Absence of electrocardiographic left ventricular hypertrophy is associated with increased mortality after transcatheter aortic valve replacement

Abstract

Background

Electrocardiographic (ECG) left ventricular hypertrophy (LVH) has been associated with increased mortality in patients with asymptomatic aortic stenosis (AS) and hypertension. However, patients with symptomatic AS undergoing transcatheter aortic valve replacement (TAVR) have higher percentages of myocardial fibrosis or amyloidosis that have been associated with decreased ECG voltage and worse outcomes.

Hypothesis

We tested the hypothesis that baseline ECG LVH is independently associated with increased all‐cause mortality after TAVR.

Methods

A total of 231 patients (96 men; mean age 84.7 ± 7.8 years) that underwent TAVR at our institution were included. Cornell voltage, defined as S_V3 + R_aVL, was used to assess for presence of ECG LVH using gender‐specific cut‐off values. We used the Kaplan‐Meier estimator to derive survival curves. Multivariate Cox regression analysis was used to compare mortality between patients without vs with ECG LVH and adjust for echocardiographic LVH and predictors of mortality in this cohort.

Results

Over a follow‐up time of 16.3 ± 10.4 months, the absence of ECG LVH was significantly associated with increased mortality (40.4% vs 23.6% at 2‐years, log rank P = 0.003). After adjusting for echocardiographic LVH and predictors of mortality in our cohort, the absence of ECG LVH remained a predictor of increased mortality (HR = 1.79, CI 95% 1.02‐3.14, P = 0.042).

Conclusions

The absence of ECG LVH was independently associated with increased mortality in patients undergoing TAVR. Baseline ECG may have an important prognostic role in these patients and could lead to further testing to evaluate for myocardial fibrosis or amyloidosis.

1. INTRODUCTION

Transcatheter aortic valve replacement (TAVR) has become a less invasive option over surgery for high‐ and intermediate‐risk surgical candidates.1, 2, 3 Despite serving as a non‐surgical option for this primarily elderly population, the work‐up before TAVR is often extensive, expensive, involves invasive procedures, and frequently requires hospitalization.4 In addition, the procedure itself can result in stroke, vascular complications, renal failure, bleeding, and paravalvular aortic insufficiency.5 Further, a subset of patients do not experience any improvement in quality of life after TAVR.6, 7 Therefore, it would be of substantial clinical impact to identify inexpensive and non‐invasive tests, such as surface 12‐lead electrocardiogram (ECG), that could help identify subsets of patients less likely to benefit from TAVR. Left ventricular hypertrophy (LVH) identified by ECG has been associated with worse outcomes in hypertensive patients,8, 9, 10, 11, 12, 13 as well as asymptomatic aortic stenosis (AS).14, 15, 16, 17 However, in elderly patients with AS that undergo TAVR, adverse concurrent pathophysiologic mechanisms such as amyloidosis or myocardial fibrosis can lead to decreased ECG voltage.18 The prognostic value of ECG LVH has not been thoroughly studied in TAVR patients.

In the current study, we therefore hypothesized that the absence of baseline ECG LVH is independently associated with increased all‐cause mortality after TAVR and performed a retrospective analysis to test that hypothesis.

2. METHODS

We evaluated 369 consecutive patients who underwent TAVR by transfemoral, transapical, or transaortic approach between March 2009 and December 2014 in a single center. Patients underwent placement of either the Edwards SAPIEN or SAPIEN XT balloon‐expandable transcatheter heart valve (Edwards Lifesciences, Irvine, California) or the Medtronic CoreValve self‐expanding transcatheter bioprosthetic heart valve (Medtronic, Inc., Minneapolis, Minnesota). Patients with a complete left bundle branch block, complete right bundle branch block, or ventricular‐paced rhythm on baseline ECG were excluded. After exclusions, 231 patients were eligible for this study.

Data were collected in an institutional TAVR database. This database included basic patient demographics, past medical history, and baseline cardiopulmonary tests before TAVR including data from ECGs, echocardiograms, right and left heart catheterizations. Pre‐procedural laboratory values as close to TAVR as possible, including baseline hemoglobin and serum creatinine, as well as procedural information were also recorded. Post‐TAVR complications were reported as per the Valve Academic Research Consortium‐2 consensus document.5 Patient follow‐up was performed at the time of clinical visits or by phone contact. The study was conducted with the approval of the local Institutional Review Board.

Baseline ECGs were performed on all patients prior to TAVR during index hospitalization. All ECGs were analyzed by dedicated software (GE MUSE, Milwaukee, Wisconsin) and were confirmed by a single experienced cardiologist who was blinded to the patient's echocardiographic data and outcome (PMO). ECG LVH was defined by Cornell voltage criteria (R wave in aVL + S wave in V3 > 28 mm in males; >20 mm in females).

All patients underwent echocardiography prior to TAVR. Transaortic velocity by continuous Doppler was used to calculate mean aortic valve (AV) gradient. The continuity equation was used to calculate AV area, which was indexed to body surface area to calculate AV area index (AVAI). Low‐flow, low‐gradient (LFLG) severe aortic stenosis was defined as a mean AV gradient ≤ 40 mmHg with AVA <1.0 cm² or AVAI <0.6 cm²/m².19 Left ventricular (LV) internal dimension in diastole (LVIDd), LV posterior wall thickness (LVPTd) and interventricular septal thickness in diastole (IVSd) were measured in accordance with ASE guidelines. LV mass was derived using previously described equations and was also indexed to body surface area to calculate LV mass index (LVMI).20 LV ejection fraction (LVEF) and chamber size were calculated by the Teichholz method. Aortic insufficiency and mitral regurgitation were graded according to ASE guidelines.21 Paravalvular aortic insufficiency after TAVR was evaluated according to the Valve Academic Research Consortum‐2 criteria on echocardiograms performed 30 days post‐TAVR. When 30‐day echocardiograms were not available, discharge echocardiograms were used.

Patients were divided into two groups based on the presence or absence of ECG LVH by Cornell voltage criteria on the baseline ECG. Mean ± SD was used to present continuous variables and these were compared using the Student's t test. Proportions were used to present categorical variables and were compared using a χ² test or Fisher's exact test where appropriate. Baseline and procedural characteristics as well as post‐TAVR complications were compared between the two groups. The relationship of all‐cause mortality to Cornell voltage LVH was assessed by Kaplan‐Meier analysis and compared by the log‐rank test. Univariate Cox regression models were used to identify baseline predictors of all‐cause mortality after TAVR, with findings presented as hazard ratios (HR) and 95% confidence intervals (CI). Univariate predictors of mortality with an alpha value less than 0.10 were included in a multivariable Cox regression model to derive independent predictors of all‐cause mortality. The degree of LVH as defined by echocardiographically calculated LVMI, the presence of LFLG AS as well as history of MI were forced into the final multivariable Cox model because of their known association with low voltage and adverse outcomes from prior studies.18, 22, 23, 24 Interaction terms were used to determine if the predictive value of Cornell voltage LVH differed significantly in relevant subsets of the population defined by LFLG AS and transfemoral vs non‐transfemoral approaches to TAVR. A two‐tail P value less than 0.05 was taken to meet statistical significance for all tests. IBM SPSS Statistics version 21 (IBM Corp., Armonk, New York) was used for all statistical analysis.

3. RESULTS

Of the 231 patients (mean age 84.7 ± 7.8 years, 42% male) in this study, 99 patients (43%) had ECG LVH by Cornell voltage criteria while 132 (57%) did not. Baseline demographic, clinical and echocardiographic characteristics of patients with and without ECG LVH by Cornell voltage are compared in Table 1. Patients without ECG LVH were more likely to be male and have had a previous myocardial infarction, however, without significantly higher prevalence of prior CABG or PCI. They also had larger AVAI and lower mean AV gradients. LFLG aortic stenosis occurred more frequently in patients without ECG LVH, however this difference did not reach statistical significance. Not surprisingly, patients without ECG LVH had lower LVMI, IVSd, and LVPWd by echocardiography, and were also less likely to have baseline aortic insufficiency classified as mild or greater. There were no significant differences in the rates of post‐TAVR complications including vascular and bleeding complications, stroke and post‐TAVR AI between the two groups (Table 2).

Table 1.

Baseline clinical and echocardiographic characteristics^a

Clinical and laboratory	All (n = 231)	ECG LVH (n = 99)	No ECG LVH (n = 132)	P
Age (y)	84.7 ± 7.8	85.4 ± 7.3	84.1 ±8.2	0.205
Male gender (%)	41.6% (96)	30.3% (30)	50.0% (66)	0.003
BMI (kg/m2)	26.9 ± 6.3	26.9 ± 5.6	26.9 ± 6.7	0.939
Diabetes mellitus	32% (74)	33.3% (33)	31.1% (41)	0.714
Hypertension	85.7% (198)	90.7% (88)	84.0% (110)	0.136
Prior MI	19% (44)	9.6% (9)	26.9% (35)	0.001
Prior PCI	42% (98)	38% (38)	46% (60)	0.282
Prior CABG	22% (51)	20% (20)	24% (31)	0.551
Prior stroke	9.5% (22)	9.1% (9)	9.8% (13)	0.846
Lung disease	29.4% (68)	27.3% (27)	31.1% (41)	0.532
NYHA class III or IV	67.5% (156)	63.3% (62)	71.8% (94)	0.173
Atrial fibrillation/flutter	19.9% (46)	17.2% (17)	22.0% (29)	0.366
Baseline Hb (g/dL)	11.1 ±1.6	11.2 ±1.6	11.0 ±1.6	0.505
Baseline Cr (mg/dL)	1.35 ± 1.28	1.30 ±1.15	1.39 ±1.37	0.637
PASP (mmHg)	46.4 ± 16.0	48.5 ± 16.7	44.8 ± 15.3	0.101
Transfemoral access	61.5% (142)	67.7% (67)	56.8% (75)	0.093
Echocardiographic
AVAI (cm2/m2)	0.407 ±0.089	0.391 ±0.076	0.419 ±0.097	0.020
Mean AV gradient (mmHg)	50.7 ± 15.2	54.8 ± 17.6	47.6 ± 12.3	0.001
Stroke volume indexed (mL/m2)	40.3 ± 11.7	41.6 ± 12.3	39.3 ± 11.5	0.130
LFLG severe AS	18.6% (43)	13.1% (13)	22.7% (30)	0.064
Baseline AI ≥ mild	51.1% (118)	61.6% (61)	43.8% (57)	0.008
Baseline MR ≥ moderate	17.3% (40)	12.1% (12)	21.5% (28)	0.063
LVEF (%)	53.0 ±13.8	53.2 ±14.3	52.8 ±13.4	0.843
LVMI (g/m2)	112.1 ± 27.9	118.9 ± 25.6	106.9 ±28.5	0.001
IVSd (cm)	1.01 ± 0.18	1.05 ± 0.20	0.99 ±0.15	0.011
LVPWd (cm)	0.93 ± 0.15	0.96 ± 0.15	0.91 ± 0.14	0.013
LVIDd (cm)	5.31 ± 0.72	5.34 ± 0.68	5.29 ± 0.75	0.619

^aAll continuous variables expressed as a mean ± SD. All categorical variables expressed as proportions.

Abbreviations: AI, aortic insufficiency; AS, aortic stenosis; AV, aortic valve; AVAI, aortic valve area index; BMI, body mass index; CABG, coronary artery bypass grafting; Cr, serum creatinine; ECG, electrocardiographic; Hb, hemoglobin; IVSd, interventricular septal dimension; LFLG, low‐flow, low‐gradient; LVEF, left ventricular ejection fraction; LVH, left ventricular hypertrophy; LVIDd, left ventricular internal diameter at end diastole; LVMI, left ventricular mass index; LVPWd, left ventricular posterior wall dimension; Lung Disease, obstructive or restrictive lung disease; MI, myocardial infarction; MR, mitral regurgitation; NYHA, New York Heart Association; PASP, catheter derived pulmonary artery systolic pressure; PCI, percutaneous coronary intervention.

Table 2.

Post‐TAVR complications^{a , b}

Complications	All (n = 231)	ECG LVH (n = 99)	No ECG LVH (n = 132)	P
Major vascular	2.2% (5)	4.0% (4)	0.8% (1)	0.167c
Minor vascular	3.0% (7)	3.0% (3)	3.0% (4)	1.000c
Major bleeding	32.5% (75)	35.4% (35)	30.3% (40)	0.417
Minor bleeding	15.6% (36)	13.1% (13)	17.4% (23)	0.373
AI > mild	10.1% (13)	13.1% (13)	7.8% (10)	0.188
CVA	1.7% (4)	2% (2)	1.5% (2)	1.0c
Stage 2 or 3 AKI	8.7% (20)	7.1% (7)	9.8% (13)	0.458

Abbreviations: AI, aortic insufficiency; AKI, acute kidney injury; CVA, cerebrovascular accident including hemorrhagic, ischemic stroke, and transient ischemic accident; ECG, electrocardiographic; LVH, left ventricular hypertrophy; TAVR, transcatheter aortic valve replacement.

^aAll categorical variables expressed as proportions.

^bAll complications reported as per definitions outlined by the Valve Academic Research Consortium‐2 consensus document⁵.

^cFisher's exact test.

During a mean follow‐up time of 16.3 ± 10.4 months after TAVR, 68 patients (29.4%) died. Over the follow‐up period, all‐cause mortality was 40.4% in patients without ECG LVH at baseline as compared to 23.6% in those who met Cornell voltage criteria (Figure 1). The absence of ECG LVH at baseline was significantly associated with increased all‐cause mortality (log‐rank test by Kaplan‐Meier estimator P = 0.003, HR 2.18, 95% CI 1.28‐3.71, P = 0.004, Figure 1). Other statistically significant baseline predictors of all‐cause mortality in this cohort were mean AV gradient, baseline serum creatinine >2 mg/dL, hemoglobin, atrial fibrillation or flutter, obstructive or restrictive lung disease, and prior history of stroke (Table 3) but not age, gender, LFLG aortic stenosis (HR 1.29 [0.73‐2.27], P = 0.385), prior MI (HR 1.38 [0.76‐2.50], P = 0.290), or LVMI (HR 1.00 [0.99‐1.01], P = 0.793). In a multivariate regression analysis that included all baseline predictors of mortality in a step‐wise, forward manner followed by forced entry of prior MI, echocardiographic LVMI and LFLG AS in the final Cox regression model, the absence of ECG LVH by Cornell voltage criteria remained an independent predictor of all‐cause mortality (HR 1.79 [1.02‐3.14], P = 0.042), as did baseline atrial fibrillation or flutter, history of lung disease, hemoglobin level, and serum creatinine >2 mg/dL (Table 3). Furthermore, there was no statistically significant difference in the predictive value of the absence of ECG LVH in patients with vs without LFLG AS or in patients with transfemoral vs non‐transfemoral approach (P = 0.502 and P = 0.283, respectively for interaction with the absence of ECG LVH).

Figure 1.

Kaplan‐Meier curves for all‐cause survival according to the presence of ECG LVH. ECG, electrocardiographic; LVH, left ventricular hypertrophy; TAVR, transcatheter aortic valve replacement

Table 3.

Univariate and multivariable prediction of mortality after TAVR^a

	Univariate HR (95% CI)	p	Multivariable HR (95% CI)b	P b
No ECG LVH	2.18 (1.28‐3.71)	0.004	1.79 (1.02‐3.14)	0.042
Male gender	1.52 (0.93‐2.48)	0.093	NS	NS
Age	0.97 (0.94‐1.00)	0.062	NS	NS
NYHA class III/IV	1.70 (0.93‐3.12)	0.087	NS	NS
Mean AV gradient (per mmHg)	0.98 (0.97‐0.99)	0.041	NS	NS
AFib/flutter	2.51 (1.48‐4.24)	0.001	2.42 (1.39‐4.21)	0.002
CVA	2.06 (1.05‐4.06)	0.036	NS	NS
Cr > 2 mg/dL	3.16 (1.67‐5.96)	<0.001	2.60 (1.31‐5.14)	0.006
Hb (per 1 g/dL)	0.79 (0.67‐0.92)	0.002	0.78 (0.66‐0.93)	0.005
Lung disease	1.76 (1.08‐2.87)	0.024	2.07 (1.24‐3.47	0.006
Prior MI	1.38 (0.76‐2.50)	0.290	NS	NS
LFLG severe AS	1.29 (0.73‐2.27)	0.385	NS	NS
LVMI	1.00 (0.99‐1.01)	0.793	NS	NS

Abbreviations: AFib, atrial fibrillation; AS, aortic stenosis; AV, aortic valve; CI, confidence interval; Cr, serum creatinine; CVA, cerebrovascular accident; ECG, electrocardiographic; Hb, hemoglobin; HR, hazard ratio; LFLG, low‐flow, low‐gradient; LVH, left ventricular hypertrophy; LVMI, left ventricular mass index; lung disease, obstructive or restrictive lung disease; MI, myocardial infarction; NYHA, New York Heart Association; TAVR, transcatheter aortic valve replacement.

^aUnivariate and multivariable predictors of mortality expressed as a hazard ratio with associated 95%CI.

^bAlso adjusted for LFLG AS, prior MI and LVMI.

4. DISCUSSION

To our knowledge, this is the first study to examine the relationship of baseline ECG LVH to mortality after TAVR. During a mean follow‐up time of 16.3 ± 10.4 months after TAVR, those patients without ECG LVH had nearly twice the adjusted mortality risk of those who met Cornell voltage criteria independently of other univariate baseline predictors of mortality in our cohort as well as LVMI, suggesting that any impact of ECG LVH on mortality was distinct from myocardial hypertrophy. Additionally, known post‐TAVR complications that are associated with increased mortality including post‐TAVR AI, stroke and vascular/bleeding complications were not significantly increased in patients without ECG LVH in this study and thus, could not account for the increased mortality of this group.

Although ECG LVH criteria have limited sensitivity for detecting actual myocardial hypertrophy, these markers that are based on QRS amplitude have good specificity for physiological LVH and studies have shown the presence of ECG LVH to be a strong predictor of both cardiovascular mortality and morbidity.9, 10, 11, 12 Similar to hypertensive patients, ECG LVH is often found in those with aortic stenosis in whom the left ventricle faces a pathologically increased afterload secondary to a narrowed AV and has been shown to be associated with worse clinical outcomes. Post‐hoc analysis of the Simvastatin and Ezetemibe in aortic stenosis (SEAS) trial, a randomized control study that explored the benefits of intensive lipid‐lowering therapy on cardiovascular outcomes in patients with asymptomatic mild to moderate aortic stenosis, showed that patients with ECG LVH were more likely to develop heart failure and progress to need AV replacement.15 A study by Lund et al showed that ECG LVH and the strain pattern of repolarization were independent predictors of mortality among patients with severe aortic stenosis awaiting surgical AV replacement.16 Based on the above, why might the absence of “electrical remodeling” be potentially detrimental after TAVR?

Loss of healthy myocardium secondary to ischemic disease is a known cause of low QRS voltage.23, 24 Patients without ECG LVH had a higher incidence of prior MI (26.9% vs 9.6%, P = 0.001) in our cohort. While prior MI was not a significant univariate predictor of mortality and our results were adjusted for it by forcing this variable into the final multivariate regression model, it is conceivable that those without “electrical remodeling” were more likely to have unreported “silent” infarctions or increased burden of microvascular disease at baseline, which could have contributed to higher mortality after TAVR.

Secondly, patients without ECG LVH in our study were more likely to have low‐gradient severe aortic stenosis (22.7% vs 13.1%, respectively, P = 0.064) and lower mean AV gradients (47.6 ± 12.3 vs 54.8 ± 17.6 mmHg, P = 0.001) suggesting that the absence of “electrical remodeling” may be a marker a decreased flow state. Although the absence of ECG LVH was associated with increased mortality in the current study regardless of mean AV gradient and LFLG AS, we believe it may be worth further investigating this interaction. Increased myocardial fibrosis, which is more frequently found in both classical and paradoxical LFLG is a known predictor of mortality after valve replacement and could be responsible for the decreased voltage on the surface ECG.25, 26 Even in the absence of a low‐flow state, advanced AS itself could be associated with the presence of increased myocardial fibrosis, similarly to patients with hypertrophic cardiomyopathy.27 Unfortunately, the extent of myocardial fibrosis itself was not evaluated in the present study.

Thirdly, cardiac amyloidosis, an infiltrative cardiomyopathy often associated with low QRS voltages, advanced age and restrictive physiology, is another potential explanation for the increased mortality in TAVR patients without ECG LVH.28 Studies suggest that transthyretin cardiac amyloidosis is relatively common in calcific aortic stenosis.29, 30 In fact, a recent study actually showed increased frequency of low‐flow state and decreased ECG voltages in AS patients with proven transthyretin cardiac amyloidosis undergoing TAVR.18

The current retrospective study has several limitations. As our purpose was to explore baseline non‐invasive characteristics that could help identify high‐risk groups, we did not compare post‐TAVR characteristics such as post‐TAVR LVH regression between the two groups, which may be of prognostic value in assessing long‐term outcomes among survivors. In addition, the presence of ECG LVH was evaluated using the Cornell criteria only. Whether other LVH criteria by ECG verify our findings remains to be seen. Furthermore, whereas low QRS voltage is known to be associated with cardiac amyloidosis and could be potentially associated with myocardial fibrosis, our study did not evaluate for these pathophysiologic mechanisms. Finally, it is possible that the results of our study will not be reproducible in younger patients given the higher prevalence of myocardial fibrosis or amyloidosis in geriatric populations.

5. CONCLUSIONS

Overall, the purpose of the current study was to explore if an ECG performed as part of a pre‐TAVR evaluation can help detect a subset of patients with severe AS who are not good candidates for TAVR, even before undergoing complex pre‐procedural workup. Our main finding was that the absence of baseline ECG LVH as defined by Cornell voltage criteria was an independent predictor of increased mortality after TAVR. Further studies with larger patient cohorts are required to examine whether low QRS voltage, particularly the lack of ECG LVH, carries worse prognosis in patients undergoing TAVR. If that is truly the case, the presence of cardiac amyloidosis or myocardial fibrosis could be explored as potential underlying mechanisms.

Conflict of interest

The authors declare no potential conflict of interests.

Kampaktsis PN, Ullal AV, Swaminathan RV, et al. Absence of electrocardiographic left ventricular hypertrophy is associated with increased mortality after transcatheter aortic valve replacement. Clin Cardiol. 2018;41:1246–1251. 10.1002/clc.23034