Electrocardiographic T Wave and its Relation With Ventricular Repolarization Along Major Anatomical AxesCLINICAL PERSPECTIVE
Background—The genesis of the electrocardiographic T wave is incompletely understood and subject to controversy. We have correlated the ventricular repolarization sequence with simultaneously recorded T waves.
Methods and Results—Nine pig hearts were Langendorff-perfused (atrial pacing, cycle length 650 ms). Local activation and repolarization times were derived from unipolar electrograms sampling the ventricular myocardium. Dispersion of repolarization time was determined along 4 anatomic axes: left ventricle (LV)–right ventricle (RV), LV:apico-basal, LV:anterior-posterior, and LV:transmural. The heart was immersed in a fluid-filled bucket containing 61 electrodes to determine Tp (Tpeak in lead of maximum integral), TpTe (Tp to Tend), and TpTe_total (first Tpeak in any lead to last Tend in any lead). Repolarization was nonlinearly distributed in time. RT25 (time at which 25% of sites were repolarized, 288±26 ms) concurred with Tp. TpTe was 38±8 ms, and TpTe_total was 75±9 ms. TpTe_total correlated with dispersion of repolarization time in the entire heart (73±18 ms), but not with dispersion of repolarization times along individual axes (LV–RV, 66±17 ms; LV:apico-basal, 51±18 ms; LV:anterior-posterior, 51±27 ms; mean LV:transmural, 14±7 ms; all n=9).
Conclusions—We provide a correlation between local repolarization and T wave in a pseudo-ECG. Repolarization differences along all anatomic axes contribute to the T wave. TpTe_total represents total dispersion of repolarization. At Tp, ≈25% of ventricular sites have been repolarized.
The T wave on the ECG (T-ECG) represents repolarization of the ventricular myocardium. Its morphology and duration are commonly used to diagnose pathology and assess risk of life-threatening ventricular arrhythmias. However, the physiological background of the T wave is incompletely understood, thereby hampering reliable interpretation of the T wave.
Clinical Perspective on p 531
Interpretation of the T wave morphology on body surface ECGs (T-ECG) is different from that on local electrograms (T-local). In a far-field recording (eg, a 12-lead ECG), the T wave represents repolarization in the entire heart.1 In a near-field recording (eg, a local electrogram), the T wave can be used to accurately determine local repolarization moments.2,3 Unlike our understanding of T-local, knowledge of the T-ECG is still incomplete.
Previous studies have focused mainly on a single dominant repolarization gradient to explain the T-ECG morphology.4–6 It has become generally accepted that concordance of the T-ECG with the QRS complex is explained by opposing sequences of transmural activation and repolarization.5 Whereas activation progresses from subendocardial to subepicardial myocardium, repolarization was assumed to progress in the opposite direction.7 Additionally, T-ECG-peak and T-ECG-end were explained by the end of repolarization of, respectively, the subepicardial and M cell layers.6 There is, however, controversy regarding the presence of a functional M cell layer in human hearts. M cells have been observed only in human wedge preparations,8,9 but were absent in vivo.10,11 In dogs, total rather than transmural dispersion of ventricular repolarization is more determinative for the peak-to-end interval of the T-ECG (TpTe).12 Other studies have suggested that an apico-basal gradient of repolarization is responsible for the T-ECG morphology.1,13 These observations have resulted in a persisting debate concerning the dominant repolarization differences giving rise to the T-ECG.14–16
We aimed to clarify the genesis of the T-ECG by correlating the ventricular repolarization sequence with simultaneously recorded T-ECGs in isolated perfused pig hearts. Pig hearts most likely lack an M cell layer, like human hearts.17
See the Data Supplement for detailed methods.
The experimental protocol was approved by the local ethical committee on animal experimentation. Male pigs (n=9; body weight, 34±12 kg) were anesthetized and a 12-lead ECG was recorded in supine position (in vivo ECG). The heart was excised and perfused according to Langendorff with a 1:1 blood–Tyrode's mixture. The apex was fixed to avoid swinging of the heart, but to allow rotation and movement in the vertical plane.
Electrophysiological Study and Data Acquisition
After removal of the sinus node area, the atrium was paced at a cycle length of 650 ms. Transmural multielectrode (0.5 mm diameter) needles were inserted in a predefined pattern to obtain transmural local unipolar electrograms throughout the heart. Exact locations of terminals were reconstructed after termination of the experiment. The heart was surrounded by a perfusion-fluid–filled bucket containing 61 regularly distributed electrodes (1 electrode at the bottom of the bucket) to obtain a pseudo body surface ECG (BS-ECG). The reference signal was the assembled average of all 61 electrodes.
In the BS-ECG and in vivo ECG, QRS duration, QRS vector, QT interval, QT interval corrected for heart rate via Fridericia18 (QTc_F), T-ECG duration (T-duration), T-ECG vector (T-vector), and TpTeinterval were determined (Figure 1). QRSonset and QRSend were defined as, respectively, the first deflection and last J-point of the QRS complex in any ECG lead. Integrals of the QRS complex and T-ECG (defined as area under the QRS and T curve) were calculated between manually set markers. These were used to identify main (in vivo) and detailed (BS-ECG) QRS and T vectors in 3 planes: frontal, transversal, and sagittal (in vivo: lead I, V6 and V2 equal 0 degrees, and rotation to lead aVF, V2, and aVF denotes a positive angle; in BS-ECG, similar definitions were used).
In the ECG lead of maximum (positive or negative) T-integral, the onset (To), peak (Tp), and end (Te) of the T-ECG were manually determined relative to QRSonset using the tangent method (Figure 1).19 QT interval and TpTe were defined as interval between QRSonset and Te, and between Tp and Te in a single lead. We additionally determined TpTe_total in BS-ECG as interval between earliest Tpeak in any lead and last Tend in any lead.
At each local recording site, local activation times and local repolarization times (RTs) were automatically determined as the interval between QRSonset to time of the minimum derivative of QRS complex and maximum derivative of T local, respectively (Figure 1). All signals and time markers were visually validated. Recordings with ST-elevation or a flat T-local were excluded from analysis of RTs.
RTmin and RTmax were start and end of repolarization, respectively. The volume distribution of repolarization was described with time points at which, respectively, n=5%, 25%, 50% (median RT), 75%, and 95% of the recording sites were repolarized (RTn). Dispersion of repolarization (dRT) was defined as range of RTs along a specific anatomic axis: left ventricle (LV)–right ventricle (RV), LV:apico-basal, LV:anterior-posterior, and LV:transmural. Maximum and mean LV:transmural dRT (dRT_maxLVtransmural, dRT_meanLVtransmural) were, respectively, maximum and mean of all dRTs per LV needle. The dRT_total was the maximum range of RTs across the entire heart (Figure 1). Signal analysis was performed offline using a custom-made data analysis program written in Matlab 2006b.20
Continuous variables were given as mean±SD if normally distributed and in median (25th–75th percentile) if not normally distributed. Evaluation of normality was based on Q-Q plots and correspondence of mean and median. Differences in characteristics between the in vivo ECG and BS-ECG, between RTn and T-ECG parameters, and between dRT values and TpTe or TpTe_total were tested using a paired t test. We determined the amount of similarity (combination of distance and correlation) between RTn and T-ECG parameters and between dRT values and TpTe or TpTe_total by calculating intraclass correlation coefficients (ICCs) using a 2-way mixed model of absolute agreement.21 Differences in RT50 between different regions along an anatomic axis were tested with a Friedman analysis. Differences and ICCs with a P value of ≤0.05 were considered statistically significant.
In Vivo Standard 12-Lead ECG and Ex Vivo BS-ECG
Figure 2 shows a typical example of a porcine in vivo 12-lead ECG in supine position (Figure 2A) and the BS-ECG of the same heart in Langendorff perfusion (Figure 2B). In general, in the in vivo ECG the T-ECG was positive in precordial and inferior leads and flat, biphasic, or slightly negative in leads I, aVR, and aVL.
Overall, the T-ECG in BS-ECG was discordant with the QRS complex in most leads and positive in anterior leads. The Table features the main characteristics of the ex vivo BS-ECG and in vivo 12-lead ECG. In comparison with the in vivo ECG of the same animal, the QRS vector in the BS-ECG is rotated in each plane (frontal: 47±74°; P=n.s.; transversal: 56±60°; P=0.024; sagittal: 49±39°; P=0.005). There was no significant change in T-vector between the in vivo ECG and BS-ECG (frontal: −3±90°, transversal: −74±105°, sagittal: 56±93°; all P=n.s.) but the change in T-vector showed a larger SD than the change in QRS vector. QRS duration, QT interval, QTc_F, T-duration, and TpTe in the BS-ECG were not statistically different from the in vivo ECG (Data Supplement).
Activation and Repolarization
Figure 3A and 3B shows typical examples of, respectively, activation and repolarization maps of 1 heart. In all hearts, activation times were obtained from 87±27 recording sites throughout the heart. Overall, activation occurred earliest at the left septum and latest at the RV basal free wall (as in Figure 3A). Transmural activation along needles was almost simultaneous. Activation was complete within 49±9 ms (see Data Supplement).
In all hearts, repolarization times were obtained from 61±21 recording sites per heart. Overall, repolarization started at the anterior or posterior RV. Repolarization in the LV progressed from basal to central (between apex and base) regions. Repolarization of apical regions varied remarkably between hearts. Final repolarization was at the central free walls (RV before LV; see also Figure 3B). Transmural repolarization was simultaneous, although there were isolated sites with larger transmural differences (asterisk in Figure 3B, crowding of isochrones).
On average, RTmin was at 251±22 ms and RTmax at 324±33 ms (n=9 hearts). RT25 was at 288±26 ms (at midrepolarization), RT50 at 303±28 ms, and RT75 at 309±30 ms. From the onset of repolarization, it took 52±13 ms to fully repolarize the first 50% of the recording sites. From that moment it took only 21±7 ms to complete repolarization.
T-ECG and the Overall Repolarization Sequence
To occurred always earlier than RTmin (−27±15 ms; P<0.05; ICC, 0.56; P<0.005). Te was not significantly different from RTmax (−3±10 ms; P=n.s.) and had a large ICC (0.95; P<0.001). We compared time of Tp on the BS-ECG (Tp=283±28 ms) with the distribution of repolarization (RTmin, RT25, RT50, RT75, and RTmax). Figure 4 shows that RT25 had the smallest difference with the time of Tp (6±12 ms; P=n.s.) and largest ICC (0.89; P<0.001). More precisely, at Tp the percentage of repolarized sites was 19.5±12.6%.
T-ECG and the Distribution of RT Along 4 Anatomic Axes
RT distribution along the 4 anatomic axes was studied in relation to T-ECG morphology. Figure 5 shows an example of the time relation per predefined axis between the T-ECG in a BS-ECG lead (with maximum T-ECG integral, opposite the RV), and RT5, RT25, RT50, RT75, and RT95. The figure (top panel) demonstrates that half of RV (RT50) had already repolarized before RT5 of LV, and that at RT95 in the RV, 50% of LV (RT50) still had to repolarize. We restricted analysis of RT distribution along other axes only to the LV, because it has the largest contributing mass and harbored more electrodes. In the figure, the dispersion of RT50 values within the LV:antero-posterior, LV:apico-basal, or LV:transmural axes was not large enough to explain the T-duration. In this example, LV repolarization starts at the basal posterior midendocardial region and ends in the basal anterior subendocardial region, illustrating that repolarization along all axes plays a role in the time relation between the T-ECG and the repolarization process. The bottom panel of the figure shows the RT distribution of the entire heart along the T-ECG. The grey histogram of repolarized recording sites re-emphasizes that the majority of sites repolarized after the moment of Tp.
Figure 5 summarizes RT50 values per anatomic axis of all animals. RT50 values were significantly different between regions within the LV:apico-basal axis (central region was 17±9 ms longer than apical or basal region; n=7), and LV:transmural axis (the subendocardial or midendocardial region was 17±13 ms longer than the subepicardial or midepicardial region; n=9).
TpTe and Dispersion in Repolarization
TpTe has been associated with heterogeneity in RT, either in wedge preparations6 or in the whole heart.12,22 We, therefore, studied dispersion of repolarization moments (dRT) along each of the 4 axes and over the entire heart in relation to TpTeon the BS-ECG (Figure 6). Locations of maximum and minimum RT across each single axis differed between animals, so the direction of the vector connecting the pair of values constituting the dRT_total was not similar among animals. TpTe_total on the BS-ECG was least different from dRT_total (Figure 6: mean difference of 1±13 ms; paired t test P=0.758), with an ICC of 0.626 (P=0.03). TpTe was least different from dRT_maxLVtransmural (Figure 6: mean difference of 4±20 ms; paired t test P=0.552). However, the ICC of 0.306 was low and statistically not significant. Other dRT values that were not statistically significant from TpTe_total or TpTe (Figure 6) had larger differences and lower ICC values with TpTe_total or TpTe.
The main findings of the present study are that (1) differences in repolarization moments along all axes contribute to the T-ECG morphology; (2) TpTe_total measured on the BS-ECG reflects total dispersion of repolarization within the heart; (3) at the moment of T-ECG peak, ≈25% of the recording sites are repolarized; and (4) onset of T-ECG precedes onset of repolarization. We provide the first complete correlation between local repolarization and the simultaneously recorded T-ECG on the surface ECG.
In Vivo ECG Versus Ex Vivo ECG
The in vivo and pseudo-ECG closely resemble each other, although transversal and sagittal QRS vectors differ. It likely results from a different position of the heart in the thorax versus the bucket (Table and Data Supplement). Therefore, we assume that activation in the hearts was unchanged between in vivo and ex vivo. This is supported by the unchanged QRS duration. In addition, the in situ ventricular activation sequence of a dog heart recorded by Durrer et al23 did not change after isolation.
Repolarization and T-ECG
We showed significant differences in RT50 along LV:apico-basal and LV:transmural axes. However, differences (≈17 ms) were too small to explain an average T-ECG duration of 97 ms. It suggests that total dispersion of RT is reflected in the entire T-ECG. Indeed, we demonstrated that the interval between first peak to last end of all T-ECGs did correspond with total dispersion of RT, corroborating results from Xia et al.22 Although maximum transmural dRT was least different from TpTe, its ICC was low and not statistically significant. The lack of correlation may be attributed to the fact that the maximum difference represents a single outlier value, which may not affect T wave morphology. The mean transmural dRT more reliably represents transmural dispersion but did not correlate with TpTe, as demonstrated before.12
To precedes the first measured RT. This is unlikely the result of a sampling error, because it occurred in all hearts. We suggest that regional differences in action potential (AP) morphology (phase 1 and 2) rather than differences in RTs contribute to genesis of the first part of the T-ECG. Te coincides with the last measured RT, consistent with the fact that RT in a unipolar electrogram relates to the steepest part of AP downstroke,2,3 which approximates APD90 (AP duration at 90% of repolarization).24
Fuller et al25 found a near equality of time of T wave peak and mean RT, which contrasts with our results that 75% of recording sites repolarizes after Tp. Methodological differences may explain this discrepancy. We used a single surface ECG lead instead of a root mean square of epicardial recordings. Not only is the former clinically more relevant, the Data Supplement indicates that the root mean square of BS-ECGs shows similar results and does not explain the discrepancy. Furthermore, Fuller used only epicardial electrograms, which may shift or transform RT distribution and underestimate RT dispersion. Differences in data presentation (medians and quartiles versus means) and species (pigs versus dogs) may cause additional discrepancy. We underscore that the finding that Tp reflects RT25 cannot indiscriminately be applied to a randomly chosen lead because each has a different view on the dominant vector. Nonetheless, we suppose it is applicable to the lead with maximum T wave area.
Our data show that the time during which the first 50% of sites repolarize is about twice as long as the time required for repolarization of the last 50% of sites. A possible physiological explanation for the nonlinear repolarization process is that repolarization accelerates exponentially as a result of radial propagation of the repolarization wave. This matches the repolarization course illustrated in Figure 5 (bottom panel). The highly skewed apparent velocity of the repolarization process has thus far not been described, although its reflection in an asymmetrical T wave is common knowledge (To to Tp interval of 59±7 versus TpTe interval of 38±8 ms).
We speculate that the first part of a normal T-ECG primarily depends on regional differences in repolarization time course (ie, changes in AP morphology) preceding the moment of full repolarization (ie, steepest part of AP downstroke). When full repolarization has started, continuation may accelerate depending on the degree of electrotonic interaction, which may determine the last part of the T wave.
Pig Versus Human: In Vivo ECG and Repolarization Sequence
The QRS configuration of the in vivo porcine ECG in our study is in agreement with another study in pigs.26 The in vivo porcine T-ECG in our study resembles a normal T-ECG in human27 because the polarity is positive in most leads and T-vectors have similar angles (if corrected for differences in position of the heart in the thorax).28 QRS duration in pig, however, is shorter than in man.23 This is because of a more extensive transmural Purkinje network29 and the smaller size of porcine hearts compared with human. Nevertheless, the general activation sequence observed in our study resembles the activation sequence described in humans23 and pigs.30,31
The porcine repolarization sequence in our study resembles that in earlier reports,22,31 although repolarization of the septal region occurred later in our study. Methodological differences (endocardial monophasic action potentials recorded sequentially at 50–70 sites instead of local transmural electrograms recorded simultaneously at 61±21 sites [our study]) may have caused the minor differences in repolarization sequences, although correlation between RT and monophasic APD90 is high.24
We provide the first complete correlation between local repolarization and the corresponding electrocardiographic T wave. A main conclusion of this study is that total dispersion of repolarization along all anatomic axes is determinative for the T wave on the pseudo-ECG of the isolated perfused pig heart. TpTe_total (first peak to last end of all T-ECGs) across ECG leads represents total dispersion of repolarization. Moreover, at the T wave peak, only ≈25% of ventricular sites are repolarized. Therefore, the start of the T wave is likely the result of regional differences in early repolarization (phase 1 and 2 of the AP), and the nonlinear repolarization process may be reflected in the final part of the T wave.
We are grateful for technical support of Carel Kools and Dr André Linnenbank.
Sources of Funding
Grant from Fundació Marató de TV3 (project 080632), Barcelona, Spain, is acknowledged.
The Data Supplement is available at http://circep.ahajournals.org/lookup/suppl/doi:10.1161/CIRCEP.113.001622/-/DC1.
- Received November 8, 2013.
- Accepted April 11, 2014.
- © 2014 American Heart Association, Inc.
- Noble D,
- Cohen I
- Coronel R,
- de Bakker JM,
- Wilms-Schopman FJ,
- Opthof T,
- Linnenbank AC,
- Belterman CN,
- Janse MJ
- Haws CW,
- Lux RL
- Franz MR,
- Bargheer K,
- Rafflenbeul W,
- Haverich A,
- Lichtlen PR
- Yan GX,
- Antzelevitch C
- Glukhov AV,
- Fedorov VV,
- Lou Q,
- Ravikumar VK,
- Kalish PW,
- Schuessler RB,
- Moazami N,
- Efimov IR
- Conrath CE,
- Wilders R,
- Coronel R,
- de Bakker JM,
- Taggart P,
- de Groot JR,
- Opthof T
- Taggart P,
- Sutton PM,
- Opthof T,
- Coronel R,
- Trimlett R,
- Pugsley W,
- Kallis P
- Patel C,
- Burke JF,
- Patel H,
- Gupta P,
- Kowey PR,
- Antzelevitch C,
- Yan GX
- Opthof T,
- Coronel R,
- Janse MJ
- Rodríguez-Sinovas A,
- Cinca J,
- Tapias A,
- Armadans L,
- Tresànchez M,
- Soler-Soler J
- Durrer D,
- van Dam RT,
- Freud GE,
- Janse MJ,
- Meijler FL,
- Arzbaecher RC
- Potse M,
- Vinet A,
- Opthof T,
- Coronel R
- Hamlin RL
- Gepstein L,
- Hayam G,
- Ben-Haim SA
The genesis of the electrocardiographic T wave is incompletely understood and a subject of controversy. Nevertheless, the morphology and duration of the T wave is commonly used by clinicians to diagnose pathology and assess the risk for life-threatening ventricular arrhythmias. This study in excised pig hearts provides the first complete correlation between local repolarization and the corresponding T wave. It shows that for all leads, the T wave is determined by repolarization heterogeneity in the entire heart and not along a single anatomic axis. We also demonstrate that repolarization is a nonlinear process in which the peak of the T wave coincides with the moment at which ≈25% of the heart is repolarized. We speculate that the first part of the normal T wave primarily depends on the regional differences in repolarization time course (ie, changes in action potential morphology) preceding the moment of full repolarization (ie, steepest part of action potential downstroke). Acceleration of repolarization, which depends on electrotonic interaction, may determine the last part of the T wave. These results contribute to understanding the genesis of the T wave and thereby may improve interpretation of the T wave in health and disease.