Diastolic Function Predicts Survival after Renal Revascularization

Abstract

Purpose

To define the relationship between left ventricular diastolic function and survival after renal revascularization.

Methods

76 adult patients (49 women, 27 men; mean age: 63 years ± 13 years) with preoperative echocardiography who underwent renal revascularization for atherosclerotic disease were identified. Diastolic function was estimated from the early diastolic transmitral flow velocity (E), the atrial transmitral flow velocity (A) and the mitral annular tissue doppler velocity (e’). Patients were divided into two groups of diastolic dysfunction as either none/mild (E/A≤0.75, E/e’<10) or moderate/severe (E/A>0.75, E/e’≥10). Perioperative and follow-up mortality were determined from a prospective vascular database and the National Death Index. Descriptive statistics were calculated. Postoperative survival was estimated by product-limit methods. Associations between preoperative factors, perioperative factors and follow-up survival were examined using proportional hazards regression models. A forward stepwise variable selection procedure was used to select a ‘best’ model to predict follow-up survival.

Results

76 patients were followed for an average of 41.9 months after renal revascularization. Within this group 47/76 (61.8%) patients were identified as having moderate or severe diastolic dysfunction. Diastolic dysfunction had no apparent association with abnormal systolic function. The mean ejection fraction for those with moderate/severe diastolic dysfunction was 57.7% +/− 11.5%. When comparing the moderate/severe and none/mild groupings of diastolic dysfunction, there was a significant difference in left ventricular mass index (151.9 +/− 48.9 vs. 125.3 +/− 31.7, P = .0087). There were 5 deaths in the perioperative period and 20 deaths on follow-up. Among perioperative survivors, hypertension was cured or improved in 82% of the none/mild group and 53% of the moderate/severe group (P = .012). In multivariable analysis, none/mild diastolic dysfunction was significantly and independently associated with an improvement in blood pressure after revascularization (OR 6.2, 95% CI 1.4-28.6, P = .018). Ejection fraction was not associated with survival. After forward variable selection, moderate/severe diastolic dysfunction (HR 5.8, 95% CI 1.4–25, P = .018) was the only variable to demonstrate a significant and independent association with follow-up survival.

Conclusion

Diastolic dysfunction, but not systolic dysfunction, was frequent in patients with renovascular disease. Blood pressure response and follow-up survival after renal revascularization demonstrated significant and independent associations with diastolic function. Consideration of diastolic function should be included in the management of patients with atherosclerotic renovascular disease.

Introduction

Preoperative cardiac risk assessment has long considered the effect of systolic function on perioperative morbidity and survival. Conversely, preoperative evaluation has largely ignored diastolic function. Recent reports describe an increase in perioperative cardiac morbidity and a decrease in survival in patients with impaired diastolic function despite normal systolic function^1-3. Given that patients with renovascular disease have demonstrated an increased prevalence of impaired diastolic function compared to the general population⁴, patients with renovascular disease undergoing surgery may demonstrate a significant association between diastolic function and operative and postoperative morbidity and mortality considering these associations. The specific aim of this report is to examine the association of diastolic function with postoperative hypertension response, postoperative renal function response, and follow-up survival among patients after open renal revascularization.

Methods

During this study period, a total of 256 renal interventions were made at our institution. These included 159 open renal artery revascularizations and 71 percutaneous renal artery stent placements. 106 patients underwent open surgical revascularization for atherosclerotic renal artery disease during this period. 76 patients had cardiac stress testing and preoperative resting echocardiography that was interpretable for diastolic function and these patients constitute this study. 22 of the 106 patients had preoperative dobutamine stress echocardiography alone, which could not be interpreted for diastolic function and these patients were omitted. The cohort of 76 adult patients (49 women, 27 men; mean age: 63 years ± 13 years) with preoperative echocardiography who underwent open operative renal revascularization for atherosclerotic disease to 114 kidneys were identified from an institutional procedural database. 74 patients underwent repair for hypertension with (26 patients) or without (48 patients) renal insufficiency. Two patients underwent renal artery repair in the course of aortic aneurysm treatment. All procedures were performed by vascular surgeons at Wake Forest University Baptist Medical Center between August 2001 and October 2006. Revascularization involved open interventions including renal artery endarterectomy (56 kidneys), renal artery bypass (53 kidneys), and nephrectomy (5 kidneys). 50% of patients had bilateral renal artery repair. Renal repair was combined with mesenteric revascularization or aortic repair in 40 patients (53%).

Echocardiography was performed and interpreted according to American Society of Echocardiography (ASE) Recommendations for Use of Echocardiography in Clinical Trials⁵. accredited by the Intersocietal Commision for the Accreditation of Echocardiography Laboratories (ICAEL) at Wake Forest University Baptist Medical Center. Sonographers met or exceeded the standard for image acquisition as required by ICAEL and institutional study specific protocols. After training by a dedicated research sonographer, two separate physician interpreters obtained measurements from digitized studies in accordance with ASE recommendations. Appropriate reference images were reviewed on a digital workstation. Analysis of these images was performed to record data on intraventricular septum thickness, left ventricular posterior wall thickness, end-diastolic left ventricular internal diameter, early (E) and late (A) transmitral diastolic flow velocities, transmitral flow deceleration time (DT), and diastolic Doppler tissue velocity of the mitral annulus (e’). Systolic function was assessed as ejection fraction (EF) and calculated using the modified Simpson’s method⁶. LV mass was determined using a necropsy-validated formula⁷, and LV hypertrophy (LVH) was considered to be present if LV mass/body surface area was >116 g/m² in men and >104 g/m² in women. Diastolic function was classified as either none/mild (E/A≤0.75, E/e’<10) or moderate/severe (E/A>0.75, E/e’≥10). When E/A and E/e’ classifications differed, E/e’ values were used.

Perioperative and follow-up mortality was determined from a prospective vascular database and the National Death Index (NDI). Descriptive statistics were calculated (Table 1). Postoperative survival was estimated by product-limit methods. Associations between preoperative and perioperative factors and follow-up survival were examined using proportional hazards regression models. Variables considered included diastolic dysfunction, age, preoperative creatinine, ejection fraction, gender, history of transient ischemic attack or cerebrovascular accident, history of congestive heart failure, history of left ventricular hypertrophy, bilateral renal artery intervention, and use of postoperative angiotensin receptor blocker or angiotensin converting enzyme inhibitor, history of diabetes, postoperative blood pressure response and postoperative serum creatinine change. Left Ventricular Mass Index was not included in the model secondary to incomplete data on eight patients. Preoperative predictors of cured or improved blood pressure response were examined using logistic regression models. For both proportional hazards and logistic regression analyses, “best” models to predict outcomes were constructed using a forward stepwise procedure which included potential predictors one-by-one until all those significant at the 10% alpha level were selected.

Table 1.

Descriptive characteristics classified according to diastolic function

	Diastolic Dysfunction
	Mild/None (29 Patients)	Moderate/Severe (47 Patients)	P-value
Age (years)1,2	58.3(±16.0)	66.0(±9.4)	0.02
SBP (mmHg) 2	192.1(±33.2)	195.3(±34.5)	0.69
DBP (mmHg) 2	106.1(±21.6)	98.3(±19.9)	0.11
Number of Meds2	2.4(±0.8)	2.5(±1.2)	0.51
Serum Creatinine (mg/dL)1,2	1.7(±1.1)	1.8(±1.0)	0.60
MDRD eGFR (mL/min/1.73m^2)	6.6(±21.5)	41.8(±19.7)	0.32
Ejection Fraction (%)1,2	59.2(±10.6)	57.7(±11.5)	0.57
LV Mass Index (g/m^2)	125.3(±31.7)	151.9(±49.0)	0.01
Female Gender1,2	21 (72%)	28 (60%)	0.26
White Race	26 (90%)	46 (98%)	0.15
Dialysis Preop	2 (7%)	2 (4%)	0.63
History of CAD	13 (46%)	19 (41%)	0.66
History of TIA or CVA1,2	3 (11%)	10 (21%)	0.35
History of Diabetes1,2	5 (17%)	12 (25%)	0.40
History of CHF1,2	1 (4%)	6 (13%)	0.24
Ejection Fraction <50%	5 (17%)	11 (23%)	0.52
Bilateral RA Repair1,2	16 (55%)	27 (57%)	0.85

¹indicates covariate in multivariable model for survival,

²indicates covariate in multivariate model for blood pressure response

Bilateral brachial artery oscillometric measurements were made during patient visits to the Wake Forest University Baptist Medical Center Vascular Surgery Clinic. Blood pressure measurements were obtained in the sitting position after 5 minutes of rest. Blood pressure response was classified based on resting blood pressure and medication requirements at eight-week postoperative follow-up. Renal function was estimated from serum creatinine measurement at 3 weeks after surgery. This follow-up time was selected based on previous experience that suggested early blood pressure response (at 8 weeks) and renal function response (at 3 weeks) demonstrated significant and independent association with death or dialysis on long-term follow-up^8,9. Blood pressure response was considered cured in patients who had a diastolic blood pressure (DBP) less than 90mmHg on no anti-hypertensive medications. Blood pressure was considered improved if any of the following criteria were satisfied: (1) Blood pressure was controlled preoperatively (i.e. DBP < 90mmHg) and there was a reduction of at least two drugs postoperatively (2) Blood pressure was not controlled preoperatively but became controlled postoperatively with a less than 20mmHg decrease in DBP with a decrease of at least one medication. (3) Blood pressure was not controlled preoperatively and became controlled with greater than 20mmHg decrease in DBP and no increase in medications. Patients that did not meet criteria for cured or improved blood pressure response were considered failed blood pressure response¹⁰.

Results

76 patients were followed for an average of 41.9 months (range: 0.2 to 77.5 months) after renal revascularization. 49(64.5%) of patients were female and the mean age was 63 ± 12.8 years. Within this group 47/76 (61.8%) patients were identified as having moderate or severe diastolic dysfunction. Systolic function was preserved. The mean systolic ejection fraction for the entire group was 58.3% ± 11.1%. Diastolic function had no apparent association with systolic function. The mean ejection fraction for those with moderate/severe diastolic dysfunction was 57.7% ± 11.5% (Figure 2, online supplement). When comparing the moderate/severe and none/mild groupings of diastolic dysfunction, there was a significant difference in left ventricular mass index (151.9 ± 48.9 g/m² vs. 125.3 ± 31.7 g/m², P = .0087) as well as age (66.0 ± 9.40 years vs. 58.3 ± 16.0 years, P = .0245). There were no significant differences among the groups in systolic blood pressure, diastolic blood pressure, serum creatinine, or estimated glomerular filtration rate. 9 patients progressed to eventual dialysis dependence during the follow-up period. This included 2 patients who underwent primary nephrectomy and 7 patients who underwent concomitant aortic repair with their open renal artery procedure. The preoperative mean serum creatinine was 3.1 (±2.0) mg/dL for patients reaching eventual dialysis dependence. There were no failures of renal artery repair observed among this group.

There were 5 deaths in the hospital and within 30 days of operation (i.e. death prior to discharge and/or within 30 days of operation) . There were 20 deaths on follow-up among perioperative survivors. Based on chart review, perioperative deaths were attributed to: cardiac causes (1), aspiration (1), CVA (1), mesenteric ischemia (1), and unknown (1). Based on chart review and NDI data, follow-up deaths were attributed to: cardiac causes (8), mesenteric ischemia (1), pneumonia (1), stroke (1), COPD exacerbation (1), trauma (1), and unknown (7). Patency without recurrent stenosis defined by renal duplex ultrasonography was 96% (±4%) at one year and 92% (±8%) at 3 years. Postoperative blood pressure response but not renal function response was related to diastolic function. Among perioperative survivors, hypertension was cured or improved in 82% of the none/mild diastolic dysfunction group and 53% of the moderate/severe group (P = .012) (Table 2). In multivariable analysis, none/mild diastolic dysfunction was significantly and independently associated with an improvement in blood pressure (OR 6.2, 95% CI 1.4-29.0, P = .018). In contrast to blood pressure, renal function was improved in 32% of the none/mild diastolic dysfunction group and 26% of the moderate/severe group (p=0.54).

Table 2.

Blood Pressure Response Data Displayed as mean (standard deviation)

Diastolic Dysfunction	Number of Patients	Preoperative			Postoperative
		Systolic Blood Pressure	Diastolic Blood Pressure	Number of Medications	Systolic Blood Pressure	Diastolic Blood Pressure	Number of Medications
None/Mild	28	190.0 (±31.8)	104.5 (±20.2)	2.4 (±0.8)	138.7 (±20.8)	79.9 (±9.1)	1.9 (±0.7)
No change	5	174 (±22.2)	74.2 (±8.3)	2.2 (±0.8)	157.6 (±32.1)	74.8 (±13.6)	2.0 (0.0)
Cured / Improved	23	193.5 (±32.9)	111.1 (±15.3)	2.4 (±0.8)	134.6 (±15.6)	81.0 (±7.8)	1.9 (±0.8)
Moderate / Severe	45	197.1 (±33.6)	100.0 (±18.5)	2.6 (±1.2)	153.3 (±29.2)	75.2 (±10.8)	2.5 (±1.0)
No change	21	191.3 (±89.9)	89.9 (±15.3)	2.1 (±0.9)	162.7 (±34.5)	78.0 (±12.8)	2.8 (±1.0)
Cured / Improved	24	202.2 (±29.7)	108.8 (±16.6)	3.0 (±1.3)	145.2 (±21.2)	72.8 (±8.3)	2.3 (±1.0)

Diastolic function but not systolic function was associated with follow-up survival. Forward stepwise selection of variables within the multivariable proportional hazards regression analysis for 71 patients demonstrated a significant and independent association between diastolic function and survival (HR 5.8, 95% CI 1.4–25, P = .018) (Figure 1, Table 3). No other variable (including age) demonstrated a significant association with survival. An additional multivariable model was created adding the covariate of age along with those chosen from the stepwise selection. In this model, diastolic function remained significantly associated with long term survival (HR 5.1, 95% CI 1.2-2.3, P=.03).

Figure 1.

Product limit estimates of long-term survival by diastolic function. Moderate/severe diastolic function appears in red; none/mild appears in black.

Table 3.

Results of forward stepwise variable selection from multivariable proportional hazards regression model of time to death

Analysis of Maximum Likelihood Estimates
Covariate	Hazard Ratio	95% Hazard Ratio Confidence Limits		P-value
Diastolic dysfunction: moderate / severe	5.84	1.35	25.23	0.018
History of transient ischemic attack and/or cerebrovascular attack	2.48	0.86	7.13	0.092

Discussion

To our knowledge, this cohort study is the first to demonstrate that diastolic function was significantly and independently associated with blood pressure response and follow-up survival in patients undergoing open renal revascularization. These findings affirm the increased association of diastolic dysfunction in patients with renovascular disease and suggest an expanded role for echocardiography as a tool for risk stratification.

Diastolic dysfunction is a clinical state of impaired ventricular filling secondary to poor ventricular relaxation and stiffness¹¹. It is often classified according to four stages: normal, mild, moderate and severe diastolic dysfunction. In a normal state, the majority of ventricular filling occurs early in diastole as a result of negative pressure created by ventricular relaxation. On echocardiography, this is depicted with the early transmitral flow velocity (E). This is supplemented late in the cardiac cycle with additional filling associated with atrial contraction (A). An E/A ratio of 0.75 to 2 is seen in patients with normal diastolic function. With mild disease there is an impairment of ventricular relaxation without increase in filling pressures. This phenomenon results in a decrease in early filling associated with an E/A ratio less than 0.75. As diastolic relaxation decreases further, intracardiac pressures become elevated and early transmitral flow improves resulting in an E/A ratio that returns to normal. This is referred to as “pseudo-normalization” and is a sign of a moderate level of diastolic dysfunction. Observations in addition to transmitral flow velocities must be considered in order to differentiate of pseudo-normalization from that of normal diastolic function. Severe diastolic dysfunction is associated with significant reductions in ventricular compliance and an E/A > 2^12-14. Mitral annular tissue Doppler measurements (e’) are associated with ventricular relaxation and can be utilized to aid in classification of diastolic function. Our study included both E/A and E/e’ for classification purposes since these measures have been extensively reported by others, the values were easily obtained by retrospective interpretation of the echocardiography studies, and the combined use of E/A and E/e’ was not subject to pseudo-normalization.

Diastolic dysfunction can be asymptomatic or manifest as diastolic heart failure. Cross-sectional studies have shown that up to half of patients with the clinical syndrome of heart failure have a preserved systolic ejection fraction¹⁵. It is believed that these cases are comprised primarily of patients with diastolic dysfunction¹⁶. A report by Persson et al. demonstrated that 67% of hospitalized patients with heart failure and preserved ejection fraction demonstrated echocardiographic evidence of diastolic dysfunction of mild to severe degree. Moreover, these authors demonstrated that diastolic dysfunction of moderate to severe degree was a significant and independent predictor of adverse outcome¹⁷. In addition, echocardiographic evidence of diastolic dysfunction has been independently associated with elevated atrial (B-type) natriuretic peptide and is a stronger predictor than either EDVI or ejection fraction of atrial peptide elevation¹³. Collectively, these observations have been attributed to ventricular hypertrophy secondary to hypertension, ischemic ventricular changes, and/or humoral effects. In this study, seven patients demonstrated a history of clinical congestive heart failure. Three of these patients had a reduced systolic ejection fraction and six had moderate to severe diastolic dysfunction.

When comparing reports on diastolic function, it is important to note the differences in the criteria for diagnosis and the classification of diastolic dysfunction. Some authors advocate composite calculations of transmitral flow estimations, tissue Doppler velocities, pulmonary vein observations, and flow deceleration times to estimate diastolic function. Moreover, the categorization of diastolic dysfunction varies among reports. Different criteria have been used for diagnosis and different categories have been used to describe the prevalence of diastolic dysfunction. These variations make comparisons of study results difficult. The patients in our study were categorized with a combination of transmitral flow velocity ratios and mitral annular tissue Doppler velocity ratios. While transmitral flow velocities can be an effective method of classification, their interpretation can be complicated by the effects of pseudonormalization and low volume states. Tissue Doppler analysis is considered less susceptible to variations in intravascular volume. By combining both of these measures to categorize diastolic function, we believe the best estimate of ventricular relaxation was provided^1,18-24.

Our findings support prior reports that have described a high prevalence of diastolic dysfunction in patients with renovascular disease. In a cross-sectional study by Redfield et al., echocardiography was performed in 2042 randomly selected subjects from the general population. Implementing a classification scheme similar to ours, these authors found the prevalence of moderate or severe diastolic dysfunction to be 7.3%¹⁴. Wright et al. found a similar prevalence (12%) in control patients, however diastolic dysfunction was present in 40.5% of patients with renovascular disease⁴. Our study demonstrated a 61.8% prevalence of moderate to severe diastolic dysfunction in patients undergoing renal revascularization. This high prevalence of diastolic dysfunction may be a result of a selection bias in which patients selected for cardiac stress testing and resting echocardiography prior to surgical intervention demonstrated more advanced disease. However, if one considered all unstudied patients free of diastolic dysfunction, a prevalence of 44% severe diastolic dysfunction would be observed for all 106 patients undergoing open operative repair during this reporting period.

Plausible explanations exist for the apparent association between renovascular disease and diastolic function. Increased afterload secondary to renovascular hypertension can result in ventricular hypertrophy and subsequent diastolic changes. In vitro studies have also suggested that direct humoral effects from the renin-angiotensin-aldosterone system may affect cardiac tissue. Specifically, angiotensin II has been associated with increased collagen accumulation/production in cardiac tissues in animal models^25,26. In concert, these effects may account for the high prevalence of diastolic dysfunction in patients with renovascular disease.

Diastolic dysfunction has been associated with an increase in morbidity and a decrease in follow-up survival independent of age after other vascular procedures. Matyal et al. found that, among 315 vascular surgery patients, diastolic dysfunction diagnosed by transmitral flow propagation velocity was an independent predictor of postoperative congestive heart failure and increased length of stay³. A recent study by Flu et al. evaluated preoperative echocardiographic data and outcome in 1005 consecutive open and endovascular procedures in which diastolic function was assessed by E/A ratio and deceleration time (DT). Among the 649 patients who underwent open surgery, the authors demonstrated an increase in 30-day adverse cardiac and vascular events (odds ratio 1.8, CI 1.1-2.9) in those with isolated left ventricular diastolic dysfunction. Long-term cardiac and vascular mortality was increased in patients with asymptomatic and isolated diastolic dysfunction (hazard ratio 3.0, CI 1.5-6.0) ¹. Collectively, these studies demonstrate the association of diastolic dysfunction with perioperative and follow-up adverse outcomes after a variety of vascular procedures.

Although diastolic dysfunction appears to be associated with adverse cardiac events, studies to date have shown inconsistent response to medical management. General recommendations for treatment of the underlying causes of diastolic heart failure include control of hypertension and diabetes as well as avoidance of myocardial ischemia and tachycardia, however there is little evidence to prove their effectiveness¹⁶. Data from the Heart and Soul Study demonstrated a 40% decrease in risk of heart failure hospitalization in patients with diastolic dysfunction treated with beta blocker therapy²⁷. Candesartan in Heart Failure Assessment of Mortality and Morbidity Preserved Study (CHARM-Preserved) examined patients with an EF >40%. CHARM failed to demonstrate improved survival free of clinical congestive heart failure¹⁷. Irbesartan in Heart Failure with Preserved Ejection Fraction (I-PRESERVE) Trial also failed to demonstrate improved outcomes with the use of irbesartan in patients with diastolic dysfunction²⁸. The value of the medical management of diastolic dysfunction in this study is unknown but deserves further study.

Although we believe our findings are of potential value, this study has several limitations. This report suffers from the biases inherent to a retrospective report originating from a cohort identified from an institutional procedural database. The small size of this study limited the test of association between diastolic function and perioperative morbidity and mortality. Moreover, this study did not include postoperative echocardiography to assess for changes in heart structure on follow-up. Almost half (48%) of patients underwent concomitant aortic intervention for significant disease and confounding is possible. Additionally, selection bias may have resulted from requiring resting echocardiography. Dobutamine stress echocardiography did not provide parameters to estimate diastolic function. Analysis of patient characteristics among those with resting echocardiography and those with only dobutamine stress echocardiography did reveal a significantly higher preoperative systolic and diastolic blood pressure in those with resting echocardiography. All other analyzed descriptive characteristics were similar between the two groups including age, number of blood pressure medications, history of coronary artery disease, history of renal insufficiency, and history of left ventricular hypertrophy. Despite these limitations, the data and the relationships demonstrated in this analysis suggest the need for additional prospective studies to confirm the strong relationship between diastolic function, blood pressure response, and survival among patients submitted to open surgical repair of atherosclerotic renovascular disease.

Conclusions

Diastolic dysfunction was prevalent in patients with atherosclerotic renovascular disease. Normal and near-normal diastolic function demonstrated a significant and independent association with blood pressure benefit after open operative repair. Moreover, diastolic dysfunction demonstrated a significant and independent relationship with follow-up mortality. Pre-operative diastolic function should be estimated in patients with renovascular disease.