Author + information
- Received July 17, 2013
- Revision received September 10, 2013
- Accepted September 23, 2013
- Published online February 25, 2014.
- Gregory G. Westin, AB, MAS∗,
- Ehrin J. Armstrong, MD, MSc, MAS†,
- Heejung Bang, PhD‡,
- Khung-Keong Yeo, MBBS†,
- David Anderson, BA∗,
- David L. Dawson, MD§,
- William C. Pevec, MD§,
- Ezra A. Amsterdam, MD† and
- John R. Laird, MD†∗ ()
- ∗School of Medicine and the Vascular Center, UC Davis Medical Center, Sacramento, California
- †Division of Cardiovascular Medicine and the Vascular Center, UC Davis Medical Center, Sacramento, California
- ‡Division of Biostatistics, Department of Public Health Sciences, UC Davis, Davis, California
- §Division of Vascular and Endovascular Surgery, the Vascular Center, Sacramento, California
- ↵∗Reprint requests and correspondence:
Dr. John R. Laird, Vascular Center, UC Davis Medical Center, 4860 Y Street, Suite 3400, Sacramento, California 95817.
Objectives The aim of this study was to determine the associations between statin use and major adverse cardiovascular and cerebrovascular events (MACCE) and amputation-free survival in critical limb ischemia (CLI) patients.
Background CLI is an advanced form of peripheral arterial disease associated with nonhealing arterial ulcers and high rates of MACCE and major amputation. Although statin medications are recommended for secondary prevention in peripheral arterial disease, their effectiveness in CLI is uncertain.
Methods We reviewed 380 CLI patients who underwent diagnostic angiography or therapeutic endovascular intervention from 2006 through 2012. Propensity scores and inverse probability of treatment weighting were used to adjust for baseline differences between patients taking and not taking statins.
Results Statins were prescribed for 246 (65%) patients. The mean serum low-density lipoprotein (LDL) level was lower in patients prescribed statins (75 ± 28 mg/dl vs. 96 ± 40 mg/dl, p < 0.001). Patients prescribed statins had more baseline comorbidities including diabetes, coronary artery disease, and hypertension, as well as more extensive lower extremity disease (all p values <0.05). After propensity weighting, statin therapy was associated with lower 1-year rates of MACCE (stroke, myocardial infarction, or death; hazard ratio [HR]: 0.53; 95% confidence interval [CI]: 0.28 to 0.99), mortality (HR: 0.49, 95% CI: 0.24 to 0.97), and major amputation or death (HR: 0.53, 95% CI: 0.35 to 0.98). Statin use was also associated with improved lesion patency among patients undergoing infrapopliteal angioplasty. Patients with LDL levels >130 mg/dl had increased HRs of MACCE and mortality compared with patients with lower levels of LDL.
Conclusions Statins are associated with lower rates of mortality and MACCE and increased amputation-free survival in CLI patients.
- lipids and lipoproteins
- major adverse cardiac event(s)
- peripheral artery disease
- secondary prevention
- statin therapy
Peripheral arterial disease (PAD) affects 4 to 8 million people in the United States (1–3). Patients with PAD have significantly increased rates of myocardial infarction (MI), cardiovascular mortality, and stroke (4). Critical limb ischemia (CLI), the most advanced form of PAD, is characterized by ischemic rest pain, nonhealing ischemic ulcers, and gangrene. Patients with CLI have a major amputation rate as high as 40% at 6 months and a mortality rate of 20% to 25% in the first year after presentation (5,6). Although CLI represents only a subset of the total PAD population, the high cardiovascular event and amputation rates in these patients result in a large overall healthcare burden (7,8).
The benefits of 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors (statins) for morbidity and mortality have been established in patients with or at high risk of ischemic heart disease (9–12). There is also evidence of the utility of statins in patients with PAD; a revision of the Adult Treatment Panel III (ATP III) guidelines designates PAD a “coronary heart disease risk equivalent,” and consensus guidelines recommend statin therapy to target a low-density lipoprotein (LDL) level of ≤100 mg/dl (2.59 mmol/l) for “very high risk” patients (4,13). However, these recommendations are based predominantly on data from patients with claudication or population screening ankle-brachial indexes (ABIs). Thus, the value of statin therapy for patients with CLI is uncertain.
We hypothesized that statin therapy would be associated with a reduced rate of major adverse cardiovascular and cerebrovascular events (MACCE) and a reduced rate of major amputation in patients with CLI. We tested this hypothesis in a large cohort of patients with CLI who were treated longitudinally at a multidisciplinary vascular center.
The PAD-UCD Registry comprises all patients with a clinical diagnosis of PAD who underwent diagnostic angiography and/or therapeutic endovascular intervention at the UC Davis Medical Center from 2006 to 2012. During this interval, 3 vascular surgeons and 1 interventional cardiologist performed all of the procedures. At the time of data analysis, the registry included 975 patients and 1,490 procedures. The study protocol was approved by the Institutional Review Board at the University of California, Davis Medical Center.
Data collection and definitions
We identified patients who had at least 1 presentation during the study period for CLI, defined as Rutherford class 4 to 6 disease (rest pain, nonhealing ulceration due to arterial insufficiency, or gangrene) (14). We retrospectively analyzed these patients’ data on the basis of a review of electronic medical record documentation. We used pre- and post-procedure hospital and clinic records to identify patient demographics, baseline health status and medical management, clinical presentation, vascular procedures, post-procedure management, and outcomes. All records were reviewed by trained chart abstractors and verified by a board-certified cardiologist. Patients were categorized into the statin group if either their hospitalization data or the most recent pre-procedure clinic visit indicated current statin use. Other medication prescriptions were determined based on the most recent pre-procedure clinic visit. Baseline serum LDL levels were determined using the most recent value within 6 months pre-procedure.
Routine practice at our institution during this period was to schedule follow-up visits within 1 month after angiographic procedures, then every 3 months for the first year and every 6 to 12 months thereafter. At these visits, patients were assessed for clinical improvement, and those who had interventions were evaluated with interval ABI measurements and duplex ultrasonography.
The primary endpoint was a composite measure of MACCE, defined as any death, MI, or stroke within 1 year post-procedure. MI was defined as symptoms of chest pressure and elevation of troponin with evidence of infarct by stress imaging or coronary angiography and ventriculography. Stroke was defined as focal neurological deficit lasting >24 h with computed tomography or magnetic resonance imaging evidence of cerebral ischemic infarct or intracerebral hemorrhage.
Secondary outcomes, all at 1 year post-procedure, included death, MI, stroke, subsequent ipsilateral lower extremity bypass grafting, and ipsilateral major amputation, defined as any amputation above the level of the ankle joint. To account for the competing hazard of death among patients at high risk of needing amputation, we also evaluated amputation-free survival as a composite endpoint.
Lesion-specific secondary outcomes included primary, primary assisted, and secondary patency of all lesions treated with endovascular intervention. Loss of primary patency was defined as a velocity ratio of ≥2.0 as assessed by duplex ultrasonography or endovascular or surgical reintervention to the target vessel. Primary assisted patency was defined as patency after treatment for restenosis, and secondary patency was defined as overall patency after restenosis or occlusion.
All outcomes were adjudicated from physician documentation in the electronic medical record. To ensure that deaths outside our institution were captured, patient vital status was also verified using the Social Security Death Index.
Values of mean ± SD were used to describe continuous variables, and frequencies and percentages were used for categorical variables. Continuous variables were compared using the Wilcoxon rank sum test or analysis of variance, and categorical values using the chi square or Fisher exact test. All analyses were performed using Stata version 11.2 (StataCorp LP, College Station, Texas).
We developed propensity scores to adjust for confounding in statin use, defined as the conditional probability of being treated with a statin given a patient’s measured demographic and clinical characteristics (15). To calculate the propensity score for statin treatment, we developed a logistic model for statin treatment using stepwise logistic regression analysis. Baseline covariates in the model included age, sex, and race; history of diabetes, coronary artery disease (CAD), MI, hypertension, heart failure, stroke, carotid artery disease, or chronic obstructive pulmonary disease; smoking status; left ventricular ejection fraction (in 5% increments from ≤10% to ≥65%); prescription of concomitant medications including angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, aspirin, and clopidogrel; and year of procedure.
Multiple methodologies were used to validate the propensity model. The C statistic of the model was 0.79. To assess covariate balance across the distribution of propensity scores, we visually compared propensity score overlap with kernel density plots (Online Fig. 1). We also calculated the odds of treatment with a statin for each of the covariates within a given quintile and then used Mantel-Haenszel estimates of common odds ratios to calculate summary odds ratios across all quintiles before and after propensity adjustment (16). Standardized mean differences were also calculated for each covariate and were verified to be balanced for each covariate after adjustment (Online Table 1) (17). To determine the best estimate with observational data of the treatment effect of statin use, proportional hazards marginal structural models were then developed via weighted regression with inverse probability of treatment weighting using the propensity score (18,19). As a sensitivity analysis, propensity modeling was also performed using nearest-neighbor matching, and results were obtained qualitatively similar to those of the inverse probability of treatment weighting model. To account for the possibility that censoring patients at the first event could obscure effects on other competing risks, we also conducted a competing risks analysis using the method of Fine and Gray (20).
To analyze lesion-level outcomes, we stratified interventions according to the anatomic level of disease (femoropopliteal vs. infrapopliteal). We then used Cox proportional hazard models to estimate hazard ratios (HRs) for the effect of statin medications on restenosis and target vessel revascularization. We included lesion length, proximal vessel reference diameter, diabetes, sex, smoking status, estimated glomerular filtration rate, and, for femoropopliteal lesions, stent placement as multivariable predictors of restenosis.
To study the relationship of baseline LDL level to outcomes, patients were stratified into 4 LDL subgroups based on ATP III guidelines: <70 mg/dl, 70 to 100 mg/dl, 100 to 130 mg/dl, and >130 mg/dl (<1.81 mmol/l, 1.81 to 2.59 mmol/l, 2.59 to 3.36 mmol/l, and >3.36 mmol/l) (13). To minimize bias in the outcomes due to missing LDL data, multiple imputation with 10 imputations was used. MACCE and mortality were then analyzed using Cox proportional hazard modeling that included LDL level as an additional covariate in a model that also adjusted for other baseline demographic factors among patients treated with statin medications.
A total of 380 patients presented with CLI and underwent peripheral angiography during the study period (Table 1). Median follow-up time derived from Kaplan-Meier estimates was 409 days. With respect to the primary outcome of MACCE, 4 patients (3.0%) in the no statin group and 9 patients (3.7%) in the statin group were lost to follow-up (p = 0.96). Overall, 65% of the patients were taking a statin at baseline, a percentage that ranged from 58% for patients entering the study in 2006 to 71% in 2012 (p = 0.07 for trend) (Fig. 1). More than 40% of the patients were female; we previously reported differences in the presentation and outcomes for this cohort on the basis of sex, including a higher rate of MACCE for women (21). Simvastatin and atorvastatin were the 2 most frequently prescribed statins, and together accounted for 73% of the prescribed statins. Additional lipid-lowering medications were prescribed in <15% of patients. Among patients not taking a statin, only 6 (4%) had a documented contraindication.
Patients prescribed statin medications had higher rates of baseline medical comorbidities, including a higher prevalence of diabetes mellitus, hypertension, CAD, and carotid artery stenosis (Table 1). Patients prescribed a statin were also more likely to have a history of MI or stroke/transient ischemic attack. Other baseline differences between the 2 groups included a higher rate of concomitant medical therapies among patients receiving statins including beta-blockers, angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, and clopidogrel. Patients taking a statin medication also had lower baseline levels of total cholesterol and LDL (Fig. 2). Rates of baseline comorbidities including a history of diabetes and CAD did not show significant trends by year (p = 0.21 and p = 0.73, respectively). The angiographic characteristics of the 2 groups were similar, with the exception of a higher rate of multilevel stenosis (defined as at least 1 stenosis >50% in both a femoropopliteal vessel and an infrapopliteal vessel) in the statin group (Table 2).
Outcomes by statin use
The event rates and HRs for all clinical outcomes are summarized in Table 3. Patients prescribed statin therapy had lower absolute event rates for all outcomes despite greater baseline comorbidities. The propensity model showed good balance in adjusted variables, with no significant difference in the pooled odds of statin prescribing after propensity weighting for all measured covariates (Fig. 3). After adjustment, MACCE (HR: 0.53, 95% confidence interval [CI]: 0.28 to 0.99) (Fig. 4A) and mortality (HR: 0.49, 95% CI: 0.24 to 0.97) (Fig. 4B) were lower in the statin group. Rates of MI (HR: 0.48, 95% CI: 0.16 to 1.44), stroke (HR: 0.18, 95% CI: 0.02 to 1.78), and amputation (HR: 0.68, 95% CI: 0.32 to 1.39) did not differ significantly, but all trended in a direction favoring statin use. The risk of death or major amputation was also significantly decreased in those prescribed statins (HR: 0.53, 95% CI: 0.35 to 0.98) (Fig. 4C). There was no association between statin use and rates of lower extremity bypass, which were 18% at 1 year in both groups. Outcomes using a competing risks analysis showed similar point estimates for the relationships between statin use and MI, stroke, bypass, and major amputation (Online Table 2).
Of the overall cohort, 295 patients (78%) underwent endovascular intervention for limb salvage at the time of baseline angiography, whereas 23 patients (6%) underwent major amputation and 38 patients (10%) underwent lower extremity bypass grafting as initial management. The overall procedural success rates for endovascular interventions were similar in both groups. The mean gain in ABI post-procedure was 0.45 ± 0.26 in the statin group and 0.39 ± 0.19 in the no statin group (p = 0.2). Among infrapopliteal lesions, statin use was associated with improved patency at 1 year (Fig. 5). Using Cox proportional modeling, statin use was associated with a nonsignificant improvement in primary patency and significantly improved primary assisted patency and secondary patency of infrapopliteal lesions (Table 4). These results remained significant after multivariable adjustment. There was no significant effect of statins on 1-year patency of femoropopliteal lesions. These results suggest that the lower amputation rates among patients treated with statins may be due in part to improved infrapopliteal vessel patency after endovascular intervention.
Outcomes by LDL level
We also explored the relationship between LDL levels and MACCE among patients prescribed statin medications. Among patients treated with statins, the group with baseline LDL levels >130 mg/dl had the highest unadjusted 1-year event rate (22%). After multivariable adjustment including age, sex, history of MI, diabetes, and carotid disease, LDL levels of 100 to 130 mg/dl (HR: 0.51, 95% CI: 0.32 to 0.80), 70 to 100 mg/dl (HR: 0.54, 95% CI: 0.36 to 0.83), and <70 mg/dl (HR: 0.71, 95% CI: 0.48 to 1.05) were each associated with a decreased HR of MACCE relative to patients taking statin medications but with baseline LDL values >130 mg/dl. Similar results were observed for composite death and amputation among patients treated with statins. Among patients not treated with statins, there was also a qualitatively similar relationship between LDL levels and MACCE.
This study has 3 major findings. First, we report the prevalence of statin use in a cohort of CLI patients treated at a tertiary care center and a trend toward increasing statin use over time. Second, statin use was associated with decreased subsequent risk MACCE (primarily due to decreased mortality), increased likelihood of amputation-free survival, and improved patency of infrapopliteal lesions. Third, we found that lower LDL levels were associated with a decreased risk of MACCE and mortality. To our knowledge, this is the first study in CLI patients to investigate the relationship between statin use and MACCE.
Rates of statin use in CLI remain low
Our data on the rates of statin use in PAD in general and CLI in particular extend previous findings on this subject. Subherwal et al. (22) recently reported on trends in statin use in a general PAD population at the time of diagnosis during the period 2000 to 2007 using Danish nationwide administrative registries. The overall statin use rate in our population (65%) is higher than the 56% rate that they reported for patients without CAD and similar to the rate for patients with both PAD and CAD. Our data show a similar, although not statistically significant, trend toward higher statin use over time. Given the lack of significant increases over time of other indications for statin therapy, this may indicate improving quality of care or increasing awareness of PAD as an indication for statin therapy. Previous reports in the CLI population have indicated even lower rates of statin use, ranging from 23% to 49% (23–25). In contrast, estimates of statin use rates in patients with CAD range from approximately 68% to 78% (26–28). Further education and awareness of the benefits of statin use among patients with CLI will be necessary to increase the rates of medication prescription. Consistent with this goal, statin prescription was recently included as a core performance measure for the treatment of patients with PAD (29).
Further support for statin therapy in CLI
Our results provide further evidence of statin therapy for patients with CLI. The strongest evidence underlying the American College of Cardiology and American Heart Association recommendations of statin therapy derive from subgroup analysis of the Heart Protection Study, which included primarily patients with a low baseline ABI or stable claudication (12,30). More recently, a large prospective, observational cohort study in the Netherlands demonstrated lower all-cause and cardiac mortality in statin users with PAD (31,32). However, these studies were conducted in general PAD populations and included only very small numbers of patients with CLI.
A small number of observational studies of CLI patients have reported outcomes by statin use with mixed results. Aiello et al. (23) recently studied patients undergoing endovascular therapy for CLI and reported 24-month results showing improved vessel patency, limb salvage, and survival with statin use, but the authors did not consider MACCE. Other studies have yielded conflicting results on the effect of statins on graft patency, limb salvage, and mortality among patients with CLI undergoing surgical bypass (24,25,33–35). Our results contribute to this developing body of literature by studying outcomes in a large representative cohort of CLI patients with and without bypass grafts, whether or not they underwent intervention. Unlike previous studies, we report MACCE, capturing important outcomes that might be neglected if only patency or mortality was studied. We also present lesion results divided by anatomic site. The efficacy of statins in promoting patency of infrapopliteal vessels may be related to the smaller caliber of these arteries and may partially explain the effects of statins on reduced amputation rates.
Higher LDL levels associated with increased MACCE in CLI
Our data also begin to explore the relationship between LDL levels and clinical outcomes in CLI patients. The American College of Cardiology/American Heart Association and ATP III guidelines recommend LDL goals on the basis of the cardiovascular risk associated with PAD and the evidence of LDL treatment goals in CAD. Feringa et al. (31) strengthened the evidence for lower LDL treatment targets in the general PAD population by reporting progressively increasing mortality with higher LDL levels. In CLI patients, Aiello et al. (23) reported baseline cholesterol and LDL levels in their statin and control groups, yet total cholesterol and LDL levels did not differ between the groups. Our results provide the first evidence that higher levels of LDL are associated with increased rates of mortality and MACCE in CLI patients, a population with an already increased baseline risk. Future studies should continue to investigate the optimal target level of LDL among this high-risk group of patients.
First, we document a single-institution experience at a tertiary care facility; patterns of care and disease may differ at other sites of care. Also, as with all observational studies, the reported associations may not represent underlying causality. Our use of propensity score weighting balanced the cohorts and reduced the effect of measured confounding variables on our outcomes, although it cannot account for unmeasured confounding variables. Statin use may be a marker for quality of or access to care, which could not be assessed with our study design. However, we included numerous measured covariates and demonstrated excellent balance after propensity weighting. We also used multiple methodologies to verify the adequacy of adjustment for differences in baseline covariates as well as sensitivity analyses that were consistent with our overall results.
Similarly, the duration of statin therapy and compliance could not be quantified in this study, and it is possible that an increasing number of patients began taking statin medications during follow-up. We also found an average LDL level of only 96 mg/dl in the group not taking statins, despite low rates of use of other lipid medications, which may indicate that some patients were using statins but discontinued use before angiography. Crossover to statin use, ceasing statin use either before baseline or during the study, or noncompliance would not negatively affect the validity of our findings, but instead strengthen our reported effect size of statin use on subsequent MACCE. Given the existing support for statin use in claudicants, we believe that evidence of statin therapy in CLI should be interpreted as support for continuous statin therapy beginning at the time of any PAD diagnosis.
Although the majority of the endpoints were driven by a decrease in overall mortality, we believe that the composite endpoints of MACCE and death or major amputation are useful because they are common, clinically meaningful endpoints. Not ascertaining the cause of death, along with losing patients to follow-up, limits our analyses. However, previous studies have found that the majority of deaths in patients with CLI are due to cardiovascular causes (36,37). That our results, like those of the PREVENT III (Project or Ex-Vivo vein graft Engineering via Transfection III) study, indicate a difference in mortality but not in sublethal vascular events may indicate effective prevention of fatal strokes and MIs. Low overall event rates for amputations and nonfatal MIs and strokes, however, limit our power to detect differences in these secondary outcomes.
This study extends our understanding of the effects of statin therapy in PAD and CLI, in particular. The improved rates of 1-year MACCE with statin use strengthens the evidence supporting the guideline recommendations of statin therapy for all PAD patients, including those with even the most advanced stages of disease. Although our data on statin use rates compare favorably with previously published values, statins remain an underused therapy (38). Our finding of superior outcomes for patients with lower LDL levels also provides support for the use of LDL as a treatment target in patients with PAD. Future studies should determine the optimal statin type and dose, further explore potential treatment targets including LDL for statins in PAD patients, and investigate barriers to more widespread use of statins among patients with CLI.
For supplementary tables and a figure, please see the online version of this article.
This project was supported in part by the National Center for Advancing Translational Sciences, National Institutes of Health, through grant number UL1 TR 000002 and linked award TL1 TR 000133. Dr. Yeo is on the speakers’ bureau of Abbott Vascular and has received research funding from Medtronic. Dr. Laird is a consultant for and advisory board member of Boston Scientific, Covidien, Abbott, Bard Peripheral Vascular, and Medtronic. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. The first 2 authors contributed equally to this manuscript.
- Abbreviations and Acronyms
- ankle-brachial index
- ATP III
- Adult Treatment Panel III
- coronary artery disease
- confidence interval
- critical limb ischemia
- hazard ratio
- low-density lipoprotein
- major adverse cardiovascular and cerebrovascular event(s)
- myocardial infarction
- peripheral arterial disease
- Received July 17, 2013.
- Revision received September 10, 2013.
- Accepted September 23, 2013.
- American College of Cardiology Foundation
- Criqui M.H.,
- Fronek A.,
- Barrett-Connor E.,
- Klauber M.R.,
- Gabriel S.,
- Goodman D.
- Selvin E.,
- Erlinger T.P.
- Hirsch A.T.,
- Haskal Z.J.,
- Hertzer N.R.,
- et al.
- Grundy S.M.,
- Cleeman J.I.,
- Merz C.N.B.,
- et al.
- Rosenbaum P.R.,
- Rubin D.B.
- Austin P.C.
- Longmore R.B.,
- Yeh R.W.,
- Kennedy K.F.,
- et al.
- McCoach C.E.,
- Armstrong E.J.,
- Singh S.,
- et al.
- Subherwal S.,
- Patel M.R.,
- Kober L.,
- et al.
- Isma N.,
- Barani J.,
- Mattiasson I.,
- Lindblad B.,
- Gottsäter A.
- Arnold S.V.,
- Spertus J.A.,
- Tang F.,
- et al.
- Olin J.W.,
- Allie D.E.,
- Belkin M.,
- et al.,
- American College of Cardiology Foundation; American Heart Association; American College of Radiology; Society for Cardiac Angiography Interventions; Society for Interventional Radiology; Society for Vascular Medicine; Society for Vascular Nursing; Society for Vascular Surgery
- Heart Protection Study Collaborative Group
- Feringa H.H.H.,
- van Waning V.H.,
- Bax J.J.,
- et al.
- Faglia E.,
- Clerici G.,
- Clerissi J.,
- et al.
- Pande R.L.,
- Perlstein T.S.,
- Beckman J.A.,
- Creager M.A.