Time-dependent responses to provocative testing with flecainide in the diagnosis of Brugada syndrome David Calvo, MD, PhD,* José M. Rubín, MD, PhD,* Diego Pérez, MD,* Juan Gómez, PhD,†‡ Juan Pablo Flórez, MD,* Pablo Avanzas, MD, PhD,y José Manuel García-Ruíz, MD,y Jesús María de la Hera, MD, PhD,‡ Julián Reguero, MD, PhD,y Eliecer Coto, MD, PhD,†‡ César Morís, MD, PhDy From the *Arrhythmia Unit, Cardiology Department, †Department of Molecular Genetics, ‡Red de Investigación Renal (REDINREN), and yCardiology Department, Hospital Universitario Central de Asturias, Oviedo, Spain. BACKGROUND Time-dependent variability of electrocardiogram (ECG) in patients with Brugada syndrome could affect the interpretation of provocative testing. OBJECTIVE The aim of this study was to characterize ECG changes during and after flecainide infusion. METHODS We studied 59 consecutive patients. The ECG was continuously analyzed during the first 30 minutes of provocative testing, and a single ECG was recorded 60 minutes later. We analyzed CYP2D6 and CYP3A5 variants affecting flecainide metabolism and performed blinded measurements at lead II. RESULTS At baseline, ECG patterns were classified as follows: type II in 31 patients (53%), type III in 15 (25%), and normal ECG in 13 (22%). Because of induction of type I ECG, the percentage of responders progressively increased with longer recording time periods (6.8% in 10 minutes vs 11.9% in 20–30 minutes vs 18.6% in 90 minutes; P o .01). Four patients displayed a late response, which was evidenced 90 minutes after the initiation of provocative testing. QRS width differentially increased between responders and nonresponders (P o .01), with a maximum QRS
Introduction Brugada syndrome (BrS) is a leading cause of sudden death (SD) in young people, with the type I electrocardiographic (ECG) pattern being an essential clue for diagnosis.1 In patients with no spontaneous type I ECG, flecainide challenge has been promoted as a commonly used test in the clinic.2 However, the timedependent course of ECG patterns in patients with BrS display a well-known variability3 that has not been appropriately analyzed in the context of provocative testing with sodium blockers. Recently, 2 case reports4,5 have described the occurrence of late positive responses after flecainide infusion. According to the standard recommendations for provocative testing
width of 110 ms during the first 30 minutes being effective for identifying possible late responders (sensitivity 100%; specificity 85.6%; positive predictive value 88%; negative predictive value 100%). The incidence of CYP2D6 variants was lower in late responders than in early or delayed responders (0% vs 75% vs 100%; P ¼ .04), while a homogeneous distribution of CYP3A5*3/*3 was observed in our population. CONCLUSION Response to flecainide exhibits time-dependent variability of ECG patterns and intervals. Longer periods of ECG recording increase the recognition probability of type I ECG. KEYWORDS Brugada syndrome; Electrocardiogram; Provocative testing; Flecainide; Cytochrome P450 ABBREVIATIONS BrS ¼ Brugada syndrome; ECG ¼ electrocardiogram/ electrocardiographic; Het-EM ¼ heterozygous extensive metabolizer; Hom-EM ¼ homozygous extensive metabolizer; QTc ¼ corrected QT; SD ¼ sudden death (Heart Rhythm 2015;12:350–357) rights reserved.
I
2015 Heart Rhythm Society. All
performance (10–30 minutes of waiting time if the test has a negative result), these tests would have been considered as negative2; however, these results are false negatives, with implications for clinical management (ie, perform lifestyle measurements, avoid drugs with potential adverse effects, and prompt treatment of fever episodes). To quantify the incidence of late responses, we analyzed the time course of ECG changes during and 90 minutes after flecainide infusion. In addition, clinical, ECG, and genetic factors affecting flecainide metabolism were analyzed to describe predictors of time-dependent responses.
Methods Address reprint requests and correspondence: Dr David Calvo Cuervo, Arrhythmia Unit, Cardiology Department, Hospital Universitario Central de Asturias, Calle Carretera de Rubín, s/n, 33011 Oviedo, Spain. E-mail address: [email protected].
1547-5271/$-see front matter B 2015 Heart Rhythm Society. All rights reserved.
Population and recording protocol In the absence of a spontaneous type I ECG pattern, patients with suspected BrS were admitted for provocative testing http://dx.doi.org/10.1016/j.hrthm.2014.10.036
Calvo et al
ECG Variability During Flecainide Testing
with flecainide. According to the standard protocol used in our institution, the clinical profile and serum levels (creatinine and electrolytes) were previously obtained at the outpatient clinic, 1–3 months before the admission. Intravenous flecainide was continuously infused at a rate of 2.0 mg/kg bodyweight over 10 minutes (limited by a maximum dosage of 150 mg). Before flecainide infusion, we checked for the absence of type I ECG, both at the standard precordial position (V1 and V2 at the third intercostal space) and at the high precordial position (V1 and V2 at the second intercostal space). Thereafter, the standard 12-lead ECG was continuously recorded during 30 minutes in the electrophysiology laboratory (sample rate 1 kHz; bandpass filtered 0.05–150 Hz; EP Tracer v1.05.v3, CardioTek, Maastricht, The Netherlands) from the initiation of flecainide infusion, which enabled us to better characterize transitions between ECG patterns because types II and III were mainly defined at the standard precordial positions.1 At the end of this period, we again explored the high precordial position for better sensitivity. Thereafter, the positions of the leads on the body surface were marked and a new ECG was recorded 60 minutes later (Figure 1). The study was approved by the ethics committee, and subjects gave informed consent. ECG tracings were analyzed by 3 independent cardiac electrophysiologists and classified by consensus according to the published recommendations as type I, II, and III.1 For the purpose of this study, we classified the normal ECG as category IV. This category also includes subtle deviations from normality as the incomplete right bundle branch block. The response to flecainide infusion was analyzed continuously during the first 30 minutes. ECG changes observed during flecainide infusion (first 10 minutes of the ECG recording) were considered to occur at the first stage of the test. The second and third stage changes were those changes that occurred between 10 and 20 minutes and between 20 and 30 minutes, respectively. Finally, the fourth stage changes were considered according to the single 12-lead ECG recorded 90 minutes after the initiation of flecainide infusion. In order to assign an ECG pattern as defining the result of the test at every stage, we selected the highest order pattern observed during the recording time, considering type I as the highest order pattern and category IV as the lowest one. Provocative testing was considered to display a positive response if the patient exhibited a type I ECG pattern anytime during the protocol. Those patients displaying a positive response during the first stage were defined as early responders. On the contrary, those patients displaying a
351 positive response during the second and third stages were defined as delayed responders. Finally, those patients displaying a positive response only in the fourth stage were defined as late responders. All other patients were defined as nonresponders. During the 30 minutes of continuous ECG recording, average heart rate, PR, QRS width, and corrected QT (QTc) intervals (Bazett’s formula) were measured every 10 minutes from baseline (using integrated calipers over the digital records; speed 100 mm/s) at lead II. Measurements of 3 consecutive cycles were averaged. To avoid potential bias, interval measurements were performed blinded to the final result of provocative testing. Also, all ECG channels other than lead II were removed from the screen at this time to avoid the researcher perceiving the final result of provocative testing.
Genetic studies According to current guidelines,6 the whole coding sequence of SCN5A was amplified and sequenced in responders as described previously.7 The nucleotide changes were classified as possible mutations if (1) reported previously (www. ensembl.org) or (2) they had an effect on the protein sequence, either by changing the amino acid or by introducing aberrant transcript sequences. Single amino acid changes were considered as mutations on the basis of the prediction programs SIFT (J. Craig Venter Institute, La Jolla, CA) and Polyphen (Harvard, Massachusetts). In the whole cohort of responders and a randomly selected subset of nonresponders (5 patients), we also determined single nucleotide polymorphisms in CYP3A5 and CYP2D6 genes because they are mainly involved in flecainide metabolism.4,8 The CYP3A5*3 allele (SNP rs776746) was genotyped with a real-time TaqMan polymerase chain reaction assay (assay ID C_25201809_30, Applied Biosystems, California, USA) as reported previously.9 To determine the CYP2D6 genotype, we amplified and sequenced the 9 coding exons with primers specific for this gene (further details are available on request to the corresponding author).
Statistical analysis Categorical variables are reported as numbers and percentages. For the purpose of the analysis and better clinical interpretation, we considered the ECG patterns as ordinal variables of arbitrary units by assigning a value from 1 to 4 to patterns I, II, III, and category IV, respectively. The ECG
Figure 1 Study flowchart showing different stages of electrocardiographic (ECG) recording. Interval measurements were performed over the digital records at baseline, 10, 20, and 30 minutes (100 mm/s).
352 patterns were analyzed using the Cochran Q or the Friedman test, as appropriate. Continuous variables are reported as mean ⫾ standard deviation. We used a mixed-effects regression model with random intercepts to test changes in the ECG intervals across stages. Statistical significance was analyzed between stages (within-subjects significance) and also according to the final result of provocative testing (responders vs nonresponders; between-subjects significance). The Bonferroni correction was used for multiple comparisons. The χ2 and Student t tests were used for univariate analysis to compare different variables as predictors of the result of provocative testing. Also, logistic binary regression was used for analysis and the predictive performance of variables was assessed by receiver operating characteristic analysis and the Hosmer-Lemeshow test. Agreement between investigators was measured by the k statistic. Analysis was performed with SPSS software (version 20, SPSS Inc, Chicago, IL), and P o .05 was defined as statistically significant.
Results From November 2012 through March 2014, we prospectively studied 60 consecutive patients (age 38.4 ⫾ 14.1 years; male sex 71.2%) because of the presence of an abnormal ECG (65%; type II or III ECG), family history of BrS (40%), and/or syncope (23%). There were no patients with history of SD or ventricular arrhythmias, renal impairment (creatinine 0.83 ⫾ 0.21 mg/dL), or electrolyte abnormalities (sodium 139.5 ⫾ 2.4 mEq/L and potassium 4.2 ⫾ 0.4 mEq/L). Flecainide infusion was prematurely interrupted in 1 case because of significant QRS widening (from 100 to 160 ms; left bundle branch block morphology). The patient was excluded from analysis and evolved with no complications. The other 59 patients completed the study protocol and conformed to the analyzed cohort.
Time course of ECG patterns The ECG patterns at baseline was classified as follows: type II in 31 patients (53%), type III in 15 (25%), and category IV in 13 (22%). Category IV included 8 patients (14%) with normal ECG and 5 (8%) with incomplete right bundle branch block. During the first stage, we observed conversion to a type I ECG pattern in 4 patients (6.8%). The proportion of responders continued to increase in the next stages, with 7 patients (11.9%) in the second stage, 6 (10.2%) in the third stage, and 9 (15.3%) in the fourth stage (P o .01; Figure 2A). Figure 2B summarizes the ECG patterns recorded at the standard lead position throughout the recording time. It shows a tendency to increase the relative contribution of the type I ECG pattern with ongoing stages (P o .01). Interestingly, in some cases showing the type I ECG pattern at the standard lead position, this pattern was observed only during a limited period of time. Individually, whenever a patient displayed an induced type I ECG at a particular stage (first, second, or third), this pattern was usually maintained in further stages (5 patients [71%]).
Heart Rhythm, Vol 12, No 2, February 2015 However, 2 patients displayed a short-lasting type I ECG pattern during the recording time: (1) in one case it was observed from the first stage to the third, but not in the fourth stage (displaying a normal ECG; category IV); (2) in the other case type I ECG was observed only during the second stage, displaying type III ECG at baseline and type II ECG during all other stages. ECG patterns at the high precordial position increased sensitivity only in the last case during the third stage (Figure 2C). Overall, 11 patients (18.6%) displayed a type I ECG pattern anytime during the study protocol (cumulative responders considering standard and high precordial positions of the leads). Figure 3A shows that the cumulative percentage of responders increased with longer periods of ECG recording. During the first stage, we classified 4 patients (6.8%) as early responders. The ratio of responders increased to 7 patients (11.9%) in the second and third stages (3 delayed responders), and it increased to 11 patients (18.6%) in the fourth stage (P o .01). This means that 4 patients (6.8%; Figure 3B) displayed no type I ECG patterns during the first 30 minutes of the ECG recording but were identified as responders 90 minutes after the initiation of flecainide infusion (late responders). The k statistics measuring agreement between observers were 0.86 and 0.89 (P o .01 for both).
Time course of ECG intervals Flecainide infusion was followed by an increase in the PR, QRS, and QTc intervals from baseline (Table 1 and Figure 4). Although no significant changes in the heart rate were observed (Figure 4A), the PR interval significantly increased between consecutive stages (Figure 4B). However, this tendency to increase was reduced at the end of the third stage. The PR interval increased by 11.8 ⫾ 2.5 ms from baseline to the 10th minute (P o .01) and by 8.1 ⫾ 2.4 ms from the 10th to the 20th minute (P ¼ .01) and remained stable with a nonsignificant decrease of 1.7 ⫾ 2.5 ms between the 20th and the 30th minute (P ¼ 1.0). As shown in Figure 4B, differences between responders and nonresponders at every stage were not statistically significant. Similarly, the QTc intervals increased by 13.2 ⫾ 3.6 ms from baseline to the 10th minute (P o .01) and by 12.8 ⫾ 3.8 ms from the 10th to the 20th minute (P o .01) and decreased by 9.9 ⫾ 2.8 ms from the 20th to the 30th minute (P o .01) (Figure 4C); differences between responders and nonresponders at every stage were not statistically significant. As observed for the PR and QTc intervals, QRS width increased by 11.2 ⫾ 1.2 ms from baseline to the 10th minute (P o .01) and by 4 ⫾ 1.2 ms from the 10th to the 20th minute (P o .01) and decreased by 3.1 ⫾ 1.1 ms from the 20th to the 30th minute (P ¼ .02) (Figure 4D). However, paired comparisons at every stage found the QRS width to be larger for responders than for nonresponders (Table 1). Also, the ratio of increase was significantly higher for responders than for nonresponders during the first 20 minutes of the
Calvo et al
ECG Variability During Flecainide Testing
353
Figure 2 Electrocardiographic (ECG) patterns recorded at the standard lead position. A: Proportion of type I ECG recognized at every stage. B: ECG patterns recorded throughout the recording time. C: Two patients displayed transient type I ECG. The panel displays a representative case in which type I ECG was transiently observed during the second stage. Other patterns were observed during all other stages (including the fourth stage: type II ECG; not shown). However, the high precordial position of the leads increased sensitivity and revealed a type I ECG pattern during the third stage.
initiation of flecainide infusion (responder’s slope 9.5; 95% CI 4.8–14 vs nonresponder’s slope 6.4; 95% CI 4.5–8.4; P ¼ .02).
Predictors of late response No differences were found between responders and nonresponders with regard to age (44.5 ⫾ 13.3 years vs 37 ⫾
Figure 3 Cumulative results obtained across stages considering both standard and high precordial positions of the leads. A: Percentage of responders. B: Four patients displayed a type I electrocardiographic (ECG) pattern only in the fourth stage. Those patients were considered as late responders, with a representative example displayed in the panel. Note that during the first 30 minutes, ECG tracings displayed no type I ECG, which became visible only in the fourth stage.
354 Table 1
Heart Rhythm, Vol 12, No 2, February 2015 ECG intervals recorded throughout the recording time
Variable
Group
Baseline
10th minute
20th minute
30th minute
HR (beats/min)
Global Responders Nonresponders Paired P value Global Responders Nonresponders Paired P value Global Responders Nonresponders Paired P value Global Responders Nonresponders Paired P value
51.1 ⫾ 8.3 52.9 ⫾ 7.7 50.7 ⫾ 8.5 .4 171.3 ⫾ 20.3 173.7 ⫾ 23.1 170.7 ⫾ 19.9 .7 402.9 ⫾ 23.2 394.2 ⫾ 24.8 404.9 ⫾ 22.7 .17 87.9 ⫾ 10.5 98.5 ⫾ 9.6 85.5 ⫾ 9.2 o.01
51 ⫾ 8.2 51.3 ⫾ 6.8 51 ⫾ 8.5 .9 184.3 ⫾ 21.5 183.6 ⫾ 20.9 184.4 ⫾ 21.8 .9 412.4 ⫾ 27.1 413.2 ⫾ 24.5 412.2 ⫾ 27.9 .9 98.3 ⫾ 11.5 112.6 ⫾ 10.9 95.1 ⫾ 9 o.01
49.4 ⫾ 7.2 50.7 ⫾ 5.7 49.1 ⫾ 7.6 .5 191.3 ⫾ 21 193.27 ⫾ 17.7 190.7 ⫾ 21.6 .8 425.4 ⫾ 26.2 425.8 ⫾ 29.5 425.4 ⫾ 25.7 .9 101.9 ⫾ 12.5 117.4 ⫾ 10.7 98.4 ⫾ 10.1 o.01
50.5 ⫾ 8 50.9 ⫾ 6 50.3 ⫾ 8.7 .8 189.3 ⫾ 20.8 192 ⫾ 18.1 188.7 ⫾ 21.6 .6 416.9 ⫾ 25.7 413.8 ⫾ 29.7 417.6 ⫾ 25 .7 99.5 ⫾ 12.3 113.2 ⫾ 12.5 96.4 ⫾ 9.9 o.01
PR interval (ms)
QTc interval (ms)
QRS width (ms)
Within-subjects P value
Between-subjects P value
.06
.65
o.01
.76
o.01
.67
o.01
o.01
For every interval, the observed value in the entire population (global) is shown, as well as according to the response to provocative testing (responders vs nonresponders). Paired P values correspond to paired comparisons between responders and nonresponders for every interval and time. The within-subjects P value evaluates the statistical significance of changes observed in every patient across stages. The between-subjects P value evaluates the statistical significance of changes observed between groups (responders vs nonresponders). ECG ¼ electrocardiographic; HR ¼ heart rate; QTc ¼ corrected QT.
14.1 years; P ¼ .12), male sex (73% vs 71%; P ¼ .9), syncope (10% vs 39%; P ¼ .07), family history of SD in individuals younger than 45 years (46% vs 46%; P ¼ .98), family history of BrS (17% vs 20%; P ¼ .8), serum creatinine level (0.83 ⫾ 0.23 mg/dL vs 0.83 ⫾ 0.16 mg/ dL; P ¼ .9), or sodium ion (140.1 ⫾ 1.7 mEq/L vs 139.2 ⫾ 2.5 mEq/L; P ¼ .36). However, the serum potassium level was observed to be higher in responders (4.57 ⫾ 0.32 mEq/L vs 4.02 ⫾ 0.34 mEq/L; P o .01), with no differences between late and early or delayed responders (4.6 ⫾ 0.42 mEq/L vs 4.56 ⫾ 0.34 mEq/L; P ¼ .89). With regard to the distribution of ECG patterns at baseline, no differences were observed between responders and nonresponders (P ¼ NS). We also evaluated predictors of late responses in relation to differences in the QRS width between patients. For this analysis, we excluded early and delayed responders (as those patients could be effectively diagnosed during the first 30 minutes of the ECG recording) and evaluated the diagnostic yield of the maximum QRS width recorded after flecainide infusion (the maximum observed between measurements at the 10th, 20th, and 30th minute). Logistic binary regression revealed a significant association between the maximum QRS width and late responses to provocative testing (P o .01) with appropriate predictive performance of the model (Hosmer-Lemeshow test, χ2¼1.3; P ¼ .99). In receiver operating characteristic analysis, the area under the curve was found as 0.96 (95% CI 0.91–1; P o .01), with a maximum QRS width of 110 ms during the first 30 minutes being effective for identifying possible late responders (sensitivity 100%, specificity 85.6%, positive predictive value 88%, and negative predictive value 00%; Figure 5).
Genetic results Genetic data are summarized in Table 2. We identified a mutation affecting the SCN5A gene in 2 patients (18%; V160F
and T1709M): 1 delayed responder and 1 late responder. The incidence of CYP2D6 variants affecting flecainide metabolism was not different between responders and nonresponders (40% vs 54.5%; P ¼ 1.0). However, we observed the incidence of CYP2D6 variants to be lower in late responders than in early or delayed responders (0% vs 75% vs 100%; P ¼ .04). With regard to CYP3A5 variants, all patients were homozygous for an allele variant causing null enzyme activity (*3/*3).
Discussion The main results of our study indicate that response to flecainide in the diagnosis of BrS exhibits width variability between patients in terms of the time course of ECG changes. In the experimental setting of this study, most of the induced type I ECG patterns are observed during the first 30 minutes of the initiation of provocative testing. However, extended periods of recording time increase the percentage of positive testing from 11.9% to 18.6% in our study population. We demonstrate that the time course of ECG changes is characterized by significant changes in the QRS width, which could be used for the prediction of late responses. In this sense, when the QRS width measured across stages after flecainide infusion was not longer than 110 ms, the incidence of late responses did not exist in our population. We conclude that in the setting of an expected variability of ECG changes, the latter could help define appropriate recording times for every patient.
Increasing the diagnostic yield of flecainide challenge Even though provocative testing is recommended for the diagnosis of BrS, some concerns about the sensitivity and specificity of the technique arise.2 Also, differences have been observed between drugs in terms of their ability to
Calvo et al
ECG Variability During Flecainide Testing
355
Figure 4 Interval measurements performed over the electrocardiographic (ECG) records (averaged every 10 minutes from baseline to the 30th minute): (A) Heart rate; (B) PR interval; (C) corrected QT interval (Bazett’s formula); and (D) QRS width.
Figure 5 A: The QRS width increase was higher in responders than in nonresponders. Two representative examples are depicted in the panel. While the late responder displayed an increased QRS width (from 96 to 120 ms), the nonresponder displayed lower variability (from 93 to 100 ms). B: The maximum QRS width appropriately identified late responders in our population. CI ¼ confidence interval; NPV ¼ negative predictive value; PPV ¼ positive predictive value; ROC ¼ receiver operating characteristic.
356
Heart Rhythm, Vol 12, No 2, February 2015
Table 2 Description of time-dependent ECG changes, genetic variants, and flecainide metabolism affected by CYP2D6 variants8 Response to Patient flecainide
SCN5A CYP3A5 CYP2D6 Flecainide genotype genotype genotype metabolism
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Normal Normal Normal Normal Normal T1709M Normal Normal Normal Normal V160F — — — — —
Early Early Early Early Delayed Delayed Delayed Late Late Late Late Nonresponder Nonresponder Nonresponder Nonresponder Nonresponder
*3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3
*2/*10 *1/*1 *2/*10 *2/*10 *2/*41 *1/*6 *2/*10 *1/*2 *1/*1 *1/*1 *1/*1 *2/*10 *1/*2 *1/*2 *1/*2 *2/*10
Het-EM Hom-EM Het-EM Het-EM Het-EM Het-EM Het-EM Hom-EM Hom-EM Hom-EM Hom-EM Het-EM Hom-EM Hom-EM Hom-EM Het-EM
ECG ¼ electrocardiographic; Het-EM ¼ heterozygous extensive metabolizer; Hom-EM ¼ homozygous extensive metabolizer.
demonstrate an induced type I ECG. Flecainide is widely used in Europe and has been proven to be less efficient than ajmaline for the diagnosis.10 However, today it is clear that responses to drug infusion in patients with BrS not only depend on drug properties but also on recording time. Gray et al4 recently reported a patient displaying a positive response 90 minutes after the infusion of flecainide, an extended period of recording time, which falls out of the recommendations.2 Thus, it is clear that at least in some patients, up to 30 minutes of recording time is not enough. At this time the question raised is: How extended must the provocative testing be? It is difficult to ascertain because, in clinical practice, often the gold standard for diagnosis is provocative testing itself. Also, the half-life of flecainide is approximately 9–13 hours in humans, which causes the serum concentration to be in the therapeutic range within several hours of drug administration.11 Under these conditions, we may speculate that during this period it is still possible to obtain induced type I ECG patterns in patients with an initially negative response (first 30 minutes after flecainide infusion). However, there is a significant link between the subjacent electrophysiological properties and the ratio of changes in the ECG that could be used for appropriate timing. Shimizu et al12 demonstrated that class I antiarrhythmic drugs significantly increase the QRS width in patients with BrS and controls, with a more marked effect in patients with BrS. This result is consistent with our data, as patients with BrS in our population denoted a higher ratio of QRS widening in response to flecainide. The relative effect in responders was highest at the end of flecainide infusion, working according to the final result of provocative testing. Taking advantage of those physiological observations, we suggest that a maximum QRS width lower than 110 ms in response to flecainide infusion probably denotes minor effects over excitability and conducting properties of the heart in our population and could be effective for identifying
nonaffected patients. On the contrary, patients with BrS respond to flecainide infusion with a higher rate of QRS widening, which predicts a higher rate of positive and late responses in our population.
Determinants of flecainide metabolism and ECG responses Flecainide exhibits a narrow therapeutic window with several factors affecting its pharmacological effects.8 Gray et al4 recently highlighted potential modulatory effects of cytochrome P450 2D6 and 3A5 gene variants on the response to flecainide infusion, which might appropriately explain the time-dependent variability of ECG changes. In the case of CYP2D6, allele variants *1 and *2 are associated with normal enzyme activity. Patients carrying those alleles (*1/*1, *1/*2, or *2/*2) were considered as homozygous extensive metabolizers (Hom-EMs). In contrast, patients carrying 1 reduced function variant (mutant; ie, *6, *10, or *41) in combination with alleles *1 or *2 were considered as heterozygous extensive metabolizers (HetEMs).8 Currently, it is debatable whether differences between Hom-EMs and Het-EMs lead to clinically significant differences in the pharmacokinetics of flecainide in patients. One study analyzed the age-dependent effect of CYP2D6 variants on flecainide metabolism.13 In patients younger than 70 years, who were studied in our work, there were no significant differences in the metabolic ratio (flecainide/flecainide metabolites) between Hom-EMs and Het-EMs. However, the coexistence of CYP3A5 variants may also affect the biodisponibility of flecainide.14 Patients carrying the loss-offunction alleles CYP3A5*3/*3 show shorter time to peak concentration and larger area under the curve of plasma concentration than do those carrying the wild-type alleles (CYP3A5*1/*1). More important to the results observed in our work, the association between CYP3A5*3/*3 and the pharmacokinetics of flecainide is even more relevant in subjects with at least one CYP2D6 mutant allele, leading to larger area under the curve concentration. All patients in our study were homozygous for the mutant allele, causing loss-of-function allele CYP3A5*3/*3, which is the most frequent allele in our population.9 In the homogeneous distribution of this genetic variant, it is not possible to explain differences between our patients. However, as stated above, there is a complex interaction between CYP2D6 and CYP3A5 variants affecting the pharmacokinetics of flecainide.14 By reducing the metabolic activity of CYP2D6 enzymes, the elimination half-life of flecainide is increased, which eventually might explain the occurrence of delayed responses, as postulated by Gray et al.4 However, higher biodisponibility of flecainide is also observed from the beginning, which in fact might increase the ratio of early responses.11 Our data suggest that a reduction of function of the CYP2D6 enzyme promotes early transformation to type I ECG changes in patients with BrS, possibly by means of an increased biodisponibility of the drug that might be clinically relevant in the presence of additional genetic
Calvo et al
ECG Variability During Flecainide Testing
variants affecting flecainide metabolism as observed for CYP3A5*3/*3.14 Furthermore, a variety of noncontrolled factors can influence the effect on ECG tracings and explain discrepancies in the observed responses. In the physiological range, we found potassium concentration to be higher in responders than in nonresponders, which might promote ECG changes after flecainide infusion.15 However, unfortunately, electrolyte determinations were not performed at the time of testing. Therefore, we cannot ascertain that because of daily variability in the serum concentrations of potassium ion, the observed results reflect the electrolyte conditions at the time of testing.
Study limitations ECG recordings were limited to the first 30 minutes and a single ECG was recorded at the 90th minute. Thus, we were unable to identify transient changes in ECG traces between the 30th and the 90th minute. Short-lasting type I ECG patterns may be observed at the standard precordial position (Figure 2), and so we cannot ascertain whether the percentage of early or delayed responses will be higher by detecting short-lasting type I ECG patterns uniquely displayed at the high precordial position during the first and second stages. With regard to the ability of the maximum QRS width to predict late responses, the number of cases studied in this work is limited and so further validation is required. SCN5A sequencing was not routinely performed in nonresponders, and so an approximation of the sensitivity and specificity of provocative testing in our population was not evaluated. The conclusions drawn from this work apply to flecainide. For other antiarrhythmic drugs (ie, ajmaline or pilsicainide), further validation is required.
Conclusion Response to flecainide infusion exhibits time-dependent variability of ECG patterns and intervals. Longer periods of ECG recording increase the recognition probability of type I ECG, with the maximum width of the QRS interval being a potential predictor of late responses.
357
Acknowledgments We thank people working at the Arrhythmia Unit of the University Hospital of Asturias for assistance in the study and Margit Sperling for editorial support.
References 1. Antzelevitch C, Brugada P, Borggrefe M, et al. Brugada syndrome: report of the second consensus conference. Heart Rhythm 2005;2:429–440. 2. Obeyesekere MN, Klein GJ, Modi S, Leong-Sit P, Gula LJ, Yee R, Skanes AC, Krahn AD. How to perform and interpret provocative testing for the diagnosis of Brugada syndrome, long-QT syndrome, and catecholaminergic polymorphic ventricular tachycardia. Circ Arrhythm Electrophysiol 2011;4:958–964. 3. Mizusawa Y, Wilde AAM. Brugada syndrome. Circ Arrhythm Electrophysiol 2012;5:606–616. 4. Gray B, McGuire M, Semsarian C, Medi C. Late positive flecainide challenge test for Brugada syndrome. Heart Rhythm 2014;11:898–900. 5. de-Riva-Silva M, Montero-Cabezas JM, Fontenla-Cerezuela A, Salguero-Bodes R, López-Gil M, Arribas-Ynsaurriaga F. Delayed positive response to a flecainide test in a patient with suspected Brugada syndrome: a worrisome finding. Rev Esp Cardiol (Engl Ed) 2014;67:674–675. 6. Priori SG, Wilde AA, Horie M, et al. Executive summary: HRS/EHRA/APHRS expert consensus statement on the diagnosis and management of patients with inherited primary arrhythmia syndromes. Heart Rhythm 2013;10:e85–e108. 7. García-Castro M, García C, Reguero JR, Miar A, Rubín JM, Alvarez V, Morís C, Coto E. The spectrum of SCN5A gene mutations in Spanish Brugada syndrome patients. Rev Esp Cardiol (Engl Ed) 2010;63:856–859. 8. Zhou S-F. Polymorphism of human cytochrome P450 2D6 and its clinical significance, part II. Clin Pharmacokinet 2009;48:761–804. 9. Tavira B, Coto E, Garciá EC, et al. Pharmacogenetics of tacrolimus after renal transplantation: analysis of polymorphisms in genes encoding 16 drug metabolizing enzymes. Clin Chem Lab Med 2011;49:825–833. 10. Wolpert C, Echternach C, Veltmann C, Antzelevitch C, Thomas GP, Spehl S, Streitner F, Kuschyk J, Schimpf R, Haase KK, Borggrefe M. Intravenous drug challenge using flecainide and ajmaline in patients with Brugada syndrome. Heart Rhythm 2005;2:254–260. 11. Mikus G, Gross AS, Beckmann J, Hertrampf R, Gundert-Remy U, Eichelbaum M. The influence of the sparteine/debrisoquin phenotype on the disposition of flecainide. Clin Pharmacol Ther 1989;45:562–567. 12. Shimizu W, Antzelevitch C, Suyama K, Kurita T, Taguchi A, Aihara N, Takaki H, Sunagawa K, Kamakura S. Effect of sodium channel blockers on ST segment, QRS duration, and corrected QT interval in patients with Brugada syndrome. J Cardiovasc Electrophysiol 2000;11:1320–1329. 13. Doki K, Homma M, Kuga K, Aonuma K, Kohda Y. Effects of CYP2D6 genotypes on age-related change of flecainide metabolism: involvement of CYP1A2-mediated metabolism. Br J Clin Pharmacol 2009;68:89–96. 14. Hu M, Yang Y-L, Fok BSP, Chan S-W, Chu TTW, Poon EWM, Yin OQP, Lee VHL, Tomlinson B. Effects of CYP2D6*10, CYP3A5*3, CYP1A2*1F, and ABCB1 C3435T polymorphisms on the pharmacokinetics of flecainide in healthy Chinese subjects. Drug Metabol Drug Interact 2012;27:33–39. 15. Campbell TJ, Wyse KR, Hemsworth PD. Effects of hyperkalemia, acidosis, and hypoxia on the depression of maximum rate of depolarization by class I antiarrhythmic drugs in guinea pig myocardium: differential actions of class Ib and Ic agents. J Cardiovasc Pharmacol 1991;18:51–59.
CLINICAL PERSPECTIVES A recent consensus statement highlights that the diagnosis of Brugada syndrome relies on the identification of a type I electrocardiographic (ECG) pattern, either occurring spontaneously or induced by sodium-blocker infusion.6 To bridge the gap between the fluctuating nature of ECG recordings and its key role in the diagnosis of this inherited disease, we provide an appropriate description of sequential changes in the ECG tracings during and after provocative testing with flecainide. The later implies not only morphological changes leading to transition between the normal ECG and patterns III, II and I, but also involve the dynamical changes in QRS width that effectively discriminate between responders and nonresponders to flecainide infusion. Together they pave the way for a more individualized performance of provocative testing, which help in improving patient care by reducing the unpredictability of time-dependent unexpected responses. The methodology displayed in this work is clinically available under the standard conditions required for provocative testing, leading to direct translation of main results into clinical practice. However, further validation of data obtained from large cohorts of patients is required to increase reliability and provide standardization.
Introduction Brugada syndrome (BrS) is a leading cause of sudden death (SD) in young people, with the type I electrocardiographic (ECG) pattern being an essential clue for diagnosis.1 In patients with no spontaneous type I ECG, flecainide challenge has been promoted as a commonly used test in the clinic.2 However, the timedependent course of ECG patterns in patients with BrS display a well-known variability3 that has not been appropriately analyzed in the context of provocative testing with sodium blockers. Recently, 2 case reports4,5 have described the occurrence of late positive responses after flecainide infusion. According to the standard recommendations for provocative testing
width of 110 ms during the first 30 minutes being effective for identifying possible late responders (sensitivity 100%; specificity 85.6%; positive predictive value 88%; negative predictive value 100%). The incidence of CYP2D6 variants was lower in late responders than in early or delayed responders (0% vs 75% vs 100%; P ¼ .04), while a homogeneous distribution of CYP3A5*3/*3 was observed in our population. CONCLUSION Response to flecainide exhibits time-dependent variability of ECG patterns and intervals. Longer periods of ECG recording increase the recognition probability of type I ECG. KEYWORDS Brugada syndrome; Electrocardiogram; Provocative testing; Flecainide; Cytochrome P450 ABBREVIATIONS BrS ¼ Brugada syndrome; ECG ¼ electrocardiogram/ electrocardiographic; Het-EM ¼ heterozygous extensive metabolizer; Hom-EM ¼ homozygous extensive metabolizer; QTc ¼ corrected QT; SD ¼ sudden death (Heart Rhythm 2015;12:350–357) rights reserved.
I
2015 Heart Rhythm Society. All
performance (10–30 minutes of waiting time if the test has a negative result), these tests would have been considered as negative2; however, these results are false negatives, with implications for clinical management (ie, perform lifestyle measurements, avoid drugs with potential adverse effects, and prompt treatment of fever episodes). To quantify the incidence of late responses, we analyzed the time course of ECG changes during and 90 minutes after flecainide infusion. In addition, clinical, ECG, and genetic factors affecting flecainide metabolism were analyzed to describe predictors of time-dependent responses.
Methods Address reprint requests and correspondence: Dr David Calvo Cuervo, Arrhythmia Unit, Cardiology Department, Hospital Universitario Central de Asturias, Calle Carretera de Rubín, s/n, 33011 Oviedo, Spain. E-mail address: [email protected].
1547-5271/$-see front matter B 2015 Heart Rhythm Society. All rights reserved.
Population and recording protocol In the absence of a spontaneous type I ECG pattern, patients with suspected BrS were admitted for provocative testing http://dx.doi.org/10.1016/j.hrthm.2014.10.036
Calvo et al
ECG Variability During Flecainide Testing
with flecainide. According to the standard protocol used in our institution, the clinical profile and serum levels (creatinine and electrolytes) were previously obtained at the outpatient clinic, 1–3 months before the admission. Intravenous flecainide was continuously infused at a rate of 2.0 mg/kg bodyweight over 10 minutes (limited by a maximum dosage of 150 mg). Before flecainide infusion, we checked for the absence of type I ECG, both at the standard precordial position (V1 and V2 at the third intercostal space) and at the high precordial position (V1 and V2 at the second intercostal space). Thereafter, the standard 12-lead ECG was continuously recorded during 30 minutes in the electrophysiology laboratory (sample rate 1 kHz; bandpass filtered 0.05–150 Hz; EP Tracer v1.05.v3, CardioTek, Maastricht, The Netherlands) from the initiation of flecainide infusion, which enabled us to better characterize transitions between ECG patterns because types II and III were mainly defined at the standard precordial positions.1 At the end of this period, we again explored the high precordial position for better sensitivity. Thereafter, the positions of the leads on the body surface were marked and a new ECG was recorded 60 minutes later (Figure 1). The study was approved by the ethics committee, and subjects gave informed consent. ECG tracings were analyzed by 3 independent cardiac electrophysiologists and classified by consensus according to the published recommendations as type I, II, and III.1 For the purpose of this study, we classified the normal ECG as category IV. This category also includes subtle deviations from normality as the incomplete right bundle branch block. The response to flecainide infusion was analyzed continuously during the first 30 minutes. ECG changes observed during flecainide infusion (first 10 minutes of the ECG recording) were considered to occur at the first stage of the test. The second and third stage changes were those changes that occurred between 10 and 20 minutes and between 20 and 30 minutes, respectively. Finally, the fourth stage changes were considered according to the single 12-lead ECG recorded 90 minutes after the initiation of flecainide infusion. In order to assign an ECG pattern as defining the result of the test at every stage, we selected the highest order pattern observed during the recording time, considering type I as the highest order pattern and category IV as the lowest one. Provocative testing was considered to display a positive response if the patient exhibited a type I ECG pattern anytime during the protocol. Those patients displaying a positive response during the first stage were defined as early responders. On the contrary, those patients displaying a
351 positive response during the second and third stages were defined as delayed responders. Finally, those patients displaying a positive response only in the fourth stage were defined as late responders. All other patients were defined as nonresponders. During the 30 minutes of continuous ECG recording, average heart rate, PR, QRS width, and corrected QT (QTc) intervals (Bazett’s formula) were measured every 10 minutes from baseline (using integrated calipers over the digital records; speed 100 mm/s) at lead II. Measurements of 3 consecutive cycles were averaged. To avoid potential bias, interval measurements were performed blinded to the final result of provocative testing. Also, all ECG channels other than lead II were removed from the screen at this time to avoid the researcher perceiving the final result of provocative testing.
Genetic studies According to current guidelines,6 the whole coding sequence of SCN5A was amplified and sequenced in responders as described previously.7 The nucleotide changes were classified as possible mutations if (1) reported previously (www. ensembl.org) or (2) they had an effect on the protein sequence, either by changing the amino acid or by introducing aberrant transcript sequences. Single amino acid changes were considered as mutations on the basis of the prediction programs SIFT (J. Craig Venter Institute, La Jolla, CA) and Polyphen (Harvard, Massachusetts). In the whole cohort of responders and a randomly selected subset of nonresponders (5 patients), we also determined single nucleotide polymorphisms in CYP3A5 and CYP2D6 genes because they are mainly involved in flecainide metabolism.4,8 The CYP3A5*3 allele (SNP rs776746) was genotyped with a real-time TaqMan polymerase chain reaction assay (assay ID C_25201809_30, Applied Biosystems, California, USA) as reported previously.9 To determine the CYP2D6 genotype, we amplified and sequenced the 9 coding exons with primers specific for this gene (further details are available on request to the corresponding author).
Statistical analysis Categorical variables are reported as numbers and percentages. For the purpose of the analysis and better clinical interpretation, we considered the ECG patterns as ordinal variables of arbitrary units by assigning a value from 1 to 4 to patterns I, II, III, and category IV, respectively. The ECG
Figure 1 Study flowchart showing different stages of electrocardiographic (ECG) recording. Interval measurements were performed over the digital records at baseline, 10, 20, and 30 minutes (100 mm/s).
352 patterns were analyzed using the Cochran Q or the Friedman test, as appropriate. Continuous variables are reported as mean ⫾ standard deviation. We used a mixed-effects regression model with random intercepts to test changes in the ECG intervals across stages. Statistical significance was analyzed between stages (within-subjects significance) and also according to the final result of provocative testing (responders vs nonresponders; between-subjects significance). The Bonferroni correction was used for multiple comparisons. The χ2 and Student t tests were used for univariate analysis to compare different variables as predictors of the result of provocative testing. Also, logistic binary regression was used for analysis and the predictive performance of variables was assessed by receiver operating characteristic analysis and the Hosmer-Lemeshow test. Agreement between investigators was measured by the k statistic. Analysis was performed with SPSS software (version 20, SPSS Inc, Chicago, IL), and P o .05 was defined as statistically significant.
Results From November 2012 through March 2014, we prospectively studied 60 consecutive patients (age 38.4 ⫾ 14.1 years; male sex 71.2%) because of the presence of an abnormal ECG (65%; type II or III ECG), family history of BrS (40%), and/or syncope (23%). There were no patients with history of SD or ventricular arrhythmias, renal impairment (creatinine 0.83 ⫾ 0.21 mg/dL), or electrolyte abnormalities (sodium 139.5 ⫾ 2.4 mEq/L and potassium 4.2 ⫾ 0.4 mEq/L). Flecainide infusion was prematurely interrupted in 1 case because of significant QRS widening (from 100 to 160 ms; left bundle branch block morphology). The patient was excluded from analysis and evolved with no complications. The other 59 patients completed the study protocol and conformed to the analyzed cohort.
Time course of ECG patterns The ECG patterns at baseline was classified as follows: type II in 31 patients (53%), type III in 15 (25%), and category IV in 13 (22%). Category IV included 8 patients (14%) with normal ECG and 5 (8%) with incomplete right bundle branch block. During the first stage, we observed conversion to a type I ECG pattern in 4 patients (6.8%). The proportion of responders continued to increase in the next stages, with 7 patients (11.9%) in the second stage, 6 (10.2%) in the third stage, and 9 (15.3%) in the fourth stage (P o .01; Figure 2A). Figure 2B summarizes the ECG patterns recorded at the standard lead position throughout the recording time. It shows a tendency to increase the relative contribution of the type I ECG pattern with ongoing stages (P o .01). Interestingly, in some cases showing the type I ECG pattern at the standard lead position, this pattern was observed only during a limited period of time. Individually, whenever a patient displayed an induced type I ECG at a particular stage (first, second, or third), this pattern was usually maintained in further stages (5 patients [71%]).
Heart Rhythm, Vol 12, No 2, February 2015 However, 2 patients displayed a short-lasting type I ECG pattern during the recording time: (1) in one case it was observed from the first stage to the third, but not in the fourth stage (displaying a normal ECG; category IV); (2) in the other case type I ECG was observed only during the second stage, displaying type III ECG at baseline and type II ECG during all other stages. ECG patterns at the high precordial position increased sensitivity only in the last case during the third stage (Figure 2C). Overall, 11 patients (18.6%) displayed a type I ECG pattern anytime during the study protocol (cumulative responders considering standard and high precordial positions of the leads). Figure 3A shows that the cumulative percentage of responders increased with longer periods of ECG recording. During the first stage, we classified 4 patients (6.8%) as early responders. The ratio of responders increased to 7 patients (11.9%) in the second and third stages (3 delayed responders), and it increased to 11 patients (18.6%) in the fourth stage (P o .01). This means that 4 patients (6.8%; Figure 3B) displayed no type I ECG patterns during the first 30 minutes of the ECG recording but were identified as responders 90 minutes after the initiation of flecainide infusion (late responders). The k statistics measuring agreement between observers were 0.86 and 0.89 (P o .01 for both).
Time course of ECG intervals Flecainide infusion was followed by an increase in the PR, QRS, and QTc intervals from baseline (Table 1 and Figure 4). Although no significant changes in the heart rate were observed (Figure 4A), the PR interval significantly increased between consecutive stages (Figure 4B). However, this tendency to increase was reduced at the end of the third stage. The PR interval increased by 11.8 ⫾ 2.5 ms from baseline to the 10th minute (P o .01) and by 8.1 ⫾ 2.4 ms from the 10th to the 20th minute (P ¼ .01) and remained stable with a nonsignificant decrease of 1.7 ⫾ 2.5 ms between the 20th and the 30th minute (P ¼ 1.0). As shown in Figure 4B, differences between responders and nonresponders at every stage were not statistically significant. Similarly, the QTc intervals increased by 13.2 ⫾ 3.6 ms from baseline to the 10th minute (P o .01) and by 12.8 ⫾ 3.8 ms from the 10th to the 20th minute (P o .01) and decreased by 9.9 ⫾ 2.8 ms from the 20th to the 30th minute (P o .01) (Figure 4C); differences between responders and nonresponders at every stage were not statistically significant. As observed for the PR and QTc intervals, QRS width increased by 11.2 ⫾ 1.2 ms from baseline to the 10th minute (P o .01) and by 4 ⫾ 1.2 ms from the 10th to the 20th minute (P o .01) and decreased by 3.1 ⫾ 1.1 ms from the 20th to the 30th minute (P ¼ .02) (Figure 4D). However, paired comparisons at every stage found the QRS width to be larger for responders than for nonresponders (Table 1). Also, the ratio of increase was significantly higher for responders than for nonresponders during the first 20 minutes of the
Calvo et al
ECG Variability During Flecainide Testing
353
Figure 2 Electrocardiographic (ECG) patterns recorded at the standard lead position. A: Proportion of type I ECG recognized at every stage. B: ECG patterns recorded throughout the recording time. C: Two patients displayed transient type I ECG. The panel displays a representative case in which type I ECG was transiently observed during the second stage. Other patterns were observed during all other stages (including the fourth stage: type II ECG; not shown). However, the high precordial position of the leads increased sensitivity and revealed a type I ECG pattern during the third stage.
initiation of flecainide infusion (responder’s slope 9.5; 95% CI 4.8–14 vs nonresponder’s slope 6.4; 95% CI 4.5–8.4; P ¼ .02).
Predictors of late response No differences were found between responders and nonresponders with regard to age (44.5 ⫾ 13.3 years vs 37 ⫾
Figure 3 Cumulative results obtained across stages considering both standard and high precordial positions of the leads. A: Percentage of responders. B: Four patients displayed a type I electrocardiographic (ECG) pattern only in the fourth stage. Those patients were considered as late responders, with a representative example displayed in the panel. Note that during the first 30 minutes, ECG tracings displayed no type I ECG, which became visible only in the fourth stage.
354 Table 1
Heart Rhythm, Vol 12, No 2, February 2015 ECG intervals recorded throughout the recording time
Variable
Group
Baseline
10th minute
20th minute
30th minute
HR (beats/min)
Global Responders Nonresponders Paired P value Global Responders Nonresponders Paired P value Global Responders Nonresponders Paired P value Global Responders Nonresponders Paired P value
51.1 ⫾ 8.3 52.9 ⫾ 7.7 50.7 ⫾ 8.5 .4 171.3 ⫾ 20.3 173.7 ⫾ 23.1 170.7 ⫾ 19.9 .7 402.9 ⫾ 23.2 394.2 ⫾ 24.8 404.9 ⫾ 22.7 .17 87.9 ⫾ 10.5 98.5 ⫾ 9.6 85.5 ⫾ 9.2 o.01
51 ⫾ 8.2 51.3 ⫾ 6.8 51 ⫾ 8.5 .9 184.3 ⫾ 21.5 183.6 ⫾ 20.9 184.4 ⫾ 21.8 .9 412.4 ⫾ 27.1 413.2 ⫾ 24.5 412.2 ⫾ 27.9 .9 98.3 ⫾ 11.5 112.6 ⫾ 10.9 95.1 ⫾ 9 o.01
49.4 ⫾ 7.2 50.7 ⫾ 5.7 49.1 ⫾ 7.6 .5 191.3 ⫾ 21 193.27 ⫾ 17.7 190.7 ⫾ 21.6 .8 425.4 ⫾ 26.2 425.8 ⫾ 29.5 425.4 ⫾ 25.7 .9 101.9 ⫾ 12.5 117.4 ⫾ 10.7 98.4 ⫾ 10.1 o.01
50.5 ⫾ 8 50.9 ⫾ 6 50.3 ⫾ 8.7 .8 189.3 ⫾ 20.8 192 ⫾ 18.1 188.7 ⫾ 21.6 .6 416.9 ⫾ 25.7 413.8 ⫾ 29.7 417.6 ⫾ 25 .7 99.5 ⫾ 12.3 113.2 ⫾ 12.5 96.4 ⫾ 9.9 o.01
PR interval (ms)
QTc interval (ms)
QRS width (ms)
Within-subjects P value
Between-subjects P value
.06
.65
o.01
.76
o.01
.67
o.01
o.01
For every interval, the observed value in the entire population (global) is shown, as well as according to the response to provocative testing (responders vs nonresponders). Paired P values correspond to paired comparisons between responders and nonresponders for every interval and time. The within-subjects P value evaluates the statistical significance of changes observed in every patient across stages. The between-subjects P value evaluates the statistical significance of changes observed between groups (responders vs nonresponders). ECG ¼ electrocardiographic; HR ¼ heart rate; QTc ¼ corrected QT.
14.1 years; P ¼ .12), male sex (73% vs 71%; P ¼ .9), syncope (10% vs 39%; P ¼ .07), family history of SD in individuals younger than 45 years (46% vs 46%; P ¼ .98), family history of BrS (17% vs 20%; P ¼ .8), serum creatinine level (0.83 ⫾ 0.23 mg/dL vs 0.83 ⫾ 0.16 mg/ dL; P ¼ .9), or sodium ion (140.1 ⫾ 1.7 mEq/L vs 139.2 ⫾ 2.5 mEq/L; P ¼ .36). However, the serum potassium level was observed to be higher in responders (4.57 ⫾ 0.32 mEq/L vs 4.02 ⫾ 0.34 mEq/L; P o .01), with no differences between late and early or delayed responders (4.6 ⫾ 0.42 mEq/L vs 4.56 ⫾ 0.34 mEq/L; P ¼ .89). With regard to the distribution of ECG patterns at baseline, no differences were observed between responders and nonresponders (P ¼ NS). We also evaluated predictors of late responses in relation to differences in the QRS width between patients. For this analysis, we excluded early and delayed responders (as those patients could be effectively diagnosed during the first 30 minutes of the ECG recording) and evaluated the diagnostic yield of the maximum QRS width recorded after flecainide infusion (the maximum observed between measurements at the 10th, 20th, and 30th minute). Logistic binary regression revealed a significant association between the maximum QRS width and late responses to provocative testing (P o .01) with appropriate predictive performance of the model (Hosmer-Lemeshow test, χ2¼1.3; P ¼ .99). In receiver operating characteristic analysis, the area under the curve was found as 0.96 (95% CI 0.91–1; P o .01), with a maximum QRS width of 110 ms during the first 30 minutes being effective for identifying possible late responders (sensitivity 100%, specificity 85.6%, positive predictive value 88%, and negative predictive value 00%; Figure 5).
Genetic results Genetic data are summarized in Table 2. We identified a mutation affecting the SCN5A gene in 2 patients (18%; V160F
and T1709M): 1 delayed responder and 1 late responder. The incidence of CYP2D6 variants affecting flecainide metabolism was not different between responders and nonresponders (40% vs 54.5%; P ¼ 1.0). However, we observed the incidence of CYP2D6 variants to be lower in late responders than in early or delayed responders (0% vs 75% vs 100%; P ¼ .04). With regard to CYP3A5 variants, all patients were homozygous for an allele variant causing null enzyme activity (*3/*3).
Discussion The main results of our study indicate that response to flecainide in the diagnosis of BrS exhibits width variability between patients in terms of the time course of ECG changes. In the experimental setting of this study, most of the induced type I ECG patterns are observed during the first 30 minutes of the initiation of provocative testing. However, extended periods of recording time increase the percentage of positive testing from 11.9% to 18.6% in our study population. We demonstrate that the time course of ECG changes is characterized by significant changes in the QRS width, which could be used for the prediction of late responses. In this sense, when the QRS width measured across stages after flecainide infusion was not longer than 110 ms, the incidence of late responses did not exist in our population. We conclude that in the setting of an expected variability of ECG changes, the latter could help define appropriate recording times for every patient.
Increasing the diagnostic yield of flecainide challenge Even though provocative testing is recommended for the diagnosis of BrS, some concerns about the sensitivity and specificity of the technique arise.2 Also, differences have been observed between drugs in terms of their ability to
Calvo et al
ECG Variability During Flecainide Testing
355
Figure 4 Interval measurements performed over the electrocardiographic (ECG) records (averaged every 10 minutes from baseline to the 30th minute): (A) Heart rate; (B) PR interval; (C) corrected QT interval (Bazett’s formula); and (D) QRS width.
Figure 5 A: The QRS width increase was higher in responders than in nonresponders. Two representative examples are depicted in the panel. While the late responder displayed an increased QRS width (from 96 to 120 ms), the nonresponder displayed lower variability (from 93 to 100 ms). B: The maximum QRS width appropriately identified late responders in our population. CI ¼ confidence interval; NPV ¼ negative predictive value; PPV ¼ positive predictive value; ROC ¼ receiver operating characteristic.
356
Heart Rhythm, Vol 12, No 2, February 2015
Table 2 Description of time-dependent ECG changes, genetic variants, and flecainide metabolism affected by CYP2D6 variants8 Response to Patient flecainide
SCN5A CYP3A5 CYP2D6 Flecainide genotype genotype genotype metabolism
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Normal Normal Normal Normal Normal T1709M Normal Normal Normal Normal V160F — — — — —
Early Early Early Early Delayed Delayed Delayed Late Late Late Late Nonresponder Nonresponder Nonresponder Nonresponder Nonresponder
*3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3 *3/*3
*2/*10 *1/*1 *2/*10 *2/*10 *2/*41 *1/*6 *2/*10 *1/*2 *1/*1 *1/*1 *1/*1 *2/*10 *1/*2 *1/*2 *1/*2 *2/*10
Het-EM Hom-EM Het-EM Het-EM Het-EM Het-EM Het-EM Hom-EM Hom-EM Hom-EM Hom-EM Het-EM Hom-EM Hom-EM Hom-EM Het-EM
ECG ¼ electrocardiographic; Het-EM ¼ heterozygous extensive metabolizer; Hom-EM ¼ homozygous extensive metabolizer.
demonstrate an induced type I ECG. Flecainide is widely used in Europe and has been proven to be less efficient than ajmaline for the diagnosis.10 However, today it is clear that responses to drug infusion in patients with BrS not only depend on drug properties but also on recording time. Gray et al4 recently reported a patient displaying a positive response 90 minutes after the infusion of flecainide, an extended period of recording time, which falls out of the recommendations.2 Thus, it is clear that at least in some patients, up to 30 minutes of recording time is not enough. At this time the question raised is: How extended must the provocative testing be? It is difficult to ascertain because, in clinical practice, often the gold standard for diagnosis is provocative testing itself. Also, the half-life of flecainide is approximately 9–13 hours in humans, which causes the serum concentration to be in the therapeutic range within several hours of drug administration.11 Under these conditions, we may speculate that during this period it is still possible to obtain induced type I ECG patterns in patients with an initially negative response (first 30 minutes after flecainide infusion). However, there is a significant link between the subjacent electrophysiological properties and the ratio of changes in the ECG that could be used for appropriate timing. Shimizu et al12 demonstrated that class I antiarrhythmic drugs significantly increase the QRS width in patients with BrS and controls, with a more marked effect in patients with BrS. This result is consistent with our data, as patients with BrS in our population denoted a higher ratio of QRS widening in response to flecainide. The relative effect in responders was highest at the end of flecainide infusion, working according to the final result of provocative testing. Taking advantage of those physiological observations, we suggest that a maximum QRS width lower than 110 ms in response to flecainide infusion probably denotes minor effects over excitability and conducting properties of the heart in our population and could be effective for identifying
nonaffected patients. On the contrary, patients with BrS respond to flecainide infusion with a higher rate of QRS widening, which predicts a higher rate of positive and late responses in our population.
Determinants of flecainide metabolism and ECG responses Flecainide exhibits a narrow therapeutic window with several factors affecting its pharmacological effects.8 Gray et al4 recently highlighted potential modulatory effects of cytochrome P450 2D6 and 3A5 gene variants on the response to flecainide infusion, which might appropriately explain the time-dependent variability of ECG changes. In the case of CYP2D6, allele variants *1 and *2 are associated with normal enzyme activity. Patients carrying those alleles (*1/*1, *1/*2, or *2/*2) were considered as homozygous extensive metabolizers (Hom-EMs). In contrast, patients carrying 1 reduced function variant (mutant; ie, *6, *10, or *41) in combination with alleles *1 or *2 were considered as heterozygous extensive metabolizers (HetEMs).8 Currently, it is debatable whether differences between Hom-EMs and Het-EMs lead to clinically significant differences in the pharmacokinetics of flecainide in patients. One study analyzed the age-dependent effect of CYP2D6 variants on flecainide metabolism.13 In patients younger than 70 years, who were studied in our work, there were no significant differences in the metabolic ratio (flecainide/flecainide metabolites) between Hom-EMs and Het-EMs. However, the coexistence of CYP3A5 variants may also affect the biodisponibility of flecainide.14 Patients carrying the loss-offunction alleles CYP3A5*3/*3 show shorter time to peak concentration and larger area under the curve of plasma concentration than do those carrying the wild-type alleles (CYP3A5*1/*1). More important to the results observed in our work, the association between CYP3A5*3/*3 and the pharmacokinetics of flecainide is even more relevant in subjects with at least one CYP2D6 mutant allele, leading to larger area under the curve concentration. All patients in our study were homozygous for the mutant allele, causing loss-of-function allele CYP3A5*3/*3, which is the most frequent allele in our population.9 In the homogeneous distribution of this genetic variant, it is not possible to explain differences between our patients. However, as stated above, there is a complex interaction between CYP2D6 and CYP3A5 variants affecting the pharmacokinetics of flecainide.14 By reducing the metabolic activity of CYP2D6 enzymes, the elimination half-life of flecainide is increased, which eventually might explain the occurrence of delayed responses, as postulated by Gray et al.4 However, higher biodisponibility of flecainide is also observed from the beginning, which in fact might increase the ratio of early responses.11 Our data suggest that a reduction of function of the CYP2D6 enzyme promotes early transformation to type I ECG changes in patients with BrS, possibly by means of an increased biodisponibility of the drug that might be clinically relevant in the presence of additional genetic
Calvo et al
ECG Variability During Flecainide Testing
variants affecting flecainide metabolism as observed for CYP3A5*3/*3.14 Furthermore, a variety of noncontrolled factors can influence the effect on ECG tracings and explain discrepancies in the observed responses. In the physiological range, we found potassium concentration to be higher in responders than in nonresponders, which might promote ECG changes after flecainide infusion.15 However, unfortunately, electrolyte determinations were not performed at the time of testing. Therefore, we cannot ascertain that because of daily variability in the serum concentrations of potassium ion, the observed results reflect the electrolyte conditions at the time of testing.
Study limitations ECG recordings were limited to the first 30 minutes and a single ECG was recorded at the 90th minute. Thus, we were unable to identify transient changes in ECG traces between the 30th and the 90th minute. Short-lasting type I ECG patterns may be observed at the standard precordial position (Figure 2), and so we cannot ascertain whether the percentage of early or delayed responses will be higher by detecting short-lasting type I ECG patterns uniquely displayed at the high precordial position during the first and second stages. With regard to the ability of the maximum QRS width to predict late responses, the number of cases studied in this work is limited and so further validation is required. SCN5A sequencing was not routinely performed in nonresponders, and so an approximation of the sensitivity and specificity of provocative testing in our population was not evaluated. The conclusions drawn from this work apply to flecainide. For other antiarrhythmic drugs (ie, ajmaline or pilsicainide), further validation is required.
Conclusion Response to flecainide infusion exhibits time-dependent variability of ECG patterns and intervals. Longer periods of ECG recording increase the recognition probability of type I ECG, with the maximum width of the QRS interval being a potential predictor of late responses.
357
Acknowledgments We thank people working at the Arrhythmia Unit of the University Hospital of Asturias for assistance in the study and Margit Sperling for editorial support.
References 1. Antzelevitch C, Brugada P, Borggrefe M, et al. Brugada syndrome: report of the second consensus conference. Heart Rhythm 2005;2:429–440. 2. Obeyesekere MN, Klein GJ, Modi S, Leong-Sit P, Gula LJ, Yee R, Skanes AC, Krahn AD. How to perform and interpret provocative testing for the diagnosis of Brugada syndrome, long-QT syndrome, and catecholaminergic polymorphic ventricular tachycardia. Circ Arrhythm Electrophysiol 2011;4:958–964. 3. Mizusawa Y, Wilde AAM. Brugada syndrome. Circ Arrhythm Electrophysiol 2012;5:606–616. 4. Gray B, McGuire M, Semsarian C, Medi C. Late positive flecainide challenge test for Brugada syndrome. Heart Rhythm 2014;11:898–900. 5. de-Riva-Silva M, Montero-Cabezas JM, Fontenla-Cerezuela A, Salguero-Bodes R, López-Gil M, Arribas-Ynsaurriaga F. Delayed positive response to a flecainide test in a patient with suspected Brugada syndrome: a worrisome finding. Rev Esp Cardiol (Engl Ed) 2014;67:674–675. 6. Priori SG, Wilde AA, Horie M, et al. Executive summary: HRS/EHRA/APHRS expert consensus statement on the diagnosis and management of patients with inherited primary arrhythmia syndromes. Heart Rhythm 2013;10:e85–e108. 7. García-Castro M, García C, Reguero JR, Miar A, Rubín JM, Alvarez V, Morís C, Coto E. The spectrum of SCN5A gene mutations in Spanish Brugada syndrome patients. Rev Esp Cardiol (Engl Ed) 2010;63:856–859. 8. Zhou S-F. Polymorphism of human cytochrome P450 2D6 and its clinical significance, part II. Clin Pharmacokinet 2009;48:761–804. 9. Tavira B, Coto E, Garciá EC, et al. Pharmacogenetics of tacrolimus after renal transplantation: analysis of polymorphisms in genes encoding 16 drug metabolizing enzymes. Clin Chem Lab Med 2011;49:825–833. 10. Wolpert C, Echternach C, Veltmann C, Antzelevitch C, Thomas GP, Spehl S, Streitner F, Kuschyk J, Schimpf R, Haase KK, Borggrefe M. Intravenous drug challenge using flecainide and ajmaline in patients with Brugada syndrome. Heart Rhythm 2005;2:254–260. 11. Mikus G, Gross AS, Beckmann J, Hertrampf R, Gundert-Remy U, Eichelbaum M. The influence of the sparteine/debrisoquin phenotype on the disposition of flecainide. Clin Pharmacol Ther 1989;45:562–567. 12. Shimizu W, Antzelevitch C, Suyama K, Kurita T, Taguchi A, Aihara N, Takaki H, Sunagawa K, Kamakura S. Effect of sodium channel blockers on ST segment, QRS duration, and corrected QT interval in patients with Brugada syndrome. J Cardiovasc Electrophysiol 2000;11:1320–1329. 13. Doki K, Homma M, Kuga K, Aonuma K, Kohda Y. Effects of CYP2D6 genotypes on age-related change of flecainide metabolism: involvement of CYP1A2-mediated metabolism. Br J Clin Pharmacol 2009;68:89–96. 14. Hu M, Yang Y-L, Fok BSP, Chan S-W, Chu TTW, Poon EWM, Yin OQP, Lee VHL, Tomlinson B. Effects of CYP2D6*10, CYP3A5*3, CYP1A2*1F, and ABCB1 C3435T polymorphisms on the pharmacokinetics of flecainide in healthy Chinese subjects. Drug Metabol Drug Interact 2012;27:33–39. 15. Campbell TJ, Wyse KR, Hemsworth PD. Effects of hyperkalemia, acidosis, and hypoxia on the depression of maximum rate of depolarization by class I antiarrhythmic drugs in guinea pig myocardium: differential actions of class Ib and Ic agents. J Cardiovasc Pharmacol 1991;18:51–59.
CLINICAL PERSPECTIVES A recent consensus statement highlights that the diagnosis of Brugada syndrome relies on the identification of a type I electrocardiographic (ECG) pattern, either occurring spontaneously or induced by sodium-blocker infusion.6 To bridge the gap between the fluctuating nature of ECG recordings and its key role in the diagnosis of this inherited disease, we provide an appropriate description of sequential changes in the ECG tracings during and after provocative testing with flecainide. The later implies not only morphological changes leading to transition between the normal ECG and patterns III, II and I, but also involve the dynamical changes in QRS width that effectively discriminate between responders and nonresponders to flecainide infusion. Together they pave the way for a more individualized performance of provocative testing, which help in improving patient care by reducing the unpredictability of time-dependent unexpected responses. The methodology displayed in this work is clinically available under the standard conditions required for provocative testing, leading to direct translation of main results into clinical practice. However, further validation of data obtained from large cohorts of patients is required to increase reliability and provide standardization.
