Author + information
- Received November 25, 2008
- Revision received March 5, 2009
- Accepted March 10, 2009
- Published online July 7, 2009.
- Paulus Kirchhof, MD⁎,
- Michael R. Franz, MD, PhD†,⁎ (, )
- Abdennasser Bardai, MD‡ and
- Arthur M. Wilde, MD‡
- ↵⁎Reprint requests and correspondence:
Dr. Michael R. Franz, VA Hospital, Cardiology, 50 Irving Street, NW, Washington, DC 40007
Objectives This study sought to identify electrocardiographic (ECG) criteria that are associated with initiation of torsades de pointes (TdP) in patients with acquired (a-) and congenital (c-) long QT syndrome (LQTS).
Background Electrocardiographic criteria used as risk predictors for TdP commonly rely on a prolonged QT interval but rarely consider abnormal T–U waves.
Methods We analyzed ECG recordings with TdP from 35 LQTS patients (15 c-LQTS and 20 a-LQTS) and compared them with premature ventricular complexes (PVCs) from 40 patients with normal QT intervals and with PVCs in 24 of the 35 LQTS patients not related to TdP.
Results Abnormal T–U waves (6.2 ± 0.9 mm) directly preceded TdP in 34 of 35 LQTS patients and were larger than T-wave amplitude (2.8 ± 0.2 mm) in control patients and larger than the largest T–U-wave in LQTS without TdP (4.7 ± 0.8 mm). The TdP-initiating beat emerged from a T–U-wave in 27 of 35 LQTS patients and in none of 40 control patients. The QRS duration of the first TdP beat (175 ± 12 ms) was longer than in control PVCs (145 ± 4 ms) and in PVCs in LQTS patients not related to TdP (138 ± 22 ms). The QRS angle was less steep before TdP than in other PVCs (all p < 0.05).
Conclusions Abnormal, giant T–U waves separate TdP initiation in LQTS patients from PVCs in other heart disease and from other PVCs in LQTS patients. These ECG analyses suggest that early afterdepolarizations initiate TdP and, if present, may help to identify an imminent risk for TdP.
The acquired long QT syndrome (a-LQTS) and congenital long QT syndrome (c-LQTS) predispose patients to torsades de pointes arrhythmias (TdP). The mechanisms that initiate TdP in LQTS patients are not well understood. Although abnormal prolongation of the QT interval identifies patients at increased risk for TdP (1–3), many patients tolerate marked QT prolongation without TdP. Thus, there must be other factors that cause TdP. In experimental settings, early afterdepolarizations (EADs) initiate TdP (4–6), which may be reflected by abnormal T–U waves in the electrocardiogram (ECG) (5,7,8).
A common feature of drug-induced TdP and of TdP in long QT syndrome type 2 (LQTS2) is that the TdP-initiating beat is preceded by a premature beat followed by a pause. Often, this pattern of premature beats and pauses, or short-long-short interval, repeats for several cycles in an incremental fashion, with TdP occurring when the pause has reached a critical length (9). A comprehensive publication from 2 decades ago (5) reviewed these ECG features from both experimental and clinical observations and suggested an eminent role of abnormal T–U waves in the triggering of TdP. In that and other subsequent studies, the use of monophasic action potential recordings showed that the U-wave in LQTS patients closely correlated with early EADs at the cellular level (10). This not only makes correct measurements of the QT interval more difficult but may also in itself contain relevant information that is more directly linked to the effects that initiate TdP than QT interval analysis alone.
These considerations and occasional observations suggest that giant T–U waves in LQTS not only are an important ECG criterion for imminent TdP, but also constitute one of the actual pathophysiologic trigger mechanisms for TdP. To study the relevance and clinical usefulness of T–U waves for identification of imminent proarrhythmia, we therefore compared ECG parameters before TdP with ECG recordings before other premature ventricular complexes (PVCs).
ECG data collection
We analyzed ECG recordings in 35 patients with a- and c-LQTS and TdP (from the Academic Medical Centre, Amsterdam, collected from 1991 to 2006) and compared them with ECGs from 40 patients with normal QT intervals and PVCs on Holter or routine ECG (from University Hospital Münster) and with ECGs from 24 of the above-mentioned LQTS patients (10 c-LQTS, 14 a-LQTS) without TdP but with PVCs (Fig. 1).
After a quality check assessing recording speed (>25 mm/s), continuous recording of the initiation of TdP or of a PVC with at least 1 normal beat before this episode, and availability of at least 2 ECG leads, all ECGs were analyzed by 2 independent observers (Fig. 1). If independently measured parameters differed (generally <5% for continuous parameters), the mean value was used for final analysis. Analysis was performed in an ECG with high-amplitude T waves. We analyzed: 1) RRprec: the RR interval preceding the last normal beat before the arrhythmia; 2) QTprec: the QT interval in the last normal beat before the arrhythmia; 3) Tprec: the T-wave amplitude in the last normal beat before the arrhythmia; 4) TUany: the largest T–U-wave amplitude in any beat of the recording strip with the exception of the beat that initiated the arrhythmia; 5) TUTdP: the T–U-wave amplitude of the T–U-wave from which the arrhythmia was initiated; 6) QRSPVC: the QRS duration of the first beat of the arrhythmia (TdP or PVC without TdP (as an indirect measure of propagation velocity); and 7) QRSangle: the angle of the first QRS upstroke (or downstroke) of the first beat of the arrhythmia (as a measure of premature activation velocity) (Fig. 2).
Because most recordings (often from monitor strips or in an emergency room setting) provided neither a 12-lead ECG nor voltage calibration, we compared all measurements in millimeters rather than millivolts (TUTdP-wave) and also normalized TUTdPto the amplitude of the “normal” T-wave (TUTdP/Tprecratio) and to the largest T- or U-wave in the entire recording (TUTdP/TUany).
Continuous parameters were normally distributed and compared between groups using unpaired Student ttests and within groups using paired ttests. No corrections were made for multiple comparisons. A 2-sided value of p < 0.05 was considered significant. All values depicted in bar graphs are indicated as mean with standard error of the mean.
Abnormal T–U waves precede TdP
Clinical data are shown in Table 1.Figure 3A,a single-monitor lead, shows giant, abnormal T–U waves directly preceding a TdP episode. In Figure 3B, a 12-lead ECG, the TdP-initiating beat arises from the end of the giant T–U-wave complex with deep inverted T–U waves (even when the R-wave is mostly positive). The TdP-initiating beat arising from the abnormal T–U shows a slow rise velocity and wide QRS complex. The amplitude of the T–U-wave preceding TdP was more than 3 times larger in LQTS patients with TdP compared with the largest T–U-wave in LQTS patients without TdP (Fig. 4A).Control patients had no U waves or only very small ones (Figs. 2A and 4A). The T–U-wave ratios also were higher before TdP than in other recordings (Figs. 4B and 4C) (ratio of T–Uprecdivided by T–Uany= 1.8 ± 0.4, p < 0.05 vs. both control groups). The T–U-wave preceding PVCs in patients with a normal QT interval was not different from other T–U waves (T–Uprec/T–Uany= 1.0 ± 0.3). The T–U-wave before a PVC without TdP in LQTS patients was even smaller than other T–U waves in the same ECG recording (T–Uprec/T–Uany= 0.4 ± 0.1) (Fig. 4C).
QRS duration and rise angle
One of our hypotheses was that a premature beat taking off from a U-wave (an EAD at the tissue level) should have a lower action potential upstroke velocity and thus translate into a slower initial QRS rise (or descent) angle, as well as a longer QRS duration. Indeed, QRS duration of the first TdP beat was longer, and QRS angle lower, before TdP than that of PVCs in patients with LQTS or in patients with a normal QT interval (Fig. 5).
Pause dependency of TdP?
The RR intervals and QT intervals preceding either TdP or PVC were longer in LQTS patients than RR intervals preceding PVCs in patients with other heart disease. Of note, neither RR interval nor QT interval were more prolonged before TdP than RR interval or QT interval before PVCs not inducing TdP in LQTS patients (Fig. 6).
Our ECG analysis in a prospectively collected ECG database of LQTS patients identified several ECG characteristics before imminent TdP: 1) Giant T–U waves directly precede TdP. The first TdP beat emerges from an abnormal T–U-wave. Abnormal T–U waves are larger than any other repolarizing wave in the available ECG recording and are not found before other types of PVCs. 2) The QRS duration of the first TdP beat was longer and the QRS angle was lower compared with QRS duration of other PVCs. Although it may seem evident to some, the first finding has never been analyzed systematically. The second finding is, to the best of our knowledge, novel.
Giant T–U waves initiate TdP
Our analysis suggests that the T–U-wave plays a critical role in the precipitation of TdP. A marked increase in T–U-wave amplitude (3-fold higher amplitude compared with ECGs without TdP, 80% increase compared with the largest repolarizing wave in the entire ECG recording) was specific for imminent TdP. Often, the blinded analyzers could not differentiate between T- and U-wave in the TdP recordings. Therefore, we chose the term T–U-wave for this phenomenon that has not been systematically studied before. We believe that both an increase in T-wave amplitude and the appearance of a closely timed abnormal U-wave added up to create giant T–U waves. Abnormal U waves have been appreciated by many clinicians and investigators before (5). A similar ECG phenomenon was described as post-extrasystolic U-wave augmentation in patients who survived ventricular fibrillation in a prior ECG analysis (11). To the best of our knowledge, this is the first systematic quantification of giant T–U waves directly before TdP.
QT interval prolongation and TdP
The LQTS patients with a prominent prolongation of the QT interval are prone to TdP (2,12). This was confirmed in our analysis. Interestingly, the degree of QT interval prolongation in LQTS patients was not different between ECGs with TdP and ECGs with PVCs but without TdP. Hence, prolongation of the QT interval did not identify an imminent TdP episode.
Facilitation of TdP onset by a preceding pause has been recognized previously (5,9,13). In this study, the initiation of a TdP episode was preceded by a longer RR interval than the prior beats, consistent with a previous report of the pause dependency of TdP in LQTS2 that had ECG traces that in part (some of the c-LQTS patients) overlapped with the ECGs used in this study (3). Interestingly, the RR interval was equally long before PVCs not initiating TdP and before the first TdP beat in the LQTS patient ECGs in this study (Fig. 6). Pause dependency therefore does not discriminate imminent TdP from other types of PVCs in this set of LQTS patients.
Normal U waves, abnormal U waves, and abnormal T–U waves before TdP
Different types of U waves may have different relevance (5,14). Small, orthotopic U waves are a normal variant in young adults, especially in the precordial leads. These normal U waves may reflect intrinsic potential differences in the terminal part of the action potential (14) or mechanoelectrical feedback with a prolonging effect on late myocardial repolarization (15).
Abnormal U waves, for example, those found in myocardial ischemia or left ventricular hypertrophy, are less well separated from the T-wave, and often show reversed polarity compared with the T-wave (5,16). These abnormal U waves may be caused by either (but not limited to) EADs, regional contractile dysfunction and subsequent stretch-induced depolarizations, regional inhomogeneities in repolarization (e.g., during regional acute ischemia), or spontaneous activity in the Purkinje network.
Abnormal intracellular calcium release and a subsequent increased activity of the sodium–calcium exchanger may trigger early-coupled depolarizations (17–20). Given the largely epicardial potentials that are recorded in the surface ECG (21), giant T–U waves are unlikely to originate from the Purkinje network. Long QRS durations of the first TdP beat in this study also suggest an origin of the first TdP beat distant from the specialized conduction system.
Giant T–U waves may trigger TdP in LQTS: a hypothetical mechanism
Occasional invasive electrophysiological recordings in patients with TdP have found that EADs correspond to abnormal T–U waves at the myocardial tissue level (10,22–24). The EADs are most likely a regional phenomenon (4,25–30), hence explaining why EADs were not found in all patients. We suggest that abnormal T–U waves on the surface ECG reflect regional EADs, supported by their exclusive presence before TdP. Initiation of TdP by EADs, that is, by a slowly rising activation wave that arises in an incompletely repolarized region of the heart, is supported by the, albeit indirect, finding that QRS duration is long and QRS angle is small in the first beat of TdP.
Although we had access to a sizeable number of ECG recordings during TdP, our analysis was confined to the available ECG recordings, often monitor strips of 2-lead ECGs and occasionally 12-lead ECGs. We could not analyze longer periods (minutes to hours) before TdP. Furthermore, we did not study subtle beat-to-beat changes in the QT interval in these ECGs. Nonetheless, the T–U-wave that preceded TdP was markedly higher in amplitude than any other T–U- or T-wave found in the TdP recordings or in recordings of ECGs with PVC, either in LQTS or in control patients. Published data suggest that EADs are the underlying myocardial electric event. We cannot conclude on mechanisms of EADs nor even prove that EADs are indeed the biological event reflected by abnormal T–U waves in this study. Because of the early take-off of the first TdP beat, we could not measure the full extent and duration of the last T–U-wave that triggered the TdP episode.
The onset of TdP is linked to abnormal giant T–U waves. Abnormal T–U waves and a slow QRS upstroke separate initiation of TdP from early PVCs in other heart diseases and in LQTS. Abnormal T–U waves support the notion that EADs are the trigger for TdP in LQTS. If found, they may be an indicator for imminent risk of TdP.
For supplementary Tables 1 through 3containing individual data, please see the online version of this article.
This study was funded in part by Deutsche Forschungsgemeinschaft (DFG, Ki/713/1-1), by the German Ministry for Research and Education (BMBF, AFNET, 01Gi0204), and by Fondation LeDucq (Alliance Against Sudden Cardiac Death and ENAFRA). Dr. Kirchhof has received consulting fees or honoraria from 3M Medica, AstraZeneca, Bayer Healthcare, Boehringer Ingelheim, MEDA Pharma, Medtronic, Sanofi-Aventis, Siemens, Sorvier, and Takeda; and research grants from Cardiovascular Therapeutics, 3M Medica/MEDA Pharma, Medtronic, Omron, and St. Jude Medical. Drs. Kirchhof and Franz contributed equally to this work.
- Abbreviations and Acronyms
- acquired long QT syndrome
- congenital long QT syndrome
- early afterdepolarization
- long QT syndrome
- premature ventricular complex
- torsades de pointes
- Received November 25, 2008.
- Revision received March 5, 2009.
- Accepted March 10, 2009.
- American College of Cardiology Foundation
- Kääb S.,
- Hinterseer M.,
- Näbauer M.,
- Steinbeck G.
- Monnig G.,
- Eckardt L.,
- Wedekind H.,
- et al.
- Tan H.L.,
- Bardai A.,
- Shimizu W.,
- et al.
- Fabritz L.,
- Kirchhof P.,
- Franz M.R.,
- et al.
- Gbadebo T.D.,
- Trimble R.W.,
- Khoo M.S.,
- et al.
- Viskin S.,
- Alla S.R.,
- Barron H.V.,
- et al.
- Viskin S.,
- Heller K.,
- Barron H.V.,
- et al.
- Locati E.H.,
- Maison-Blanche P.,
- Dejode P.,
- Cauchemez B.,
- Coumel P.
- Surawicz B.
- Ter Keurs H.E.,
- Boyden P.A.
- Pogwizd S.M.,
- Qi M.,
- Yuan W.,
- Samarel A.M.,
- Bers D.M.
- Ozdemir S.,
- Bito V.,
- Holemans P.,
- et al.
- Rubart M.,
- Pressler M.L.,
- Pride H.P.,
- Zipes D.P.
- Vos M.A.,
- Verduyn S.C.,
- Gorgels A.P.,
- Lipcsei G.C.,
- Wellens H.J.
- Shimizu W.,
- Ohe T.,
- Kurita T.,
- et al.