Lack of correlation between the responses to tilt testing and adenosine triphosphate test and the mechanism of spontaneous neurally mediated syncope

Department of Cardiology, Arrhythmologic Centre, Ospedali del Tigullio, Via don Bobbio 24, 16033 Lavagna, Italy; Department of Cardiology, Royal Brompton and Harefield Hospitals, London, UK; Department of Cardiology, Ospedale S Maria Nuova, Reggio Emilia, Italy; Department of Cardiology, H. Clı́nico, Valencia, Spain; Department of Cardiology, Hospital General Vall dHebrón, Barcelona, Spain; Department of Internal Medicine, Academisch Medisch Centrum, Amsterdam, The Netherlands; Department of Cardiology, Krankenhaus Am Urban, Berlin, Germany; Department of Cardiology, University of Minnesota, Minneapolis, USA; Clinical and New Business Development Division, Medtronic Italia, Rome, Italy; and Department of Cardiology, University Hospital of Crete, Herakleion, Greece


Introduction
The correlation of spontaneous syncopal episodes with an abnormal finding detected by an implantable loop recorder (ILR) can be regarded as a reference standard when an arrhythmia is suspected to have a role in the genesis of syncope. In previous ILR studies, [1][2][3][4] the observations at the time of syncope were heterogeneous, with asystole accounting for up to a half of the syncopal events. The International Study on Syncope of Uncertain Etiology 2 (ISSUE-2) 5 showed that a new strategy based on simple initial evaluation, early application of an ILR, and therapy delayed until ILRdocumentation of the mechanism of syncope is safe and can guide effective therapy in patients with recurrent suspected neurally mediated syncope.
Tilt testing (TT) and adenosine triphosphate (ATP) test are generally regarded as useful tests for the diagnosis of suspected neurally mediated syncope. 6,7 The capability of these tests to predict the exact mechanism of spontaneous syncope should have practical, diagnostical, and therapeutical importance. However, it has been questioned in a few small studies. Moya et al. 1 found a correlation between TT and spontaneous findings in four out of eight patients and Deharo et al. 8 in seven of 11 patients. Donateo et al. 9 found a correlation between ATP test and spontaneous findings in eight out of 16 patients and Deharo et al. 8 in four out 11 patients.
In this study, we prospectively evaluated whether the responses to TT and ATP test were correlated and therefore could predict the clinical outcome and the mechanism of ILR-documented spontaneous syncope employing the much larger population of the ISSUE-2. 5

Methods
The ISSUE-2 5 was a multicentre, prospective, observational study enrolling consecutive patients who underwent an ILR implantation for suspected neurally mediated syncope. Patients were enrolled at 63 European and American centres; enrolment began in June 2002 and ended in July 2004. The Steering Committee designed the trial. The Medtronic Corporation funded the trial and provided a study manager to supervise its conduct. Data were sent by investigators through the web to an independent clinical research organization (RDES SL, Barcelona, Spain) that maintained the database, issued data clarification forms and, assisted by Medtronic clinical monitors, verified source documents. The sponsor had no access to the database and did not participate in the analysis of the results or in the writing of the article. All the analyses were performed by the Endpoints Committee members with the assistance of a statistician. The study was approved by the institutional review boards, and patients provided informed consent.

Patients
Eligible patients were at least 30 years of age, had suffered, in the last 2 years, from three or more syncope episodes of suspected neurally mediated syncope with a severe clinical presentation (because of high number of episodes that affect patient's quality of life or high risk for physical injury due to unpredictable occurrence) requiring treatment initiation. Patients with carotid sinus syncope were excluded.
In accordance with current guidelines, 6,7 a neurally mediated mechanism was considered likely when, on the initial evaluation, there were suggestive data and the following competitive diagnoses could be ruled out: (i) suspected or definite heart disease and high likelihood of cardiac syncope, i.e. syncope during exercise; overt heart failure; ejection fraction 40%; old or recent myocardial infarction; hypertrophic cardiomyopathy; dilated cardiomyopathy; significant valvular disease; sinus bradycardia ,50 bpm or sino-atrial block; Mobitz I second-degree atrioventricular block; Mobitz II second-or third-degree atrioventricular block; bundle branch block; rapid paroxysmal supraventricular tachycardia or ventricular tachycardia; pre-excited QRS complexes; prolonged QT interval; right bundle branch block pattern with ST-elevation in leads V1-V3 (Brugada syndrome); negative T waves in right precordial leads, epsilon waves, and ventricular late potentials suggestive of arrhythmogenic right ventricular dysplasia; (ii) symptomatic orthostatic hypotension diagnosed by standing blood pressure measurement; (iii) non-syncopal loss of consciousness (e.g. epilepsy, psychiatric, metabolic, drop-attack, cerebral transient ischaemic attack, intoxication, cataplexy); (iv) subclavian steal syndrome. All patients received carotid sinus massage of 10 s duration, supine and upright, and those with carotid sinus syncope were excluded.
TT and ATP test were recommended during the screening phase but they were not mandatory for the inclusion of the patients in the study group. The methodology of execution was that recommended in the recent guidelines. 6,7 A positive response to TT was defined as the induction of syncope in the presence of bradycardia, hypotension, or both, and positive responses were classified according to the New VASIS classification. 10 A positive response to ATP test was defined as the induction of complete AV block (or sinus pause) with a ventricular pause 6.0 s. 11

Study protocol
Eligible patients were enrolled in the study if they had undergone ILR implantation (Reveal Plus, Medtronic). The recommended programmed mode was one manual event and 13 automatically recorded events for a total duration of 42 min of storage. Patients were instructed to activate the device after every episode of syncope. A screening log of eligible not-implanted patients was also collected. After ILR implantation, patients were followed quarterly until the first electrographically (ECG) documented syncope or for a maximum of 24 months. Neither any therapy nor specific counselling was prescribed during the follow-up. The mechanism of syncope was assigned according to the ISSUE classification by the Endpoints Committee members who analysed the records of all episodes. 12 The study protocol has been previously published. 13 Objectives Objectives of the study were to define the mechanism of syncope in patients with clinical diagnosis of neurally mediated syncope and to evaluate prospectively whether the responses to TT and ATP test were correlated and therefore could predict the clinical outcome and the mechanism of spontaneous syncope.

Outcome measures
The time of the first ECG-documented syncopal recurrence after ILR implantation and the time of the first documented or undocumented syncopal event were collected. ECG-documented presyncopal and asymptomatic episodes were not considered as endpoints.

Statistical methods
A sample size of 350 patients was calculated to be sufficiently large in order to derive an accurate assessment of positive and negative predictive accuracy of TT and ATP test to predict bradycardic/asystolic syncope. 13 Comparison between groups was performed with Student's t-test or the non-parametric Mann-Whitney U test for continuous variables and with Fisher's exact or McNemar test for proportions, as appropriate. All reported P-values were two-tailed and the value less than 0.05 was considered significant. The time to the first recurrence of syncope was analysed by means of Kaplan-Meier survival curves, which were compared using the log-rank test.

Results
The clinical characteristics of 392 analysed patients are listed in Table 1. Of these, 343 underwent TT, which was positive in 164 (48%), and 182 the ATP test, which was positive in 54 (30%) (Figure 1).
During a mean of 12+8 months of follow-up, 143 patients (36%) had a syncopal recurrence. Syncope was documented by ILR in 106 (26%) patients after a median of 3 months (interquartile range, 1-7), and the mechanism of syncope was classified according to the ISSUE classification ( Table 2). 12 The most frequent finding, which was observed in 57/106 patients (54%), was one or more prolonged asystolic pauses ranging from 3 to 51 s, the maximum pause being a median of 11.5 s (interquartile range 6.3-18.5 s). Of the patients with an ILR documentation of spontaneous syncope, the correlation with TT response could be evaluated in 94 patients; in 52 of these also the correlation with ATP response could be evaluated.

Correlation between TT and spontaneous syncope
The patients with positive TT showed similar baseline characteristics, with the exception of higher prevalence of males in TT negative patients ( Table 1), and syncopal recurrence rate ( Figure 2) than those with negative test. The mechanism of syncope was also similar in both groups, but those with positive TT had more frequently no or slight rhythm variations during spontaneous syncope [45 vs. 21%, odds ratio 3.0 (95% CI 1.2-7.3), P ¼ 0.02] ( Table 3).

Correlation between ATP and spontaneous syncope
This could be evaluated in 52 patients who had undergone ATP tests at enrolment and subsequently had an ILR documentation of a spontaneous syncope. The 14 patients who had a positive response had AV block of 12.8 + 7.9 s duration with a median maximum pause of 8.1 + 2.5 s. The patients with positive and negative ATP had similar syncopal recurrence rate ( Figure 4) and mechanism of syncope ( Table 5 and Figure 5). Responses consistent with spontaneously documented syncope were found in only 26   (54%) patients ( Figure 5). Thus, ATP test results were unrelated to the mechanism of the spontaneously documented syncope. ATP results were also independent from TT results. Compared with the patients with negative test, those with positive ATP test were older, and there were more females and more patients with hypertension ( Table 1).

Discussion Electrocardiographic findings in patients with suspected neurally mediated syncope
This study shows that a long asystole is the most frequent finding at the time of spontaneous syncope in patients with suspected neurally mediated syncope and that asystolic responses are more frequent during spontaneous syncope than during induced syncope. These findings are consistent with those of two previous smaller studies. 1,8 This study also shows that the heterogeneous findings at the time of syncope can easily be classified according to major categories such as those of the ISSUE classification. 12 Despite the enhancement of diagnostic capabilities by the ILR in providing a correlation between electrocardiographic findings and syncope, the device is still unable to provide any information about arterial blood pressure and cerebral blood flow, which are involved in causing syncope. Moreover, the exact nature of a documented arrhythmia, i.e. intrinsic cardiac vs. extrinsic reflex, may remain uncertain. For all these reasons, the underlying mechanism of syncope is still largely uncertain and aetiology can only be inferred. 12 For example, on the one hand, the findings of progressive sinus bradycardia, most often followed by ventricular asystole due to sinus arrest, or progressive tachycardia followed by progressive bradycardia and, eventually, ventricular asystole due to sinus arrest (type 1A, 1B, and 2 of ISSUE classification 12 ) suggest that the aetiology of syncope is neurally mediated 1 ; on the other hand, the observation that a minority of patients had documentation of an atrial or ventricular tachyarrhythmia at the time of recurrence of syncope, a mechanism that is probably inconsistent with the initial neurally mediated diagnosis, suggests that a primary cardiac aetiology was the major determinant of syncope (type 4B, 4C, and 4D). 12 Nevertheless, the study population was of patients who, in accordance with Figure 5 Correlation between the ATP test response and the mechanism of syncope, as documented by ILR in 52 patients. Table 4 Correlation between asystolic responses to TT (VASIS type 2B) and the mechanism of syncope, as documented by ILR in 38 patients ILR response TT positive P-value Figure 3 Correlation between the TT response and the mechanism of syncope, as documented by ILR in 38 patients.   6,7 are usually diagnosed as affected by a likely neurally mediated syncope and, in absence of an ILR observation, are managed with conventional treatments of neurally mediated syncope.

Predictive value of TT
This study shows that the mechanism of spontaneous syncope as documented by an ILR is poorly correlated with the results of TT. Owing to the high rate of discordance between test and reference standard, the clinical utility of TT in predicting the spontaneous event is modest. The syncopal recurrence rate was not predicted by the results of TT (Figure 1), as shown previously by others. 1,14,15 However, the patients with positive TT more frequently had no or slight rhythm variations during spontaneous syncope than those with negative TT ( Table 3). The reason for this small difference is unclear, and its clinical utility is useless because of the unacceptably low positive predictive value (58%) and sensitivity (45%) (Figure 2).
Asystolic responses are more frequent during spontaneous syncope than during induced syncope. In the present study, an asystolic response (i.e. type 2B of the New VASIS classification) is present in 21% of positive tests, the same percentage observed in the original New VASIS description. 10 Although with ILR an asystolic syncope was equally frequent in the younger as in the older patients, during TT, an asystolic response was more frequent in the younger patients.
As a consequence, TT has a low sensitivity in detecting asystolic responses during spontaneous syncope. However, there was a trend for patients with asystolic responses during TT to have an asystolic event during spontaneous syncope ( Table 4 and Figure 2). This correlation gives a 75% probability of an asystolic response during TT to predict an asystolic event during spontaneous syncope with a low sensitivity of 35% This latter observation is reinforced by pooling the data of this study with those of the first ISSUE study 1 : eight out of 10 patients with an asystolic response during TT also showed an asystolic event during spontaneous syncope compared with 14/36 with mixed or vasodepressor response [P ¼ 0.02, odds ratio 6.3 (95% CI 1.2-34), positive predictive value 80%]. In other words, it seems that an asystolic response during TT is able to predict an asystolic response during spontaneous syncope with a 20% risk of misdiagnosis, but a mixed or vasodepressor response during TT cannot exclude an asystolic spontaneous syncope.
This study was not aimed to evaluate the effect of TT to guide therapy. However, on the basis of these findings, it has been found that a mechanism-specific therapy guided by TT results has potential limitations and seems less likely to be effective than ILR-based therapy. These findings might give an explanation of the controversial results of controlled trials of tilt-guided therapy, specifically those evaluating the effect of pacing therapy. The efficacy of pacemaker therapy was questioned after two recent controlled blind trials failed to prove superiority of cardiac pacing over placebo of unselected patients with positive TT. 16,17 Conversely, other controlled unblind pacemaker trials in which patient inclusion was largely made by the presence of an asystolic TT response showed efficacy of cardiac pacing. 18,19 This is not surprising if we consider that, in this as well as in previous ILR studies, [1][2][3][4] the mechanism of syncope was heterogeneous, with bradycardia or asystole accounting for up to a half of the syncopal events.

Predictive value of ATP
In the present study, ATP test was positive in 30% of the patients. This figure is similar to the 28% rate reported in the literature in a study using the same methodology. 11 The syncopal recurrence rate bore no relationship with the ATP results (Figure 3), and the mechanism of spontaneous syncope as documented by ILR was not predicted at all by the results of the ATP test ( Table 5 and Figure 4). These results are consistent with those of previous studies. 8,9 Therefore, the ATP test is definitely of no value in guiding ILR-based therapy. The observation in this study, as well in previous ones, 11,20 that ATP-positive patients had a few peculiar clinical characteristics that differentiated them from the others ( Table 1) remains largely unexplained and seems to have no practical value.

Limitations
The ISSUE classification 12 is based on observations during syncope. We do not know whether similar findings are also present in the general population. For example, what is the prevalence of pauses .3 s among subjects without history of syncope? This question is of practical importance because the absence of pauses in patients without syncope would increase the value of the same findings in patients with syncope. We know from previous studies on 24 h Holter monitoring that pauses .3 s are very rare in subjects without syncope and these were never observed among 259 healthy older adults pooled from three studies. [21][22][23] Admittedly, this finding cannot be automatically translated to ILR, which has much more powerful capabilities of detecting abnormalities, if any. An analysis from ISSUE 2 study (data not yet published) showed that an asymptomatic asystolic pause .3 s was detected in only 1/158 (0.6%) documented events among patients with diagnosis of nonasystolic syncope compared with 13/25 (52%) documented events among patients with a diagnosis of asystolic syncope (P , 0.001). However, even if these findings seem to suggest a good specificity of asystolic pauses ,3 s, uncertainty still persist on their value.
This study shows that syncope is difficult to predict and most patients did not have recurrence during the 12-month follow-up period. Therefore, a longer follow-up would have been probably helpful in order to determine the mechanism of syncope in other patients. Theoretically, these late-recurrent patients could have mechanisms different from the early-recurrent patients. Longer follow-up was limited by the battery longevity of the present ILR generation. By extrapolating the recurrence curve shown in Figure 2, we can expect that 4 years of follow-up would be necessary to detect syncope in about 80% of the patients.
The study correlated one test with the first ILR-documented episode. Reproducibility of spontaneous episodes was not systematically evaluated; a weak reproducibility of spontaneous responses would impair the importance of the results. An analysis of those patients who had multiple syncopal episodes in the ISSUE 2 study (data not yet published) showed that an asystolic syncope was present in three of the four (75%) patients who had a first asystolic syncope vs. none of the nine (0%) patients who had a first non-asystolic syncope (P ¼ 0.01); moreover, ISSUE 2 study therapy, 5 a proven effective-specific therapy, was administered on the basis of the first ILR finding, thus indirectly supporting a good reproducibility of spontaneous mechanisms of syncope.

Conclusions and practical implications
In patients with neurally mediated syncope, clinical characteristics, outcome, and mechanism of syncope are poorly correlated with and not predicted by the results of TT and ATP test. ISSUE 2 therapy study 5 showed that pacing is effective when asystole is documented at the time of syncope and that a strategy based on therapy delayed until ILR documentation is useful. The capability of TT and ATP to predict ILR spontaneous syncope would allow to anticipate therapy and avoid ILR implant. Unfortunately, this correlation being weak or absent, these tests are probably of little or no value for guiding specific therapy with the exception, perhaps, of asystolic responses during TT.