepatic specificity, both atorvastatin and simvastatin are lipophilic drugs, which can readily diffuse across both hepatic and extra-hepatic cell membranes (21, 22). On the other hand, like rosuvastatin, atorvastatin has enhanced affinity for HMG-CoA reductase (two additional bonds with rosuvastatin and one with atorvastatin) and an extended half-life, compared to all other statins (e.g., 15–30 hours for atorvastatin and rosuvastatin vs. 2–3 hours for simvastatin) (21). Therefore, atorvastatin combines high lipophilicity, increased affinity for HMG-CoA reductase and extended duration of action. There are multiple published reports about the anti-stenotic effects of atorvastatin (15, 16). Beyond its lipid-lowering capacity, atorvastatin therapy has been shown to improve endothelial function within 24 hours of treatment (24) and increase brachial artery vasodilation in two weeks of therapy (25). Interestingly, while no randomized controlled trials (RCT) have been designed to assess the effects of statins on newly created AVFs, a post hoc analysis of both the AURORA and SHARP trials suggests that neither rosuvastatin nor the simvastatin/ezetimibe combination decreases the risk of vascular access occlusive events in patients with working fistulas (26). These results support the lack of benefit that we have also revealed for these two drugs in our study.

The reason behind the differential effects of statins on the two types of fistulas is one of the most intriguing findings of the study. In addition to the different veins used for each type of procedure, the patient population that received a two-stage AVF had a significantly higher proportion of African Americans and of patients with a history of a prior AVF. The black race was associated with an increased risk of primary failure in some logistic regression models, but not in others (Supplementary Tab. S2). Similarly, a history of previous AVF was consistently identified as a determinant of primary failure in both one-stage and two-stage fistulas.

The limitations of this study include a relatively low number of patients in the individual drug treatment groups and the potential contribution of unassessed clinical variables to AVF outcomes. Despite these limitations, we identified three important issues for future confirmatory studies: i) all statins are not alike when it comes to improving fistula outcomes, ii) atorvastatin is the only statin that has shown a reduction in AVF primary failure, and iii) the effects of statins on AVF performance is also influenced by the type of fistula. Our data provide the basis for the design of future RCT and identify atorvastatin as a potential drug candidate for the much needed reduction of primary failure in fistulas.