- Email: [email protected]
QT intervals and QT dispersion as measures of left ventricular hypertrophy in an unselected hypertensive population
AJH
2001; 14:455– 462
QT Intervals and QT Dispersion as Measures of Left Ventricular Hypertrophy in an Unselected Hypertensive Population Neil Chapman, Jamil Mayet, Muhidin Ozkor, Fiona C. Lampe, Simon A.McG. Thom, and Neil R. Poulter Electrocardiographic (ECG) QT intervals and dispersion correlate with echocardiographic left ventricular mass index (LVMI) in groups of selected essential hypertensives. We tested the strength of this relationship in a large group of unselected hypertensives to assess whether QT measurements may be a simple screening test for LVH in clinical practice. In a cross-sectional study of 386 unselected hypertensive subjects, maximal QT intervals (QTmax), QT dispersion (QTdisp), and ECG voltages (Sokolow-Lyon and Cornell sex-specific voltages) were measured from 12-lead ECG. The LVMI correlated most strongly with Cornell voltage (linear regression r ⫽ 0.44, P ⬍ .001). The strongest relationship between LVMI and QT parameters was with QTmax, (r ⫽ 0.25, P ⬍ .001). This relationship weakened using heart rate– corrected QTmax. Correlations between LVMI and QTdisp
E
were weak, whether or not they were corrected for heart rate. Sokolow-Lyon voltages, Cornell voltage and QTmax, but not QTdisp, were independently predictive of LVMI after adjustment for age, sex, race, and the other ECG parameters. Receiver operating characteristic (ROC) curve analyses demonstrated that no QT parameter performed better than simple ECG voltage criteria in the detection of LVH. In conclusion, QTmax, the QT parameter most strongly associated with LVMI, was independently associated with LVMI after adjustment for standard ECG voltage criteria. However, as an isolated measure it was no better than simple ECG voltage criteria as a screening test for LVH in clinical practice. Am J Hypertens 2001;14: 455– 462 © 2001 American Journal of Hypertension, Ltd. Key Words: QT intervals, QT dispersion, ECG voltage criteria, left ventricular hypertrophy, hypertension.
lectrocardiographic QT dispersion, a measure of heterogeneity of QT intervals across the 12 ECG leads, is thought to represent regional differences in myocardial repolarization,1 and it has been proposed that increased variability in myocardial repolarization times may provide the substrate for re-entry ventricular arrhythmias. Indeed, many published studies have shown that prolonged QT intervals or increased QT dispersion (QTdisp) are associated with an adverse prognosis in a variety of pathological conditions, including myocardial ischemia,2,3 heart failure,4 hypertrophic cardiomyopathy,5 and long QT syndromes.6 However, the literature is inconsistent and conflicting, particularly with respect to myocardial infarction.7–9 In a Dutch population, prolonged QT intervals predicted death in apparently healthy individuals.10 However, these data conflict with those from the Framingham cohort in which QT intervals failed to predict overall or cardiac mortality.11
Left ventricular hypertrophy (LVH) in hypertensives predicts cardiovascular mortality.12,13 Left ventricular hypertrophy is associated with excess cardiac arrhythmias,14 and this has been proposed as a mechanism linking LVH with sudden death. Unfortunately, standard ECG criteria for the detection of LVH, although highly specific, are insensitive.15 Several studies have looked at the association between QT parameters and left ventricular hypertrophy. In a small study, QT dispersion was increased in subjects with, compared to those without, LVH.16 In another study, the degree of LVH in patients with hypertrophic obstructive cardiomyopathy correlated with QT intervals but not QT dispersion.17 More recently, several studies have looked specifically at the relationship between QT parameters and left ventricular mass in small groups of selected hypertensive subjects. Two studies found that echocardiographic left ventricular mass index (LVMI) correlated significantly
Received May 31, 2000. Accepted September 12, 2000. From The Peart-Rose Clinic and Department of Clinical Pharmacology (NC, JM, MO, SAMcGT, NRP), Imperial College School of Medicine at St Mary’s, London, United Kingdom; Department of Cardiology (JM), Imperial College School of Medicine at St Mary’s, London, United Kingdom; and Department of Primary Care and Population Sciences
(FCL), Royal Free and UCL Medical School, London, United Kingdom.
© 2001 by the American Journal of Hypertension, Ltd. Published by Elsevier Science Inc.
Address correspondence and reprint requests to Dr N. Chapman, Department of Clinical Pharmacology, Imperial College School of Medicine at St Mary’s, South Wharf Road, London W2 1NY, UK; e-mail: [email protected] 0895-7061/01/$20.00 PII S0895-7061(00)01292-9
456
QT AND LEFT VENTRICULAR HYPERTROPHY
with both maximum QT intervals and QT dispersion.18,19 Another found a very close correlation between QT dispersion and left ventricular mass (r ⫽ 0.74, P ⬍ .001). 20 Based on these data, it has been suggested18 that stratification of hypertensive patients by QT dispersion might allow the echocardiographic identification of a subgroup of patients with LVH. In view of these results in small groups of selected hypertensives, we investigated the relationship between QT parameters, left ventricular mass, and LVH in an unselected hypertensive population. If there was evidence of a strong relationship, then a measure of QT interval or dispersion might provide a suitable alternative ECG method for the detection of LVH in clinical practice, thereby possibly obviating the need for echocardiography in a proportion of patients.
Methods Subjects This was a cross-sectional study of patients attending a hypertension and cardiovascular disease prevention clinic. A total of 386 consecutive patients (189 male and 197 female) referred for assessment of hypertension, with acceptable M-mode echocardiographic images and no bundle branch block on the echocardiogram, were included. None of the subjects had been included in a previous study of QT intervals and hypertension conducted at this institution.19 Most had essential hypertension. Others had borderline, white-coat, or secondary hypertension. The latter groups of subjects were included to assess the generalizability of using QT parameters as a predictor of LVH in unselected hypertensive subjects. Echocardiography Echocardiography was performed using a phased array sector scanner (General Electric Pass II, 3.5-MHz transducer) according to a standard protocol. M-mode images were obtained from the short-axis parasternal view at the level of the tips of the mitral valve leaflets and were printed onto light-sensitive paper for subsequent analysis. Care was taken to ensure that a precise “on-axis” view was obtained. Measurements of septal wall thickness (SWT), posterior wall thickness (PWT), and left ventricular internal dimension (LVID) were made at the peak of the R wave according to the Penn convention.21 Left ventricular mass index (LVMI) (in g/m2) was calculated as: LVMI ⫽ 共1.04关共SWT ⫹ PWT ⫹ LVID兲 3 ⫺ 共LVID兲 3 兴 ⫺ 14兲/Bod y Surface Area Cut-off values for the presence of LVH were taken as ⱖ110 g/m2 for women and ⱖ134 g/m2 for men. Blood Pressure Measurement Blood pressure was measured in the right arm using a suitably sized cuff and an automated recorder (Sentron,
AJH–May 2001–VOL. 14, NO. 5, PART 1
C.R. Bard, Haverhill, MA). A single measurement was taken after approximately 10 min supine rest, immediately after echocardiography. Electrocardiography Routine clinical echocardiography was used (Marquette 12SL, Marquette Electronics, Milwaukee, WI). Recordings were performed at 25 mm/sec with standard lead positions and nonsimultaneous lead recordings. ECG Voltage Criteria Two criteria for the ECG detection of LVH were used, and a single blinded observer made all measurements. According to the Sokolow-Lyon criteria,22 LVH is present if SV1 ⫹ RV5 ⱖ 3.5 mV. According to the Cornell sexspecific criteria (designated here as simply “Cornell”),23 LVH is present if SV3 ⫹ RaVL ⱖ 2.0 mV in women or 2.8 mV in men. Voltages in individual leads were calculated as the mean of three complexes whenever possible. QT Analysis QT analysis was performed by two trained observers who were blind to all clinical details, using a digitizing pad (NewSketch 1212, Genius, KYE Systems UK, Wallington, UK) linked to a PC (Viglen, London, UK). The QT intervals were measured on all possible leads. Where possible, the mean of three complexes was taken as the QT interval for each lead. The QT intervals were measured from the onset of the QRS complex to the end of the T wave, defined as the return to the T-P baseline. Where U waves were present, the QT interval was measured to the nadir of the curve between T and U waves. Leads in which the end of the T wave was unclear were excluded. The mean (⫾ SD) number of evaluable leads was 11 (⫾ 1) leads. The QT intervals were also corrected (QTC) for the known effect of heart rate using Bazett’s formula (QTC ⫽ QT/(RR)1/2).24 The QT dispersion (QTdisp) was calculated as the difference between the maximum QT interval (QTmax) and minimum QT interval in each subject. Heart rate– corrected QT dispersion (QTCdisp) was calculated as the difference between the maximum QTC interval (QTCmax) and minimum QTC interval in each subject. Coefficients of variation for intra- and interobserver variability for QTmax were 1.6% and 4.5%, and for QTdisp were 12% and 21%, respectively. Statistics Continuous data are summarized as mean (⫾ SD) and compared between two groups using unpaired t tests. Categoric data were compared between groups using the 2 test. The Pearson correlation coefficient was used to assess the associations of a range of variables with LVMI. Multiple linear regression was used to assess the independent contribution of clinical and ECG variables to the prediction of LVMI. Results are presented as regression coefficients with 95% confidence intervals, and as partial correlation coefficients. The
AJH–May 2001–VOL. 14, NO. 5, PART 1
Table 1.
QT AND LEFT VENTRICULAR HYPERTROPHY
Subject characteristics
Characteristic n Male/female (%) White/Afro-Caribbean/Other race (%) Treated/untreated (%) Age (years) Systolic BP (mm Hg) Diastolic BP (mm Hg) Heart rate (beats/min) Inter ventricular septum (mm) Left ventricular internal diameter (mm) Posterior wall thickness (mm) Left ventricular mass index (g/m2) SV1 ⫹ RV5 (mV) SV3 ⫹ RaVL (mV) QTmax (msec) QTCmax (msec) QTdisp (msec) QTCdisp (msec)
% or Mean (SD) 386 49/51 65/27/8 49/51 48 (14) 160 (22) 94 (12) 73 (13) 10.4 (1.9) 47.5 (5.5) 10.3 (1.7) 114 2.77 1.67 432 472 68 78
(34) (0.97) (0.72) (47) (47) (22) (25)
SV1 ⫹ RV5, Sokolow-Lyon voltage; SV3 ⫹ RaVL, Cornell sex-specific voltage. Data shown as percentage of subjects or mean (SD).
overall performance of ECG voltages and QT parameters in detecting LVH was assessed using ROC curve analyses. The sensitivities of tests in detecting LVH at matched specificities were compared performed using the McNemar paired 2 test. All analyses were performed using the SPSS statistics package (SPSS Software, Chicago, IL).
Results Subject Characteristics Table 1 shows the characteristics of the 386 subjects in the study population. Of the 191 subjects taking antihypertensive medication, 105 were on monotherapy (11 on diuretics, 12 -blockers, 48 calcium channel blockers, 26 angiotensin converting enzyme inhibitors, and eight ␣-blockers) and 86 were on combinations of two or more drugs. Relationship Between ECG Variables and LVMI Correlation coefficients (r) between LVMI and other measured parameters are shown in Table 2 and scatter plots in Fig. 1. Among the ECG variables tested, LVMI correlated most strongly with the Cornell voltage (linear regression r ⫽ 0.44, P ⬍ .001). Of the QT parameters measured, LVMI correlated most strongly with QTmax (r ⫽ 0.25, P ⬍ .001). However, this correlation was weakened after correction of maximal QT intervals for heart rate (QTCmax; r ⫽ 0.11, P ⫽ .037). Associations of BP, heart rate, ECG voltages, and uncorrected QT measures with
457
Table 2. Unadjusted and adjusted correlation coefficients (r) of LVMI with other variables Partial Unadjusted Correlation* r
P
R
P
Systolic BP (mm Hg) 0.26 ⬍.001 0.19 ⬍.001 Diastolic BP (mm Hg) 0.21 ⬍.001 0.16 .002 Heart rate (beats/ min) ⫺0.17 .001 ⫺0.12 .016 SV1 ⫹ RV5 (mV) 0.24 ⬍.001 0.20 ⬍.001 SV3 ⫹ RaVL (mV) 0.44 ⬍.001 0.35 ⬍.001 QTmax (msec) 0.25 ⬍.001 0.22 ⬍.001 QTCmax (msec) 0.11 .037 0.12 .025 QTdisp (msec) 0.15 .004 0.13 .013 QTCdisp (msec) 0.06 .21 0.08 .130 Pearsons correlation coefficients of hemodynamic and ECG variables with LVMI. * Adjusted for age, sex, and race.
LVMI were slightly weakened by adjustment for age, sex, and race, although all remained significant (Table 2). A multiple regression model of factors potentially predictive of LVMI is shown in Table 3. Increasing age, male gender, SBP, heart rate, ECG voltages, and QTmax were all independently associated with increased LVMI, whereas heart rate was inversely associated with LVMI. However, the strength of the association between QTmax and LVMI was weaker than for either the Sokolow-Lyon or Cornell voltages. The QTdisp was not independently associated with LVMI. Overall, the factors included in the regression model explained 34% of the variation in LVMI. QT Intervals in Subjects With and Without LVH Characteristics of subjects with and without LVH are shown in Table 4. There was a significantly higher proportion of Afro-Caribbean subjects among those with LVH compared with those without. Those with LVH were older and had significantly higher BP and ECG voltages. Mean QTmax was significantly greater in those with LVH compared with those without LVH, but the standard deviation was large and there was a large degree of overlap between the two groups. Heart rate– corrected QTCmax did not differ significantly between those with and without LVH. The QTdisp was significantly greater in those with LVH compared with those without LVH. However, as with QTmax, the spread of values was wide, and significance was lost when heart rate– corrected QTCdisp was compared between the two groups. Receiver Operating Characteristic Curve Analyses of the Performance of ECG Variables in the Detection of LVH Using ROC curve analysis, the Cornell voltage was the parameter with the highest overall predictive capacity. Areas under the curve were as follows: Sokolow-Lyon
458
QT AND LEFT VENTRICULAR HYPERTROPHY
AJH–May 2001–VOL. 14, NO. 5, PART 1
FIG. 1. Scatter plots of echocardiographic (ECG) parameters (QT intervals, QT dispersion, and ECG voltages) plotted against left ventricular mass index (LVMI). ECG voltages (Sokolow-Lyon and Cornell sex-specific), QTmax, and QTdisp plotted against LVMI.
voltage 0.590, Cornell voltage (male and female combined) 0.664, QTmax 0.615, QTCmax 0.566, QTdisp 0.567, and QTCdisp 0.541. Subsequent comparisons of test perTable 3.
formance compared only the Cornell voltage and QTmax as the best-performing examples of voltage criteria and QT parameters, respectively. Sex-specific ROC curves
Multiple regression model of factors predictive of LVMI Variable
Age Gender (male) Ethnic origin (Afro-Caribbean) SBP (mm Hg) SV1 ⫹ RV5 (mV) SV3 ⫹ RaVL (mV) HR (beats/min) QTmax (msec) QTdisp (msec)
Regression Coefficient (B)
95% Confidence Interval
Partial Correlations
P Value
0.24 14.9 5.1 0.25 4.7 12.9 ⫺0.27 78 39
⫺0.01, 0.49 8.6, 21.2 ⫺1.5, 11.6 0.09, 0.40 1.4, 8.0 8.3, 17.5 ⫺0.54, 0.00 2, 155 ⫺106, 184
0.10 0.24 0.08 0.17 0.15 0.28 ⫺0.11 0.11 0.03
.062 ⬍.001 .13 .002 .005 ⬍.001 .051 .045 .60
LVMI ⫽ left ventricular mass index; SBP ⫽ systolic blood pressure; HR ⫽ heart rate. Multiple regression model of factors predicting LVMI (the dependent variable). All factors above are included in the model.
AJH–May 2001–VOL. 14, NO. 5, PART 1
Table 4.
QT AND LEFT VENTRICULAR HYPERTROPHY
459
Characteristics of subjects without versus with left ventricular hypertrophy (LVH)
Number of subjects LVMI (g/m2) Male/female (%) White/Afro-Caribbean/Other race (%) Treated/untreated (%) Age (years) Systolic BP (mm Hg) Diastolic BP (mm Hg) Heart rate (beats/min) SV1 ⫹ RV5 (mV) SV3 ⫹ RaVL (mV) QTmax (msec) QTCmax QTdisp (msec) QTCdisp
No LVH
LVH
P Value
244 95 (18) 52/48 68/23/9 47/53 46 (12) 156 (19) 93 (11) 74 (14) 2.65 (0.89) 1.53 (0.67) 426 (47) 469 (49) 66 (21) 77 (24)
142 147 (32) 44/56 59/36/5 55/45 51 (15) 167 (25) 96 (13) 71 (13) 2.99 (1.06) 1.92 (0.74) 442 (47) 476 (42) 71 (23) 81 (26)
.177 .037 .101 .001 ⬍.001 .009 .025 .001 ⬍.001 .001 .159 .026 .175
Difference Between % or Means (95% CI)
⫺7 ⫺10 8 5 11 3 ⫺3 0.34 0.40 16 7 5 4
(⫺18, 3) (⫺20, ⫺1) (⫺2, 19) (2, 8) (6, 15) (1, 6) (⫺6, 0) (0.14, 0.54) (0.26, 0.55) (6, 26) (⫺3, 17) (1, 10) (⫺1, 9)
LVMI ⫽ left ventricular mass index; BP ⫽ blood pressure; CI ⫽ confidence interval. Data shown as percentages, means (SD), and differences between percentages or means (95% confidence intervals) of the two groups.
showed that the performance of QTmax was no better than Cornell voltage in detecting LVH in either men or women (Fig. 2). Sensitivity and Positive Predictive Value of ECG Variables in the Detection of LVH The standard Cornell cut-off voltages (2.8 and 2.0 mV for men and women, respectively) give high specificity (96% and 84%, respectively) but low sensitivity (19.4% and 25.6%, respectively). To compare directly the sensitivity of Cornell voltages and QTmax in the detection of LVH, sex-specific partition points were adjusted for Cornell voltages and chosen for QTmax to match the specificity of each test at approximately 90%. As shown in Table 5, at this fixed specificity, the sensitivity of QTmax tended to be lower than that for Cornell voltages, although this was not statistically significant for either men or women (P ⫽ .13 and 0.21, respectively). Similarly, positive predictive values tended to be lower using QTmax than using Cornell voltages. QT Intervals in Subjects Taking Antihypertensive Medication Compared with untreated subjects, those subjects on antihypertensive medication were significantly older and had greater Cornell voltages. There were no significant differences in heart rate, BP, LVMI, Sokolow-Lyon voltage, or any QT parameter (data not shown). Compared with all other subjects, subjects on -blockers were significantly older and had slower heart rates (68 ⫾ 15 v 74 ⫾ 13 beats/min, P ⬍ .01). Maximum QT intervals were prolonged in subjects taking -blockers (450 ⫾ 39 v 429 ⫾ 48 msec, P ⬍ .005), but this difference was no longer apparent in heart rate– corrected QT intervals (QTCmax 473 ⫾ 42 v 472 ⫾ 48 msec, P ⫽
.96). There were no differences in QTdisp (68 ⫾ 23 v 68 ⫾ 22 msec, P ⫽ .98), BP, LVMI, ECG voltages, or QTCdisp between those patients on -blockers and all other subjects studied.
Discussion The primary aim of the present study was to assess the strength of the relationship between echocardiographic QT measurements and LVH. This is pertinent to recently published data in selected groups of hypertensives suggesting that QT intervals and dispersion correlate closely with echocardiographic LVMI.18 –20 If the relationship proved to be sufficiently robust, then some measure of QT interval or dispersion might provide a sensitive and specific ECG method of LVH detection superior to currently used ECG voltage criteria, thereby reducing numbers of patients requiring echocardiography. An important issue in the development of any screening technique is applicability in a real clinical environment. Our data from an unselected hospital-based hypertensive population show that correlations between LVMI and QT parameters are generally weaker than those between LVMI and ECG voltage criteria. Although maximal QT intervals and QTdisp were significantly greater in those with LVH compared with those without LVH, there was a large degree of overlap between groups. Moreover, the associations were weakened and the statistical significance of the differences was lost after correction for heart rate. Multiple linear regression and ROC curve analyses also suggest that although QTmax does provide some additional information beyond that given by standard ECG criteria, neither it nor any of the other QT parameters tested in isolation are any better than simple ECG voltages in screening for LVH. Previous studies18 –20 found much stronger relation-
460
QT AND LEFT VENTRICULAR HYPERTROPHY
AJH–May 2001–VOL. 14, NO. 5, PART 1
FIG. 2. Sex-specific receiver operating characteristic curves for the performance of a) Cornell voltages and b) QTmax in the prediction of left ventricular hypertrophy (LVH). Sex-specific sensitivities of Cornell voltage criteria and maximal QT intervals in the detection of LVH at varying specificities. QTmax, maximal QT interval.
ships between QT parameters and LVMI than were seen in the current study. In two studies of both uncorrected and rate-corrected QTdisp, correction for heart rate was associated with little worsening of the relationship with LVMI,18,19 and, unlike the current study, all found QTCdisp to be significantly correlated with LVMI. However, two of the previous studies used well characterized groups of untreated essential hypertensive subjects and
excluded those with intercurrent cardiac and noncardiac diseases. The third study did not specify inclusion or exclusion criteria beyond the fact that the patients had a diagnosis of confirmed systemic hypertension.20 In contrast, the current study was designed to include those with elevated clinic BP irrespective of existing drug treatment or other disease processes that frequently coexist in hypertensives. It has been documented that several forms of
AJH–May 2001–VOL. 14, NO. 5, PART 1
QT AND LEFT VENTRICULAR HYPERTROPHY
Table 5. Sensitivity and positive predictive (PPV) value of Cornell sex-specific voltage and QTmax at matched specificity in the detection of LVH
Cornell Specificity Sensitivity PPV QTmax Specificity Sensitivity PPV
Men
Women
91.2 32.3 64.5
91.5 21.3 61.5
90.5 19.4 50.5
90.8 11.5 45.0
Sensitivity and positive predictive values of Cornell voltage criteria and QTmax at matched specificities. Sex-specific partition points were adjusted for Cornell voltages (to 2.6 and 2.2 mV in men and women, respectively) and chosen for QTmax (461 msec in both men and women) to compare sensitivity and positive predictive values of each method at matched specificity.
cardiac disease affect QT parameters3,4 and that QT measurements are less reliable in abnormal ECG.25 The presence in this population of subjects with cardiac disease, and in others with repolarization changes associated with LVH, is likely to have weakened the relationship between QT parameters and LVH. Our population also included treated hypertensives and those with borderline or secondary hypertension, and is likely to have included some subjects with white-coat hypertension. The inclusion of these groups, although justified by the primary aim of the study, is also likely to have weakened the relationship between LVMI and QT parameters. Finally, in contrast to previous studies,26 we found ECG-determined heart rate to be weakly inversely correlated with LVMI. This unexpected finding may be due to skewing of the data by those with physiological LVH and bradycardia due to physical training, as well as individuals with white-coat hypertension and normal LVMI but a marked chronotropic response to stress. Any such effects would contribute to the attenuation of correlations seen between LVMI and QT parameters on correction for heart rate. In our population, no differences were seen in any QT parameter between subjects taking, versus those not taking, antihypertensive drugs. However, when compared with all other subjects, those taking -blockers had prolonged QTmax, but no difference was seen in QTdisp. However, the difference in QTmax may be explained by the relative bradycardia in subjects taking -blockers, as no difference was seen in the difference in QTCmax. Numerous papers have reported on the associations and prognostic implications of QT intervals and dispersion. However, results have been inconsistent,7–11 possibly reflecting variability in the methodology of QT interval measurement.27 Although interobserver variability in this study was comparable with that in previous reports,28 significant variability exists, particularly in measurements of dispersion. Furthermore, measurement variability is greater in abnormal, compared with normal, ECG.25 Standardization of measurement criteria and computerized analysis may improve
461
the repeatability and usefulness of the technique in the future, although studies have shown marked variation in measurements between different automated systems.29 Finally, the interpretation and significance of QT dispersion as a marker of ventricular repolarization has recently been called into question.30 In conclusion, unlike previous studies in small groups of selected hypertensives, correlations between QT parameters and echocardiographic LVMI were weak in an unselected hypertensive population, probably due to a number of confounding factors and variability in QT measurements. The overall performance of QT parameters in the detection of LVH was no better than simple ECG voltage techniques. Although QTmax (but not QTdisp) predicted LVMI independently of ECG voltage criteria, no QT measure in isolation is superior to ECG voltage criteria in the routine detection of LVH setting. Whether QT intervals or dispersion predict prognosis in hypertensives independently of LVH is a further issue that can only be properly addressed by a prospective study.
Acknowledgments We thank Melissa Poulter and Cecilia Tse for their invaluable assistance with ECG analysis.
References 1.
Higham PD, Campbell RWF: QT dispersion. Br Heart J 1994;71: 508 –510. 2. Taylor GJ, Crampton RS, Gibson RS, Stebbins PT, Waldman MTG, Beller GA: Prolonged QT interval at onset of acute myocardial infarction in predicting early phase ventricular tachycardia. Am Heart J 1981;102:16 –24. 3. Higham PD, Furniss SS, Campbell RWF: QT dispersion and components of the QT interval in ischaemia and infarction. Br Heart J 1995;73:32–36. 4. Barr CS, Naas A, Freeman M, Lang CC, Struthers AD: QT dispersion and sudden unexpected death in chronic heart failure. Lancet 1994;343:327–329. 5. Buja G, Miorelli M, Turrini P, Melacini P, Nava A: Comparison of QT dispersion in hypertrophic cardiomyopathy between patients with and without ventricular arrhythmias and sudden death. Am J Cardiol 1993;72:973–976. 6. Day CP, McComb JM, Campbell RWF: QT dispersion: an indication of arrhythmia risk in patients with long QT intervals. Br Heart J 1990;63:342–344. 7. Pohjola-Sintonen S, Siltanen P, Haapakoski J: Usefulness of QTc interval on the discharge electrocardiogram for predicting survival after acute myocardial infarction. Am J Cardiol 1986;57:1066 – 1068. 8. Glancy JM, Garratt CJ, Woods KL, de Bono DP: QT dispersion and mortality after myocardial infarction. Lancet 1995;345:945–948. 9. Zabel M, Klingenheben T, Franz MR, Hohnloser SH: Assessment of QT dispersion for prediction of mortality or arrhythmic events after myocardial infarction. Results of a prospective, long-term follow-up study. Circulation 1998;97:2543–2550. 10. Schouten EG, Dekker JM, Meppelink P, Kok FJ, Vandenbrouke JP, Pool J: QT interval prolongation predicts cardiovascular mortality in an apparently healthy population. Circulation 1991;84:1516 –1523.
462
QT AND LEFT VENTRICULAR HYPERTROPHY
11. Golberg RJ, Bengtson J, Chen ZY, Anderson KM, Locati E, Levy D: Duration of the QT interval and total and cardiovascular mortality in healthy persons (the Framingham Heart Study experience). Am J Cardiol 1991;67:55–58. 12. Kannel WB, Gordon T, Castelli WP, Margolis JR: Electrocardiographic left ventricular hypertrophy and risk of coronary heart disease: the Framingham Heart Study. Ann Intern Med 1970;72: 813– 822. 13. Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP: Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Study. N Engl J Med 1990;322: 1561–1566. 14. McLenachan JM, Henderson E, Morris KI, Dargie HJ: Ventricular arrhythmias in patients with hypertensive left ventricular hypertrophy. N Engl J Med 1987;317:787–792. 15. Chapman N, Mayet J, Sever P, Foale R, Thom S, Poulter N: Ethnic differences in the identification of left ventricular hypertrophy in the hypertensive patient. Am J Hypertens 1999;12:437– 442. 16. Davey PP, Bateman J, Mulligan IP, Forfar C, Barlow C, Hart G: QT interval dispersion in chronic heart failure and left ventricular hypertrophy: relation to autonomic nervous system and Holter tape abnormalities. Br Heart J 1994;71:268 –273. 17. Dritsas A, Sbarouni E, Gilligan D, Nihoyannopoulos P, Oakley CM: QT-interval abnormalities in hypertrophic cardiomyopathy. Clin Cardiol 1992;15:739 –742. 18. Clarkson PBM, Naas AAO, McMahon A, MacLeod C, Struthers AD, MacDonald TM: QT dispersion in essential hypertension. Q J Med 1995;88:327–332. 19. Mayet J, Shahi M, McGrath K, Poulter NR, Sever PS, Foale RA, Thom SAMcG: Left ventricular hypertrophy and QT dispersion in hypertension. Hypertension 1996;28:791–796. 20. Ichkhan K, Molnar J, Somberg J: Relation of left ventricular mass and QT dispersion in patients with systematic hypertension. Am J Cardiol 1997;79:8508 – 8511.
AJH–May 2001–VOL. 14, NO. 5, PART 1
21. Devereux RB, Reichek N: Echocardiographic determination of left ventricular mass in man. Anatomic validation of the method. Circulation 1977;55:613– 618. 22. Sokolow M, Lyon TP: The ventricular complex in left ventricular hypertrophy as obtained by unipolar precordial and limb leads. Am Heart J 1949;37:161–186. 23. Casale PN, Devereux RB, Alonso DR, Campo E, Kligfield P: Improved sex-specific criteria of left ventricular hypertrophy for clinical and computer interpretation of echocardiograms: validation with autopsy findings. Circulation 1987;75:565–572. 24. Ahnve S: Correction of the QT interval for heart rate. Review of different formulas and the use of Bazetts formula in myocardial infarction. Am Heart J 1985;109:568 –574. 25. McLaughlin NB, Campbell RWF, Murray A: Accuracy of four automatic QT measurement techniques in cardiac patients and healthy subjects. Heart 1996;76:422– 426. 26. Mayet J, Shahi M, Hughes AD, Stanton AV, Poulter NR, Sever PS, Foale RA, Thom SAMcG: Left ventricular structure and function in previously treated hypertensive patients: the importance of blood pressure, the nocturnal blood pressure dip and heart rate. J Cardiovasc Risk 1995;2:255–261. 27. Statters DJ, Malik M, Ward DE, Camm AJ: QT dispersion: problems of methodology and clinical significance. J Cardiovasc Electrophysiol 1994;5:672– 685. 28. Kautzner J, Gang Y, Camm J, Malik M: Short- and long-term reproducibility of QT, QTc and QT dispersion measured in healthy subjects. PACE 1994;17:928 –937. 29. McLaughlin NB, Campbell RWF, Murray A: Comparison of automatic QT measurement techniques in the normal 12 lead echocardiogram. Br Heart J 1995;74:84 – 89. 30. Coumel P, Maison-Blanche P, Badilini F: Dispersion of ventricular repolarization. Reality? Illusion? Significance? Circulation 1998;97: 2491–2493.
2001; 14:455– 462
QT Intervals and QT Dispersion as Measures of Left Ventricular Hypertrophy in an Unselected Hypertensive Population Neil Chapman, Jamil Mayet, Muhidin Ozkor, Fiona C. Lampe, Simon A.McG. Thom, and Neil R. Poulter Electrocardiographic (ECG) QT intervals and dispersion correlate with echocardiographic left ventricular mass index (LVMI) in groups of selected essential hypertensives. We tested the strength of this relationship in a large group of unselected hypertensives to assess whether QT measurements may be a simple screening test for LVH in clinical practice. In a cross-sectional study of 386 unselected hypertensive subjects, maximal QT intervals (QTmax), QT dispersion (QTdisp), and ECG voltages (Sokolow-Lyon and Cornell sex-specific voltages) were measured from 12-lead ECG. The LVMI correlated most strongly with Cornell voltage (linear regression r ⫽ 0.44, P ⬍ .001). The strongest relationship between LVMI and QT parameters was with QTmax, (r ⫽ 0.25, P ⬍ .001). This relationship weakened using heart rate– corrected QTmax. Correlations between LVMI and QTdisp
E
were weak, whether or not they were corrected for heart rate. Sokolow-Lyon voltages, Cornell voltage and QTmax, but not QTdisp, were independently predictive of LVMI after adjustment for age, sex, race, and the other ECG parameters. Receiver operating characteristic (ROC) curve analyses demonstrated that no QT parameter performed better than simple ECG voltage criteria in the detection of LVH. In conclusion, QTmax, the QT parameter most strongly associated with LVMI, was independently associated with LVMI after adjustment for standard ECG voltage criteria. However, as an isolated measure it was no better than simple ECG voltage criteria as a screening test for LVH in clinical practice. Am J Hypertens 2001;14: 455– 462 © 2001 American Journal of Hypertension, Ltd. Key Words: QT intervals, QT dispersion, ECG voltage criteria, left ventricular hypertrophy, hypertension.
lectrocardiographic QT dispersion, a measure of heterogeneity of QT intervals across the 12 ECG leads, is thought to represent regional differences in myocardial repolarization,1 and it has been proposed that increased variability in myocardial repolarization times may provide the substrate for re-entry ventricular arrhythmias. Indeed, many published studies have shown that prolonged QT intervals or increased QT dispersion (QTdisp) are associated with an adverse prognosis in a variety of pathological conditions, including myocardial ischemia,2,3 heart failure,4 hypertrophic cardiomyopathy,5 and long QT syndromes.6 However, the literature is inconsistent and conflicting, particularly with respect to myocardial infarction.7–9 In a Dutch population, prolonged QT intervals predicted death in apparently healthy individuals.10 However, these data conflict with those from the Framingham cohort in which QT intervals failed to predict overall or cardiac mortality.11
Left ventricular hypertrophy (LVH) in hypertensives predicts cardiovascular mortality.12,13 Left ventricular hypertrophy is associated with excess cardiac arrhythmias,14 and this has been proposed as a mechanism linking LVH with sudden death. Unfortunately, standard ECG criteria for the detection of LVH, although highly specific, are insensitive.15 Several studies have looked at the association between QT parameters and left ventricular hypertrophy. In a small study, QT dispersion was increased in subjects with, compared to those without, LVH.16 In another study, the degree of LVH in patients with hypertrophic obstructive cardiomyopathy correlated with QT intervals but not QT dispersion.17 More recently, several studies have looked specifically at the relationship between QT parameters and left ventricular mass in small groups of selected hypertensive subjects. Two studies found that echocardiographic left ventricular mass index (LVMI) correlated significantly
Received May 31, 2000. Accepted September 12, 2000. From The Peart-Rose Clinic and Department of Clinical Pharmacology (NC, JM, MO, SAMcGT, NRP), Imperial College School of Medicine at St Mary’s, London, United Kingdom; Department of Cardiology (JM), Imperial College School of Medicine at St Mary’s, London, United Kingdom; and Department of Primary Care and Population Sciences
(FCL), Royal Free and UCL Medical School, London, United Kingdom.
© 2001 by the American Journal of Hypertension, Ltd. Published by Elsevier Science Inc.
Address correspondence and reprint requests to Dr N. Chapman, Department of Clinical Pharmacology, Imperial College School of Medicine at St Mary’s, South Wharf Road, London W2 1NY, UK; e-mail: [email protected] 0895-7061/01/$20.00 PII S0895-7061(00)01292-9
456
QT AND LEFT VENTRICULAR HYPERTROPHY
with both maximum QT intervals and QT dispersion.18,19 Another found a very close correlation between QT dispersion and left ventricular mass (r ⫽ 0.74, P ⬍ .001). 20 Based on these data, it has been suggested18 that stratification of hypertensive patients by QT dispersion might allow the echocardiographic identification of a subgroup of patients with LVH. In view of these results in small groups of selected hypertensives, we investigated the relationship between QT parameters, left ventricular mass, and LVH in an unselected hypertensive population. If there was evidence of a strong relationship, then a measure of QT interval or dispersion might provide a suitable alternative ECG method for the detection of LVH in clinical practice, thereby possibly obviating the need for echocardiography in a proportion of patients.
Methods Subjects This was a cross-sectional study of patients attending a hypertension and cardiovascular disease prevention clinic. A total of 386 consecutive patients (189 male and 197 female) referred for assessment of hypertension, with acceptable M-mode echocardiographic images and no bundle branch block on the echocardiogram, were included. None of the subjects had been included in a previous study of QT intervals and hypertension conducted at this institution.19 Most had essential hypertension. Others had borderline, white-coat, or secondary hypertension. The latter groups of subjects were included to assess the generalizability of using QT parameters as a predictor of LVH in unselected hypertensive subjects. Echocardiography Echocardiography was performed using a phased array sector scanner (General Electric Pass II, 3.5-MHz transducer) according to a standard protocol. M-mode images were obtained from the short-axis parasternal view at the level of the tips of the mitral valve leaflets and were printed onto light-sensitive paper for subsequent analysis. Care was taken to ensure that a precise “on-axis” view was obtained. Measurements of septal wall thickness (SWT), posterior wall thickness (PWT), and left ventricular internal dimension (LVID) were made at the peak of the R wave according to the Penn convention.21 Left ventricular mass index (LVMI) (in g/m2) was calculated as: LVMI ⫽ 共1.04关共SWT ⫹ PWT ⫹ LVID兲 3 ⫺ 共LVID兲 3 兴 ⫺ 14兲/Bod y Surface Area Cut-off values for the presence of LVH were taken as ⱖ110 g/m2 for women and ⱖ134 g/m2 for men. Blood Pressure Measurement Blood pressure was measured in the right arm using a suitably sized cuff and an automated recorder (Sentron,
AJH–May 2001–VOL. 14, NO. 5, PART 1
C.R. Bard, Haverhill, MA). A single measurement was taken after approximately 10 min supine rest, immediately after echocardiography. Electrocardiography Routine clinical echocardiography was used (Marquette 12SL, Marquette Electronics, Milwaukee, WI). Recordings were performed at 25 mm/sec with standard lead positions and nonsimultaneous lead recordings. ECG Voltage Criteria Two criteria for the ECG detection of LVH were used, and a single blinded observer made all measurements. According to the Sokolow-Lyon criteria,22 LVH is present if SV1 ⫹ RV5 ⱖ 3.5 mV. According to the Cornell sexspecific criteria (designated here as simply “Cornell”),23 LVH is present if SV3 ⫹ RaVL ⱖ 2.0 mV in women or 2.8 mV in men. Voltages in individual leads were calculated as the mean of three complexes whenever possible. QT Analysis QT analysis was performed by two trained observers who were blind to all clinical details, using a digitizing pad (NewSketch 1212, Genius, KYE Systems UK, Wallington, UK) linked to a PC (Viglen, London, UK). The QT intervals were measured on all possible leads. Where possible, the mean of three complexes was taken as the QT interval for each lead. The QT intervals were measured from the onset of the QRS complex to the end of the T wave, defined as the return to the T-P baseline. Where U waves were present, the QT interval was measured to the nadir of the curve between T and U waves. Leads in which the end of the T wave was unclear were excluded. The mean (⫾ SD) number of evaluable leads was 11 (⫾ 1) leads. The QT intervals were also corrected (QTC) for the known effect of heart rate using Bazett’s formula (QTC ⫽ QT/(RR)1/2).24 The QT dispersion (QTdisp) was calculated as the difference between the maximum QT interval (QTmax) and minimum QT interval in each subject. Heart rate– corrected QT dispersion (QTCdisp) was calculated as the difference between the maximum QTC interval (QTCmax) and minimum QTC interval in each subject. Coefficients of variation for intra- and interobserver variability for QTmax were 1.6% and 4.5%, and for QTdisp were 12% and 21%, respectively. Statistics Continuous data are summarized as mean (⫾ SD) and compared between two groups using unpaired t tests. Categoric data were compared between groups using the 2 test. The Pearson correlation coefficient was used to assess the associations of a range of variables with LVMI. Multiple linear regression was used to assess the independent contribution of clinical and ECG variables to the prediction of LVMI. Results are presented as regression coefficients with 95% confidence intervals, and as partial correlation coefficients. The
AJH–May 2001–VOL. 14, NO. 5, PART 1
Table 1.
QT AND LEFT VENTRICULAR HYPERTROPHY
Subject characteristics
Characteristic n Male/female (%) White/Afro-Caribbean/Other race (%) Treated/untreated (%) Age (years) Systolic BP (mm Hg) Diastolic BP (mm Hg) Heart rate (beats/min) Inter ventricular septum (mm) Left ventricular internal diameter (mm) Posterior wall thickness (mm) Left ventricular mass index (g/m2) SV1 ⫹ RV5 (mV) SV3 ⫹ RaVL (mV) QTmax (msec) QTCmax (msec) QTdisp (msec) QTCdisp (msec)
% or Mean (SD) 386 49/51 65/27/8 49/51 48 (14) 160 (22) 94 (12) 73 (13) 10.4 (1.9) 47.5 (5.5) 10.3 (1.7) 114 2.77 1.67 432 472 68 78
(34) (0.97) (0.72) (47) (47) (22) (25)
SV1 ⫹ RV5, Sokolow-Lyon voltage; SV3 ⫹ RaVL, Cornell sex-specific voltage. Data shown as percentage of subjects or mean (SD).
overall performance of ECG voltages and QT parameters in detecting LVH was assessed using ROC curve analyses. The sensitivities of tests in detecting LVH at matched specificities were compared performed using the McNemar paired 2 test. All analyses were performed using the SPSS statistics package (SPSS Software, Chicago, IL).
Results Subject Characteristics Table 1 shows the characteristics of the 386 subjects in the study population. Of the 191 subjects taking antihypertensive medication, 105 were on monotherapy (11 on diuretics, 12 -blockers, 48 calcium channel blockers, 26 angiotensin converting enzyme inhibitors, and eight ␣-blockers) and 86 were on combinations of two or more drugs. Relationship Between ECG Variables and LVMI Correlation coefficients (r) between LVMI and other measured parameters are shown in Table 2 and scatter plots in Fig. 1. Among the ECG variables tested, LVMI correlated most strongly with the Cornell voltage (linear regression r ⫽ 0.44, P ⬍ .001). Of the QT parameters measured, LVMI correlated most strongly with QTmax (r ⫽ 0.25, P ⬍ .001). However, this correlation was weakened after correction of maximal QT intervals for heart rate (QTCmax; r ⫽ 0.11, P ⫽ .037). Associations of BP, heart rate, ECG voltages, and uncorrected QT measures with
457
Table 2. Unadjusted and adjusted correlation coefficients (r) of LVMI with other variables Partial Unadjusted Correlation* r
P
R
P
Systolic BP (mm Hg) 0.26 ⬍.001 0.19 ⬍.001 Diastolic BP (mm Hg) 0.21 ⬍.001 0.16 .002 Heart rate (beats/ min) ⫺0.17 .001 ⫺0.12 .016 SV1 ⫹ RV5 (mV) 0.24 ⬍.001 0.20 ⬍.001 SV3 ⫹ RaVL (mV) 0.44 ⬍.001 0.35 ⬍.001 QTmax (msec) 0.25 ⬍.001 0.22 ⬍.001 QTCmax (msec) 0.11 .037 0.12 .025 QTdisp (msec) 0.15 .004 0.13 .013 QTCdisp (msec) 0.06 .21 0.08 .130 Pearsons correlation coefficients of hemodynamic and ECG variables with LVMI. * Adjusted for age, sex, and race.
LVMI were slightly weakened by adjustment for age, sex, and race, although all remained significant (Table 2). A multiple regression model of factors potentially predictive of LVMI is shown in Table 3. Increasing age, male gender, SBP, heart rate, ECG voltages, and QTmax were all independently associated with increased LVMI, whereas heart rate was inversely associated with LVMI. However, the strength of the association between QTmax and LVMI was weaker than for either the Sokolow-Lyon or Cornell voltages. The QTdisp was not independently associated with LVMI. Overall, the factors included in the regression model explained 34% of the variation in LVMI. QT Intervals in Subjects With and Without LVH Characteristics of subjects with and without LVH are shown in Table 4. There was a significantly higher proportion of Afro-Caribbean subjects among those with LVH compared with those without. Those with LVH were older and had significantly higher BP and ECG voltages. Mean QTmax was significantly greater in those with LVH compared with those without LVH, but the standard deviation was large and there was a large degree of overlap between the two groups. Heart rate– corrected QTCmax did not differ significantly between those with and without LVH. The QTdisp was significantly greater in those with LVH compared with those without LVH. However, as with QTmax, the spread of values was wide, and significance was lost when heart rate– corrected QTCdisp was compared between the two groups. Receiver Operating Characteristic Curve Analyses of the Performance of ECG Variables in the Detection of LVH Using ROC curve analysis, the Cornell voltage was the parameter with the highest overall predictive capacity. Areas under the curve were as follows: Sokolow-Lyon
458
QT AND LEFT VENTRICULAR HYPERTROPHY
AJH–May 2001–VOL. 14, NO. 5, PART 1
FIG. 1. Scatter plots of echocardiographic (ECG) parameters (QT intervals, QT dispersion, and ECG voltages) plotted against left ventricular mass index (LVMI). ECG voltages (Sokolow-Lyon and Cornell sex-specific), QTmax, and QTdisp plotted against LVMI.
voltage 0.590, Cornell voltage (male and female combined) 0.664, QTmax 0.615, QTCmax 0.566, QTdisp 0.567, and QTCdisp 0.541. Subsequent comparisons of test perTable 3.
formance compared only the Cornell voltage and QTmax as the best-performing examples of voltage criteria and QT parameters, respectively. Sex-specific ROC curves
Multiple regression model of factors predictive of LVMI Variable
Age Gender (male) Ethnic origin (Afro-Caribbean) SBP (mm Hg) SV1 ⫹ RV5 (mV) SV3 ⫹ RaVL (mV) HR (beats/min) QTmax (msec) QTdisp (msec)
Regression Coefficient (B)
95% Confidence Interval
Partial Correlations
P Value
0.24 14.9 5.1 0.25 4.7 12.9 ⫺0.27 78 39
⫺0.01, 0.49 8.6, 21.2 ⫺1.5, 11.6 0.09, 0.40 1.4, 8.0 8.3, 17.5 ⫺0.54, 0.00 2, 155 ⫺106, 184
0.10 0.24 0.08 0.17 0.15 0.28 ⫺0.11 0.11 0.03
.062 ⬍.001 .13 .002 .005 ⬍.001 .051 .045 .60
LVMI ⫽ left ventricular mass index; SBP ⫽ systolic blood pressure; HR ⫽ heart rate. Multiple regression model of factors predicting LVMI (the dependent variable). All factors above are included in the model.
AJH–May 2001–VOL. 14, NO. 5, PART 1
Table 4.
QT AND LEFT VENTRICULAR HYPERTROPHY
459
Characteristics of subjects without versus with left ventricular hypertrophy (LVH)
Number of subjects LVMI (g/m2) Male/female (%) White/Afro-Caribbean/Other race (%) Treated/untreated (%) Age (years) Systolic BP (mm Hg) Diastolic BP (mm Hg) Heart rate (beats/min) SV1 ⫹ RV5 (mV) SV3 ⫹ RaVL (mV) QTmax (msec) QTCmax QTdisp (msec) QTCdisp
No LVH
LVH
P Value
244 95 (18) 52/48 68/23/9 47/53 46 (12) 156 (19) 93 (11) 74 (14) 2.65 (0.89) 1.53 (0.67) 426 (47) 469 (49) 66 (21) 77 (24)
142 147 (32) 44/56 59/36/5 55/45 51 (15) 167 (25) 96 (13) 71 (13) 2.99 (1.06) 1.92 (0.74) 442 (47) 476 (42) 71 (23) 81 (26)
.177 .037 .101 .001 ⬍.001 .009 .025 .001 ⬍.001 .001 .159 .026 .175
Difference Between % or Means (95% CI)
⫺7 ⫺10 8 5 11 3 ⫺3 0.34 0.40 16 7 5 4
(⫺18, 3) (⫺20, ⫺1) (⫺2, 19) (2, 8) (6, 15) (1, 6) (⫺6, 0) (0.14, 0.54) (0.26, 0.55) (6, 26) (⫺3, 17) (1, 10) (⫺1, 9)
LVMI ⫽ left ventricular mass index; BP ⫽ blood pressure; CI ⫽ confidence interval. Data shown as percentages, means (SD), and differences between percentages or means (95% confidence intervals) of the two groups.
showed that the performance of QTmax was no better than Cornell voltage in detecting LVH in either men or women (Fig. 2). Sensitivity and Positive Predictive Value of ECG Variables in the Detection of LVH The standard Cornell cut-off voltages (2.8 and 2.0 mV for men and women, respectively) give high specificity (96% and 84%, respectively) but low sensitivity (19.4% and 25.6%, respectively). To compare directly the sensitivity of Cornell voltages and QTmax in the detection of LVH, sex-specific partition points were adjusted for Cornell voltages and chosen for QTmax to match the specificity of each test at approximately 90%. As shown in Table 5, at this fixed specificity, the sensitivity of QTmax tended to be lower than that for Cornell voltages, although this was not statistically significant for either men or women (P ⫽ .13 and 0.21, respectively). Similarly, positive predictive values tended to be lower using QTmax than using Cornell voltages. QT Intervals in Subjects Taking Antihypertensive Medication Compared with untreated subjects, those subjects on antihypertensive medication were significantly older and had greater Cornell voltages. There were no significant differences in heart rate, BP, LVMI, Sokolow-Lyon voltage, or any QT parameter (data not shown). Compared with all other subjects, subjects on -blockers were significantly older and had slower heart rates (68 ⫾ 15 v 74 ⫾ 13 beats/min, P ⬍ .01). Maximum QT intervals were prolonged in subjects taking -blockers (450 ⫾ 39 v 429 ⫾ 48 msec, P ⬍ .005), but this difference was no longer apparent in heart rate– corrected QT intervals (QTCmax 473 ⫾ 42 v 472 ⫾ 48 msec, P ⫽
.96). There were no differences in QTdisp (68 ⫾ 23 v 68 ⫾ 22 msec, P ⫽ .98), BP, LVMI, ECG voltages, or QTCdisp between those patients on -blockers and all other subjects studied.
Discussion The primary aim of the present study was to assess the strength of the relationship between echocardiographic QT measurements and LVH. This is pertinent to recently published data in selected groups of hypertensives suggesting that QT intervals and dispersion correlate closely with echocardiographic LVMI.18 –20 If the relationship proved to be sufficiently robust, then some measure of QT interval or dispersion might provide a sensitive and specific ECG method of LVH detection superior to currently used ECG voltage criteria, thereby reducing numbers of patients requiring echocardiography. An important issue in the development of any screening technique is applicability in a real clinical environment. Our data from an unselected hospital-based hypertensive population show that correlations between LVMI and QT parameters are generally weaker than those between LVMI and ECG voltage criteria. Although maximal QT intervals and QTdisp were significantly greater in those with LVH compared with those without LVH, there was a large degree of overlap between groups. Moreover, the associations were weakened and the statistical significance of the differences was lost after correction for heart rate. Multiple linear regression and ROC curve analyses also suggest that although QTmax does provide some additional information beyond that given by standard ECG criteria, neither it nor any of the other QT parameters tested in isolation are any better than simple ECG voltages in screening for LVH. Previous studies18 –20 found much stronger relation-
460
QT AND LEFT VENTRICULAR HYPERTROPHY
AJH–May 2001–VOL. 14, NO. 5, PART 1
FIG. 2. Sex-specific receiver operating characteristic curves for the performance of a) Cornell voltages and b) QTmax in the prediction of left ventricular hypertrophy (LVH). Sex-specific sensitivities of Cornell voltage criteria and maximal QT intervals in the detection of LVH at varying specificities. QTmax, maximal QT interval.
ships between QT parameters and LVMI than were seen in the current study. In two studies of both uncorrected and rate-corrected QTdisp, correction for heart rate was associated with little worsening of the relationship with LVMI,18,19 and, unlike the current study, all found QTCdisp to be significantly correlated with LVMI. However, two of the previous studies used well characterized groups of untreated essential hypertensive subjects and
excluded those with intercurrent cardiac and noncardiac diseases. The third study did not specify inclusion or exclusion criteria beyond the fact that the patients had a diagnosis of confirmed systemic hypertension.20 In contrast, the current study was designed to include those with elevated clinic BP irrespective of existing drug treatment or other disease processes that frequently coexist in hypertensives. It has been documented that several forms of
AJH–May 2001–VOL. 14, NO. 5, PART 1
QT AND LEFT VENTRICULAR HYPERTROPHY
Table 5. Sensitivity and positive predictive (PPV) value of Cornell sex-specific voltage and QTmax at matched specificity in the detection of LVH
Cornell Specificity Sensitivity PPV QTmax Specificity Sensitivity PPV
Men
Women
91.2 32.3 64.5
91.5 21.3 61.5
90.5 19.4 50.5
90.8 11.5 45.0
Sensitivity and positive predictive values of Cornell voltage criteria and QTmax at matched specificities. Sex-specific partition points were adjusted for Cornell voltages (to 2.6 and 2.2 mV in men and women, respectively) and chosen for QTmax (461 msec in both men and women) to compare sensitivity and positive predictive values of each method at matched specificity.
cardiac disease affect QT parameters3,4 and that QT measurements are less reliable in abnormal ECG.25 The presence in this population of subjects with cardiac disease, and in others with repolarization changes associated with LVH, is likely to have weakened the relationship between QT parameters and LVH. Our population also included treated hypertensives and those with borderline or secondary hypertension, and is likely to have included some subjects with white-coat hypertension. The inclusion of these groups, although justified by the primary aim of the study, is also likely to have weakened the relationship between LVMI and QT parameters. Finally, in contrast to previous studies,26 we found ECG-determined heart rate to be weakly inversely correlated with LVMI. This unexpected finding may be due to skewing of the data by those with physiological LVH and bradycardia due to physical training, as well as individuals with white-coat hypertension and normal LVMI but a marked chronotropic response to stress. Any such effects would contribute to the attenuation of correlations seen between LVMI and QT parameters on correction for heart rate. In our population, no differences were seen in any QT parameter between subjects taking, versus those not taking, antihypertensive drugs. However, when compared with all other subjects, those taking -blockers had prolonged QTmax, but no difference was seen in QTdisp. However, the difference in QTmax may be explained by the relative bradycardia in subjects taking -blockers, as no difference was seen in the difference in QTCmax. Numerous papers have reported on the associations and prognostic implications of QT intervals and dispersion. However, results have been inconsistent,7–11 possibly reflecting variability in the methodology of QT interval measurement.27 Although interobserver variability in this study was comparable with that in previous reports,28 significant variability exists, particularly in measurements of dispersion. Furthermore, measurement variability is greater in abnormal, compared with normal, ECG.25 Standardization of measurement criteria and computerized analysis may improve
461
the repeatability and usefulness of the technique in the future, although studies have shown marked variation in measurements between different automated systems.29 Finally, the interpretation and significance of QT dispersion as a marker of ventricular repolarization has recently been called into question.30 In conclusion, unlike previous studies in small groups of selected hypertensives, correlations between QT parameters and echocardiographic LVMI were weak in an unselected hypertensive population, probably due to a number of confounding factors and variability in QT measurements. The overall performance of QT parameters in the detection of LVH was no better than simple ECG voltage techniques. Although QTmax (but not QTdisp) predicted LVMI independently of ECG voltage criteria, no QT measure in isolation is superior to ECG voltage criteria in the routine detection of LVH setting. Whether QT intervals or dispersion predict prognosis in hypertensives independently of LVH is a further issue that can only be properly addressed by a prospective study.
Acknowledgments We thank Melissa Poulter and Cecilia Tse for their invaluable assistance with ECG analysis.
References 1.
Higham PD, Campbell RWF: QT dispersion. Br Heart J 1994;71: 508 –510. 2. Taylor GJ, Crampton RS, Gibson RS, Stebbins PT, Waldman MTG, Beller GA: Prolonged QT interval at onset of acute myocardial infarction in predicting early phase ventricular tachycardia. Am Heart J 1981;102:16 –24. 3. Higham PD, Furniss SS, Campbell RWF: QT dispersion and components of the QT interval in ischaemia and infarction. Br Heart J 1995;73:32–36. 4. Barr CS, Naas A, Freeman M, Lang CC, Struthers AD: QT dispersion and sudden unexpected death in chronic heart failure. Lancet 1994;343:327–329. 5. Buja G, Miorelli M, Turrini P, Melacini P, Nava A: Comparison of QT dispersion in hypertrophic cardiomyopathy between patients with and without ventricular arrhythmias and sudden death. Am J Cardiol 1993;72:973–976. 6. Day CP, McComb JM, Campbell RWF: QT dispersion: an indication of arrhythmia risk in patients with long QT intervals. Br Heart J 1990;63:342–344. 7. Pohjola-Sintonen S, Siltanen P, Haapakoski J: Usefulness of QTc interval on the discharge electrocardiogram for predicting survival after acute myocardial infarction. Am J Cardiol 1986;57:1066 – 1068. 8. Glancy JM, Garratt CJ, Woods KL, de Bono DP: QT dispersion and mortality after myocardial infarction. Lancet 1995;345:945–948. 9. Zabel M, Klingenheben T, Franz MR, Hohnloser SH: Assessment of QT dispersion for prediction of mortality or arrhythmic events after myocardial infarction. Results of a prospective, long-term follow-up study. Circulation 1998;97:2543–2550. 10. Schouten EG, Dekker JM, Meppelink P, Kok FJ, Vandenbrouke JP, Pool J: QT interval prolongation predicts cardiovascular mortality in an apparently healthy population. Circulation 1991;84:1516 –1523.
462
QT AND LEFT VENTRICULAR HYPERTROPHY
11. Golberg RJ, Bengtson J, Chen ZY, Anderson KM, Locati E, Levy D: Duration of the QT interval and total and cardiovascular mortality in healthy persons (the Framingham Heart Study experience). Am J Cardiol 1991;67:55–58. 12. Kannel WB, Gordon T, Castelli WP, Margolis JR: Electrocardiographic left ventricular hypertrophy and risk of coronary heart disease: the Framingham Heart Study. Ann Intern Med 1970;72: 813– 822. 13. Levy D, Garrison RJ, Savage DD, Kannel WB, Castelli WP: Prognostic implications of echocardiographically determined left ventricular mass in the Framingham Study. N Engl J Med 1990;322: 1561–1566. 14. McLenachan JM, Henderson E, Morris KI, Dargie HJ: Ventricular arrhythmias in patients with hypertensive left ventricular hypertrophy. N Engl J Med 1987;317:787–792. 15. Chapman N, Mayet J, Sever P, Foale R, Thom S, Poulter N: Ethnic differences in the identification of left ventricular hypertrophy in the hypertensive patient. Am J Hypertens 1999;12:437– 442. 16. Davey PP, Bateman J, Mulligan IP, Forfar C, Barlow C, Hart G: QT interval dispersion in chronic heart failure and left ventricular hypertrophy: relation to autonomic nervous system and Holter tape abnormalities. Br Heart J 1994;71:268 –273. 17. Dritsas A, Sbarouni E, Gilligan D, Nihoyannopoulos P, Oakley CM: QT-interval abnormalities in hypertrophic cardiomyopathy. Clin Cardiol 1992;15:739 –742. 18. Clarkson PBM, Naas AAO, McMahon A, MacLeod C, Struthers AD, MacDonald TM: QT dispersion in essential hypertension. Q J Med 1995;88:327–332. 19. Mayet J, Shahi M, McGrath K, Poulter NR, Sever PS, Foale RA, Thom SAMcG: Left ventricular hypertrophy and QT dispersion in hypertension. Hypertension 1996;28:791–796. 20. Ichkhan K, Molnar J, Somberg J: Relation of left ventricular mass and QT dispersion in patients with systematic hypertension. Am J Cardiol 1997;79:8508 – 8511.
AJH–May 2001–VOL. 14, NO. 5, PART 1
21. Devereux RB, Reichek N: Echocardiographic determination of left ventricular mass in man. Anatomic validation of the method. Circulation 1977;55:613– 618. 22. Sokolow M, Lyon TP: The ventricular complex in left ventricular hypertrophy as obtained by unipolar precordial and limb leads. Am Heart J 1949;37:161–186. 23. Casale PN, Devereux RB, Alonso DR, Campo E, Kligfield P: Improved sex-specific criteria of left ventricular hypertrophy for clinical and computer interpretation of echocardiograms: validation with autopsy findings. Circulation 1987;75:565–572. 24. Ahnve S: Correction of the QT interval for heart rate. Review of different formulas and the use of Bazetts formula in myocardial infarction. Am Heart J 1985;109:568 –574. 25. McLaughlin NB, Campbell RWF, Murray A: Accuracy of four automatic QT measurement techniques in cardiac patients and healthy subjects. Heart 1996;76:422– 426. 26. Mayet J, Shahi M, Hughes AD, Stanton AV, Poulter NR, Sever PS, Foale RA, Thom SAMcG: Left ventricular structure and function in previously treated hypertensive patients: the importance of blood pressure, the nocturnal blood pressure dip and heart rate. J Cardiovasc Risk 1995;2:255–261. 27. Statters DJ, Malik M, Ward DE, Camm AJ: QT dispersion: problems of methodology and clinical significance. J Cardiovasc Electrophysiol 1994;5:672– 685. 28. Kautzner J, Gang Y, Camm J, Malik M: Short- and long-term reproducibility of QT, QTc and QT dispersion measured in healthy subjects. PACE 1994;17:928 –937. 29. McLaughlin NB, Campbell RWF, Murray A: Comparison of automatic QT measurement techniques in the normal 12 lead echocardiogram. Br Heart J 1995;74:84 – 89. 30. Coumel P, Maison-Blanche P, Badilini F: Dispersion of ventricular repolarization. Reality? Illusion? Significance? Circulation 1998;97: 2491–2493.