Time-dependent responses to provocative testing withflecainide in the diagnosis of Brugada syndromeDavid 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 deInvestigación 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 inter-pretation of provocative testing.

OBJECTIVE The aim of this study was to characterize ECG changesduring and after flecainide infusion.

METHODS We studied 59 consecutive patients. The ECG wascontinuously analyzed during the first 30 minutes of provocativetesting, and a single ECG was recorded 60 minutes later. Weanalyzed CYP2D6 and CYP3A5 variants affecting flecainide metab-olism and performed blinded measurements at lead II.

RESULTS At baseline, ECG patterns were classified as follows: typeII in 31 patients (53%), type III in 15 (25%), and normal ECG in 13(22%). Because of induction of type I ECG, the percentage ofresponders progressively increased with longer recording timeperiods (6.8% in 10 minutes vs 11.9% in 20–30 minutes vs18.6% in 90 minutes; P o .01). Four patients displayed a lateresponse, which was evidenced 90 minutes after the initiation ofprovocative testing. QRS width differentially increased betweenresponders and nonresponders (P o .01), with a maximum QRS

Address reprint requests and correspondence: Dr David Calvo Cuervo,Arrhythmia Unit, Cardiology Department, Hospital Universitario Central deAsturias, Calle Carretera de Rubín, s/n, 33011 Oviedo, Spain. E-mailaddress: dcalvo307@secardiologia.es.

1547-5271/$-see front matter B 2015 Heart Rhythm Society. All rights reserved.

width of 110 ms during the first 30 minutes being effective foridentifying possible late responders (sensitivity 100%; specificity85.6%; positive predictive value 88%; negative predictive value100%). The incidence of CYP2D6 variants was lower in lateresponders than in early or delayed responders (0% vs 75% vs100%; P¼ .04), while a homogeneous distribution of CYP3A5*3/*3was observed in our population.

CONCLUSION Response to flecainide exhibits time-dependentvariability of ECG patterns and intervals. Longer periods of ECGrecording increase the recognition probability of type I ECG.

KEYWORDS Brugada syndrome; Electrocardiogram; Provocativetesting; Flecainide; Cytochrome P450

ABBREVIATIONS BrS¼ Brugada syndrome; ECG¼ 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) I 2015 Heart Rhythm Society. Allrights reserved.

IntroductionBrugada 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 nospontaneous type I ECG, flecainide challenge has been promotedas a commonly used test in the clinic.2 However, the time-dependent course of ECG patterns in patients with BrS display awell-known variability3 that has not been appropriately analyzedin the context of provocative testing with sodium blockers.

Recently, 2 case reports4,5 have described the occurrenceof late positive responses after flecainide infusion. Accordingto the standard recommendations for provocative testing

performance (10–30 minutes of waiting time if the test has anegative result), these tests would have been considered asnegative2; however, these results are false negatives, withimplications for clinical management (ie, perform lifestylemeasurements, avoid drugs with potential adverse effects,and prompt treatment of fever episodes). To quantify theincidence of late responses, we analyzed the time course ofECG changes during and 90 minutes after flecainideinfusion. In addition, clinical, ECG, and genetic factorsaffecting flecainide metabolism were analyzed to describepredictors of time-dependent responses.

MethodsPopulation and recording protocolIn the absence of a spontaneous type I ECG pattern, patientswith suspected BrS were admitted for provocative testing

http://dx.doi.org/10.1016/j.hrthm.2014.10.036

351Calvo et al ECG Variability During Flecainide Testing

with flecainide. According to the standard protocol used inour institution, the clinical profile and serum levels (creatinineand electrolytes) were previously obtained at the outpatientclinic, 1–3 months before the admission. Intravenous flecai-nide was continuously infused at a rate of 2.0 mg/kgbodyweight over 10 minutes (limited by a maximum dosageof 150 mg). Before flecainide infusion, we checked for theabsence of type I ECG, both at the standard precordialposition (V1 and V2 at the third intercostal space) and at thehigh precordial position (V1 and V2 at the second intercostalspace). Thereafter, the standard 12-lead ECG was continu-ously recorded during 30 minutes in the electrophysiologylaboratory (sample rate 1 kHz; bandpass filtered 0.05–150Hz; EP Tracer v1.05.v3, CardioTek, Maastricht, The Nether-lands) from the initiation of flecainide infusion, whichenabled us to better characterize transitions between ECGpatterns because types II and III were mainly defined at thestandard precordial positions.1 At the end of this period, weagain explored the high precordial position for better sensi-tivity. Thereafter, the positions of the leads on the bodysurface were marked and a new ECG was recorded 60minutes later (Figure 1). The study was approved by theethics committee, and subjects gave informed consent.

ECG tracings were analyzed by 3 independent cardiacelectrophysiologists and classified by consensus according tothe published recommendations as type I, II, and III.1 For thepurpose of this study, we classified the normal ECG ascategory IV. This category also includes subtle deviationsfrom normality as the incomplete right bundle branch block.The response to flecainide infusion was analyzed continu-ously during the first 30 minutes. ECG changes observedduring flecainide infusion (first 10 minutes of the ECGrecording) were considered to occur at the first stage of thetest. The second and third stage changes were those changesthat occurred between 10 and 20 minutes and between 20and 30 minutes, respectively. Finally, the fourth stagechanges were considered according to the single 12-leadECG recorded 90 minutes after the initiation of flecainideinfusion. In order to assign an ECG pattern as defining theresult of the test at every stage, we selected the highest orderpattern observed during the recording time, considering typeI as the highest order pattern and category IV as the lowestone. Provocative testing was considered to display a positiveresponse if the patient exhibited a type I ECG patternanytime during the protocol. Those patients displaying apositive response during the first stage were defined as earlyresponders. On the contrary, those patients displaying a

Figure 1 Study flowchart showing different stages of electrocardiographic (ECGbaseline, 10, 20, and 30 minutes (100 mm/s).

positive response during the second and third stages weredefined as delayed responders. Finally, those patients dis-playing a positive response only in the fourth stage weredefined as late responders. All other patients were defined asnonresponders.

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 minutesfrom baseline (using integrated calipers over the digitalrecords; speed 100 mm/s) at lead II. Measurements of 3consecutive cycles were averaged. To avoid potential bias,interval measurements were performed blinded to the finalresult of provocative testing. Also, all ECG channels otherthan lead II were removed from the screen at this time toavoid the researcher perceiving the final result of provocativetesting.

Genetic studiesAccording to current guidelines,6 the whole coding sequenceof SCN5A was amplified and sequenced in responders asdescribed previously.7 The nucleotide changes were classi-fied as possible mutations if (1) reported previously (www.ensembl.org) or (2) they had an effect on the proteinsequence, either by changing the amino acid or by introduc-ing aberrant transcript sequences. Single amino acid changeswere considered as mutations on the basis of the predictionprograms SIFT (J. Craig Venter Institute, La Jolla, CA) andPolyphen (Harvard, Massachusetts).

In the whole cohort of responders and a randomly selectedsubset of nonresponders (5 patients), we also determinedsingle nucleotide polymorphisms in CYP3A5 and CYP2D6genes because they are mainly involved in flecainidemetabolism.4,8 The CYP3A5*3 allele (SNP rs776746) wasgenotyped with a real-time TaqMan polymerase chainreaction assay (assay ID C_25201809_30, Applied Biosys-tems, California, USA) as reported previously.9 To deter-mine the CYP2D6 genotype, we amplified and sequenced the9 coding exons with primers specific for this gene (furtherdetails are available on request to the corresponding author).

Statistical analysisCategorical variables are reported as numbers and percen-tages. For the purpose of the analysis and better clinicalinterpretation, we considered the ECG patterns as ordinalvariables of arbitrary units by assigning a value from 1 to 4 topatterns I, II, III, and category IV, respectively. The ECG

) recording. Interval measurements were performed over the digital records at

Heart Rhythm, Vol 12, No 2, February 2015352

patterns were analyzed using the Cochran Q or the Friedmantest, as appropriate. Continuous variables are reported asmean � standard deviation. We used a mixed-effectsregression model with random intercepts to test changes inthe ECG intervals across stages. Statistical significance wasanalyzed between stages (within-subjects significance) andalso according to the final result of provocative testing(responders vs nonresponders; between-subjects signifi-cance). The Bonferroni correction was used for multiplecomparisons. The χ2 and Student t tests were used forunivariate analysis to compare different variables as pre-dictors of the result of provocative testing. Also, logisticbinary regression was used for analysis and the predictiveperformance of variables was assessed by receiver operatingcharacteristic analysis and the Hosmer-Lemeshow test.Agreement between investigators was measured by the kstatistic. Analysis was performed with SPSS software(version 20, SPSS Inc, Chicago, IL), and P o .05 wasdefined as statistically significant.

ResultsFrom November 2012 through March 2014, we prospec-tively studied 60 consecutive patients (age 38.4 � 14.1years; male sex 71.2%) because of the presence of anabnormal ECG (65%; type II or III ECG), family historyof BrS (40%), and/or syncope (23%). There were no patientswith history of SD or ventricular arrhythmias, renal impair-ment (creatinine 0.83 � 0.21 mg/dL), or electrolyte abnor-malities (sodium 139.5 � 2.4 mEq/L and potassium 4.2 �0.4 mEq/L). Flecainide infusion was prematurely interruptedin 1 case because of significant QRS widening (from 100 to160 ms; left bundle branch block morphology). The patientwas excluded from analysis and evolved with no complica-tions. The other 59 patients completed the study protocol andconformed to the analyzed cohort.

Time course of ECG patternsThe ECG patterns at baseline was classified as follows: typeII in 31 patients (53%), type III in 15 (25%), and category IVin 13 (22%). Category IV included 8 patients (14%) withnormal ECG and 5 (8%) with incomplete right bundle branchblock. During the first stage, we observed conversion to atype I ECG pattern in 4 patients (6.8%). The proportion ofresponders continued to increase in the next stages, with7 patients (11.9%) in the second stage, 6 (10.2%) in thethird stage, and 9 (15.3%) in the fourth stage (P o .01;Figure 2A). Figure 2B summarizes the ECG patternsrecorded at the standard lead position throughout the record-ing time. It shows a tendency to increase the relativecontribution of the type I ECG pattern with ongoing stages(P o .01). Interestingly, in some cases showing the type IECG pattern at the standard lead position, this pattern wasobserved only during a limited period of time. Individually,whenever a patient displayed an induced type I ECG at aparticular stage (first, second, or third), this pattern wasusually maintained in further stages (5 patients [71%]).

However, 2 patients displayed a short-lasting type I ECGpattern during the recording time: (1) in one case it wasobserved from the first stage to the third, but not in the fourthstage (displaying a normal ECG; category IV); (2) in theother case type I ECG was observed only during the secondstage, displaying type III ECG at baseline and type II ECGduring all other stages. ECG patterns at the high precordialposition increased sensitivity only in the last case during thethird stage (Figure 2C).

Overall, 11 patients (18.6%) displayed a type I ECGpattern anytime during the study protocol (cumulativeresponders considering standard and high precordial posi-tions of the leads). Figure 3A shows that the cumulativepercentage of responders increased with longer periods ofECG recording. During the first stage, we classified 4patients (6.8%) as early responders. The ratio of respondersincreased 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 4patients (6.8%; Figure 3B) displayed no type I ECG patternsduring the first 30 minutes of the ECG recording but wereidentified as responders 90 minutes after the initiation offlecainide infusion (late responders). The k statistics meas-uring agreement between observers were 0.86 and 0.89 (Po.01 for both).

Time course of ECG intervalsFlecainide infusion was followed by an increase in the PR,QRS, and QTc intervals from baseline (Table 1 andFigure 4). Although no significant changes in the heart ratewere observed (Figure 4A), the PR interval significantlyincreased between consecutive stages (Figure 4B). How-ever, this tendency to increase was reduced at the end of thethird stage. The PR interval increased by 11.8 � 2.5 msfrom baseline to the 10th minute (Po .01) and by 8.1� 2.4ms from the 10th to the 20th minute (P¼ .01) and remainedstable with a nonsignificant decrease of 1.7 � 2.5 msbetween the 20th and the 30th minute (P ¼ 1.0). As shownin Figure 4B, differences between responders and non-responders at every stage were not statistically significant.Similarly, the QTc intervals increased by 13.2 � 3.6 msfrom 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) anddecreased by 9.9� 2.8 ms from the 20th to the 30th minute(P o .01) (Figure 4C); differences between responders andnonresponders at every stage were not statisticallysignificant.

As observed for the PR and QTc intervals, QRS widthincreased 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 20thminute (P o .01) and decreased by 3.1 � 1.1 ms from the20th to the 30th minute (P ¼ .02) (Figure 4D). However,paired comparisons at every stage found the QRS width to belarger for responders than for nonresponders (Table 1). Also,the ratio of increase was significantly higher for respondersthan for nonresponders during the first 20 minutes of the

Figure 2 Electrocardiographic (ECG) patterns recorded at the standard lead position.A: Proportion of type I ECG recognized at every stage. B: ECG patternsrecorded throughout the recording time. C: Two patients displayed transient type I ECG. The panel displays a representative case in which type I ECG wastransiently 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.

353Calvo et al ECG Variability During Flecainide Testing

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).

Figure 3 Cumulative results obtained across stages considering both standard anpatients displayed a type I electrocardiographic (ECG) pattern only in the fourth stexample displayed in the panel. Note that during the first 30 minutes, ECG tracin

Predictors of late responseNo differences were found between responders and non-responders with regard to age (44.5 � 13.3 years vs 37 �

d high precordial positions of the leads.A: Percentage of responders. B: Fourage. Those patients were considered as late responders, with a representativegs displayed no type I ECG, which became visible only in the fourth stage.

Table 1 ECG intervals recorded throughout the recording time

Variable Group Baseline 10th minute 20th minute 30th minuteWithin-subjectsP value

Between-subjectsP value

HR (beats/min) Global 51.1 � 8.3 51 � 8.2 49.4 � 7.2 50.5 � 8 .06 .65Responders 52.9 � 7.7 51.3 � 6.8 50.7 � 5.7 50.9 � 6Nonresponders 50.7 � 8.5 51 � 8.5 49.1 � 7.6 50.3 � 8.7Paired P value .4 .9 .5 .8

PR interval (ms) Global 171.3 � 20.3 184.3 � 21.5 191.3 � 21 189.3 � 20.8 o.01 .76Responders 173.7 � 23.1 183.6 � 20.9 193.27 � 17.7 192 � 18.1Nonresponders 170.7 � 19.9 184.4 � 21.8 190.7 � 21.6 188.7 � 21.6Paired P value .7 .9 .8 .6

QTc interval (ms) Global 402.9 � 23.2 412.4 � 27.1 425.4 � 26.2 416.9 � 25.7 o.01 .67Responders 394.2 � 24.8 413.2 � 24.5 425.8 � 29.5 413.8 � 29.7Nonresponders 404.9 � 22.7 412.2 � 27.9 425.4 � 25.7 417.6 � 25Paired P value .17 .9 .9 .7

QRS width (ms) Global 87.9 � 10.5 98.3 � 11.5 101.9 � 12.5 99.5 � 12.3 o.01 o.01Responders 98.5 � 9.6 112.6 � 10.9 117.4 � 10.7 113.2 � 12.5Nonresponders 85.5 � 9.2 95.1 � 9 98.4 � 10.1 96.4 � 9.9Paired P value o.01 o.01 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 vsnonresponders). Paired P values correspond to paired comparisons between responders and nonresponders for every interval and time. The within-subjects Pvalue evaluates the statistical significance of changes observed in every patient across stages. The between-subjects P value evaluates the statistical significanceof changes observed between groups (responders vs nonresponders).

ECG ¼ electrocardiographic; HR ¼ heart rate; QTc ¼ corrected QT.

Heart Rhythm, Vol 12, No 2, February 2015354

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 inindividuals younger than 45 years (46% vs 46%; P ¼ .98),family history of BrS (17% vs 20%; P ¼ .8), serumcreatinine 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 levelwas observed to be higher in responders (4.57� 0.32 mEq/Lvs 4.02 � 0.34 mEq/L; P o .01), with no differencesbetween late and early or delayed responders (4.6 � 0.42mEq/L vs 4.56 � 0.34 mEq/L; P ¼ .89). With regard to thedistribution of ECG patterns at baseline, no differences wereobserved between responders and nonresponders (P ¼ NS).

We also evaluated predictors of late responses in relation todifferences in the QRSwidth between patients. For this analysis,we excluded early and delayed responders (as those patientscould be effectively diagnosed during the first 30 minutes of theECG recording) and evaluated the diagnostic yield of themaximum QRS width recorded after flecainide infusion (themaximum observed between measurements at the 10th, 20th,and 30th minute). Logistic binary regression revealed asignificant association between the maximum QRS width andlate responses to provocative testing (Po .01) with appropriatepredictive 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;Po .01), with a maximumQRSwidth of 110 ms during the first30 minutes being effective for identifying possible late responders(sensitivity 100%, specificity 85.6%, positive predictive value88%, and negative predictive value 00%; Figure 5).

Genetic resultsGenetic data are summarized in Table 2. We identified amutation affecting the SCN5A gene in 2 patients (18%; V160F

and T1709M): 1 delayed responder and 1 late responder. Theincidence of CYP2D6 variants affecting flecainide metabolismwas not different between responders and nonresponders (40%vs 54.5%; P ¼ 1.0). However, we observed the incidence ofCYP2D6 variants to be lower in late responders than in earlyor delayed responders (0% vs 75% vs 100%; P ¼ .04). Withregard to CYP3A5 variants, all patients were homozygous foran allele variant causing null enzyme activity (*3/*3).

DiscussionThe main results of our study indicate that response toflecainide in the diagnosis of BrS exhibits width variabilitybetween patients in terms of the time course of ECG changes.In the experimental setting of this study, most of the inducedtype I ECG patterns are observed during the first 30 minutes ofthe initiation of provocative testing. However, extendedperiods of recording time increase the percentage of positivetesting from 11.9% to 18.6% in our study population. Wedemonstrate that the time course of ECG changes is charac-terized by significant changes in the QRS width, which couldbe used for the prediction of late responses. In this sense, whenthe QRSwidth measured across stages after flecainide infusionwas not longer than 110 ms, the incidence of late responses didnot exist in our population. We conclude that in the setting ofan expected variability of ECG changes, the latter could helpdefine appropriate recording times for every patient.

Increasing the diagnostic yieldof flecainide challengeEven though provocative testing is recommended for thediagnosis of BrS, some concerns about the sensitivity andspecificity of the technique arise.2 Also, differences havebeen observed between drugs in terms of their ability to

Figure 5 A: The QRS width increase was higher in responders than in nonresponders. Two representative examples are depicted in the panel. While the lateresponder displayed an increased QRS width (from 96 to 120 ms), the nonresponder displayed lower variability (from 93 to 100 ms). B: The maximum QRSwidth 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.

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.

355Calvo et al ECG Variability During Flecainide Testing

Table 2 Description of time-dependent ECG changes, geneticvariants, and flecainide metabolism affected by CYP2D6 variants8

PatientResponse toflecainide

SCN5Agenotype

CYP3A5genotype

CYP2D6genotype

Flecainidemetabolism

1 Early Normal *3/*3 *2/*10 Het-EM2 Early Normal *3/*3 *1/*1 Hom-EM3 Early Normal *3/*3 *2/*10 Het-EM4 Early Normal *3/*3 *2/*10 Het-EM5 Delayed Normal *3/*3 *2/*41 Het-EM6 Delayed T1709M *3/*3 *1/*6 Het-EM7 Delayed Normal *3/*3 *2/*10 Het-EM8 Late Normal *3/*3 *1/*2 Hom-EM9 Late Normal *3/*3 *1/*1 Hom-EM10 Late Normal *3/*3 *1/*1 Hom-EM11 Late V160F *3/*3 *1/*1 Hom-EM12 Nonresponder — *3/*3 *2/*10 Het-EM13 Nonresponder — *3/*3 *1/*2 Hom-EM14 Nonresponder — *3/*3 *1/*2 Hom-EM15 Nonresponder — *3/*3 *1/*2 Hom-EM16 Nonresponder — *3/*3 *2/*10 Het-EM

ECG ¼ electrocardiographic; Het-EM ¼ heterozygous extensive meta-bolizer; Hom-EM ¼ homozygous extensive metabolizer.

Heart Rhythm, Vol 12, No 2, February 2015356

demonstrate an induced type I ECG. Flecainide is widelyused in Europe and has been proven to be less efficient thanajmaline for the diagnosis.10 However, today it is clear thatresponses to drug infusion in patients with BrS not onlydepend on drug properties but also on recording time. Grayet al4 recently reported a patient displaying a positiveresponse 90 minutes after the infusion of flecainide, anextended period of recording time, which falls out of therecommendations.2 Thus, it is clear that at least in somepatients, up to 30 minutes of recording time is not enough.

At this time the question raised is: How extended must theprovocative testing be? It is difficult to ascertain because, inclinical practice, often the gold standard for diagnosis isprovocative testing itself. Also, the half-life of flecainide isapproximately 9–13 hours in humans, which causes theserum concentration to be in the therapeutic range withinseveral hours of drug administration.11 Under these con-ditions, we may speculate that during this period it is stillpossible to obtain induced type I ECG patterns in patientswith an initially negative response (first 30 minutes afterflecainide infusion). However, there is a significant linkbetween the subjacent electrophysiological properties andthe ratio of changes in the ECG that could be used forappropriate timing. Shimizu et al12 demonstrated that class Iantiarrhythmic drugs significantly increase the QRS width inpatients with BrS and controls, with a more marked effectin patients with BrS. This result is consistent with our data, aspatients with BrS in our population denoted a higher ratio ofQRS widening in response to flecainide. The relative effectin responders was highest at the end of flecainide infusion,working according to the final result of provocative testing.Taking advantage of those physiological observations, wesuggest that a maximum QRS width lower than 110 ms inresponse to flecainide infusion probably denotes minoreffects over excitability and conducting properties of theheart in our population and could be effective for identifying

nonaffected patients. On the contrary, patients with BrSrespond to flecainide infusion with a higher rate of QRSwidening, which predicts a higher rate of positive and lateresponses in our population.

Determinants of flecainide metabolismand ECG responsesFlecainide exhibits a narrow therapeutic window withseveral factors affecting its pharmacological effects.8 Grayet al4 recently highlighted potential modulatory effects ofcytochrome P450 2D6 and 3A5 gene variants on theresponse to flecainide infusion, which might appropriatelyexplain the time-dependent variability of ECG changes.

In the case of CYP2D6, allele variants *1 and *2 areassociated with normal enzyme activity. Patients carryingthose alleles (*1/*1, *1/*2, or *2/*2) were considered ashomozygous 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 wereconsidered as heterozygous extensive metabolizers (Het-EMs).8 Currently, it is debatable whether differences betweenHom-EMs and Het-EMs lead to clinically significant differ-ences in the pharmacokinetics of flecainide in patients. Onestudy analyzed the age-dependent effect of CYP2D6 variantson flecainide metabolism.13 In patients younger than 70years, who were studied in our work, there were no significantdifferences in the metabolic ratio (flecainide/flecainidemetabolites) between Hom-EMs and Het-EMs. However,the coexistence of CYP3A5 variants may also affect thebiodisponibility of flecainide.14 Patients carrying the loss-of-function alleles CYP3A5*3/*3 show shorter time to peakconcentration and larger area under the curve of plasmaconcentration than do those carrying the wild-type alleles(CYP3A5*1/*1). More important to the results observed inour work, the association between CYP3A5*3/*3 and thepharmacokinetics of flecainide is even more relevant insubjects with at least one CYP2D6 mutant allele, leading tolarger area under the curve concentration.

All patients in our study were homozygous for the mutantallele, causing loss-of-function allele CYP3A5*3/*3, whichis the most frequent allele in our population.9 In thehomogeneous distribution of this genetic variant, it is notpossible to explain differences between our patients. How-ever, as stated above, there is a complex interaction betweenCYP2D6 and CYP3A5 variants affecting the pharmacoki-netics of flecainide.14 By reducing the metabolic activity ofCYP2D6 enzymes, the elimination half-life of flecainide isincreased, which eventually might explain the occurrence ofdelayed responses, as postulated by Gray et al.4 However,higher biodisponibility of flecainide is also observed fromthe beginning, which in fact might increase the ratio of earlyresponses.11 Our data suggest that a reduction of functionof the CYP2D6 enzyme promotes early transformation totype I ECG changes in patients with BrS, possibly by meansof an increased biodisponibility of the drug that might beclinically relevant in the presence of additional genetic

357Calvo et al ECG Variability During Flecainide Testing

variants affecting flecainide metabolism as observed forCYP3A5*3/*3.14

Furthermore, a variety of noncontrolled factors caninfluence the effect on ECG tracings and explain discrep-ancies in the observed responses. In the physiological range,we found potassium concentration to be higher in respondersthan in nonresponders, which might promote ECG changesafter flecainide infusion.15 However, unfortunately, electro-lyte determinations were not performed at the time of testing.Therefore, we cannot ascertain that because of daily varia-bility in the serum concentrations of potassium ion, theobserved results reflect the electrolyte conditions at the timeof testing.

Study limitationsECG recordings were limited to the first 30 minutes and asingle ECG was recorded at the 90th minute. Thus, we wereunable to identify transient changes in ECG traces betweenthe 30th and the 90th minute. Short-lasting type I ECGpatterns may be observed at the standard precordial position(Figure 2), and so we cannot ascertain whether the percent-age of early or delayed responses will be higher by detectingshort-lasting type I ECG patterns uniquely displayed at thehigh precordial position during the first and second stages.With regard to the ability of the maximum QRS width topredict late responses, the number of cases studied in thiswork is limited and so further validation is required. SCN5Asequencing was not routinely performed in nonresponders,and so an approximation of the sensitivity and specificity ofprovocative testing in our population was not evaluated. Theconclusions drawn from this work apply to flecainide. Forother antiarrhythmic drugs (ie, ajmaline or pilsicainide),further validation is required.

ConclusionResponse to flecainide infusion exhibits time-dependentvariability of ECG patterns and intervals. Longer periodsof ECG recording increase the recognition probability oftype I ECG, with the maximum width of the QRS intervalbeing a potential predictor of late responses.

AcknowledgmentsWe thank people working at the Arrhythmia Unit of theUniversity Hospital of Asturias for assistance in the studyand Margit Sperling for editorial support.

References1. 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 ofBrugada syndrome, long-QT syndrome, and catecholaminergic polymorphicventricular tachycardia. Circ Arrhythm Electrophysiol 2011;4:958–964.

3. Mizusawa Y, Wilde AAM. Brugada syndrome. Circ Arrhythm Electrophysiol2012;5:606–616.

4. Gray B, McGuire M, Semsarian C, Medi C. Late positive flecainide challenge testfor Brugada syndrome. Heart Rhythm 2014;11:898–900.

5. de-Riva-Silva M, Montero-Cabezas JM, Fontenla-Cerezuela A, Salguero-BodesR, López-Gil M, Arribas-Ynsaurriaga F. Delayed positive response to a flecainidetest in a patient with suspected Brugada syndrome: a worrisome finding. Rev EspCardiol (Engl Ed) 2014;67:674–675.

6. Priori SG, Wilde AA, Horie M, et al. Executive summary: HRS/EHRA/APHRSexpert consensus statement on the diagnosis and management of patients withinherited 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 syndromepatients. Rev Esp Cardiol (Engl Ed) 2010;63:856–859.

8. Zhou S-F. Polymorphism of human cytochrome P450 2D6 and its clinicalsignificance, part II. Clin Pharmacokinet 2009;48:761–804.

9. Tavira B, Coto E, Garciá EC, et al. Pharmacogenetics of tacrolimus after renaltransplantation: analysis of polymorphisms in genes encoding 16 drug metabo-lizing 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 drugchallenge using flecainide and ajmaline in patients with Brugada syndrome. HeartRhythm 2005;2:254–260.

11. Mikus G, Gross AS, Beckmann J, Hertrampf R, Gundert-Remy U, EichelbaumM. The influence of the sparteine/debrisoquin phenotype on the disposition offlecainide. Clin Pharmacol Ther 1989;45:562–567.

12. Shimizu W, Antzelevitch C, Suyama K, Kurita T, Taguchi A, Aihara N, TakakiH, Sunagawa K, Kamakura S. Effect of sodium channel blockers on ST segment,QRS duration, and corrected QT interval in patients with Brugada syndrome. JCardiovasc Electrophysiol 2000;11:1320–1329.

13. Doki K, Homma M, Kuga K, Aonuma K, Kohda Y. Effects of CYP2D6genotypes on age-related change of flecainide metabolism: involvement ofCYP1A2-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, LeeVHL, Tomlinson B. Effects of CYP2D6*10, CYP3A5*3, CYP1A2*1F, andABCB1 C3435T polymorphisms on the pharmacokinetics of flecainide in healthyChinese subjects. Drug Metabol Drug Interact 2012;27:33–39.

15. Campbell TJ, Wyse KR, Hemsworth PD. Effects of hyperkalemia, acidosis, andhypoxia on the depression of maximum rate of depolarization by class Iantiarrhythmic drugs in guinea pig myocardium: differential actions of class Iband Ic agents. J Cardiovasc Pharmacol 1991;18:51–59.

CLINICAL PERSPECTIVESA recent consensus statement highlights that the diagnosis of Brugada syndrome relies on the identification of a type Ielectrocardiographic (ECG) pattern, either occurring spontaneously or induced by sodium-blocker infusion.6 To bridge thegap between the fluctuating nature of ECG recordings and its key role in the diagnosis of this inherited disease, we providean 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 nonrespondersto flecainide infusion. Together they pave the way for a more individualized performance of provocative testing, which helpin improving patient care by reducing the unpredictability of time-dependent unexpected responses. The methodologydisplayed in this work is clinically available under the standard conditions required for provocative testing, leading to directtranslation of main results into clinical practice. However, further validation of data obtained from large cohorts of patientsis required to increase reliability and provide standardization.