Usefulness of exercise test in the diagnosis of short QT syndrome

Abstract

Introduction

Short QT syndrome (SQTS) is a rare congenital ion channel disease with autosomal dominant inheritance, characterized by an abnormally short QT interval on the surface electrocardiogram (ECG) and an increased susceptibility to life-threatening arrhythmias. Clinical presentation is heterogeneous; in the series of 53 patients published from our group in 2011¹ over 60% of the subjects presented with symptoms: the most frequent was cardiac arrest (CA), which represented the first clinical manifestation in one-third of the patients. Syncope or palpitations, often with evidence of atrial fibrillation (AF) represent other common clinical findings.

The diagnosis of SQTS is based on the detection of a constantly short QT interval on ECG; however, it is unclear which is the highest value of QT compatible with the syndrome. In the first described cases,^2,3 the QTc interval never exceeded 300 ms; later, subjects with QTc intervals of 340 ms have been reported in the literature.^4,5 More recently, a QTc interval up to 360 ms has been associated with CA.¹ From the analysis of several population studies that evaluated the QTc interval distribution in the general population, Viskin⁶ suggested that values less than 360 ms in men and 370 ms in women are relatively infrequent and thus should be considered as abnormally short. However, in those studies borderline or slightly shortened QT values were not associated with arrhythmic events. In contrast, Miyamoto et al.⁷ recently analysed a hospital-based population of 105 824 individuals, and reported two subjects with documented ventricular fibrillation and syncope associated with QTc intervals of 332 and 355 ms; however, the patients also displayed, respectively, early repolarization and Brugada ECG pattern.

Owing to the overlapping range of QT intervals in affected individuals and general population, it seems improbable that a single QTc value can distinguish all cases of SQTS from healthy individuals. This could make the diagnosis of SQTS sometimes difficult, while it is vitally important to distinguish between subjects with a borderline short QT interval and those with true SQTS, due to the malignant potential of the disease.

In the first described SQTS cases a characteristic QT interval behaviour during exercise was observed, with only a slight shortening during heart rate (HR) increase;³ later Wolpert et al.⁸ reported that the QT peak/HR correlation was much weaker in three SQTS patients than in a control group.

The aim of this study was to evaluate the QT interval behaviour during exercise in a cohort of SQTS patients, and to compare it with that of a control group from the general population, in order to evaluate the usefulness of exercise testing in the diagnosis of the syndrome.

Methods

Study population

Twenty-one subjects with diagnosis of SQTS from five European Centers were included in the study (16 males, 76%). Inclusion criteria comprised QTc values below 360 ms (or 88% of QTp) in patients symptomatic for aborted SD or syncope of arrhythmic origin or with family history of SQTS. Patients with QTc values below 340 ms were included even if asymptomatic. Almost all the patients had a high probability of having SQTS according to the recently proposed Gollob's score,⁹ with a total score ≥ 4 points; only one patient resulted in a low probability of having SQTS with only 2 points (Table 1).

Three subjects came to clinical observation with aborted CA, other three for a history of syncope and six for palpitations, in some cases with evidence of AF (Table 1). Mean QTc interval at rest was 306 ± 23 ms (range 268–345 ms). In eight subjects from four unrelated families a mutation in KCNH2 gene was found, while genetic screening revealed no known mutation in genes encoding for potassium channels (KCNH2, KCNQ1, KCNJ2) in the others. The mean age at the time of exercise test was 31 ± 15 years. Most of the subjects, 17, have been cited in previous studies.^1,3–5

Twenty healthy subjects with QTc values ≤ 390 ms and normal 12 lead ECG at rest (no evidence of ventricular hypertrophy, intraventricular conduction defects or repolarization phase alterations), were recruited as control group. All of them were asymptomatic, without clinical evidence of cardiovascular disease. Mean QTc value was 375 ± 12 ms (range 354–390 ms).

None was receiving any medication at the time of exercise test.

QT interval measurement and exercise protocol

All subjects performed an exercise stress test: depending on the Center, treadmill (Bruce protocols) or bicycle ergometer were used. All the patients gave their informed written consent to the test. Twelve-lead ECG was monitored throughout the entire test and recorded at the end of each stage at a paper speed of 25 mm/s; blood pressure was measured at appropriate intervals. QT intervals were measured manually from the precordial lead with the highest T wave amplitude (usually V2 or V3) by two independent examiners, both at rest and during effort up to peak exercise. Three intermediate HR values (HR25%-HR50%-HR75%) were obtained between rest and peak. As the study was mainly retrospective and the five HR points (HR at rest, HR25%-HR50%-HR75%, and HR at peak) were not available for all the patients, a minimum of three HR points was required to include the test. During the recovery phase, the QT interval was measured at HR similar to those considered during exercise; only HR75% and HR50% were considered in the analysis, as data at lower HRs were not available for most patients.

When the ECGs were available, the QT interval was measured both in supine and in upright position. The QT interval was measured from the onset of the first QRS deflection to the end of the T wave according to the tangential method,¹⁰ as illustrated in Figure 1.

Figure 1

QT interval measurement according to the tangential method. (A) 12 lead ECG during stress test of a 17-year-old male patient with short QT syndrome. (B) The QT interval was measured in lead V2 from the onset of the first QRS deflection to the end of the T wave according to the tangential method.

Open in new tabDownload slide

Statistical analysis

The relation between QT interval and HR was evaluated by linear regression analysis according to the formula: QT = β×HR + α, where QT is the uncorrected QT interval in milliseconds, HR is the HR in beats/min, β is the slope of the linear relationship, and α is the intercept expressed in milliseconds. Regression was calculated from the data points acquired from rest to peak exercise as illustrated above.

We used multilevel mixed-effects linear regression models accounting for QT and HR time-dependent variables, with a random intercept and a random coefficient for modelling heterogeneity between individuals and an unstructured variance-covariance structure. All regression analyses were also multivariable and adjusted for age and sex. All analyses were performed using Stata version 12 and used 2-sided tests for significance at the 0.05 level, with 95% confidence interval (CI).

Results

Electrocardiographic findings, at rest and during effort, as well as regression analysis results of the QT/HR relationship for affected subjects and controls are listed in Table 2. In the exercise phase, in seven patients and five control subjects data points for regression analysis were fewer than 5 (4 measurements in 11 subjects and 3 in one patient), due to the lack of recordings at all the desired HRs.

There were no differences between the two groups in rest and peak effort HRs. Short QT syndrome patients showed a lower adaptation of the QT interval at increasing HRs: the mean variation of the QT interval from rest to peak effort was 48 ± 14 ms, compared with a mean of 120 ± 20 ms in the control group, P < 0.0001 (Figure 2, left). As a result, regression analysis of the QT/HR relationship revealed a less steep slope for SQTS patients as compared with controls (−0.53 ± 0.15 vs. −1.29 ± 0.30, P value < 0.0001), which never exceeded the value of −0.90 ms/beat/min (range −0.33; −0.90) (Figure 3), while in the control group the slope was always over −1.0 ms/beat/min (range −1.05; −1.72). This holds true also for patient 20 (see Table 1), who had reached a total of only 2 points in Gollob's score, resulting in a low probability of having SQTS, but showed a slope of −0.74 ms/beat/min. When the analysis was repeated also considering the time from rest to peak HR, the slope for SQTS patients was −0.45 ± 0.86 and for controls −1.27 ± 0.06, P < 0.0001.

Figure 2

Heart rate/QT relationship in SQTS (red line) and controls (turquoise line) during exercise (left) and during recovery (right). Short QT syndrome patients showed a lower adaptation of the QT interval at increasing heart rates. QT points represent the mean values at 25%, 50%, 75%, and 100% of the HR range from baseline to peak exercise and back to 75% and 50% of the HR during the recovery phase. Bars indicate 95% confidence intervals, HR = heart rate.

Open in new tabDownload slide

Figure 3

Regression analysis between QT interval and heart rate during exercise in SQTS patients (A) as compared with the control group (B). Values on the vertical axis represent the predicted QT following the relation QT = β×HR + α. HR = heart rate.

Open in new tabDownload slide

Data of the recovery phase were available for 14 SQTS and 18 controls (Table 3). The mean variation of the QT interval from peak effort to HR50% was 23 ± 17 ms, compared with a mean of 43 ± 20 ms in the control group, P < 0.01 (Figure 2, right and Figure 4). Also in this phase SQTS subjects displayed a flatter QT/HR slope (mean value 0.50 ± 0.19 vs. 0.97 ± 0.35, P < 0.0001). When considering in the multilevel regression analysis the time from peak to HR50%, the slope for SQTS patients was 0.52 ± 0.11 and for controls 1.14 ± 0.12, P < 0.0001.

Figure 4

Regression analysis between QT interval and heart rate during recovery in SQTS patients (A) as compared with the control group (B). Values on the vertical axis represent the predicted QT following the relation QT = β×HR + α. HR = heart rate.

Open in new tabDownload slide

The analysis was repeated excluding the eight SQTS patients with KCNH2 mutations, who were those with the shortest QT and QTc intervals, both during exercise (mean QT variation from rest to peak exercise 51 ± 16 ms; mean slope −0.55 ± 0.14, P < 0.0001) and in the recovery phase (mean QT variation from peak HR to HR50% 27 ± 18 ms, mean slope 0.58 ± 16, P < 0.0001).

In a multilevel analysis controlled for age and sex, we found that age did not influence the QT/HR slope during exercise. In the SQTS group, female subjects exhibited steeper slopes than males (β = −0.58, 95%CI: −0.46, −0.70, vs. β = −0.52, 95%C.I:−0.43, −0.60, P = 0.02). To analyse whether KCNH2 carriers and patients with unknown genotype had a different behaviour, we repeated the analysis not considering the female subjects, as all of them had KCNH2 mutation. Males with KCNH2 mutation showed a flatter slope as compared with those with unknown genotype: in the first group β was –0.36 (95% CI −0.25, −0.47) vs β = −0.55 in the second (95% CI −0.45, −0.62, P = 0.02).

In our study, we observed six patients with a history of cardiac arrest (n = 3) or syncope (n = 3). In a multilevel regression model adjusted also for age and sex, a previous occurrence of CA or syncope didn't show a significant influence on QT slope (P = 0.78).

In 10 SQTS patients in which rest ECGs were available both in supine and standing position, mean HR increased of 15 ± 7 bpm, mean QT and QTc variation was −10 ± 8 ms and 15 ± 14 ms, respectively.

Discussion

Several studies have analysed the QT and QTc behaviour during exercise test in healthy individuals from the general population. In 1996 Kligfield et al.¹¹ observed that the unadjusted QT interval varies linearly with HR, and for this reason the determination of QT/HR slope, quantifying the QT adaptation to HR, could represent an easy method of representing their relationship during exercise. In a paper by Magnano et al.¹² the autonomic nervous system influence on repolarization was evaluated, investigating the relationship between HR and QT interval during exercise and during atropine or isoproterenol infusion. In these two studies, the mean QT/HR slopes during exercise in control subjects from the general population were, respectively, −1.45 ± 0.34 and −1.29 ± 0.23 ms/beats/min in men and −1.74 ± 0.32 and −1.43 ± 0.21 ms/beats/min in women.

In 1991, Vincent et al.¹³ reported abnormal QT responses to exercise in patients with type 1 LQTS as compared with normal subjects; Swan et al.¹⁴ observed that during the recovery phase after exercise the QT/HR slopes were steeper in children with long QT syndrome than in controls. It has also been reported that the adaptation of the QT interval is different in the three major LQTS subtypes, suggesting that it could be helpful in predicting and directing the genetic testing in affected patients,¹⁵ and in distinguishing asymptomatic LQTS mutation carriers with no or borderline QT lengthening from normal subjects.^16,17

Isolated reports have reported the behaviour of the QT interval during exercise in a limited number of SQTS cases. Wolpert et al.⁸ analysed the rate dependence of the QT interval during exercise in three SQTS patients with a mutation in KCNH2 gene and found a weak correlation between HR and QT peak interval. Moreover, they found that the administration of quinidine partially restored the HR dependence of the QT peak interval toward the range of adaptation reported for normal subjects. Antzelevitch et al.¹⁸ reported QT/HR slopes from −0.54 to −0.99 ms/beats/min in three subjects carrying a mutation in CACNA1C or CACNB2b genes, which are responsible for a mixed phenotype of Brugada pattern ECG and shorter than normal QT intervals. Sun et al.¹⁹ reported a similar behaviour in two subjects with a novel mutation in the KCNH2 gene, with QT/HR slopes of −0.55 and −0.74 ms/bpm.

Currently, the diagnosis of SQTS is based on the finding of a constantly short QT interval on 12 lead ECG. Our group in 2011 published the long-term follow-up of 53 SQTS patients.¹ Inclusion criteria comprised a QTc interval ≤ 340 ms even in the absence of symptoms, or a QTc interval up to 360 ms in association with a history of cardiac arrest or syncope of arrhythmic origin, or belonging to SQTS families. A scoring system based on electrocardiographic, clinical, and genetic parameters has been shortly after proposed in order to define the probability of having SQTS.⁹ The analysis of the QT/HR relationship was not included in the variables considered in that study, due to the absence of validation of the above-mentioned observations in a large cohort of patients.

In consideration of the known limitations of the Bazett's formula at HRs >100 bpm, during exercise the evaluation of the relationship between uncorrected QT values and HR seems preferable. Our data on 21 SQTS patients compared with controls show a clear difference in QT adaptation during exercise, because the QT interval in the affected subjects does not shorten as expected with increasing HR. As a consequence, the regression lines show that the QT intervals in the two groups come closer to each other at the highest HR (Figure 2A).

Although a less dynamic QT adaptation to HR has been observed also in healthy subjects,¹¹ values of the QT/HR slope less than −0.9 ms/beat/min have not been reported. As a consequence, in subjects with QTc intervals between 340 and 360 ms, the presence of a QT/HR relationship slope under −0.9 ms/beat/min could help distinguish affected subjects from healthy individuals. Using this value as cut-off, we were able to recognize a subject with short QT interval in several ECG recordings, that would have not been considered as affected based on Gollob's score, but who exhibited values of QT/HR relationship that, as far as we know, are not common in the general population.

Other findings in this study are that the QT/HR slope is significantly flatter in the subgroup with a KCNH2 mutation, which shows the shortest QT intervals, as compared with the subgroup with unknown genotype. This analysis was performed only in male patients as all the female subjects had KCNH2 mutation. The slope is steeper in female, as compared with male SQTS patients, similarly to what has been reported for normal subjects.¹¹

Also in the recovery phase the QT/HR slope is flatter in the SQTS patients than in the controls.

The response of the QT interval to brisk standing has been evaluated in subjects with LQTS,²⁰ demonstrating an impaired adaptation of the QT to the acceleration of the HR. Our data suggest that also in SQTS the QT adaptation to standing is reduced, as compared with the values for a normal population reported by Viskin et al.

Study limitations

Owing to the mostly retrospective nature of this study, the exercise protocol was not standardized and the ECG data, especially during the recovery phase, were not complete for all patients, although sufficient for the analysis. The evaluation of the reproducibility of our observations is affected by the fact that most patients began a pharmacological treatment.

The preliminary observation of an altered response of QT interval to postural changes in SQTS patients needs to be further validated in a larger controlled study.

Conflict of interest: None declared.

References