Progression of atherosclerotic renovascular disease: a prospective population-based study

Methods 

Results 

Between January 1995 and February 1997, 1245 Forsyth County participants returned to the CHS Field Center for annual examination. Among returning participants, 870 gave consent for their first RDS, providing 69.9% recruitment efficiency. The comparison of CHS participants with nonparticipants and the prevalence of RVD among recruited CHS participants have been described in detail.3, 11, 18 Of 834 participants successfully studied with RDS, 57 (6.8%) had significant RVD at the first examination. Significant RVD was unilateral in 50 patients and bilateral in seven. Of 64 renal arteries with significant RVD, 57 kidneys had hemodynamically significant renal artery stenosis and seven arteries were occluded. Prevalent RVD had no significant association with either gender or ethnicity, but did demonstrate a significant and independent association with increasing participant age, decreased high-density lipoprotein cholesterol, and increased systolic blood pressure.3

Of the 834 technically satisfactory first RDS examinations, 610 surviving CHS participants were invited to return for a second RDS examination. Between 2002 and 2005, 119 (14.3%) from the first RDS cohort returned for a second RDS examination (mean study interval, 8.0 ± 0.8 years). The repeat study cohort was a mean age of 82.8 ± 3.4 years at the second RDS examination. The group included 65% women and 35% men, and 77% were white and 23% were African American. Table I defines atherosclerotic risk factors that were examined and compares surviving CHS participants who returned for a second RDS study with those who did not.

Table I. Demographics and baseline atherosclerotic risk factors among 610 Cardiovascular Health Study participants

	Variable/Definition⁎	Recruited (n = 119)	Non-recruited (n = 491)	P
	Age in years†	81.9±3.5	83.3±4.4	<.001
	Race			.870
	African-American (%)	27(33)	108(22)
	White or other (%)	92(77)	383(78)	—
	Gender			.781
	Female (%)	77(65)	311(63)	—
	Male (%)	42(35)	180(37)	—
	Blood pressure (mm HG)
	Systolic	134±19	136±20	.489
	Diastolic	72±9	72±10	.775
	Clinical hypertension	14(12)	64(13)	.710
	SBP >60 mm Hg or
	DBP >95 mm Hg or
	Anti-hypertensive meds (%)	65(55)	251(51)	.493
	Statin use (%)	10(8)	28(6)	.274
	ACE inhibitor use (%)	6(5)	51(10)	.072
	Diabetes mellitus (%)‡	19(16)	71(14)	.678
	Cigarette smoking†
	Ever	49(41)	222(45)	0.427
	Hypercholesterolemia
	Total cholesterol (mg/dL)	197±31	205±38	0.020
	Elevated LDL-C (mg/dL)	123±32	128±33	0.140
	Decreased HDL-C (mg/dL)	53±15	54±13	0.281
	Obesity† (%) >130% ideal body weight	37(31)	156(32)	0.886

SBP, Systolic blood pressure; DPB, diastolic blood pressure, ACE, angiotensin converting enzyme; LDL-C, low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol.

⁎ Data are presented as % or mean ± SD. † At November 1, 2002. ‡ Diabetes mellitus is defined as fasting glucose >140 mg/dL or 2-hr post-glucose load >200 mg/dL or insulin or oral hypoglycemic agent.

Among 235 kidneys with a technically satisfactory RDS examination, significant RVD was present at the first examination in 13 kidneys (6%; 95% confidence interval [CI], 3.2% to 11.2%). No significant change of prevalent RVD was found on the second RDS study. No ≥60% diameter-reducing renal artery stenosis at the first examination progressed to occlusion on the second RDS.

At the second RDS, new hemodynamically significant RVD was identified in nine kidneys (incident disease: 8 new renal artery stenoses; 1 new renal artery occlusion). A significant increase in RA-PSV was observed in 29 kidneys, including the nine cases of incident RVD. A significant decrease in PSV was observed in 23 kidneys, including six kidneys with velocities indicative of RVD (ie, PSV ≥1.8 m/s) at the initial RDS exam had PSV values ≤1.8 m/s at follow-up exam.

In patients with repeat RDS, mean blood pressure was 136 ± 21/80 ± 9 mm Hg at first RDS and 145 ± 23/72 ± 12 mm Hg at the second examination. Mean serum creatinine concentration was 1.0 ± 0.3 mg/dL at first RDS vs 1.3 ± 0.3 mg/dL at follow-up. This reflected a mean increase of 0.29 mg/dL (SEM, 0.04; 95% CI, 0.21 to 0.37). For 228 kidneys with length measurements at the first and second RDS, maximum pole-to-pole renal length decreased by 0.37 ± 1.17 cm (SEM, 0.9; 95% CI, 0.20 to 0.55). By univariate analysis, longitudinal increase in diastolic blood pressure (P = .01) and decrease in renal length (P < .001) were significantly associated with incident RVD (ie, progression to hemodynamically significant RVD) but not prevalent RVD or significant change in RVD that equated to <60% renal artery stenosis or occlusion.

Discussion 

To our knowledge, this is the first prospective, population-based estimate of incident RVD and progression of prevalent RVD among free-living, elderly Americans. Anatomic progression to hemodynamically significant RVD was observed in only 4% of participants on a mean follow-up of 8 years, an annualized progression rate of 0.5% per year. Moreover, no cohort participant with significant RVD at the first examination progressed to renal artery occlusion on follow-up.

These study results contrast sharply with retrospective angiographic case series, prospective angiographic studies, and prospective duplex studies, which describe progression of RVD. Retrospective reports of serial aortography and progression of RVD are summarized in Table II.19, 20, 21, 22, 23, 24 Considered collectively, the retrospective angiographic case series suggest that atherosclerotic lesions of the renal arteries demonstrate significant anatomic progression over relatively short periods of follow-up. On mean follow-up of 44 months, roughly one third of renal artery lesions demonstrated radiographic progression, and 10% progressed to occlusion.

Table II. Retrospective angiographic studies of medically managed atherosclerotic renal artery stenosis

	Author	Year	Patients (n)	Renal arteries (n)	Mean follow-up (months)	Anatomic progression (% patients)	Progression to occlusion (% arteries)	Blood pressure change	Decrease in renal length (% patients)	SCr increase (% patients)	GFR decline (% patients)
	Wollenweber19	1968	109	252	42	59	—	—	—	—	—
	Meaney20	1968	39	78	34	36	4	—	—	—	—
	Schreiber21	1984	85	126	52	44	11	NS	46⁎	38	—
	Tollefson22	1991	48	—	54	53†	9†	—	—	—	—
	Crowley23	1998	1178	—	30	11	0.3	—	—	‡	—
	Chabova24	2000	68	—	39	—	—	NS	—	15	—

SCr, Serum creatinine; GFR, glomerular filtration rate; NS, not significant.

⁎ 1.5 cm discrepancy in renal length. † Percentage of renal arteries with baseline stenosis or stenosis in follow-up. ‡ SCr increased among patients with anatomic progression to >75% stenosis.

Although these early studies describe a dramatic progression of RVD, the application of these observations to the population at large is probably flawed. Each of these studies reported on highly select groups of patients with significant clinical disease that warranted serial aortography. With the exception of the report by Crowley et al,23 renovascular hypertension was suspected in nearly all of the subjects and worsening hypertension was the most frequent indication for repeat study. Consequently, it is doubtful that the same rate of progression would apply to all individuals with RVD.

Prospective angiographic clinical studies of RVD are summarized in Table III.25, 26, 27, 28, 29 More than 25 years ago, Dean et al24 reported on patients with renovascular hypertension randomized to medical management or surgical revascularization. Randomized to medical management were 41 patients with high grade atherosclerotic renal artery stenosis and renovascular hypertension proven by renal vein renin assay, or split renal function studies, or both. These patients were monitored for an average of 44 months, during which 17 patients (41%) crossed over to the surgical arm. Although 15 of the 17 patients had controlled hypertension, each patient had declining renal function as defined by a 10% loss of renal length, a 100% increase in serum creatinine level, or a 50% reduction in measured glomerular filtration rate (GFR) or creatinine clearance (CrCl). Among the patients treated medically, 22 (54%) had no increase in serum creatinine, 47% of those who underwent isotopic measurement of GFR or CrCl had no significant change, 37% had a <50% decline in GFR, and 2% (1 patient) experienced a >50% decline. Interestingly, four patients (13%) demonstrated improvement in measured CrCl or GFR when measured serially. Despite the severity of renovascular disease, the decline in renal function was variable, with 97% of patients losing <50% of measured GFR.

Table III. Prospective angiographic natural history studies of atherosclerotic renal artery stenosis

		Year	Patients (n)	Renal arteries (n)	Mean follow-up (months)	Anatomic progression (% patients)	Progression to occlusion (% arteries)	Blood pressure change	Decrease in renal length (% patients)	SCr increase (% patients)	GFR decline (% of patients)
	Dean25	1981	41	—	44	17	12	—	37	46	3⁎
	Plouin26	1998	26	—	6	—	—	−24/+12	—	NS	NS
	Webster27	1998	30	—	—	13†	0†	−28/−16‡	—	NS	—
	van Jaarsveld28	2000	50	100	12	20	5	−17/−7	—	NS	NS
	Pillay29	2002	85	159	30	—	—	NS	NS	§	—

SCr, Serum creatinine; GFR, glomerular filtration rate; NS, not significant.

⁎ >50% increase, data for 30 patients. † Of eight patients with serial angiography. ‡ From referral to last follow-up. § Unilateral group had significant increase, bilateral group did not.

A recent prospective study by Pillay et al29 described the change in blood pressure and serum creatinine among patients with RVD. In this multicenter, nonrandomized, observational study, 98 patients were noted to have ≥50% renal artery stenosis during aortography obtained to evaluate peripheral vascular disease. Complete data were available for 85 patients during a minimum 2-year follow-up. Among these, 64 patients with unilateral stenosis and 21 patients with bilateral stenosis were managed medically. Twelve patients with bilateral disease underwent angioplasty or open revascularization.

The overall 2-year estimated mortality was 32%. Mortality was the same for patients treated with renal artery intervention or treated medically. Most of the deaths were due to coronary artery disease; however, three (11%) died of complications from renal failure. Two of these deaths occurred in patients with unilateral RVD, suggesting renal parenchymal disease rather than ischemic nephropathy.

No change occurred in median blood pressure, number of antihypertensive agents, or renal size on follow-up among survivors. During the study period, a small but statistically significant increase in serum creatinine levels was observed in patients with both unilateral and bilateral RVD who underwent renal artery intervention. Patients with bilateral RVD treated medically had stable serum creatinine levels over 2 years. Although this study lacked a specific measure of glomerular filtration, it demonstrated stable renal length and stable serum creatinine levels, with controlled hypertension in medically managed patients who had bilateral RVD but no clinical indication for renal artery intervention.

The available prospective duplex ultrasound studies of RVD are summarized in Table IV. A series of consecutive reports described prospective investigation performed at the University of Washington.14, 30, 31, 32 These authors first reported on serial renal duplex sonography examinations performed on 80 patients with hypertension.30 Renal arteries were classified according to four categories: normal, stenosis of <60%, stenosis of >60%, or renal artery occlusion. The rate of progression to >60% stenosis during 3 years of follow-up was 8% for renal arteries that were initially classified as normal and 43% for arteries initially classified as having <60% diameter-reducing stenosis. Incident renal artery occlusions were observed only in arteries previously classified as having >60% diameter-reducing stenosis. The 3-year risk for occlusion among the group was 7%. Factors associated with lesion progression included increasing patient age, increasing systolic blood pressure, smoking, female sex, and poorly controlled hypertension. Unfortunately, the number of study participants and the low rate of progression of RVD in our report did not allow meaningful exploration of associations with significant change.

Table IV. Prospective duplex sonography natural history studies of atherosclerotic renal artery stenosis

		Year	Patients (n)	Kidneys (n)	Mean follow-up (months)	Anatomic progression (% arteries)	Progression to occlusion (% arteries)	Blood pressure change	Decrease in renal length >1 cm (% arteries)	SCr increase (% patients)	GFR decline (% patients)
	Zierler30	1994	80	134	13	8⁎	3	—	8	—	—
	Zierler14	1996	76	132	32	20	7	—	—	—	—
	Caps31	1998	170	295	33	31	3	—	—	—	—
	Caps32	1998	122	204	33	—	2	—	16	†	—

SCr, Serum creatinine; GFR, glomerular filtration rate.

⁎ Progression at 12 months. † Seven subjects with bilateral atrophy increased 0.33 mg/dL/year; remainder were NS.

In this study from the University of Washington, a principle criterion for disease progression from <60% renal artery stenosis to >60% renal artery stenosis was an increase in the renal-aortic ratio (the ratio of RA-PSVs to aortic PSV) value to >3.5 among subjects with RA-PSV >1.8 m/s.30 However, we have observed no association between RA-PSV and aortic PSV in the presence or absence of significant RVD in either population-based or clinical studies. Rather, the renal-aortic ratio can be considered an example of a spurious correlation. The association with the presence or absence of RVD resides entirely with RA-PSV. This interpretation is supported from 834 renal duplex exams performed initially among the Forsyth County CHS cohort (Fig).3 Analysis from this cohort showed no relationship between aortic PSV and RA-PSV. Considered in light of these data, the patients with estimated stenosis of <60% based on a renal-aortic ratio of <3.5 but with a RA-PSV >1.8 m/sec could be considered to have significant RVD at baseline.

View Large Image Download to PowerPoint

Renal artery peak systolic velocity (PVS) vs aortic PSV for the 834 Forsyth County participants in the Cardiovascular Health Study.

Perhaps the most informative prospective study of RVD using RDS was provided by Capps et al,31 who described 5-year follow-up on 170 patients and 295 kidneys. In this extended report, disease progression was defined by a 100 cm/s increase RA-PSV or progression to occlusion. By these criteria, disease progression was detected in 91 (31%) of the renal arteries in this study. Nine arteries (3%) progressed to occlusion, and all of these were considered to be diseased at the baseline RDS. The authors created a model to predict the 2-year cumulative incidence of RVD progression. For renal arteries without ipsilateral or contralateral stenosis in a nondiabetic patient with systolic blood pressure (SBP) <160 mm Hg, the calculated risk of progression at 2 years was 7%. For arteries with high-grade ipsilateral and contralateral disease in a diabetic patient with a SBP >160 mm Hg, the risk was estimated at 65%. In contrast to these data among hypertensive patients, our population-based study suggests a much lower rate of significant change in RVD even though the criterion for significant change (ie, 45 cm/s change in RA-PSV) was less than half the criterion adopted by these authors.

The data reviewed in this report are frequently cited as rationale for intervention for RVD discovered incidentally during assessment of cardiac or vascular disease.1 However, if one considers the potential value of prophylactic renal artery intervention in light of the associated risks in a normotensive individual without renal insufficiency, the unique benefit from prophylactic renal artery intervention is not justified.33 Moreover, when considered in terms of the data provided by the current study, there appears to be no justification for prophylactic renal artery intervention, either as an open operative procedure performed in combination with aortic repair or as an independent catheter-based intervention.

Although this prospective study provides unique data, it suffers from a number of limitations. The index participants for CHS were selected through a two-step random process that included eligible members from the sampled individual’s household. Although this strategy was adopted to enhance recruitment and retention, the final CHS cohort consisted of 70% of individuals initially sampled and 30% who shared the same household. Significant differences existed between randomly selected participants and those who chose not to participate in the CHS. The refusal rate was higher among women compared with men, although a notably higher percentage of women participated in the study.

The enrolled participants were younger, more highly education, more likely to be married, and less likely to be smokers.11 This “healthy cohort effect” may have contributed to a decreased rate of progression of RVD and may have led to a potential survivorship effect. In previous reports, prevalent RVD has demonstrated significant and independent associations with both prevalent cardiovascular disease and subsequent adverse cardiovascular events.18, 34, 35

Finally, this prospective study was constructed with assumptions not met by observation. This study was originally constructed to include 300 participants, anticipating a 20% rate of progression during follow-up. However, a number of factors, including outdated follow-up information with the closure of the CHS, advanced participant age, and changes in participants’ overall clinical conditions hampered our patient recruitment for the second RDS. Consequently, participant recruitment was less than half of that projected. When considered collectively, these limitations may have led to underestimation of RVD progression.

Another area of potential limitation with our methodology is the progression of RVD as defined by changes in RA-PSV. Progression of RVD at the subcritical level was defined by changes in PSV greater than twice the standard deviation of predicted change (RA-PSV ≥45 cm/s). This predicted change in PSV was determined by using 281age-matched individuals with two or more studies over a median follow-up of 2 years. Ninety-five percent of these patients were hypertensive, 37% had a history of tobacco use, and men comprised higher percentage of participants (49%). Thus, the comparison group likely had a high rate of atherosclerotic vascular disease and may presumably have had a greater expected rate of RVD progression. Moreover, the variability of the estimated change in PSV among the independent cohort may be greater than that among the CHS group, thereby underestimating progression of RVD by this definition.

The observation of a significant decrease in RA-PSV in 23 kidneys deserves special comment. Because of the number of participants and the extended period of follow-up, three registered vascular technologists participated in the study. One technologist participated in both the first and second RDS examination, but two technologists participated in only one exam. Examination of interobserver differences at the first RDS suggested significant differences between technologists. Although the apparent bias associated with each sonographer was consistent, this bias may have contributed to an observed decrease in PSV when different sonographers performed the first and second RDS. In addition, significant technical improvements occurred in RDS during the course of study. Compared with the first RDS examination, the second study utilized technology that provided greater spatial resolution. Moreover, color-flow Doppler imaging, which was not available at the initial study, was used routinely to estimate the angle of insonation at the second study. Considered collectively, these differences may have contributed to the observed decline in RA-PSV.

Despite these potential limitations, the study of the Forsyth CHS cohort allows for longitudinal follow-up of a large, diffuse group of community-dwelling black and white men and women. As our population ages and as advanced imaging techniques define incidental vascular disease, anticipated RVD progression may be best estimated by change in RVD observed in this CHS cohort.

Conclusion 

This prospective, population-based evaluation of prevalent RVD among free-living, elderly Americans suggests that the rate of significant changes in RVD is low. This low rate of progression, combined with no observed progression in significant RVD to renal artery occlusion, does not justify prophylactic intervention for asymptomatic renovascular disease in the elderly.

Author contributions