Abstract
ABSTRACT. Renal insufficiency is a risk factor for coronary heart disease and stroke, but whether it predicts lower extremity peripheral arterial disease (PAD) is unknown. The authors evaluated was the association of baseline renal insufficiency with future PAD events in the Heart and Estrogen/Progestin Replacement Study (HERS) and follow-up study (HERS II). A total of 2763 postmenopausal women with known coronary heart disease were enrolled in HERS and randomly assigned to receive hormone therapy with conjugated estrogens and medroxyprogesterone acetate or placebo and followed for up to 8 yr for clinical end points. The outcome was time from randomization to first occurrence of either a lower extremity amputation, revascularization (surgical or percutaneous), or lumbar sympathectomy during follow-up. Incident lower extremity PAD event rates among women with creatinine clearances ≥60, 30 to 59, and <30 ml/min/1.73 m2 were, respectively, 0.55%, 0.92%, and 2.73% per year. After multivariable proportional-hazard adjustment for potential confounders and other known risk factors for PAD, women with a creatinine clearance 30 to 59 ml/min/1.73 m2 (hazard ratio [HR], 1.63; 95% confidence interval [CI], 1.04 to 2.54, P = 0.032) and <30 ml/min/1.73 m2 (HR, 3.24; 95% CI, 1.20 to 8.78, P = 0.021) had a significantly increased risk of PAD compared with participants with a creatinine clearance ≥60 ml/min/1.73 m2. Renal insufficiency was independently associated with future PAD events among postmenopausal women with coronary heart disease. Future studies should determine whether this association is present in other populations and investigate its potential mechanisms.
The prevalence of peripheral arterial disease (PAD) is high among patients with renal insufficiency and end-stage renal disease, and hemodialysis patients experience lower extremity amputation rates that are tenfold greater than among elderly persons with diabetes in the general population (1–5⇓⇓⇓⇓). Many studies have now demonstrated that renal insufficiency is a risk factor for cardiovascular death, coronary heart disease, and stroke (6–10⇓⇓⇓⇓), but most did not examine its association with lower extremity PAD; conversely, epidemiologic studies of PAD have not evaluated renal insufficiency as a potential predictor (11–16⇓⇓⇓⇓⇓). Thus, although renal disease might be expected to have a longitudinal association with PAD, as it has for other forms of cardiovascular disease, an association has not been demonstrated between renal insufficiency and incident PAD events, such as amputation and lower extremity revascularization.
Studies of PAD risk factors have been further limited by the fact that early disease is minimally symptomatic and often escapes clinical attention and documentation (17). Measurable PAD clinical end points are relatively uncommon and are usually limited to severe disease complications (e.g., rest pain, foot ulceration, and gangrene) that require surgical intervention via amputation or revascularization. Because few clinical studies have rigorously measured PAD end points, the identification of unique risk factors for PAD has been challenging.
The Heart and Estrogen/Progestin Replacement Study (HERS) was a unique cohort of postmenopausal women with established coronary artery disease who were randomly assigned to receive hormone therapy with estrogens and progestin or placebo and followed for up to 8 yr for the occurrence of cardiovascular end points (including lower extremity PAD) and death (18). Because all HERS participants had coronary artery disease, we believed that they would also have an increased incidence of PAD events that would permit us to explore the association of renal insufficiency with PAD. We hypothesized that among HERS participants, baseline renal insufficiency would be an independent risk factor for the future development of lower extremity PAD complications.
Materials and Methods
Study Design
HERS was a randomized double-blinded, placebo-controlled trial of the effect of daily conjugated estrogens plus medroxyprogesterone acetate on coronary heart disease risk among 2763 postmenopausal women with documented coronary heart disease (18). Study participants were enrolled between February 1993 and September 1994, and the trial was completed in 1998. At the end of the study period, participants were informed of their treatment status and instructed to make a decision regarding postmenopausal hormone use in consultation with their personal physician. At this time, all surviving participants were invited to enroll in follow-up (HERS II) (19). A total of 2321 women (93% of those surviving) enrolled in HERS II, which extended follow-up to January 2001. The primary outcome for HERS and HERS II was nonfatal myocardial infarction and death due to coronary heart disease. PAD was a prespecified secondary outcome. Study participants were contacted every 4 mo during follow-up. When potential cardiovascular events were reported, hospital records were requested and independently reviewed by two physicians blinded to treatment arm (for HERS) and to open-label hormone use (for HERS II) at the HERS coordinating center. Event adjudication required reviewer consensus or a third physician to resolve discordant classifications.
Definition of Renal Insufficiency
Estimated creatinine clearance was the main predictor variable for all analyses. We used the Cockroft-Gault formula to estimate creatinine clearance at the time of enrollment in HERS (20). Estimated creatinine clearance was standardized to body surface area by using the formula of Dubois and Dubois (21). We modeled creatinine clearance as a categorical variable whereby women with an estimated creatinine clearance between 30 and 59 ml/min/1.73 m2 (moderate renal insufficiency) and <30 ml/min/1.73 m2 (severe renal insufficiency) were compared with the reference category of women with an estimated creatinine clearance ≥60 ml/min/1.73 m2 (normal or mildly reduced renal function). We also examined the graded association of creatinine clearance with the outcome within the moderate renal insufficiency category by using cutpoints of 30 to 44 and 45 to 59 ml/min/1.73 m2. Finally, we also report the association with serum creatinine by using cutpoints corresponding to the 75th and 90th percentiles of creatinine distribution in HERS (1.2 and 1.4 mg/dl, respectively). We selected these cutpoints to be consistent with an earlier analysis of renal insufficiency and cardiovascular outcomes in HERS (10).
Baseline Covariates
At the time of study enrollment, HERS participants underwent baseline physical examination, phlebotomy, and interview in which they were asked about their behavioral risk factors and medical history. For the purposes of this analysis, race was categorized as white, black, or other. Women were considered to have diabetes if they had ever been told they had diabetes or if they had a fasting glucose >125 mg/dl at baseline. To adjust for diabetes severity, women were further classified according to whether their diabetes was treated with oral hypoglycemic agents, was treated with oral hypoglycemic agents, insulin, or was untreated. We considered participants to have hypertension if they reported ever being told they had hypertension or if they had a systolic BP ≥140 mmHg and/or a diastolic BP ≥90 mmHg at the baseline examination. HDL, triglycerides, and lipoprotein(a) were obtained on a fasting blood sample. LDL was estimated via the Friedewald equation. Body mass index, current smoking, aspirin, and HMG-CoA reductase inhibitor use were ascertained at baseline. We also included HERS treatment group assignment (conjugated estrogens plus medroxyprogesterone acetate versus placebo). Alcohol use at the time of study enrollment was dichotomized as any alcohol in the last 30 d versus none; educational level was dichotomized as high school graduate versus non-high school graduate; and participants were considered to exercise if they reported participating in a regular exercise program or walked at least occasionally for more than 10 min at a time. Information on the number of prior myocardial infarctions and history of coronary artery bypass graft and percutaneous transluminal angioplasty (PTCA) was ascertained by questionnaire and validated by detailed medical record review.
Outcome
The outcome for the present analysis was time from randomization to first lower extremity PAD outcome. Such outcomes included lower extremity bypass with distal anastomosis below the level of the iliac arteries (n = 54), lower extremity angioplasty, endarterectomy, repair/reconstruction, or thrombectomy at the level of the distal aorta or below (n = 58), lower extremity amputation (including toe) in the absence of bypass or angioplasty (n = 13), and lumbar sympathectomy (n = 2).
Statistical Analyses
In descriptive analyses, we compared each renal insufficiency group to the referent category with an estimated clearance ≥60 ml/min/1.73 m2 by either a χ2 test (for categorical variables), a t test (for normally distributed numeric variables) or a Mann-Whitney U test (for nonnormally distributed numeric variables). We used Cox’s proportional-hazard analysis to measure the univariate and multivariate associations of renal insufficiency with time from randomization to first hospitalization for a PAD event. Women were censored at the time of death or most recent follow-up through January 1, 2001. In multivariable analysis, we adjusted for age, race, and cardiovascular risk factors, including diabetes (categorized as untreated, treated with oral agents, or treated with insulin), hypertension, fasting LDL, HDL, triglycerides, lipoprotein(a), current smoking, body mass index alcohol use, and exercise. We also adjusted for aspirin use, HMG-CoA reductase inhibitor use, treatment arm, education, and baseline severity of coronary artery disease (number of prior myocardial infarctions and prior coronary artery bypass graft or PTCA). Body mass index and lipoprotein(a) were nonlinearly associated with the outcome and were thus modeled as linear splines to accommodate the observed nonlinearity. For all variables included in the final model, we tested for interactions with renal function category. The proportional-hazard assumption was tested by standard residual-based techniques.
Results
The study sample consisted of 2757 of the 2763 women enrolled in HERS (six women were excluded because we lacked complete data required for calculation of standardized creatinine clearance). Among these, almost half had an estimated creatinine clearance <60 ml/min/1.73 m2; 1438 women had clearance between 30 and 59; and 44 women had a clearance <30 ml/min/1.73 m2. Table 1 summarizes baseline characteristics across levels of renal function.
Table 1. Baseline characteristics by level of estimated creatinine clearance among women enrolled in HERSa
Overall, 127 lower extremity PAD events occurred during the follow up period, 44 among women with a clearance ≥60 ml/min/1.73 m2, 78 among women with a clearance between 30 and 59 ml/min/1.73 m2, and 5 among women with a creatinine clearance <30 ml/min/1.73 m2. Figure 1 shows rates of incident PAD events for incremental categories of creatinine clearance, with results reported for the intermediate categories of 30 to 44 and 45 to 59 ml/min/1.73 m2. In unadjusted proportional-hazard analysis, creatinine clearance of 30 to 59 and <30 ml/min/1.73 m2 were associated, respectively, with a 1.6- and almost 5-fold risk for PAD events during follow-up compared with the reference category (Table 2).
Figure 1. Incident peripheral arterial disease (PAD) events (percentage per year) by level of creatinine clearance for participants in the Heart and Estrogen/Progestin Replacement Study (HERS).
Table 2. Association of estimated creatinine clearance with development of a lower extremity PAD eventa
The magnitude of the association was not greatly altered by adjustment for other established cardiovascular risk factors, potential confounders, and baseline severity of coronary artery disease. Additionally, there was also a graded increase in risk among those with moderate renal insufficiency: women with a clearance between 30 to 44 ml/min/1.73 m2 (hazard ratio [HR], 1.85; 95% confidence interval [CI], 1.02 to 3.37, P = 0.04) had a higher risk of PAD than women with a clearance of 45 to 60 ml/min/1.73 m2 (HR, 1.57; 95% CI, 1.00 to 2.48; P = 0.06). When creatinine cutpoints were used, relative to women with a serum creatinine <1.2 mg/dl, the adjusted risk of PAD was similar for those with a serum creatinine between 1.2 and 1.4 mg/dl (HR, 1.00; 95% CI, 0.63 to 1.60; P = 0.96) but increased for those with a serum creatinine >1.4 mg/dl (HR, 2.28; 95% CI, 1.34 to 3.89; P = 0.002).
We detected no statistically significant interactions (P < 0.05) between any of the variables included in the multivariable model and renal insufficiency. Specifically, renal insufficiency was associated with PAD even among those without diabetes (HR 1.91 for a clearance 30 to 60 ml/min/1.73 m2, 95% CI, 1.04 to 3.48; P = 0.04 and HR 1.51 for a clearance <30 ml/min/1.73 m2, 95% CI, 0.18 to 12.40). The proportional-hazard assumption was not violated in any of the above models (P = 0.99 for the primary analysis).
Discussion
We herein report strong associations of even moderate renal insufficiency with incident lower extremity PAD events in the HERS and HERS II studies. Compared with women with clearance ≥60 ml/min/1.73 m2, the risk of reaching a PAD end point was increased by 60% for women with clearance between 30 and 59 ml/min/1.73 m2 and more than threefold for women with clearance <30 ml/min/1.73 m2. These associations persisted after adjustment for demographic characteristics, other cardiovascular risk factors, potential confounders, and baseline severity of coronary heart disease, as indicated by the number of prior myocardial infarctions and history of coronary artery bypass surgery or PTCA.
In previous studies, risk factors for PAD have included male gender, older age, diabetes, smoking, hypertension, hyperlipidemia, lipoprotein(a), hyperhomocysteinemia, and chronic inflammation, whereas alcohol intake and physical activity appear to be protective (11–16,22–30⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓). However, renal insufficiency—a risk factor for many other forms of cardiovascular disease—had not been considered as a risk factor for lower extremity PAD. This is in part because most previous studies examining the association of renal insufficiency with future cardiovascular events have either not included lower extremity PAD as an outcome or have only included lower extremity PAD events in a composite cardiovascular death or event outcome (7–10⇓⇓⇓). Strikingly, the adjusted association of creatinine clearance <30 ml/min/1.73 m2 with PAD was similar in magnitude to previously reported associations of established PAD risk factors such as diabetes and current smoking (30).
The finding that persons with moderate renal insufficiency are at increased risk for PAD complications is consistent both with previous studies documenting a high prevalence of PAD and high incidence of lower extremity amputation among patients with end-stage renal disease and with a prior report linking chronic renal insufficiency with the development of claudication (1,2,6⇓⇓). High complication rates after both lower extremity amputation and revascularization in patients with even moderate renal insufficiency further underscore the substantial clinical significance of lower extremity PAD in this population (1,31,32⇓⇓). However, lower extremity PAD has unfortunately received little attention from the nephrology community, with surprisingly few reports evaluating the risk of PAD events or evaluating targeted interventions to prevent PAD complications in this high-risk population (33–35⇓⇓).
In the general population, interventions such as exercise and statin therapy appear to improve the functional limitation associated with lower extremity PAD, and regular foot care and patient and provider education can be extremely effective in preventing amputations among persons with diabetes (36–39⇓⇓⇓). Our findings support concerted efforts to develop, test, and implement strategies to prevent lower extremity PAD complications in persons with renal insufficiency, particularly among those at highest risk for these complications. The increased incidence of PAD complications among persons with renal insufficiency also reinforces the importance of increasing PAD awareness among nephrologists and others caring for these patients
Because the kidney is a highly vascular organ, one possible explanation for the association of renal insufficiency with PAD reported here is that renal disease may merely be a marker for more generalized atherosclerosis. However, the present analysis demonstrates a strong association between renal insufficiency and PAD in a cohort of women with known coronary atherosclerosis despite adjustment for the severity of baseline coronary artery disease (number of prior myocardial infarctions and history of coronary artery bypass graft or PTCA). Potential pathophysiological mechanisms by which decreased GFR might directly predispose to PAD include altered homocysteine metabolism, anemia, oxidative stress, and alterations in inflammatory pathways (40,41⇓). Our findings highlight the need for research to identify possible mechanisms for the association of renal insufficiency with the development of PAD. Investigation along these lines may lead to interventions that would slow disease progression and/or prevent disease complications in this group.
Our study does have certain limitations. First, we were unable to adjust for baseline severity or prevalence of PAD among study participants because information on PAD history was not collected and the ankle-brachial index was not measured in HERS. Because the prevalence of PAD in patients with chronic renal insufficiency is high, further studies are needed to determine whether the association documented here between renal insufficiency and PAD reflects a higher prevalence and greater severity of PAD at baseline among those with renal insufficiency or whether renal insufficiency is in fact a true risk factor for PAD. Second, confounding by severity of other comorbidities included in the model is also a concern. In particular, objective measurements of baseline severity of coronary artery disease such as exercise tolerance testing were not obtained, and data on diabetes severity were limited to serum glucose measurements and method of treatment (i.e., nontreatment, oral agent, or insulin). Third, whereas the surgical end points for lower extremity PAD used in HERS and HERS II are specific and relevant to patient care, use of surgical end points may capture only a subset of persons affected by PAD. For example, a surgical bias against women with renal insufficiency could have affected our results, particularly in the case of elective revascularization procedures performed for noncritical ischemia. However, this should have biased our results toward the null hypothesis. In addition, use of a severe end point such as amputation or revascularization rather than other measures of disease progression, such as the development of intermittent claudication or change in the ankle-brachial index, does not allow us to extrapolate our findings to milder forms of PAD. Finally, although we would hypothesize that renal insufficiency is also associated with PAD in men and in those without known coronary artery disease, because of the HERS study design, the present analysis only supports an association in women with established coronary artery disease.
Conclusion
Our findings indicate that even persons with moderate renal insufficiency are at increased risk for future PAD events. Further studies are needed to determine whether this reflects a higher prevalence and/or greater severity of PAD at baseline in those with renal insufficiency or a true independent association between renal insufficiency and development and progression of PAD. Taken together with previous reports of the high prevalence of PAD and high morbidity and mortality associated with both amputation and lower extremity revascularization in patients with renal insufficiency, our results indicate the need for a stronger focus on the prevention and treatment of this disease in persons with even moderate renal insufficiency.
Acknowledgments
HERS and HERS II were supported by Wyeth-Ayerst Laboratories. Drs. O’Hare and Shlipak are supported by Career Development Awards from the Department of Veterans Affairs Health Services Research and Development Program.
- © 2004 American Society of Nephrology