Prognostic Usefulness of Left Ventricular Hypertrophy by Electrocardiography in Patients With Atrial Fibrillation (from the Randomized Evaluation of Long-Term Anticoagulant Therapy Study) Paolo Verdecchia, MD a, *, Gianpaolo Reboldi, MD, PhD b , Giuseppe Di Pasquale, MD c , Giovanni Mazzotta, MD d , Giuseppe Ambrosio, MD d , Sean Yang, PhD e , Janice Pogue, PhD e , Lars Wallentin, MD f , Michael D. Ezekowitz, MD g,h , Stuart J. Connolly, MD e , and Salim Yusuf, MD e , On behalf of the RE-LY Study Investigators It is unknown whether left ventricular hypertrophy (LVH) diagnosis by electrocardiography improves risk stratification in patients with atrial fibrillation (AF). We investigated the prognostic impact of LVH diagnosis by electrocardiography in a large sample of anti- coagulated patients with AF included in the Randomized Evaluation of Long-Term Anti- coagulant Therapy (RE-LY) Study. We defined electrographic LVH (ECG-LVH) by strain pattern or Cornell voltage (R wave in aVL plus S wave in V 3 ) >2.0 mV (women) or >2.4 mV (men). LVH prevalence was 22.7%. During a median follow-up of 2.0 years, 303 patients developed a stroke, 778 died (497 from cardiovascular causes), and 140 developed a myocardial infarction. LVH was associated with a greater risk of stroke (1.99% vs 1.32% per year, hazard ratio [HR] 1.51, 95% confidence interval [CI] 1.18 to 1.93, p <0.001), cardio- vascular death (4.52% vs 1.80% per year, HR 2.56, 95% CI 2.14 to 3.06, p <0.0001), all-cause death (6.03% vs 3.11% per year, HR 1.95, 95% CI 1.68 to 2.26, p <0.0001), and myocardial infarction (1.11% vs 0.55% per year, HR 2.07, 95% CI 1.47 to 2.92, p <0.0001). In multivariate analysis, the prognostic value of LVH was additive to CHA 2 DS 2 -VASc score and other covariates. The category-free net reclassification index and integrated discrimination improvement increased significantly after adding LVH to multivariate models. In conclusion, our study demonstrates for the first time that ECG-LVH, a simple and easily accessible prognostic indicator, improves risk stratification in anticoagulated patients with AF. Ó 2014 Elsevier Inc. All rights reserved. (Am J Cardiol 2014;113:669e675) Left ventricular hypertrophy (LVH) diagnosed by stan- dard 12-lead electrocardiogram is an independent predictor of all-cause and cardiovascular (CV) death, 1,2 sudden car- diac death, 3 coronary artery disease, 4 and stroke. 5 However, to the best of our knowledge, the adverse prognostic impact of LVH diagnosed by electrocardiography has never been tested in cohorts of patients with atrial fibrillation (AF). AF might limit the potential for electrocardiography to detect LVH because the use of P-wave abnormalities associated with left atrial dilatation 6 is obviously restricted to patients in sinus rhythm. Because of the growing burden of AF, 7 the importance of risk stratification in these patients, 7 and the important role of hypertension and LVH 8,9 as predictors of AF, we sought to clarify the prognostic value of LVH diagnosis by electrocardiography in these patients. Electro- cardiography has the advantage of being less expensive and more widely available than echocardiography or magnetic resonance imaging and, hence, more suited for application to population-wide AF management. Thus, we reviewed for LVH the ECG tracings of patients enrolled in the Ran- domized Evaluation of Long-Term Anticoagulant Therapy (RE-LY) trial 10 and tested the hypothesis that LVH at entry would identify patients at increased risk of stroke and other major events. Methods Details of the RE-LY trial have been published. 10,11 It was a noninferiority trial of 2 doses of dabigatran compared with warfarin for prevention of stroke or systemic embolism in patients with AF and at least 1 additional risk factor for stroke. 10,11 Overall, 18,113 patients were randomized to receive dabigatran 110 mg twice daily, dabigatran 150 mg twice daily, or warfarin (dose adjusted to a target interna- tional normalized ratio 2.0 to 3.0) and monitored for a a Department of Medicine, Hospital of Assisi, Assisi, Italy; Departments of b Internal Medicine and d Cardiology and Cardiovascular Pathophysi- ology, University of Perugia, Perugia, Italy; c Department of Cardiology, Maggiore Hospital, Bologna, Italy; e Population Health Research Institute, Hamilton Health Sciences and McMaster University, Hamilton, Ontario, Canada; f Uppsala Clinical Research Center and Department of Medical Sciences, Uppsala University, Uppsala, Sweden; g Jefferson Medical College, Wynnewood, Pennsylvania; and h Atrial Fibrillation Research and Education, Cardiovascular Research Foundation, New York, New York. Manuscript received August 14, 2013; revised manuscript received and accepted October 15, 2013. This study is registered at www.clinicaltrials.gov (NCT00262600). This study was funded by Boehringer-Ingelheim, Ingelheim am Rein, Germany and the Fondazione Umbra Cuore e IpertensioneeONLUS, Perugia, Italy. See page 674 for disclosure information. *Corresponding author: Tel: (þ39) 0758139336; fax: (þ39) 0758139301. E-mail address: [email protected](P. Verdecchia). 0002-9149/13/$ - see front matter Ó 2014 Elsevier Inc. All rights reserved. www.ajconline.org http://dx.doi.org/10.1016/j.amjcard.2013.10.045
Transcript
Prognostic Usefulness of Left Ventricular Hypertrophy byElectrocardiography in Patients With Atrial Fibrillation (from the
Randomized Evaluation of Long-Term Anticoagulant Therapy Study)
Paolo Verdecchia, MDa,*, Gianpaolo Reboldi, MD, PhDb, Giuseppe Di Pasquale, MDc,Giovanni Mazzotta, MDd, Giuseppe Ambrosio, MDd, Sean Yang, PhDe, Janice Pogue, PhDe,
Lars Wallentin, MDf, Michael D. Ezekowitz, MDg,h, Stuart J. Connolly, MDe, and Salim Yusuf, MDe,On behalf of the RE-LY Study Investigators
It is unknown whether left ventricular hypertrophy (LVH) diagnosis by electrocardiography
improves risk stratification in patients with atrial fibrillation (AF). We investigated theprognostic impact of LVH diagnosis by electrocardiography in a large sample of anti-coagulated patients with AF included in the Randomized Evaluation of Long-Term Anti-coagulant Therapy (RE-LY) Study. We defined electrographic LVH (ECG-LVH) by strainpattern or Cornell voltage (R wave in aVL plus S wave in V3) >2.0 mV (women) or >2.4 mV(men). LVH prevalence was 22.7%. During a median follow-up of 2.0 years, 303 patientsdeveloped a stroke, 778 died (497 from cardiovascular causes), and 140 developed amyocardial infarction. LVH was associated with a greater risk of stroke (1.99% vs 1.32% peryear, hazard ratio [HR] 1.51, 95% confidence interval [CI] 1.18 to 1.93, p <0.001), cardio-vascular death (4.52% vs 1.80% per year, HR 2.56, 95% CI 2.14 to 3.06, p <0.0001), all-causedeath (6.03% vs 3.11% per year, HR 1.95, 95% CI 1.68 to 2.26, p <0.0001), and myocardialinfarction (1.11% vs 0.55% per year, HR 2.07, 95%CI 1.47 to 2.92, p <0.0001). Inmultivariateanalysis, the prognostic value of LVH was additive to CHA2DS2-VASc score and othercovariates. The category-free net reclassification index and integrated discriminationimprovement increased significantly after adding LVH tomultivariate models. In conclusion,our study demonstrates for the first time that ECG-LVH, a simple and easily accessibleprognostic indicator, improves risk stratification in anticoagulated patients withAF. � 2014 Elsevier Inc. All rights reserved. (Am J Cardiol 2014;113:669e675)
Left ventricular hypertrophy (LVH) diagnosed by stan-dard 12-lead electrocardiogram is an independent predictorof all-cause and cardiovascular (CV) death,1,2 sudden car-diac death,3 coronary artery disease,4 and stroke.5 However,to the best of our knowledge, the adverse prognostic impactof LVH diagnosed by electrocardiography has never beentested in cohorts of patients with atrial fibrillation (AF). AFmight limit the potential for electrocardiography to detect
f Medicine, Hospital of Assisi, Assisi, Italy; Departmentscine and dCardiology and Cardiovascular Pathophysi-of Perugia, Perugia, Italy; cDepartment of Cardiology,l, Bologna, Italy; ePopulation Health Research Institute,Sciences and McMaster University, Hamilton, Ontario,Clinical Research Center and Department of Medicala University, Uppsala, Sweden; gJefferson Medicalood, Pennsylvania; and hAtrial Fibrillation Research andvascular Research Foundation, New York, New York.ed August 14, 2013; revised manuscript received and15, 2013.egistered at www.clinicaltrials.gov (NCT00262600).s funded by Boehringer-Ingelheim, Ingelheim am Rein,Fondazione Umbra Cuore e IpertensioneeONLUS,
for disclosure information.g author: Tel: (þ39) 0758139336; fax: (þ39)
see front matter � 2014 Elsevier Inc. All rights reserved.0.1016/j.amjcard.2013.10.045
LVH because the use of P-wave abnormalities associatedwith left atrial dilatation6 is obviously restricted to patientsin sinus rhythm. Because of the growing burden of AF,7 theimportance of risk stratification in these patients,7 and theimportant role of hypertension and LVH8,9 as predictors ofAF, we sought to clarify the prognostic value of LVHdiagnosis by electrocardiography in these patients. Electro-cardiography has the advantage of being less expensive andmore widely available than echocardiography or magneticresonance imaging and, hence, more suited for applicationto population-wide AF management. Thus, we reviewed forLVH the ECG tracings of patients enrolled in the Ran-domized Evaluation of Long-Term Anticoagulant Therapy(RE-LY) trial10 and tested the hypothesis that LVH at entrywould identify patients at increased risk of stroke and othermajor events.
Methods
Details of the RE-LY trial have been published.10,11 Itwas a noninferiority trial of 2 doses of dabigatran comparedwith warfarin for prevention of stroke or systemic embolismin patients with AF and at least 1 additional risk factor forstroke.10,11 Overall, 18,113 patients were randomized toreceive dabigatran 110 mg twice daily, dabigatran 150 mgtwice daily, or warfarin (dose adjusted to a target interna-tional normalized ratio 2.0 to 3.0) and monitored for a
Figure 1. Flow diagram of the study. LBBB ¼ left bundle branch block; LV ¼ left ventricular; RBBB ¼ right bundle branch block.
670 The American Journal of Cardiology (www.ajconline.org)
median of 2 years. The primary outcome measure was acomposite of stroke or systemic embolism and the primarysafety end point was major hemorrhage.11 The protocol wasapproved by national regulatory authorities and ethicscommittees, and all patients provided written informedconsent before study entry. The investigators of this posthoc analysis had full access to the data and designed thestatistical analysis plan. For the purpose of the present study,we excluded patients with sinus rhythm at entry and usedthe selection criteria outlined in Figure 1.
Participants underwent a 25-mm/s 12-lead electrocardi-ography at entry and then annually and at the final follow-upvisit or permanent discontinuation of study drugs. Copies ofECG tracings carried out at scheduled visits were forwardedby participating centers to the coordinating center. ECGtracings of suboptimal quality (see Figure 1) were excluded.An expert reader, (GM) blinded to the patients’ features andrandomized treatment, examined the baseline ECG tracingsof all patients. We defined LVH diagnosis by electrocardi-ography in a binary (yes or no) fashion by one or both of thefollowing: (1) sum of the R wave in lead aVL and depth ofthe S wave in lead V3 >2.0 mV in women and >2.4 mV inmen and (2) strain pattern in at least one of the followingleads: I, II, aVL, or V4 to V6. Strain pattern was consideredpresent if there was ST-segment depression of at least0.5 mm and inverted T wave in any of the aforementioned
leads in the direction opposite the polarity of the QRS. Thisdefinition of ECG-LVH is simple and applicable in largepopulations.12 In a large validation study in hypertensivepatients, it yielded 34% sensitivity and 91% specificitywith echocardiographic LVH as reference and identifiedsubjects at increased risk of major CV events with a greaterpopulation-attributable risk (16.1%) when compared withtraditional ECG criteria of LVH.13
Study outcomes were stroke, CV death, all-cause death,and myocardial infarction (MI) because all these outcomeshave been associated with ECG-LVH in previous stud-ies.1e5 Sources of outcomes data were clinical and hospitalrecords and death certificates, when appropriate. Two in-dependent experts who were unaware of randomized treat-ment adjudicated outcome events. Definition of events hasbeen previously published.11 Deaths were adjudicated asbeing vascular or nonvascular, due to other specified causessuch as cancer, or of unknown cause.
For data analysis, we used the SAS software, version9.1.3 (SAS Institute, Cary, North Carolina). Continuous dataare given as mean � SD and categorical data as frequencies.We compared the characteristics of patients with andwithout LVH at entry by the Student t test for continuousvariables and chi-square test for proportions. LVH wascategorized as present or absent. For subjects who experi-enced multiple events, analysis was restricted to the first
Data are reported as mean � SD or n (%).ACE ¼ angiotensin-converting enzyme; ARBs ¼ angiotensin receptor
blockers; BP ¼ blood pressure; CNS ¼ central nervous system; VKA ¼vitamin K antagonist.
Arrhythmias and Conduction Disturbances/Cardiac Hypertrophy in Atrial Fibrillation 671
event. Survival curves were estimated by the Kaplan-Meierproduct-limit method and compared by the Mantel (log-rank) test. We report the risk of events as percentage peryear, estimated by dividing the total number of patients with
events by the total number of patient-years of follow-up.The effect of prognostic factors on survival was evaluatedby using the Cox model.14 In multivariate analysis, weadjusted for the CHA2DS2-VASc score15 (score 0 to 2 asreference) and other covariates not included in the score.These were current smoking at entry (0 ¼ no; 1 ¼ yes), useof digoxin at entry (0 ¼ no; 1 ¼ yes), permanent AF at entry(0 ¼ no; 1 ¼ yes), and randomized treatment (dabigatran110 mg twice daily ¼ 0; warfarin ¼ 1; dabigatran 150 mgtwice daily ¼ 0; warfarin ¼ 1). We used the CHA2DS2-VASc score, in place of its 7 components taken separately,because of the growing use of this score in the clinicalpractice7 and also to preserve model parsimony and preventoverfitting. We also tested the interaction of LVH statuswith CHA2DS2-VASc score to test the hypothesis that LVHmay be differently predictive at different levels of the pre-viously established risk score.
The incremental predictive value of LVH was estimatedby computing the difference in c-statistics between models.Confidence intervals (CIs) for the difference in c-statisticsbetween models were calculated by bootstrap resampling.16
To quantify the clinical impact of adding LVH to the modelspredicting outcome events, we calculated the net reclassifi-cation improvement (NRI) and the integrated discriminationimprovement (IDI). In our context, clinically relevant cut-offvalues for the risk of outcome events are absent. Thus, weused category-free reclassification measures (category-freeNRI), which are independent of arbitrarily defined riskthresholds. The category-free NRI is computed as the netproportion of events in which the new model correctly pre-dicts a higher probability plus the net proportion of non-events in which the new model correctly predicts a lowerprobability.17,18 The IDI expresses the improvement in theprobability of discrimination between models, and wascalculated by the formula (PY1 � Px1) � (PY0 � Px0), wherePY1 and PY0 are the mean estimated probabilities of eventsand nonevents with the model that includes LVH, and Px1and Px0 are the mean estimated probabilities of events andnonevents with the model without LVH. Because IDI, as anabsolute difference in probabilities, is highly sensitive toevent rates, its absolute magnitude may be difficult tointerpret quantitatively. Therefore, to facilitate interpreta-tion, we calculated the relative IDI as [(PY1 � PY0)/(Px1 �Px0)] � 1. For instance, if the addition of LVH to a basemodel results in a relative IDI of 0.30, this indicates that theaddition of LVH results in a 30% relative improvement inseparation of events and nonevents. In these analyses, theoccurrence or nonoccurrence of outcome events during thewhole follow-up period (median duration: 2.0 years10) wasused as a binary response. A probability value of<0.05 from2-sided tests was considered to be statistically significant.
Results
The patient flow diagram is shown in Figure 1. Patientcharacteristics at baseline are summarized in Table 1.Prevalence of LVH was 22.7%. When compared with pa-tients without LVH, those with LVH were more frequentlywomen and of Asian ethnicity. Permanent AF was morefrequent in the subgroup with LVH. Other relevant featuresof the population including systolic blood pressure at
Unadjusted HR = 1.51 (95% CI: 1.18-1.93) p < 0.001
Figure 2. Cumulative hazard for stroke, CV mortality, all-cause mortality, and MI in patients with AF with and without LVH.
672 The American Journal of Cardiology (www.ajconline.org)
baseline and history of diabetes, coronary artery disease, MI,and heart failure were more frequent in the subgroup withLVH (all p <0.001). Prevalence of hypertension wasslightly, but not significantly, greater in the subset with thanwithout LVH. The CHA2DS2-VASc score was higher in thesubset with than without LVH (p <0.001).
During a median follow-up period of 2.0 years, 303 pa-tients (2.9%) developed a stroke, 497 (4.8%) died from CVcauses, 778 (7.5%) died from any cause, and 140 (1.3%)developed MI. LVH at baseline was associated with anincreased risk of stroke (1.99% vs 1.32% per year, unad-justed hazard ratio [HR] 1.51, 95% CI 1.18 to 1.93, p<0.001), CV death (4.52% vs 1.80% per year, unadjustedHR 2.56, 95% CI 2.14 to 3.06, p <0.0001), all-cause death(6.03% vs 3.11% per year, unadjusted HR 1.95, 95% CI1.68 to 2.26, p <0.0001), and MI (1.11% vs 0.55% per year,unadjusted HR 2.07, 95% CI 1.47 to 2.92, p <0.0001). TheKaplan-Meier curves for stroke, CV death, all-cause death,and MI in the groups with and without LVH are reported inFigure 2 (all p <0.001).
As shown in Figure 3, the excess risk of stroke (p ¼ 0.29)and MI (p ¼ 0.82) associated with LVH did not show anysignificant interaction with the CHA2DS2-VASc score, thus,
suggesting that the predictive impact of LVH is comparableacross levels of CHA2DS2-VASc score. In contrast, theexcess risk of CV death (p ¼ 0.0042) and all-cause death(p ¼ 0.0188) associated with LVH was greater in patientswith CHA2DS2-VASc score 0 to 2 than in those with ascore �3.
The independent predictive value of LVH beyond theCHA2DS2-VASc score on stroke, CV death, all-cause death,and MI is listed in Table 2. After adjustment for theCHA2DS2-VASc score, and other variables not included inthe score, LVH was associated with an increased risk forstroke (HR 1.34, 95% CI 1.04 to 1.73), CV death (HR 2.22,95% CI 1.85 to 2.67), all-cause death (HR 1.75, 95%CI 1.51 to 2.04), and MI (HR 1.92, 95% CI 1.35 to 2.73).As listed in Table 3, adding LVH to the multivariatemodel (model B) yielded generally modest increases in thec-statistics that were significant for CV death and all-causedeath, but formally not for MI and stroke. Conversely, asshown in Figure 4, both the category-free NRI (stroke, p ¼0.0094; CV death, p <0.0001; all-cause death, p <0.0001;and MI, p ¼ 0.0011) and the IDI (stroke, p ¼ 0.0202;CV death, p <0.0001; all-cause death, p <0.0001; and MI,p ¼ 0.0010) were significant for all outcomes.
Figure 3. Prognostic impact of LVH in patients with low and high CHA2DS2-VASc score. N denotes the total number of patients and n the number of patientswith case-specific event. Squares denote the point estimate (� 95% confidence interval).
Table 2Risk of stroke, cardiovascular death, all-cause death, and myocardialinfarction in association with electrocardiographic left ventricularhypertrophy
* Adjusted for CHA2DS2-VASc score (0 to 2 as reference), randomizedtreatment in the RE-LY study (dabigatran 110 mg twice daily vs warfarin;dabigatran 150 mg twice daily vs warfarin), cigarette smoking (no vs yes),digoxin use (no vs yes), and permanent AF (no vs yes).
Arrhythmias and Conduction Disturbances/Cardiac Hypertrophy in Atrial Fibrillation 673
Discussion
This study provides the first evidence of the prognosticusefulness of ECG-LVH in anticoagulated patients withnonvalvular AF. The excess risk of stroke, CV death, all-cause death, and MI associated with LVH was significantand additive to the CHA2DS2-VASc score and other prog-nostic factors not included in the score. LVH improved riskdiscrimination between patients with AF with and withoutfuture events. The excess risk of CV death and all-cause
death associated with LVH was greater in the patients withlower (from 0 to 2) than higher (�3) CHA2DS2-VAScscore, whereas the risk of stroke and MI did not show anyinteraction with the CHA2DS2-VASc score. This suggeststhat ECG-LVH may be more suitable as a prognostic markerfor hard events in patients with lower CHA2DS2-VAScscore.
Patients with AF have been pooled with patients affectedby other atrial arrhythmias,2,19 or excluded,20,21 in studiesthat tested the prognostic impact of LVH. In the CoronaryDrug Project, conducted in MI survivors, ECG-LVH was asignificant predictor of 3-year death even after adjustmentfor AF at baseline, but no specific analysis was conducted inthe subset with AF.22 In the Losartan Intervention for EndPoint reduction study, new-onset AF was less frequent withlosartan than with atenolol and was associated with anexcess risk of stroke and hospitalization for heart failure.23
Because of the growing burden of AF in the communityand the limited accuracy of the equations used to assess therisk of stroke and other major events in patients with AF,7
validation of ECG-LVH as a simple and easily accessibleprognostic indicator in these patients would be valuable.However, extrapolation to patients with AF of the prognosticvalue of ECG-LVH on the basis of studies conducted in thegeneral population,1 or hypertensive cohorts in sinus rhythm,5
may not be justified because of the peculiarity of the riskprofile of patients with AF treated with anticoagulants.24e26
We used a composite score for diagnosis of ECG-LVH.Because of the limited sensitivity of electrocardiography,27
inclusion of a few specific ECG criteria in a score mayimprove sensitivity without seriously affecting specificity.6
The score used in the present study (“Perugia score”) refinedrisk stratification in large cohorts of hypertensive patients13
Table 3Results of the c-statistics
OutcomeEvent
c-Statistics Differences
Model 1 Model 2 Model 3 Model 1 vs 2 Model 2 vs 3 Model 1 vs 3
Stroke 0.601 (0.573e0.638) 0.632 (0.599e0.665) 0.636 (0.603e0.669) 0.032 (0.010e0.053) 0.005 (�0.003 to 0.012) 0.036 (0.015e0.058)p Value 0.00381 0.23252 0.00073
CV death 0.584 (0.559e0.609) 0.617 (0.591e0.643) 0.662 (0.637e0.686) 0.036 (0.018e0.054) 0.045 (0.024e0.066) 0.081 (0.054e0.108)p Value 0.00011 0.00003 <0.00001
All-cause death 0.573 (0.556e0.596) 0.598 (0.577e0.618) 0.624 (0.604e0.644) 0.025 (0.011e0.040) 0.026 (0.011e0.041) 0.051 (0.032e0.071)p Value 0.00039 0.00094 <0.0001
MI 0.570 (0.522e0.619) 0.593 (0.544e0.641) 0.620 (0.571e0.669) 0.033 (�0.003 to 0.069) 0.025 (�0.007 to 0.056) 0.058 (0.014e0.102)p Value 0.06920 0.12470 0.01012
0.398
0.277 0.279
0.151
0.953
0.655
0.947
0.672
2.22
1.75
1.92
1.34
1.85
1.51
1.35
1.04
2.67
2.04
2.73
1.73
0.00
0.20
0.40
0.60
0.80
1.00
1.20
1.40
1.60
1.80
2.00
0.2
0.6
1.0
1.4
1.8
2.2
2.6
CV Death All-Cause
Death
MI Stroke
Ad
ju
ste
d H
R fo
r p
atie
nts
w
ith
L
VH
Category-free NRI Relative IDI
Ca
te
go
ry
-fre
e N
RI - R
ela
tiv
e ID
I
Figure 4. Category-free NRI, relative IDI, and adjusted HRs after addingLVH to the multivariate Cox model. Circles denote the point estimate(� 95% confidence interval).
674 The American Journal of Cardiology (www.ajconline.org)
and high risk vascular patients.12 Other approaches havebeen tested to improve the performance of electrocardiog-raphy for diagnosis of LVH. For example, total 12-leadelectrocardiogram QRS voltage (from the peak of the Rwave to the nadir of either the Q or the S wave, whicheverwas deeper) was superior to traditional ECG criteria of LVHin a validation study with cardiac weight at necropsy or aftertransplantation as reference.28 In this study, LVH diagnosedby electrocardiography was associated with a significantlygreater risk of all-cause death, CV death, MI, and strokebefore and after adjustment for the CHA2DS2-VASc scoreand other predictors not included in the score. The c-statisticsincreased slightly, albeit significantly, after adding LVH tothe multivariate model composed by the CHA2DS2-VAScscore and other covariates, whereas the increase did not reachsignificance for MI and stroke. However, reliance on the gainin the c-statistics after adding a novel risk marker may beinadequate to assess its overall predictive ability.29 Thereceiver operating characteristic curve and c-statistic may beinsensitive in assessing the impact of adding new predictors toa score or predictivemodel when other strong predictors are inthemodel.29 In the present study, LVHsignificantly improved
2 metrics used to quantify risk prediction improvementoffered by new markers: the IDI and NRI.17 LVH improveddiscrimination between patients with AF with and withoutfuture events. In other terms, the difference in the probabilityof stroke, CV death, all-cause death, andMI between patientswith AF who later experienced an event and patients who didnot, increased significantly after adding LVH to the predictingmodel. Furthermore, based on results of the category-freeNRI, addition of LVH to multivariate models yielded an up-ward versus downward movement in model-based predictedprobabilities of CV death, all-cause death, MI, and stroke thatbetter reflected their subsequent actual occurrence.
Overall, these findings indicate that LVH improves thepredictive ability of a model composed by the CHA2DS2-VASc score, and other variables not included in the score,over all possible cut points of risk in patientswithAF.Becausedigoxin may cause abnormalities of repolarization that maymimic LVH diagnosed by the Perugia score,13 the prognosticvalue of ECG-LVH has also been adjusted for digoxin use.
A RE-LY substudy suggested that some biomarkers,including troponin I and N-terminal proeB-type natriureticpeptide, contribute to stratify the risk of stroke and death.30
Future analyses should test ECG-LVH and cardiac bio-markers together in an attempt to improve risk stratificationin patients with AF.
This study has a number of limitations. Because all patientsreceived, by protocol, oral anticoagulants, our results do notclarify whether ECG-LVH contributes to select patients fororal anticoagulation. Furthermore, the present study lacksimaging studies of left ventricular structure and function thatmight have refined prognostic stratification. A strength of thisstudy was that all ECG tracings were examined by a singlereader in blind conditions with regard to clinical features andrandomized treatment. Overall, the present study supports theuse of electrocardiography to detect LVH in anticoagulatedpatients with AF and encourages future studies aimed to testthe hypothesis that ECG-LVH may contribute to select pa-tients with AF who are suitable for anticoagulation therapy.
Disclosures
Drs. Verdecchia, Di Pasquale, Ambrosio, Ezekowitz,Wallentin, Connolly, and Yusuf received research grants,consulting fees, and lecture fees from Boehringer-IngelheimPharmaceuticals Inc., Ingelheim am Rein, Germany. Theother authors report no conflicts.
Arrhythmias and Conduction Disturbances/Cardiac Hypertrophy in Atrial Fibrillation 675
1. Kannel WB, Gordon T, Offutt D. Left ventricular hypertrophy byelectrocardiogram. Prevalence, incidence, and mortality in the Fra-mingham Study. Ann Intern Med 1969;71:89e105.
2. De Bacquer D, De Backer G, Kornitzer M, Blackburn H. Prognosticvalue of ECG findings for total, cardiovascular disease, and coronaryheart disease death in men and women. Heart 1998;80:570e577.
3. Kreger BE, Cupples LA, Kannel WB. The electrocardiogram in pre-diction of sudden death: Framingham Study experience. Am Heart J1987;113:377e382.
4. Kannel WB, Gordon T, Castelli WP, Margolis JR. Electrocardiographicleft ventricular hypertrophy and risk of coronary heart disease. TheFramingham Study. Ann Intern Med 1970;72:813e822.
5. Verdecchia P, Porcellati C, Reboldi G, Gattobigio R, Borgioni C,Pearson TA, Ambrosio G. Left ventricular hypertrophy as an inde-pendent predictor of acute cerebrovascular events in essential hyper-tension. Circulation 2001;104:2039e2044.
6. Romhilt DW, Estes EH Jr. A point-score system for the ECG diagnosisof left ventricular hypertrophy. Am Heart J 1968;75:752e758.
7. Camm AJ, Kirchhof P, Lip GY, Schotten U, Savelieva I, Ernst S, VanGelder IC,Al-AttarN,HindricksG, Prendergast B,HeidbuchelH,AlfieriO, Angelini A, Atar D, Colonna P, De Caterina R, De Sutter J, Goette A,Gorenek B, Heldal M, Hohloser SH, Kolh P, Le Heuzey JY, PonikowskiP, Rutten FH. Guidelines for the management of atrial fibrillation: theTask Force for the Management of Atrial Fibrillation of the EuropeanSociety of Cardiology (ESC). Eur Heart J 2010;31:2369e2429.
8. Kannel WB, Wolf PA, Benjamin EJ, Levy D. Prevalence, incidence,prognosis, and predisposing conditions for atrial fibrillation: population-based estimates. Am J Cardiol 1998;82:2Ne9N.
9. Verdecchia P, Reboldi G, Gattobigio R, Bentivoglio M, Borgioni C,Angeli F, Carluccio E, Sardone MG, Porcellati C. Atrial fibrillation inhypertension: predictors and outcome.Hypertension 2003;41:218e223.
10. Connolly SJ, Ezekowitz MD, Yusuf S, Eikelboom J, Oldgren J, ParekhA, Pogue J, Reilly PA, Themeles E, Varrone J, Wang S, Alings M,Xavier D, Zhu J, Diaz R, Lewis BS, Darius H, Diener HC, Joyner CD,Wallentin L. Dabigatran versus warfarin in patients with atrial fibril-lation. N Engl J Med 2009;361:1139e1151.
11. Ezekowitz MD, Connolly S, Parekh A, Reilly PA, Varrone J, Wang S,Oldgren J, Themeles E, Wallentin L, Yusuf S. Rationale and design ofRE-LY: Randomized evaluation of long-term anticoagulant therapy,warfarin, compared with dabigatran. Am Heart J 2009;157:805e810;810.e1-e2.
12. Verdecchia P, Sleight P, Mancia G, Fagard R, Trimarco B, SchmiederRE, Kim JH, Jennings G, Jansky P, Chen JH, Liu L, Gao P, ProbstfieldJ, Teo K, Yusuf S. Effects of telmisartan, ramipril, and their combi-nation on left ventricular hypertrophy in individuals at high vascularrisk in the Ongoing Telmisartan Alone and in Combination WithRamipril Global End Point Trial and the Telmisartan RandomizedAssessment Study in ACE Intolerant Subjects With CardiovascularDisease. Circulation 2009;120:1380e1389.
13. Verdecchia P, Angeli F, Reboldi G, Carluccio E, Benemio G, Gatto-bigio R, Borgioni C, Bentivoglio M, Porcellati C, Ambrosio G.Improved cardiovascular risk stratification by a simple ECG index inhypertension. Am J Hypertens 2003;16:646e652.
14. Cox DR. Regression models and life-tables. J R Stat Soc [B] 1972;34:187e220.
15. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinicalrisk stratification for predicting stroke and thromboembolism in atrial
fibrillation using a novel risk factor-based approach: the Euro HeartSurvey on Atrial Fibrillation. Chest 2010;137:263e272.
16. Efron B. Better bootstrap confidence intervals. J Am Stat Assoc1987;82:171e185.
17. Pencina MJ, D’Agostino RB Sr, D’Agostino RB Jr, Vasan RS. Eval-uating the added predictive ability of a new marker: from area under theROC curve to reclassification and beyond. Stat Med 2008;27:157e172;discussion 207e112.
18. Pencina MJ, D’Agostino RB Sr, Steyerberg EW. Extensions of netreclassification improvement calculations to measure usefulness of newbiomarkers. Stat Med 2011;30:11e21.
19. MRFIT-Investigators. Baseline rest electrocardiographic abnormalities,antihypertensive treatment, and mortality in the Multiple Risk FactorIntervention Trial. Multiple Risk Factor Intervention Trial ResearchGroup. Am J Cardiol 1985;55:1e15.
20. Larsen CT, Dahlin J, Blackburn H, Scharling H, AppleyardM, Sigurd B,Schnohr P. Prevalence and prognosis of electrocardiographic left ven-tricular hypertrophy, ST segment depression and negative T-wave; theCopenhagen City Heart Study. Eur Heart J 2002;23:315e324.
21. Hsieh BP, Pham MX, Froelicher VF. Prognostic value of electrocar-diographic criteria for left ventricular hypertrophy. Am Heart J2005;150:161e167.
22. Coronary-Drug-Project-Research-Group-Investigators. Left ventricularhypertrophy patterns and prognosis. Experience postinfarction in theCoronary Drug Project. Circulation 1974;49:862e869.
23. Wachtell K, Lehto M, Gerdts E, Olsen MH, Hornestam B, Dahlof B,Ibsen H, Julius S, Kjeldsen SE, Lindholm LH, Nieminen MS, Dever-eux RB. Angiotensin II receptor blockade reduces new-onset atrialfibrillation and subsequent stroke compared to atenolol: the LosartanIntervention For End Point Reduction in Hypertension (LIFE) study.J Am Coll Cardiol 2005;45:712e719.
24. LipGY.Thebalance between stroke prevention andbleeding risk in atrialfibrillation: a delicate balance revisited. Stroke 2008;39:1406e1408.
25. Lip GY, Lane DA. Does warfarin for stroke thromboprophylaxisprotect against MI in atrial fibrillation patients? Am J Med 2010;123:785e789.
26. Raunso J, Selmer C, Olesen JB, Charlot MG, Olsen AM, Bretler DM,Nielsen JD, Dominguez H, Gadsboll N, Kober L, Gislason GH, Torp-Pedersen C, Hansen ML. Increased short-term risk of thrombo-embo-lism or death after interruption of warfarin treatment in patients withatrial fibrillation. Eur Heart J 2012;33:1886e1892.
27. Casale PN, Devereux RB, Alonso DR, Campo E, Kligfield P. Improvedsex-specific criteria of left ventricular hypertrophy for clinical andcomputer interpretation of electrocardiograms: validation with autopsyfindings. Circulation 1987;75:565e572.
28. Roberts WC, Filardo G, Ko JM, Siegel RJ, Dollar AL, Ross EM,Shirani J. Comparison of total 12-lead QRS voltage in a variety ofcardiac conditions and its usefulness in predicting increased cardiacmass. Am J Cardiol 2013;112:904e909.
29. Cook NR. Use and misuse of the receiver operating characteristic curvein risk prediction. Circulation 2007;115:928e935.
30. Hijazi Z, Oldgren J, Andersson U, Connolly SJ, Ezekowitz MD,Hohnloser SH, Reilly PA, Vinereanu D, Siegbahn A, Yusuf S, Wal-lentin L. Cardiac biomarkers are associated with an increased risk ofstroke and death in patients with atrial fibrillation: a RandomizedEvaluation of Long-term Anticoagulation Therapy (RE-LY) substudy.Circulation 2012;125:1605e1616.
