Lengthening of corrected QT during epileptic seizures

Authors


Address correspondence to Ruth Brotherstone, Department of Clinical Neurophysiology, Western General Hospital, Crewe Road, Edinburgh EH4 2XU, U.K. E-mail: ruth.brotherstone@luht.scot.nhs.uk

Summary

Purpose: To measure the corrected QT cardiac repolarization time before and during epileptic seizures.

Methods: Thirty-nine video-EEG/ECG/SAO2 (electroencephalography/electrocardiography/oxygen saturation) telemetry patients were included in this prospective study. Epileptic seizures were identified both clinically and electrographically. RR intervals and associated QT intervals were measured 5 min prior to the onset of the identified seizure. Consecutive RR and associated QT intervals were then measured from the seizure onset until the seizure had ended and the EEG had resumed its preseizure trace. Averaged RR and QT intervals over nine consecutive beats were applied to Bazett’s, Hodge’s, Fridericia’s, and Framingham’s formulas to compare the corrected QT values before and during the seizures.

Results: A total of 156 seizures had corrected QT analysis performed. Nine generalized tonic–clonic seizures (5 patients), 34 absences (6 patients), 12 tonic seizures (6 patients), 27 temporal lobe seizures (14 patients), 58 frontal lobe seizures (4 patients), and 16 subclinical seizures (4 patients). All formulae reported a statistically significant difference in corrected QT (p < 0.001) during total seizure data compared to total preseizure values. According to Bazett’s formula, 21 seizures (nine patients) transiently increased their corrected QT beyond normal limits, with a maximum corrected QT of 512 ms during a right temporal lobe seizure.

Conclusion: Significant lengthening of corrected QT cardiac repolarization time occurred during some epileptic seizures in this study. Prolonged corrected QT may have a role in sudden unexplained death in epilepsy (SUDEP).

The diagnosis of epilepsy carries an excess mortality that is 2–3 times higher than that of the general population (Cockerall et al., 1994; Tavernor et al., 1996). The National Sentinel of Epilepsy-Related Deaths Report 2002 (Hanna et al., 2002) states that almost 1,000 deaths are attributed to epilepsy in the United Kingdom each year, of which 60% are believed to be sudden unexplained death in epilepsy (SUDEP). In chronic epilepsy, SUDEP is the main cause of excess mortality, and risk of premature death from SUDEP in individuals with severe intractable epilepsy is 1:200 person years (Nashef et al., 1995).

SUDEP has been defined as “sudden, unexpected, witnessed or unwitnessed, nontraumatic and nondrowning death in patients with epilepsy, with or without evidence of a seizure and excluding documented status epilepticus, where postmortem examination does not reveal a toxicological or anatomical cause of death” (Nashef, 1997).

SUDEP has been recognized for more than a 100 years, but there still remains little evidence to explain the mechanisms involved (Nashef, 2000), although several hypotheses have been postulated mainly concerning respiratory, cardiac, and genetic involvement (Nashef et al., 2007). Autonomic symptoms frequently occur during epileptic seizures and range from subtle to life-threatening manifestations (Baumgartner et al., 2001). A study of patients with severe epilepsy and learning difficulties showed that 60% had abnormalities on electrocardiography (ECG), raising the question whether underlying cardiac rhythm disturbance increases the risk of SUDEP (Jeffrey, 2006).

Cardiac arrhythmias are considered to have a possible role in SUDEP (Tavernor et al., 1996; Kwon et al., 2003). An increase in repolarization time can lead to cardiac arrhythmias with any combination of bradycardia, atrial fibrillation, hypokalemia, hypomagnesemia, hypoxia, congestive heart failure, or ischemia, resulting in an induced polymorphic ventricular tachycardia (Morganroth & Pyper, 2001). Circulating adrenaline can produce lengthening of the corrected QT by a direct effect on the myocardium (Lee et al., 2003). In addition, some seizures are associated with various levels of hypoxia and ischemic changes, potentially contributing to a prolonged corrected QT, cardiac arrhythmia, and sudden death.

Measurements of the corrected QT during interictal electroencephalography (EEG) discharges indicated that the corrected QT (QTc) increased during interictal epileptic discharges, and it was proposed that a prolonged QTc could be a possible risk factor for SUDEP in patients with epilepsy (Tavernor et al., 1996). One previous study has examined the corrected QT during epileptic seizures (Nei et al., 2000). This study, however, only considered using a total mean average for group data and did not analyze individual seizures in each subject. They reported that no corrected QT lengthening occurred during seizures based on a grouped mean average. When total data is averaged for all patients in this study a normal mean value of 427 ms is calculated. However, we hypothesize that when data are analyzed for individual seizures in each subject, that lengthening of the corrected QT can occur during some epileptic seizures and consider that this may be one of the possible intrinsic mechanisms involved in SUDEP.

Methods

Thirty-nine consenting patients, most of whom had previously experienced video-EEG/ECG/SAO2 (oxygen saturation) telemetry took part in the ethically approved prospective study (LREC/2003/6/22). Patients diagnosed with any seizure type were accepted as part of the inclusion criterion. No restrictions on data inclusion were made due to patient pharmacology. The study design and sample size made it inappropriate to categorize and analyze data in terms of any perceived effects of antiepileptic drugs.

All patients were in sinus rhythm and none demonstrated a conductance disturbance during the preseizure resting trace. The study was inclusive for all ages except for neonates. The predicted observer error of QT measurement was considered to be too high to reliably measure the corrected QT in this group because of the high heart rates in neonatal tachycardia. Pediatric age in this study is defined as 1 month to 16 years. Data resulting in excessive electromyography (EMG) artifact obscuring the ECG signal were excluded from the study.

Lead II (Tavernor et al., 1996; Benatar & Decraene, 2001; Pater, 2005) ECG was recorded from disposable ECG electrodes and 23 silver silver-chloride electroencephalographic electrodes were applied in accordance to the International 10:20 system (Cooper et al., 1980). All wires were tied together to minimize electrical noise pick-up from the ward environment and to provide comfort and mobility for the patient. Continuously integrated Masimo oxygen saturation monitoring and ECG signals were networked via the EEG head-box to a main computer server with simultaneous video of the patient. Electrographic physiological signals were sampled at 500 Hz (Rijnbeek et al., 2001) and displayed on review using a maximum of 32 channels depending on selected montage on full screen of 12 s display using a chart speed of 30 m/ms.

Seizures and subclinical seizures were classified on the basis of clinical semiology from video evidence and simultaneous EEG changes. If the EEG showed changes prior to the clinical onset, the start of the seizure was measured at the start of the electrographic change. If the clinical manifestations appeared prior to the electrical seizure, then the start of the seizure was taken at the start of the clinical onset. The data were annotated to the nearest second.

The QT interval by definition represents the time required for depolarization and repolarization of the ventricular musculature (Yu et al., 1950). QT intervals were measured manually (Benatar & Decraene, 2001; Morganroth & Pyper, 2001; Malik, 2004; Fossa et al., 2005; Pater, 2005) from lead II. Consecutive measurements of QT intervals and their associated RR intervals were measured over a 9-s epoch (Tavernor et al., 1996; Benatar & Decraene, 2001; Luo et al., 2004). The QT interval was measured at the start of the QRS complex to the end of the T-wave, defined as the intersection of isoelectric line and the tangent of maximal downward limb of the T-wave (Lepeschken & Surawicz, 1952; Surawicz, 1998; Benatar & Decraene, 2001), and entered onto a Microsoft Excel spreadsheet (Microsoft Corp. Redmond, WA, U.S.A.). Careful measurement of the T-wave did not include the U-wave as the final QT measurement (Bazett, 1920; Lepeschken & Surawicz, 1952; Surawicz, 1998). The mean average values of nine consecutive RR and QT intervals were calculated over a 9-s epoch to average out the effect of normal sinus arrhythmia (Benatar & Decraene, 2001; Luo et al., 2004; Malik, 2004).

Corrective formulas were applied to each consecutive 9-s epoch from Excel formula computation. Effectively every heartbeat was measured manually during every seizure until the seizure ended clinically and the EEG had resumed its preevent trace. Data were also measured and analyzed prior to the seizure. From the onset of the seizure, data were measured 5 min (T-5m), 4 min (T-4m), 3 min (T-3m), 2 min (T-2m), 1 min (T-1m), and 9 s (T-9s) before the seizure. During the seizure, the 9-s epoch with the longest QTc was chosen to compare with the preseizure QTc value calculated 9 s prior to the seizure (T-9). Inter- and intraobserver analysis comparison was performed to assess accuracy of measurement. Clinically significant hypoxia was considered to be less than 85% oxygen saturation in this study (Bateman et al., 2008).

Four commonly used correction formulas were applied to all data:

Two nonlinear formulas:

Bazett (Bazett, 1920). inline image.

Fridericia (Fridericia, 1920). inline image

Two linear formulas:

Hodges (Hodges et al., 1983; Hodges, 1997). QTc = QT + 1.75 (heart rate − 60)

Framingham (Sagie et al., 1992). QTc = QT + 0.154 (1 − RR)

Statistical analysis was performed using Minitab version 14 (Minitab Ltd, Coventry, U.K.) with paired Student t-tests comparing the maximum corrected QT value prior to the seizure with the maximum corrected QT value during the seizure. Paired t-testing was considered appropriate as the analysis method to compare the corrected QT prior to the seizure with the corrected QT during the seizure. Due to the requirement of searching for transient increases in corrected QT, linear regression models were not appropriate statistical analyses for this study. Cross-tabulation of data comparing all formulas were analyzed to give sensitivity and specificity values for each formula. Bazett’s formula was used as the “gold standard” to enable uniformity and comparison to other studies (Yu et al., 1950; Morganroth & Pyper, 2001; Fossa et al., 2005; Pater, 2005). The Pearson chi-square test (Chernoff & Lehmann, 1954) was applied to determine statistical significance in the comparison of the corrected QT during generalized seizures, right hemisphere focal seizures, and left hemisphere focal seizures.

Normal limits of corrected QT during a seizure were considered to be comparable to corrected QT values during exercise. In children the normal limit of corrected QT during exercise is 440 ms (Benatar & Decraene, 2001; Rijnbeek et al., 2001). Adult male and adult female corrected QT limits are dependent on the upper normal limits for each formula and heart rate (Luo et al., 2004) summarized in Table 1. Increases in corrected QT from the baseline exceeding 60 ms are considered to present a small risk of a fatal arrhythmia (Morganroth & Pyper, 2001; Fenichel et al., 2004; European Agency for the Evaluation of Medicinal Products, 2005; Pater, 2005). Patients in this study were investigated for corrected QT increases exceeding 60 ms from preseizure values during epileptic seizures and categorized into seizure types.

Table 1. Upper limits of normal values for corrected QT in adult males and adult females using Bazett’s (QTcB), Hodge’s (QTcH), Fridericia’s (QTcFri), and Framingham’s formulas (QTcFra) Luo et al., 2004
GenderUpper normal limits (98% in ms)
HRQTcBQTcFriQTcFraQTcH
  1. HR, heart rate.

Both (10,303)All HR483460457457
HR < 60454459459466
HR 60–99483461458456
HR > 99492445436451
Male (5,420)All HR480457454454
HR < 60450455455465
HR 60–99480457454452
HR > 99490445436450
Female (4,883)All HR486463461460
HR < 60460463463470
HR 60–99486465462459
HR > 99492448434452

Results

The subject gender mix was 25 males and 14 females. The age range was 2 years 5 months to 60 years 3 months. The mean age was 17 years 2 months, with a median age of 11 years 5 months. A total of 156 seizures had corrected QT analysis performed (371 seizures from 11 patients had to be excluded from the study due to EMG artifact obscuring the ECG trace). The 156 analyzed seizures were composed of: 9 generalized tonic–clonic seizures (5 patients), 34 absences (6 patients), 12 tonic seizures (6 patients), 27 temporal lobe seizures (14 patients), 58 frontal lobe seizures (4 patients), and 16 subclinical temporal lobe seizures (4 patients).

Whole group analysis

All formulas agreed statistically significant lengthening of the corrected QT for the total group p < 0.001 during epileptic seizures compared to preseizure values (Fig. 1). A summary of results for all four QTc formulae during specific seizure types are presented in Table 2. The following results concentrate largely on Bazett’s formula, as this is the most commonly used formula in clinical practice. The QTc values during individual seizure types using Bazett’s formula are presented in Fig. 2. Nine patients exceeded normal QTc limits during 21 individual seizures (Table 3). The classification of the number of seizures and patients for each seizure type in which QTc exceeded normal limits is shown in Fig. 3. Six of these patients also exceeded their preseizure corrected QT by more than 60 ms during seizures. Two additional patients (four seizures) also increased their corrected QT exceeding 60 ms and remained within QTc normal limits (Table 4). Significant oxygen desaturation occurred in four patients (eight seizures). One of these patients (19F) prolonged the corrected QT beyond normal limits, and another patient (43M) increased the corrected QT by 130 ms (Table 5).

Figure 1.


Corrected QT before and during total seizures according to Bazett’s, Hodge’s, Fridericia’s, and Framingham’s corrective formulas.

Table 2. Summary of corrected QT results during epileptic seizures according to Bazett’s, Hodge’s, Fridericia’s, and Framingham’s corrective formulae
n = 156Bazett’s QTcHodge’s QTcFridericia’s QTcFramingham’s QTc
  1. P, pediatric; M, male; F, female; HR, heart rate; CI, 95% confidence interval.

Temporal lobe seizures n = 27 CI 39.467–68.682
p < 0.001
CI 24.022–37.978
p < 0.001
CI 16.709–36.891
p < 0.001
CI 10.330–26.707
p < 0.001
Heart rate
 Max. QTc ms
110 (HR)
512 (M)
66.7 (HR)
479 (M)
103.2 (HR)
464 (F)
65.9 (HR)
462 (M)
 Prolonged QTc seizures (patients)13 (5)1 (1)4 (2)2 (2)
 No. seizures QTc > 60 ms (patients)11 (4)0 (0)3 (2)1 (1)
Subclinical seizures n = 16CL 15.777–47.972
p = 0.001
CI 5.152–26.097
p = 0.006
CI 7.726–29.524
p = 0.002
CI 8.140–28.860
= 0.002
Heart rate
 Max. QTc ms
120 (HR)
493 (P)
127.3 (HR)
475 (P)
92.8 (HR)
454 (P)
88.8 (HR)
451 (P)
 Prolonged QTc seizures (patients)6 (2)3 (1)1 (1)1 (1)
 No. seizures QTc > 60 ms (patients)2 (1)1 (1)1 (1)1 (1)
Frontal lobe seizures n = 58CI 27.074–36.374
p < 0.001
CI 21.549–29.723
p < 0.001
CI 10.123–19.532
p < 0.001
CI 11.603–20.466
p < 0.001
Heart rate
 Max. QTc ms
117.9 (HR)
462 (M)
117.9 (HR)
437 (M)
117.9 (HR)
423 (M)
117.9 (HR)
422 (M)
 Prolonged QTc seizures (patients)0 (0)0 (0)0 (0)0 (0)
 No. seizures QTc > 60 ms (patients)3 (1)0 (0)0 (0)0 (0)
GTCS n = 9CI 25.380–79.065
p = 0.002
CI 32.097–96.125
p = 0.002
CI −2.320–46.765
p = 0.07
CI −12.320–36.765
p = 0.284
Heart rate
 Max.QTc ms
148.1 (HR)
490 (P)
148.1 (HR)
466 (P)
131.8 (HR)
464 (P)
131.8 (HR)
440 (P)
 Prolonged QTc seizures (patients)1 (1)1 (1)1 (1)0 (0)
 No. seizures QTc > 60 ms (patients)3 (2)3 (3)0 (0)1 (1)
Tonic seizures n = 12CI −0.715–29.946
p = 0.060
CI 10.336–36.164
p = 0.002
CI −18.424–15.347
p = 0.846
CI −19.936–12.244
p = 0.612
Heart rate
 Max. QTc ms
124.2 (HR)
465 (M)
124.2 (HR)
463 (M)
121.9 (HR)
421 (M)
121.9 (HR)
414 (M)
 Prolonged QTc seizures (patients)0 (0)2 (1)0 (0)0 (0)
 No. seizures QTc > 60 ms (patients)0 (0)0 (0)0 (0)0 (0)
Absence seizures n = 34CI −10.181–3.514
p = 0.329
CI −6.379–1.320
p = 0.191
CI −3.563–2.975
p = 0.856
CI −3.674–3.086
p = 0.861
Heart rate
 Max. QTc ms
117.6 (HR)
462 (P)
117.6 (HR)
432 (P)
90.1 (HR)
434 (F)
90.1 (HR)
431 (F)
 Prolonged QTc seizures (patients)1 (1)0 (0)0 (0)0 (0)
 No. seizures QTc > 60 ms (patients)0 (0)0 (0)0 (0)0 (0)
Figure 2.


Corrected QT during grouped seizures according to Bazett’s corrective formula.

Table 3. Summary of Bazett’s corrected QT values for patients during seizures exceeding normal limits according to Luo et al., 2004
Patient study identification number and genderHeart rate during BpmBazett’s QTc before (ms)Bazett’s QTc during (ms)Bazett’s QTc change (ms)Seizure type
  1. M, male; F, female.

9M (pediatric)119.544149251Right subclinical temporal
9M (pediatric)111.944447430Right subclinical temporal
9M (pediatric)109.139047181Right subclinical temporal
9M (pediatric)120376493117Right subclinical temporal
9M (pediatric)140.544547227Right subclinical temporal
22F (pediatric)72.242545429Left subclinical temporal
20M66.739649599Right temporal
20M119.541250189Right temporal
20M100.641049484Right temporal
20M100.842149372Right temporal
20M88.440149089Right temporal
20M10241150291Right temporal
20M100.343649458Right temporal
20M110390512122Right temporal
20M10241049282Right temporal
21F10141349481Right temporal
24F110.142050585Right temporal
19F10939649498Left temporal
17M76.843349158Left temporal
18M (pediatric)148.140349087Generalized tonic–clonic
28M (pediatric)117.644046222Myoclonic absence
Figure 3.


Total patients and seizures exceeding Bazett’s corrected QT normal limits (according to Luo et al., 2004).

Table 4. Summary of patients demonstrating an increase in corrected QT exceeding 60 ms during a seizure according to Bazett’s formula
Patient study identification number and genderHeart rate during (bpm)QTc before (ms)QTc during (ms)QTc change (ms)Seizure type
  1. M, male; F, female.

9M (pediatric)109.139047181Right subclinical temporal
9M (pediatric)120376493117Right subclinical temporal
20M119.541250189Right temporal
20M100.641049484Right temporal
20M100.842149372Right temporal
20M88.440149089Right temporal
20M10241150291Right temporal
20M110390512122Right temporal
20M10241049282Right temporal
20M66.739649599Right temporal
21F10141349481Right temporal
24F110.142050585Right temporal
19F10939649498Left temporal
30M98.133641781Left frontal
30M105.233341380Left frontal
30M90.933040171Left frontal
18M (pediatric)148.140349087Generalized tonic–clonic
18M (pediatric)123.935142170Generalized tonic–clonic
43M79.8264394130Generalized tonic–clonic
Table 5. Seizures associated with oxygen desaturation
Patient study identification number and genderMinimum % oxygen saturation during seizurePeriod of oxygen desaturation (s)Seizure duration (s)Heart rate during (bpm)Bazett’s QTc before (ms)Bazett’s QTc during (ms)Bazett’s QTc change (ms)Seizure type
  1. M, male; F, female; GTCS, generalized tonic–clonic seizure.

31M Pediatric7181252101.542143110Right temporal
31M Pediatric7157255118.841144029Right temporal
19F757811110939649498Left temporal
43M678716579.8264394130Left temporal progressing to GTCS
43M7213220154.54004000Left temporal
43M764815059.438641731Left temporal
44F7433120103.841047060Left temporal
44F753910273.239044050Left temporal

Temporal lobe seizures and subclinical temporal lobe seizures

Twenty-seven temporal lobe seizures (14 patients) and 16 subclinical seizures (4 patients) were recorded and analyzed. Seventy percent (19 of 27) of temporal lobe seizures measured an increase in corrected QT during the seizure compared to QTc measurement prior to the seizure (p < 0.001). Bazett’s formula indicated 48% (13 of 27) of these events from five patients exceeded normal limits, with the most dramatic corrected QT lengthening occurring from 390 to 512 ms during a right temporal lobe seizure lasting 1 min 27 s. Four patients (eight seizures) became hypoxic during their temporal lobe seizures. During two right temporal lobe seizures (one patient), oxygen saturation dropped to 71% during the desaturation period lasting up to 81 s, but the corrected QT remained within normal limits. During one of the left temporal lobe seizures, oxygen saturation dropped to 75% during the desaturation period lasting 78 s, and the corrected QT lengthened to 494 ms, with an increase of 98 ms (Table 5). Sixteen subclinical temporal lobe seizures were recorded and analyzed from four patients. Six subclinical temporal lobe seizures from two patients were found to lengthen the corrected QT beyond normal limits according to Bazett’s formula, with a maximum increase of 117 ms during one subclinical seizure, from 376 to 493 ms.

Generalized tonic–clonic seizures

There was a total of nine generalized tonic–clonic seizures recorded (five patients). Overall there was an increase in corrected QT during these seizures compared to QTc values prior to the generalized tonic–clonic seizures (p < 0.001). One seizure (patient 18M) resulted in corrected QT exceeding normal limits, with a maximum increase of 87 ms from 403 to 490 ms. Hypoxia occurred during one generalized tonic–clonic seizure (patient 43M), which was a secondary generalized left temporal lobe seizure. Oxygen saturation dropped to 67% during the desaturated period lasting 87 s. Although the corrected QT stayed within normal limits, it increased by 130 ms Table 5. The remaining six patients in the study who had generalized tonic–clonic seizures did not become hypoxic.

Frontal lobe seizures

Fifty-eight frontal lobe seizures in total were analyzed (two derived from the right hemisphere and 56 derived from the left hemisphere) from four patients. Generally, slight increases in corrected QT lengthening were observed in 64% (37 of 58) of frontal lobe seizures resulting in high significance p < 0.001. However, the QTc values remained within normal limits. No hypoxia occurred during frontal lobe seizures in this study.

Absence seizures

A total of 35 absences were recorded from six patients. Generally there was no significant change in corrected QTc in this seizure type (p = 0.3). Seventy-four percent (26 of 35) of absence seizures either showed no increase in corrected QT or slightly shortened the corrected QT. However, one myoclonic absence demonstrated corrected QT lengthening by a small amount of 22 ms resulting in corrected QT prolongation of 462 ms in a child with no hypoxia.

Tonic seizures

Twelve tonic seizures in total were recorded from six patients. According to Bazett’s formula, none of the tonic seizure events increased the corrected QT values beyond normal limits; p = 0.06. No QTc lengthening exceeding 60 ms or significant hypoxia occurred in this group. Only Hodge’s formula identified two tonic seizures (one patient) as prolonging the corrected QT of up to 464 ms (Table 2).

Corrected QT lengthening and laterality of seizure types

Patients who had temporal lobe seizures or subclinical seizures derived from the right hemisphere that produced a prolongation in corrected QT demonstrated a tendency to have multiple seizures compared to those patients experiencing left hemisphere seizures.

A total of 27 temporal lobe seizures (14 patients) were recorded and analyzed: 17 right temporal lobe seizures (7 patients) and 10 left temporal lobe seizures (7 patients). Thirteen temporal lobe seizure events were found to lengthen the corrected QT beyond normal limits according to Bazett’s formula. Eleven of these seizures (11 of 13) were derived from the right temporal lobe (3 patients) compared to two seizures (2 of 13) from the left temporal lobe (2 patients). Sixteen subclinical temporal lobe seizures were recorded from 4 patients and analyzed (6 derived from the right hemisphere from 2 patients and 10 derived from the left hemisphere from 2 patients). Six subclinical temporal lobe seizures were found to lengthen the corrected QT beyond normal limits according to Bazett’s formula. Five of these subclinical seizures (one patient) were derived from the right temporal lobe compared to one subclinical seizure from the left temporal lobe (one patient).

Although patients with seizures arising from the right hemisphere showed more frequent lengthening of the corrected QT, there is no statistical significance when comparing right and left hemisphere seizures when the analysis is performed in terms of number of patients (p = 0.422).

Corrected QT prolongation exceeding 60 ms

An increase of the corrected QT by more than 60 ms is considered to present a small risk of a fatal arrhythmia (European Agency for the Evaluation of Medicinal Products, 2005; Fenichel et al., 2004; Morganroth & Pyper, 2001; Pater, 2005). In this study, 19 seizures (eight patients) demonstrated an increase in corrected QT exceeding 60 ms using Bazett’s formula (Table 4). Of these 19 seizures there were 10 right temporal lobe seizures (three patients), one left temporal lobe seizure (one patient), three generalized tonic–clonic seizures (two patients), three left frontal lobe seizures (one patient), and two right temporal subclinical seizures (one patient).

Ten of the temporal lobe seizures (10 of 19) were derived from the right temporal lobe (three patients) compared to one (1 of 19) derived from the left temporal lobe (one patient). The left temporal lobe seizure (patient 19F) increased the corrected QT by 98 ms to a prolonged measurement of 494 ms. In addition, the patient became hypoxic, with an oxygen saturation of 75% (Tables 3–5).

Two (2 of 19) right temporal lobe subclinical seizures were identified with an increased QTc by more than 60 ms compared to no left temporal subclinical seizures. Five seizures (three patients) identified as having an increase from baseline by 60 ms were still within normal QTc limits; two of these were generalized tonic–clonic seizures (two patients) and three were left frontal lobe seizures (one patient). The types of seizures resulting in maximum increases in corrected QT lengthening occurred in a generalized tonic–clonic seizure (130 ms) in a patient who became hypoxic with oxygen saturation decreasing to 67%, right temporal lobe seizure (122 ms), and right temporal subclinical seizure (117 ms). Overall, increases in corrected QT exceeding 60 ms in this study occurred in 33% (3 of 9) of generalized tonic–clonic seizures (from 2 of 5 patients), 59% (10 of 17) of right temporal lobe seizures (from 3 of 7 patients), and in 33% (2 of 6) of right temporal subclinical seizures (from 1 of 2 patients) compared to 10% (1 of 10) of left temporal lobe seizures (from 1 of 7 patients) and 5% (3 of 56) of left frontal lobe seizures (from 1 of 2 patients).

Comparison of corrected QT formulae

Cross-tabulation was performed using Minitab “version 14” to calculate the sensitivity and specificity of Hodges (Hodges et al., 1983; Hodges, 1997); Fridericia (Fridericia, 1920), and Framingham’s formulae (Sagie et al., 1992), with Bazett’s formula (Bazett, 1920) used as the gold standard (Yu et al., 1950). All formulae gave good specificity values, with Framingham’s giving the highest specificity of 100% correlation with Bazett’s formula in identifying normal values. Fridericia’s specificity was slightly lower at 99.3%, and Hodge’s specificity was 98.5%.

Sensitivity of identifying seizures in agreement with Bazett’s formula was poor from all comparative formulae, with Hodge’s and Fridericia’s formula at 35.7%, and Framingham’s at 21%.

The results of identified cases of lengthening of the corrected QT beyond normal limits are dependent on which formula is used. Bazett’s formula identified a total of 21 seizures from 9 patients, Hodge’s formula identified 7 seizures from 4 patients, Fridericia’s identified 6 seizures from 4 patients, and Framingham’s formula identified 3 seizures from 3 patients (Fig. 4 and Table 2). However, all formulae agreed on the identification of two patients as prolonging their corrected QT during seizures. The corrected QT prolongation occurred as the seizure progressed, and as the heart rate changed each formula demonstrated prolongation during different epochs within the same seizure. The first patient was a 2-years-old male during a right subclinical seizure. Bazett’s 492 ms (heart rate of 119.5/min), Hodge’s 475 ms (heart rate 127.3/min), Fridericia’s 454 ms (heart rate 92.8/min), and Framingham’s 451 ms (heart rate 88.8/min). The second patient was a 21-year-old man during a right temporal lobe seizure. Bazett’s 495 ms (heart rate 66.7/min), Hodge’s 479 ms (heart rate 66.7/min), Fridericia’s 463 ms (heart rate 65.9/min), and Framingham’s 462 ms (heart rate 65.9/min).

Figure 4.


Total patients and seizures exceeding normal corrected QT limits (according to Luo et al., 2004) using Bazett’s, Hodge’s, Fridericia’s, and Framingham’s corrective formulae.

Inter- and intraobserver analysis

The margin of maximum observer disagreement in corrected QT measurement was 10 ms and introduced a low error rate of 0.02%. This result is comparable to those accepted within guidelines agreed in drug trial regulations of inter- and intraobserver variability of ±10 ms (Pater, 2005). The Kappa statistic of 0.726 was calculated of the difference of the observed agreement from two independent reviewers. This constitutes a “substantial agreement” and “precision” between two observers in measuring normal and prolonged corrected QT even with expected chance measurements (Viera & Garrett, 2005).

Discussion

Impulses from higher brain regions in the cerebral cortex, limbic system, and hypothalamus affect the cardiovascular center (Tortora & Grabowski, 1992). During cases of SUDEP it has been hypothesized that an intrinsic channelopathy may occur with sudden and intense impulses directed from cortical areas, driven by a seizure source, which overstimulate either the cardioinhibitory center causing asystole or cause a massive adrenergic effect on the sympathetic pathways resulting in tachycardia (Wannamaker, 1985; Lathers et al., 1987; Cheung & Hachinski, 2000; Ansakorpi et al., 2004). At autopsy, raised catecholamine detection has been identified on occasion, indicating massive beta-adrenergic catecholamine effect (Nei et al., 2004). It is considered that this physiologic mechanism may not be amenable to intervention (Dashieff & Dickinson, 1986).

The effects of sustained tachycardia during epileptic seizures in the presence of prolonged cardiac repolarization could leave some individuals vulnerable to ventricular tachycardia and sudden death. This is particularly true for some individuals with long QT syndrome or catecholaminergic polymorphic ventricular tachycardia who have a congenital predisposition and who are susceptible to lengthening of their cardiac repolarization time during exercise. In some individuals, exercise testing may be helpful in differentiating between seizure effect and exercise effect on corrected QT.

This study has shown overall small increases in corrected QT during epileptic seizures, which generally may not be of any clinical importance to most individuals. However, we have demonstrated that there are some seizures that result in transiently prolonged QTc beyond normal limits even during self-resolving seizures. Transient QTc prolongation (Kandler et al., 2005) occurred in 7 of 30 pediatric patients during a postconvulsive period of 2 h following generalized seizures.

It is considered that when the corrected QT lengthens by more than 60 ms there is a small risk of a fatal arrhythmia. In this study, 10 right temporal lobe seizures (three patients), two right temporal subclinical seizures (one patient), and three generalized tonic–clonic seizures (two patients) demonstrated the highest proportions of increases in corrected QT exceeding 60 ms compared to other seizure types. It is reported that generalized tonic–clonic seizures appear more commonly responsible for seizure-related deaths (Opherk et al., 2002). Lengthening of the corrected QT beyond normal limits occurred transiently during this type of seizure in one child. Two further generalized tonic–clonic seizures were found to cause large increases of corrected QT of 130 ms during a seizure in an adult male and 70 ms during a seizure in a child, but both remained within normal QTc limits during exercise according to Bazett’s formula. Although values remain within normal limits, these increases in corrected QT may be a possible contributing factor in triggering cardiac arrhythmias in some individuals and additionally could be further compounded by cases of associated hypoxia.

Lengthening of the corrected QT during subclinical seizures strengthens the possibility that this could be involved in SUDEP where no evidence of a seizure has clinically taken place. Subclinical seizure activity could stimulate intrinsic adrenergic pathways and produce a cardiologic effect of QT prolongation. Animal models of epilepsy suggest even interictal epileptogenic activity may induce changes in the autonomic nervous system, which could result in cardiac arrhythmia and risk of sudden death (Persson et al., 2003).

Nei et al., 2004, describe sudden and extreme changes in heart rate during nocturnal seizures and consider that marked fluctuations in autonomic tone could precipitate cardiac arrhythmias. These researchers reported that 14 of 21 people died from SUDEP in sleep. When they studied the seizure EEG/ECG data of these patients retrospectively they discovered that much greater increases in heart rate occurred during seizures arising from sleep than seizures during wakefulness in the SUDEP group compared to the non-SUDEP group. In addition, cardiac repolarization and rhythm abnormalities occurred in 56% of the SUDEP patients during seizures.

Sudden forced awakening itself causes marked increases in heart rate (Kaida et al., 2003). Marked changes in T-wave morphology, signaling an alteration in ventricular repolarization, were demonstrated when subjects were unexpectedly awakened from deep sleep (Dweck et al., 2006). Increased autonomic instability is described during early waking hours corresponding to a reported period of increased vulnerability to ventricular tachycardia and sudden death (Pater, 2005). Forced awakening occurs during nocturnal seizures, and the effects of sudden waking with associated increase in heart rate is compounded by some seizures at a time of suspected autonomic instability and, therefore, could leave some individuals more vulnerable to an induced cardiac arrhythmia during early waking hours.

During nocturnal seizures when vagal tone is high in sleep, sudden changes in heart rate could leave the heart vulnerable because of a phenomenon known as QT hysteresis. QT hysteresis is defined as the recovery QT peak interval subtracted from the exercise QT peak interval (Krahn et al., 2002). A significant lag time relationship between the RR interval and QT interval associated with a sudden tachycardia during a seizure could theoretically lengthen the corrected QT and increase the risk of cardiac arrhythmia. Acute effects on cardiac ionic flux during seizures from sustained tachycardia can result in acidosis; acidosis in turn causes decreased contractility and Ca+ overload. Hypoxia exacerbates this negative inotropic effect and reperfusion ironically can potentially result in arrhythmias (Bers, 2001). Hypoxia is well recognized in some seizures and was also observed in this study group, particularly associated with temporal lobe seizures.

Although a high number of seizures in the study showed a transient increase in corrected QT, none of the patients had died of SUDEP at the time of this publication. The duration of QTc prolongation was typically brief (up to 9 s), with all seizures self-resolving without the requirement of rescue medication, as the longest seizure was 4 min 15 s. This study indicates that transient prolongation of the corrected QT occurs during some seizures in some individuals. It is proposed that there may be some individuals who are more susceptible than others to developing a cardiac arrhythmia during seizures because of the effects of QT hysteresis and more sustained or increased prolongation of the corrected QT, particularly during prolonged unwitnessed nocturnal seizures with associated hypoxia.

Subclinical temporal lobe seizures and temporal lobe seizures arising from the right hemisphere demonstrated marked lengthening of the corrected QT from two patients in this study according to all four corrective formulae. Lathers et al., (1987) describe a mechanism called the “lockstep phenomenon” based on a study of anesthetized cats, where sympathetic discharge was synchronized with epileptogenic activity and associated with changes in ECG. The changes during an epileptiform discharge were shown to alter the sympathetic activity and caused an alteration of peripheral efferent discharge to the heart. Lathers and colleagues postulated that the lockstep phenomenon maybe a factor of SUDEP where there has been no evidence of a clinical seizure around the time of death. Increases in corrected QT during subclinical seizures are not related to exercise, as no physical exertion takes place, and therefore must have an intrinsic basis. It is possible, therefore, that a sudden and intense electrical discharge caused by a subclinical seizure could directly alter the peripheral efferent discharge to the heart causing lengthening of the corrected QT and lead to a cardiac arrhythmia. Cardiac arrhythmias and repolarization abnormalities have been identified in epilepsy. It is considered possible that sudden unexpected death in epilepsy could be due to fatal neurogenic cardiac arrhythmias (Rugg-Gunn et al., 2004).

Many researchers believe that cardiologic effects during seizures are possibly by the result of a genetically determined channelopathy, which becomes triggered by the seizure, directly affecting the heart (Wannamaker, 1985; Lathers et al., 1987; Cheung & Hachinski, 2000; Ansakorpi et al., 2004). These susceptible individuals may develop an intracellular ionic flux disturbance in the myocyte because of sudden cardiologic physiologic demands sustained by an intense seizure-induced train of electrical impulses via a channelopathy. This disturbance in ionic flux may deteriorate further over time, leading to an arrhythmia during an undetected prolonged nocturnal seizure, and could hypothetically be another contributing factor in some cases of SUDEP.

Nine patients were identified as having transiently lengthened their corrected QT values during epileptic seizures according to Bazett’s formula, compared to only three patients identified by Framingham’s formula. Bazett’s formula is the most common corrective formula and has been used widely for more than 80 years (Bazett, 1920; Yu et al., 1950; Pater, 2005). For consistency of data collection in this study and to allow a comparison to other studies, Bazett’s has been the main formula used to compare to other formulae results. However, it is important to note and is well recognized that Bazett’s formula overcorrects and, therefore, overestimates QTc at fast heart rates, resulting in a tendency toward falsely prolonged corrected QT values, and undercorrects and underestimates QTc at slow heart rates resulting in a tendency toward falsely shortened corrected QT values (Sagie et al., 1992; Hodges, 1997; Pater, 2005; Sredniawa et al., 2005). Conversely, Framingham’s formula can lead to falsely shortened corrected QT values at fast heart rates and falsely lengthened QTc values at slow heart rates (Sagie et al., 1992; Benatar & Decraene, 2001). No single mathematical transformation can adjust for the rapidly changing nonlinear dynamics of the QT/RR interval relationship (Fossa et al., 2005). Individual correction formulae have been proposed (Malik, 2004); however, this is difficult in practice as it would be excessively time-consuming (Luo et al., 2004; Fossa et al., 2005; Pater, 2005).

Some authors warn of artificially prolonged QTc calculated from rapidly accelerating heart rate disproportionately affecting QTc measurements (Yu et al., 1950; Malik, 2004). This potential error was objectively assessed in this study and it was found that lengthening of the corrected QT occurred during some seizures whether heart rate increased or decreased. Furthermore, during some subclinical seizures where there is no element of physical exertion, again measurements of QT prolongation occurred whether heart rate increased or decreased. During a left temporal subclinical seizure heart rate decreased from 80 to 70/min and the corrected QT increased from 425 to 454 ms. Therefore, although there is a strong heart rate correlation with corrected QT calculations when using a linear formula like Bazett’s, we have found that during seizures corrected QT prolongation is not simply the result of the previous RR history. Magnano et al., (2002) demonstrated that autonomic conditions affect the ventricular myocardium and cause differences in QT that are independent of heart rate.

Measuring corrected QT during exercise is used as a clinical method for distinguishing and identifying patients with latent long QT syndrome from healthy subjects who all have normal QT at rest. Bazett’s formula is used widely in clinical practice and remains the best predictor of carrier status of latent long QT compared to other formulae (Walker et al., 2005). Some epileptic seizures cause more rapid acceleration or deceleration than others, and this is highly variable, both between individuals and within the same individual. Nocturnal seizures causing forced arousal compounds heart rate acceleration more than simple arousal itself. “Artificial prolongation” of the corrected QT can occur in some cases where heart rate increases dramatically and short epochs are used. Many researchers, including drug trial researchers, only average three beats and apply these values to corrective formulae. However, because “artificial prolongation” of the corrected QT is possible in accelerated heart rates when analyzing the effects during seizures, we used a longer epoch of 9 s to average out this effect.

A previous study analyzed corrected QT during epileptic seizures and reported that all mean QTc values using Bazett’s formula were normal for total grouped data (Nei et al., 2000). Different epoch lengths were used to calculate the preseizure data with the during seizure data; lead I was measured instead of lead II and there was no explanation as to whether different normal limits were applied to gender and age differences by these authors. By comparison, this study used a standard epoch length of 9 s to measure nine consecutive beats in order to calculate the corrected QT prior to the seizure and throughout the seizure as described in the methodology. Lead II was selected, as it is generally recommended instead of lead I so that the U wave was not included in the QT measurement. We believe that transient QTc increases or individual patients demonstrating an increase in corrected QT are lost if calculated as a mean average of total patient grouped “preseizure” and “during seizure” data.

Study Limitations

Studies of larger sample sizes with controlled drug monotherapy and polytherapy analyses may be useful in the investigation of any contributing effect on lengthening of the corrected QT. However, a previous study reported that there was no evidence of QTc prolongation detected in patients receiving antiepileptic drug monotherapy or polytherapy in a pediatric group (Kwon et al., 2003). Effects of diurnal variation, effects of winter months on adult male subjects, and stages of sleep were not investigated in this study.

Continued research is required in the analysis of corrected QT measurement during various seizure types to consider the effects in a larger population study. Synchronous measurement of cardiac vagal tone may provide an insight into intrinsic mechanisms involved in SUDEP.

Conclusion

This study has indicated that significant lengthening of the corrected QT beyond normal limits in children and adults transiently occurs during some epileptic seizures. Prolonged corrected QT is associated with an increased risk of cardiac arrhythmia, and this study supports the possibility that this cardiologic mechanism may be involved in SUDEP.

Acknowledgment

We would like to thank our Clinical Neurophysiology colleagues at the Royal Hospital for Sick Children and Western General Hospital, Edinburgh for their assistance in the collection of patient data using video EEG/ECG/SAO2 telemetry. Also we acknowledge Cat Graham, Biostatistician, Wellcome Trust, Western General Hospital, Edinburgh for her assistance and guidance in the statistical analysis performed in this study.

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Disclosure: None of the authors has any conflict of interest to disclose.

Ancillary