Article Text

Download PDFPDF
Original article
Defective recovery of QT dispersion predicts late cardiac mortality after percutaneous coronary intervention
  1. Marco Zimarino,
  2. Alessandro Corazzini,
  3. Alfonso Tatasciore,
  4. Stefania Marazia,
  5. Giuseppe Torge,
  6. Cesare Di Iorio,
  7. Raffaele De Caterina
  1. Institute of Cardiology and Center of Excellence on Ageing, “G. d'Annunzio” University, Chieti, Italy
  1. Correspondence to Professor Raffaele de Caterina, Institute of Cardiology and Center of Excellence on Ageing, “G. d'Annunzio” University, Chieti, Ospedale SS. Annunziata, Via dei Vestini, 30, Chieti 66013, Italy; rdecater{at}


Background It has been suggested that corrected QT dispersion (cQTD) provides a measure of repolarisation inhomogeneity; however, the existence of a relationship between cQTD and cardiac outcomes is controversial.

Objective To assess whether changes in cQTD following percutaneous coronary intervention (PCI) predict long-term survival.

Design Prospective observational study.

Setting Single tertiary care centre.

Main outcome measures Cardiac mortality.

Patients 612 patients had a 12-lead ECG recorded before and 6 h after PCI, and were followed-up for 49±10 months.

Results PCI was associated with a significant overall reduction of cQTD at 6 h versus baseline (p<0.001); a reduction in cQTD occurred in 343 patients (56%). During the follow-up, 46 deaths (7.5%) were recorded, 21 of which for non-cardiac and 25 for cardiac causes. At Cox regression analysis, a reduced ΔcQTD (cQTD baseline − 6 h) was an independent predictor of cardiac mortality (HR=1.497; 95% CI 1.081 to 2.075 for each 20 ms decrease; p=0.015), together with age (HR=1.672; 95% CI 1.039 to 2.691 per 10 years increase; p=0.034), diabetes (HR=2.622; 95% CI 1.112 to 6.184; p=0.028), peak CK-MB (HR=1.798; 95% CI 1.063 to 3.039 per each unit increase over normal level; p=0.029), three-vessel coronary artery disease (HR=3.626; 95% CI 1.079 to 12.187; p=0.037) and the number of treated lesions (HR=2.066; 95% CI 1.208 to 3.532; p=0.008). Patients in the lowest tertile of ΔcQTD and having a post-procedural increase of CK-MB had a considerably higher cardiac mortality than the remaining population (14.6 vs 2.4%, p<0.001).

Conclusions cQTD decreases after PCI. A defective cQTD recovery, suggesting the persistence of repolarisation inhomogeneities, predicts long-term cardiac mortality.

  • Percutaneous coronary interventions
  • QT dispersion
  • cardiac mortality
  • sudden cardiac death
  • sudden adult death syndrome
  • coronary angioplasty (PCI)

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The QT interval in the standard surface electrocardiogram (ECG) is a measure of ventricular repolarisation, with the 12 leads capturing information from different myocardial areas.1 It has been suggested that the interlead difference in QT, known as QT dispersion (QTD), provides a measure of repolarisation inhomogeneity.2

The existence of a relationship between QTD and cardiac outcome is controversial.3 4 Since QTD decreases after effective myocardial revascularisation,5 QTD changes, which are thought to reflect the efficacy of revascularisation, may predict outcomes after revascularisation.

We hypothesised that a defective QTD recovery after percutaneous coronary intervention (PCI) relates to long-term adverse cardiac outcomes. We have tested this hypothesis in a cohort of patients undergoing elective PCI.

Patients and methods

Study population

This was a prospective, observational study among patients undergoing successful elective PCI.

Patient disposition is detailed in figure 1: we screened 888 consecutive patients undergoing PCI at our institution between April 2001 and December 2003. Patients were required to have a 12-lead ECG recording in sinus rhythm and a reliable QT assessment (see below) at baseline, 6 and 18 h after PCI. All subjects with ECG findings of QRS ≥120 ms were excluded from the analysis.

Figure 1

Screening and assessment of the study population. NSTE-ACS, non-ST-elevation acute coronary syndrome; PCI, percutaneous coronary intervention; STEMI, ST-elevation myocardial infarction.

Blood samples to measure creatine kinase-MB (CK-MB) were collected before PCI, at 6, 12 and 18 h after PCI and, in the case of abnormal values, at 6 h intervals thereafter until normalisation. CK-MB (mass) levels were obtained using the Access-2 immunochemiluminescence method (Beckman Coulter, Brea, California, USA), with normal limits, defined as the 99th centile of a normal population, ≤4 ng/ml.

As a result of this screening procedure, 644 patients (73%) were discharged after PCI and 612 were followed-up.

In this population the indication for coronary angiography and subsequent myocardial revascularisation was based on the presence of unstable angina (n=108, 17%), effort angina (n=263, 41%), a recent (>48 h, <7 days) myocardial infarction (MI; n=118, 18%), and stress test-inducible ischaemia in the absence of symptoms (155 cases, 24%).

The study was approved by the local ethics committee and written informed consent was obtained from all patients.

Percutaneous coronary intervention

PCI was performed according to standard clinical practice, preferably with stent implantation. All patients were receiving aspirin (100 mg/day) and ticlopidine (250 mg twice daily) or clopidogrel—either with a loading dose of 300 mg just before PCI or started at 75 mg/day at least 3 days before the procedure and continued for 1 month thereafter in patients receiving a bare metal stent, and for at least 6 months in subjects receiving a drug-eluting stent—these becoming available at our institution since May 2002. Intravenous unfractionated heparin was administered with a 70 IU/kg bolus at the beginning of the procedure, followed by additional boluses to maintain an activated clotting time in the range of 250–300 s (200–250 s for patients receiving glycoprotein IIb/IIIa inhibitors). The use of glycoprotein IIb/IIIa inhibitors was left to the operator's discretion.

Patients were recorded as receiving statin treatment if it had been started at least 5 days before PCI.

PCI failure was defined as an angiographic residual stenosis >30% at the end of the procedure or the presence of a TIMI <3 grade flow.6

ECG measurements

A resting 12-lead ECG was obtained at 25 mm/s paper speed and 10 mm/mV amplitude on each participant before, at 6 h and 18 h after PCI. All ECGs were analysed and submitted for manual interval measurements performed with the aid of callipers by two experienced investigators (AC and AT), who were blinded to patients' clinical status, PCI outcomes and ECG timing.

The QT interval was measured manually in all 12 leads from the beginning of the QRS complex to the end of the T wave, defined as the return of its descending limb to baseline. When the T wave was followed by a U wave, the end of the T wave was defined as the nadir between the T and the U wave. If the T wave could not be reliably determined (for amplitudes <100 μV), QT measurements were not done, and these leads were excluded from the analysis. Inter- and intraobserver variabilities in our laboratory were 18±31 and 14±29 ms, respectively, as estimated in a first cohort of 118 patients who underwent PCI in the period September to December 20007; concordance between independent readings by the two operators (AC and AT) was also tested (r=0.91). QTD was calculated as the difference between the maximum (QTmax) and minimum QT (QTmin). QT, QTmax and QTmin intervals were corrected for the heart rate with the Bazett's formula8:


Corrected QTD (cQTD) was therefore calculated for each resting 12-lead ECG as cQTmax – cQTmin. The cQTD variation (ΔcQTD) was calculated as the difference between cQTD before PCI and cQTD at 6 h. The difference cQTD before PCI – cQTD at 18 h was also calculated.

QTD was also calculated using the Fridericia9 and Framingham10 formulae for QT correction, as follows:


The heart rate variation was calculated as the difference between both 6 and 18 h and baseline heart rate.


Patients' follow-up was performed by outpatient visits and exercise stress testing at 1 and 6 months. Patients were contacted every 6 months thereafter by a visit or through telephone interviews by trained investigators (AC, GT and SM), and, in cases of rehospitalisation, by reviewing the patient's hospital records. Patients who underwent further coronary revascularisation were assessed at the time of the new procedure.

When it was learnt by telephone call to relatives that death had occurred, the cause of death was investigated by interviewing the patient's relatives and the doctor signing the death certificate. Deaths were classified as either cardiac or non-cardiac11; cardiac deaths were further classified as sudden (presumed arrhythmic) and non-sudden.12 Sudden cardiac death was defined as death of cardiac origin that occurred unexpectedly within 1 h of new symptoms onset or that was unwitnessed and unexpected, unless a specific non-cardiac cause was confirmed. Non-sudden cardiac death was defined as death of cardiac origin but not within the sudden cardiac death definition. MI was diagnosed according to standard criteria: detection of elevations of cardiac biomarkers with signs or symptoms compatible with new-onset myocardial ischaemia, or newly developed Q waves, or imaging evidence of new loss of viable myocardium.13 Coronary angiography was performed only when ischaemia-driven. Target vessel revascularisation (TVR) was defined as revascularisation with either PCI or coronary artery bypass grafting performed on the same coronary vessel treated at the index procedure, within and beyond the target-lesion limits. Major adverse cardiac events (MACE) were defined as death (cardiac and non-cardiac), non-fatal MI and TVR.

The end of the follow-up period was 1 January 2008.

The vital status could not be determined for 32 subjects (4.9% of the original 644 subjects); their baseline and procedural characteristics were not, however, significantly different from those of the remaining 612 subjects studied in this analysis (data not shown).

Statistical analysis

We estimated that a sample size of 474 subjects could provide a study power of 90% to detect a 50% increase in the risk of cardiac mortality between the first and the third tertiles of ΔcQTD (158 subjects for each tertile), in a population with an estimated cardiac mortality of 2%/year, assuming a two-sided type I error of 0.05. With an expected dropout rate of 5%, we estimated that a population of 500 subjects with complete 3-year follow-up was necessary. After 2 years, the observed cardiac mortality was about 1.5%, lower than expected. We therefore increased the study population by 20% and prolonged the follow-up for 1 additional year to meet the above-mentioned criteria.

Data are expressed as mean±SD for normally distributed continuous variables, as median and range for non-normally distributed continuous variables and as counts and percentages for discrete variables. Normal distribution was tested by the one-sample Kolmogorov–Smirnov test. For non-normally distributed variables, a mathematical transformation was performed as appropriate.

Subjects were analysed for baseline demographic and procedural data according to vital status at follow-up (survivors, cardiac and non-cardiac deaths). To test for differences among groups of vital status, we used a one-way analysis of variance for continuous variables, and a χ2 or Fisher exact test for discrete variables, as appropriate. For post hoc multiple comparisons of continuous variables among groups, we set the threshold for significance at a preassigned probability value of 0.05/3 (Bonferroni's correction, therefore at 0.016).

Subjects were also divided into tertiles on the basis of ΔcQTD—that is, patients with an increased cQTD after PCI in the lowest tertile, and those with a decreased cQTD in the highest tertile; intertertile range of ΔcQTD was −2 and +17 ms.

Curves of cumulative incidence for the various outcomes were generated by the Kaplan–Meier method, and differences between curves of ΔcQTD tertiles were assessed by the log-rank test. Variables identified by univariate analysis as related (with at least a p<0.1) to cardiac death were further analysed by multivariable analysis: a forward stepwise multivariable Cox regression analysis was performed to identify independent correlates of cardiac death, as measured by the hazard ratio (HR) and its associated 95% CIs. Proportion hazards assumption was assessed by a log-minus-log survival plot.

Finally, we evaluated the predictive accuracy of the model for cardiac death by means of the area under the receiver operating characteristics curve for both ΔcQTD and peak CK-MB.

A probability (p) value <0.05 was considered significant, unless otherwise indicated. All analyses were performed with the SPSS release 15.0 software package (SPSS Inc).


Patient and procedural characteristics

Among the 612 patients recruited (median age 63 years, range 29–87), 46 deaths (7.5%) occurred during the follow-up: 25 deaths (4.1% of patients) were for cardiac causes, and 21 (3.4% of patients) for other causes. All cardiac deaths were ‘sudden’. Baseline and procedural characteristics of the study population according to the vital status are given in tables 1 and 2.

Table 1

Baseline characteristics and their association with vital status during follow-up

Table 2

Procedural characteristics and their association with vital status during follow-up

ECG findings

Findings from ECG recordings are summarised in table 3. In the overall population, QTmax and the RR interval decreased from baseline to 6 h after PCI; cQTmin and heart rate increased from baseline to 6 h after PCI; QTD and cQTD both decreased both from baseline to 6 h. A reduction of cQTD after PCI was documented in 343 patients (56%).

Table 3

ECG findings

QT-interval measurements at baseline, at 6 and 18 h were similar whether calculated using the Bazzet (reference formula), Fridericia (p=NS) or Framingham (p=NS) formulae.

Patients receiving (n=282, 46%) and not receiving concomitant β-blocker therapy (n=330, 54%) had similar heart rates at baseline (62±11 vs 62±12 bpm, p=NS), 6 h (64±12 vs 65±11 bpm, p=NS) and 18 h (64±11 vs 65±10, p=NS). ΔcQTD as measured at baseline – 6 h after PCI was similar among patients assuming (3±27 ms) and not assuming β blockers (6±26 ms, p=NS). This was also the case for ΔcQTD at baseline – 18 h after PCI (3±31 vs 4±28 ms in patients assuming and not assuming β blockers, respectively, p=NS).

All parameters investigated showed no change between measurements at 6 and 18 h. The ΔcQTD for this study was therefore calculated as cQTD measured at baseline − cQTD at 6 h after PCI.

Distribution of the vital status according to modifications of cQTD is shown in figure 2.

Figure 2

Distribution of vital status according to modification of QT dispersion. ∆cQTD, modification of 6 h corrected QT dispersion (cQTD) as compared with baseline value (in ms).


During a mean period of 49±10 months, 135 patients (22%) experienced a MACE: 46 patients died (7.5%); 67 subjects (11%) had a non-fatal MI and 69 (14%) a TVR. Among patients who died, 25 (4.1% of the overall population) had a sudden cardiac death, and 21 (3.4%) died owing to non-cardiac causes. No patient received a successful resuscitation for a cardiac arrest; no patient was reported to have had a non-sudden cardiac death. Times from index procedure to cardiac and non-cardiac death were 29±16 months and 31±12 months, respectively (p=NS).

Repeat angiography was performed in 132 patients (22%) at a mean interval of 10±12 months from the index PCI; 69 such patients underwent a TVR.

Results of Kaplan–Meier analysis of vital status and causes of death according to ΔcQTD are described in figure 3. There were no differences among tertiles of ΔcQTD in the incidence of overall mortality (figure 3A), non-cardiac mortality (figure 3C), MI (figure 3D), TVR (figure 3E) and MACE (figure 3F). Patients in the lowest (first) tertile of ΔcQTD—that is, patients who had an increase in cQTD from baseline to 6 h recording—had an increased risk of cardiac death (figure 3B), as compared with both the second and the third tertiles of ΔcQTD.

Figure 3

Kaplan–Meier curves of adverse events among tertiles of modification of corrected QT dispersion. Incidence of overall mortality (A), cardiac mortality (B), non-cardiac mortality (C), myocardial infarction (MI) (D), target vessel revascularisation (TVR) (E) and major adverse cardiac events (MACE) (F) according to tertiles of modification of 6 h cQTD as compared with the baseline value (∆cQTD).

Multivariable analysis

After adjusting for clinical, ECG and procedural variables, a reduced ΔcQTD—a defective recovery of cQTD at 6 h compared with baseline value—was an independent predictor of long-term cardiac mortality, together with age, diabetes, the increase in peak CK-MB, the number of treated lesions, and the presence of a three-vessel coronary artery disease (table 4).

Table 4

Clinical, angiographic and procedural multivariable predictors of long-term cardiac mortality

The area under the receiver operating characteristics curve for cardiac mortality was 0.712 for ΔcQTD, with an optimised cut-off point of 2 ms, and 0.645 for peak CK-MB, with an optimised cut-off point of >1× upper normal value.

When assessing long-term cardiac mortality according to aggregate ΔcQTD and peak CK-MB after PCI, patients in the first tertile of ΔcQTD and experiencing a post-procedural increase of CK-MB had a 4-year cardiac mortality significantly higher than that of the remaining population (figure 4).

Figure 4

4-Year cardiac death rate according to variation of corrected QT dispersion (∆cQTD) and peak CK-MB release after PCI. n, number of patients at risk; UNV, upper normal values.


In agreement with previous findings,14 we here show that myocardial revascularisation with PCI is associated with—and probably, by inference, causes—a reduction in cQTD, mainly through a reduction in QTmax. The novel finding of this study is that a defective cQTD recovery early after PCI is a strong and independent predictor of long-term cardiac mortality.

As expected from other studies on outcomes following PCI,15 subjects with advanced age, diabetes, larger periprocedural CK-MB release, a higher number of treated lesions and the presence of three-vessel coronary artery disease were here more likely to die owing to a cardiac cause in the long term. The relationship between ∆cQTD and cardiac death, however, was here independent of such other variables, and subjects who failed to recover cQTD 6 h after PCI—those in the lowest tertile of ΔcQTD—had a 50% higher risk of cardiac death than the remaining population.

QTD increases significantly during ischaemia,16 induced by either balloon inflation during angioplasty17 or stress testing,18 and decreases after successful reperfusion with thrombolysis19 or after myocardial revascularisation.14 However, clinical studies aiming at evaluating the prognostic value of QTD for adverse cardiac events have given controversial results.3 4 A major criticism of the QT interval is that most prognostic information on mortality seems to reside in the T-wave morphology.20 Another criticism may be the wide range of QT dispersion values, together with a complete absence of established reference values.21 The poor reproducibility of QTD with both manual and automatic measurements,22 as well as the variable morphology of the T wave largely contribute to the poor interobserver reproducibility. Therefore, contrary to initial expectations, and despite numerous attempts, a single measurement of QTD obtained at a certain time point in the natural history of heart disease did not prove to be a useful clinical tool. Our finding is substantially different from those attempts at predicting events on the basis of single measurements, because it is based on peri-PCI changes in QTD.

The increased dispersion of electrical recovery is a key factor in the development of serious and fatal arrhythmias,23 and ‘dynamic’ changes of related indices might be more useful than static “snapshop” assessments to predict the outcome of treatments. Impaired adaptation of the QT interval to changes in heart rate has been shown to predict sudden death in patients with chronic heart failure.24 More recently, a direct correlation has been documented between increased QTD after biventricular pacing and the occurrence of sudden cardiac death and/or resuscitation from a potentially fatal ventricular tachyarrhythmia in patients undergoing cardiac resynchronisation therapy.25 In this study, the association between defective QTD recovery and cardiac death was strong and—apart from a marginally increased risk of MI—neither dependent on progression of the disease nor on restenosis. Moreover, the documentation of a defective cQTD recovery proved to be predictive independently of, and additive to, CK-MB elevation—until now the single and most powerful determinant of long-term mortality after an otherwise angiographically judged “successful” PCI15: the combined presence of these two prognostic markers identified a subgroup of patients with a 4-year cardiac mortality about six times higher than in the remaining population.

Study limitations

We acknowledge a few limitations of this study. (1) The study is observational, and deals with a low-risk population of patients undergoing PCI. Apart from the extent of coronary artery disease and the number of treated vessels, no data have been collected on the adequacy and completeness of revascularisation that could affect both acute and long-term outcomes.26 Such results cannot be generalised to patients with acute coronary syndromes. (2) The size of the population is limited, but the relatively long follow-up yielded a number of cardiac deaths sufficient to achieve the intended statistical power in survival analyses. (3) ECG were analysed manually, and only manual measurements of the cQT were available. Manual measurement of cQT, however, has been judged to be at least as reproducible as automatic measurements.27 The availability of reliable automatic QT measurements in most current ECG instrumentations has, on the other hand, the potential to expand the applicability of the current findings. (4) ECG was recorded at 25 mm/s paper speed, and T-wave morphology was not specifically analysed; a higher paper speed would have allowed a more accurate analysis; similarly, we cannot exclude additional prognostic information deriving from a detailed evaluation of the T-wave morphology. (5) The influence of antiarrhythmic drugs on QTD was not specifically investigated; moreover, the use of antiarrhythmic drugs was recorded only during hospitalisation, and therefore the impact of patient adherence to treatment with antiarrhythmic or other life-saving drugs (antiplatelet agents, statins, ACE inhibitors) on the outcome cannot be assessed. (6) An echocardiogram was not systematically performed before discharge, and therefore no speculations can be done on the relationship between QTD and left ventricular ejection fraction modifications after PCI. (7) QTD was not specifically measured during follow-up visits, and therefore the association between long-term QTD changes and outcome was not tested. A defective 6 month recovery of QTD has been associated with increased risk of failed target-vessel patency after PCI,28 but in our population the association between post-PCI modification of QTD and cardiac mortality was independent of other adverse cardiovascular events. (8) Post-procedural troponin measurements were not systematically available, and therefore assessments of troponins were not included in this analysis. In the recent definition of myocardial infarction13 elevations of any cardiac biomarkers (including troponins) more than three times above the 99th centile of upper normal values after PCI has been suggested to identify periprocedural myocardial necrosis. However, the prognostic value of a post-procedural troponin elevation, independent of baseline values and of the increase in CK-MB, is at the moment, at best, uncertain.15


PCI has variable effects on cQTD, leading to its reduction in the majority of patients and to no change or an increase in others. Patients who fail to recover cQTD following PCI, suggesting the persistence of myocardial areas with repolarisation inhomogeneities, are at increased risk of long-term sudden cardiac death. The risk of cardiac death is sixfold higher among patients with an increase in both cQTD and peak CK-MB after PCI. The combined post-PCI increase of cQTD and peak CK-MB might be used to risk-stratify patients who would benefit from more aggressive preventive strategies.



  • Funding Italian Ministry of the University—to RDC—“G. d'Annunzio” University.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the ethics committee, The University of Chieti, Chieti, Italy.

  • Provenance and peer review Not commissioned; externally peer reviewed.