Skip to main content

Mechanism of syncope in patients with positive adenosine triphosphate tests

Clinical study: cardiac arrhythmias

J Am Coll Cardiol, 41 (1) 93–98



We prospectively evaluated the mechanism of syncope in patients with positive adenosine triphosphate (ATP) tests (defined as the induction of atrioventricular [AV] block with a ventricular pause ≥6 s after an intravenous bolus of 20 mg ATP).


Patients with unexplained syncope tend to have more positive ATP tests results than those without syncope.


An implantable loop recorder (ILR) was inserted in 36 ATP-positive patients (69 ± 10 years; 22 women; median of 6 syncopal episodes); 15 of them also had a positive response to tilt testing.


During the follow-up of 18 ± 9 months, 18 patients (50%) had syncopal recurrence and 16 (44%) had an electrocardiographically documented episode: AV block (n = 3: paroxysmal in 2 and permanent in 1), AV block followed by sinus arrest (n = 1), sinus arrest (n = 5), sinus bradycardia <40 beats/min (n = 2), normal sinus rhythm (n = 2), sinus tachycardia (n = 1), rapid atrial fibrillation (n = 1), and ectopic atrial tachycardia (n = 1). Bradycardia was documented in a total of 11 cases (69%), and a long ventricular pause (4 to 29 s) was present in eight cases (50%). All three patients with ILR-documented AV block had previously had a negative tilt test, whereas seven of eight with ILR-documented sinus bradycardia or sinus arrest had previously had a positive tilt test.


In patients with adenosine-sensitive syncope, the mechanism of syncope is heterogeneous, although bradycardia is the most frequent finding. Adenosine triphosphate–induced AV block predicts AV block as the mechanism of spontaneous syncope in only a few tilt-negative patients.


Intravenous injection of adenosine triphosphate (ATP) has recently been proposed as an investigative tool in patients with unexplained syncope (1,2). In predisposed patients with unexplained syncope, the cardiac action of adenosine, which has a powerful dromotropic effect on the atrioventricular (AV) node (3), causes prolonged ventricular pauses that are due to AV block and that are suspected of being responsible for spontaneous attacks. The action of ATP is due to its rapid catabolism to adenosine and the subsequent action of adenosine at purinoceptor sites. Adenosine and ATP have similar effects in humans (3), even if some vagal effect may be exerted by ATP but not by adenosine in the canine heart (4). Patients with unexplained syncope are reported to show greater susceptibility to ATP, as demonstrated by their positive ATP tests, than those without syncope (1,2). Also, ATP testing has been shown to reproduce AV block (and suggest the mechanism of syncope) in a few patients with fortuitous documentation of spontaneous paroxysmal AV block, especially in those without abnormalities of AV conduction or of the autonomic nervous system (2). The logical inference is that ATP testing can identify patients with syncope due to transient AV block, even when electrophysiologic findings and other conventional tests are unremarkable. However, this interesting hypothesis remains to be confirmed. In the present prospective study, we identified and followed up a consecutive group of patients with adenosine-sensitive syncope, with the aim of demonstrating the causal relationship between increased susceptibility to adenosine and syncope due to paroxysmal AV block.


Study protocol

Consecutive patients with adenosine-sensitive syncope who met the inclusion criteria received an implantable loop recorder (ILR) and were followed up to obtain electrocardiographic (ECG) documentation of a spontaneous syncopal relapse.

Patients were considered eligible if they were >40 years old, had previously had three or more syncopal episodes with an interval of >6 months between the first and last episode, and had a clinical presentation severe enough (because of a high number of spells or high risk) to require treatment, if any was available. Patients were included in the study only if a careful history, physical examination, baseline ECG, carotid sinus massage, echocardiogram, and 24-h ambulatory recording were not diagnostic of the etiology of syncope. Any other test necessary for a definitive diagnosisof the cause of syncope was performed when clinically indicated. In particular, an electrophysiologic study was performed in the case of cardiac abnormalities or a history of palpitations.

Because of the overlap between adenosine-sensitive syncope and tilt-induced syncope (5,6), all patients underwent tilt testing. The tilt-test protocol always included a 20 min passive phase and a subsequent 15 min phase with nitroglycerin provocation, according to the Italian protocol (7). Patients with positive response to both ATP and tilt testing were included in the study.

Protocol of the ATP test

The ATP test consisted of injecting a 20-mg bolus (<3 s) of ATP dissolved in 10 ml of saline solution into a brachial vein, followed by a 20-ml flush of dextrose solution (1,2). During injection, patients remained supine, and continuous ECG recordings were made just before and 2 min after drug administration. Blood pressure was monitored non-invasively. Interpretation of the test result was based exclusively on the duration of the maximum ventricular pause (maximum RR interval). A value ≥6 s was defined as abnormal (2). In a previous study (2), this value was found in 28% of patients with syncope of unknown origin and in 5% of control subjects without syncope.

End point

When patients were deemed eligible, an ILR (Reveal, Medtronic) was implanted subcutaneously. The recommended program mode was one event, 21 min before activation and 1 min after activation. Patients were instructed to activate the device after every episode of syncope or pre-syncope. The records of all episodes were retrieved, printed, and analyzed. The primary end point of this study was the analysis of the ECG tracing obtained during the first syncopal episode that was correctly recorded by the device.

Control group

As control group, 15 patients (61 ± 13 years; 10 men) who had negative ATP and tilt tests were evaluated. They were all consecutive patients who had received an ILR device during the same period of recruitment because of unexplained syncope; they had received the same workup as described for the study patients. An ATP injection had induced a pause of 2.6 ± 0.8 s.

Statistical methods

A comparison between groups was carried out by means of the Student ttest for continuous variables and the Fisher exact test for proportions. The time to onset of events was analyzed by means of Kaplan-Meier survival curves.


From January 1998 to December 2000, 36 patients were included. The clinical characteristics of the patients are shown in Table 1. Patients were seen at the outpatient clinic every three months until the primary end point was reached or the study ended. The mean follow-up was of 18 ± 9 months (minimum 3 months) and was completed in December 2001. Syncope recurred in 18 patients (50%); their actuarial estimates of recurrence were 11%, 26%, and 42% at 3, 9, and 15 months, respectively.

Table 1. Patient Characteristics (n = 36)

Age (yrs)69 ± 10 (range 41–82)
Female gender22 (61%)
History of syncope
Duration of syncope in years (median, interquartile range)4 (2–10)
Number of syncopal episodes (range)6 (4–10)
Number of syncopes per last year (median, interquartile range)3 (2–5)
Patients with presyncopal episodes22 (61%)
Situational or vasovagal symptoms before episode10 (28%)
No warnings18 (50%)
Associated structural heart disease10 (28%)
ATP response
AV block, maximal ventricular pause (s)8.3 ± 2.4
AV block, total duration (s)16.4 ± 8.6

Data are presented as the mean value ± SD, number (%) of patients, or median value (interquartile range).

ATP = adenosine triphosphate; AV = atrioventricular.

Among the 15 patients comprising the control group, syncope recurred and could be documented by the ILR in nine (60%) during a mean follow-up of 16 ± 9 months.

Primary end point

An ILR-documented syncopal event occurred in 16 adenosine-sensitive patients (44%) after a median nine months (interquartile range 3 to 13) (Table 2). Two other patients were unable to activate the device at the time of their syncopal relapse. Atrioventricular block was recorded in three cases (19%) (paroxysmal in two and permanent in one), AV block followed by sinus arrest in one case, sinus arrest in five cases (mostly preceded by progressive bradycardia and/or tachycardia), progressive sinus bradycardia <40 beats/min in two cases, normal sinus rhythm in two cases, progressive sinus tachycardia in one case, rapid atrial fibrillation in one case, and ectopic atrial tachycardia in one case. Bradycardia was documented in 11 cases (69%), and a long ventricular pause (4 to 29 s) was present in eight cases (50%). The ATP test results were similar in the patients who had syncopal recurrence and in those who did not (maximum pause of 8.4 vs. 8.2 s, respectively) and also in the subgroup who had recurrence due to bradycardia (maximum pause of 8.6 s).

Table 2. Patients With Syncope Documented by Implantable Loop Recorder

Patient No.Time Since Enrollment (months)Description of EventVentricular Pauses (s)Total Duration (s)ATP Maximal RR Interval (s)Tilt Test Result
12Sudden-onset paroxysmal third-degree AV block12188.6
211Sudden-onset paroxysmal third-degree AV block7+2+2156
336Sudden-onset third-degree AV block persisting for a few hoursNoneUndefined14
410Sudden-onset third-degree AV block with a 5 s pause followed by progressive sinus bradycardia and SA of 29 s and then junctional bradycardia5+29406+
51Progressive sinus bradycardia followed by SA71606.6+
618Progressive sinus bradycardia followed by SA9+7+66012.4+
718Progressive sinus bradycardia followed by SA3+3+4306.2+
83Sinus tachycardia followed by progressive sinus bradycardia, SA, and atrial fibrillation8+4+49011
91Sudden-onset SA25+9347+
102Progressive sinus bradycardia up to 35 beats/min4208+
1129Progressive sinus bradycardia then junctional rhythm at 40 beats/min3308.6+
123Sinus rhythm, no variationsUndefined8
134Sinus rhythm, no variationsUndefined6.2
1411Progressive sinus tachycardia at 120 beats/min1808.8+
1511Rapid atrial fibrillation*Undefined10
168Sudden-onset ectopic atrial tachycardia at 200 beats/min1206

SA = sinus arrest; + = positive response; − = negative response; other abbreviations as in Table 1.

* During a suspected epileptic attack (witnessed).

Three patients had AV block. Patient no. 1 was a 63-year-old woman who had had seven unexplained syncopal attacks over the previous 10 years; conventional investigations, including tilt testing, were negative, and no signs of structural heart disease were detected. Adenosine triphosphate caused AV block of 15-s duration, with a main pause of 8.6 s. The recorded event (Fig. 1) was sudden-onset AV block, with a primary ventricular pause lasting 12 s and followed by further secondary pauses. The overall episode lasted 18 s before normal sinus rhythm suddenly resumed; syncope was also sudden and without prodromes. Patient no. 2 was a 54-year-old woman with a history of 15 unexplained syncopal attacks over the previous six years; conventional investigations, including tilt testing, were negative, and there was no sign of structural heart disease. The result of the ATP test is shown in Figure 2. The ILR-recorded event (Fig. 3) was very similar to that induced during the ATP test, in that there was sudden-onset AV block with a primary ventricular pause lasting 7 s, followed by further secondary pauses; the overall episode lasted 15 s before normal sinus rhythm suddenly resumed. Syncope was preceded for a few seconds by a feeling of imminent fainting. A few months before syncope, she had had an ILR-documented pre-syncopal episode due to 2:1 second-degree AV block of 10 s duration. Patient no. 3 was a 78-year-old woman affected by hypertension who had a history of four syncopal attacks over the previous three years; conventional investigations, including tilt testing, were negative. Adenosine triphosphate caused AV block lasting 43 s, with a main pause of 14 s. The recorded syncopal event was sudden-onset, third-degree AV block (35 beats/min), which persisted for a few hours (in-hospital ECG documentation).

Figure 1.
Figure 1.

The figure shows the syncopal event observed in Patient no. 1. (A)Heart rate trend during the whole 21-min loop recording. The heart rate is initially stable at ∼80 beats/min and then suddenly falls at the time of syncope. (B)Expanded electrocardiographic recording at the time of the syncopal event. The onset of atrioventricular block is sudden, with no change in the PP interval; the main pause lasted 12 s.

Figure 2.
Figure 2.

The figure shows the result of the adenosine triphosphate (ATP) test performed in Patient no. 2. Continuous tracings of the electrocardiogram and non-invasive blood pressure (BP) (Finapres method) are shown. The bolus of ATP causes abrupt third-degree atrioventricular (AV) block with long ventricular asystoles of 5.2, 6.0, and 4.5 s. The sinus rate increases during the block. Thus, the effect of the drug is limited to the AV conduction system. Systolic blood pressure dropped from an initial value of 140 mm Hg to values ranging between 80 and 110 mm Hg.

Figure 3.
Figure 3.

The figure shows the electrocardiographic recording of the syncopal event observed in the same patient as in Figure 2(Patient no. 2). The onset of atrioventricular block is sudden, with a long ventricular asystole of 7 s. The sinus rate increases during the spontaneous event. The increase in sinus rate during the episode argues against the presence of a vagal reflex. The noise recorded probably reflects jerking movements of the patient during syncope.

Secondary end points

A positive response during tilt testing was observed in 15 cases (42%): cardioinhibitory in six, mixed in seven, and vasodepressive in two. With regard to the baseline clinical features (Table 1), these patients differed from the 21 with negative responses only in terms of their higher prevalence of situational or vasovagal symptoms before loss of consciousness (47% vs. 14%, p < 0.05). Eight patients in each of the two groups had ILR-documented recurrence of syncope during follow-up (Table 2 and 3). Seven (87%) of the eight patients with a positive tilt test showed sinus bradycardia or sinus arrest. By contrast, the three patients who had isolated AV block belonged to the group of eight patients with negative tilt tests (37%).

Table 3. Mechanism of Syncope, as Detected by Implantable Loop Recorder, in Patients and Control Subjects With Adenosine-Sensitive Syncope (Both Tilt-Negative and B4-Positive)

Patients with Adenosine-Sensitive SyncopeControl Subjects
Tilt-Negative (n = 8)Tilt-Positive (n = 8)Tilt-Negative (n = 9)
AV block3 (37%)01 (11%)
Sinus arrest1 (12%)5 (62%)4 (44%)
Sinus bradycardia02 (25%)1 (11%)
No rhythm variations2 (25%)02 (22%)
Tachycardia2 (25%)1 (12%)1 (11%)

Data are presented as the number (%) of patients or control subjects. AV = atrioventricular.

Also, in the control group, as well as in the tilt-positive group, sinus bradycardia or sinus arrest was the most frequent finding (Table 3). Overall, among the 17 patients who had bradycardia in the three groups (Table 3), AV block was significantly more frequent in the ATP-positive, tilt-negative group (37% vs. 8%), whereas sinus bradycardia and sinus arrest were more frequent in the tilt-positive and control groups (92% vs. 12%; p = 0.02).

Multiple syncopal episodes occurred in six patients (Patients no. 5, 6, 7, 8, 9, and 11; Table 2). In all patients, very similar findings were observed during their second events, and in one patient during his third event.

No patient suffered severe injury due to syncopal relapse, and none died. An interview of a witness to the episode recorded in Patient no. 15 suggested a diagnosis of an epileptic attack. As a consequence of the results of the study, seven patients had a permanent pacemaker implanted, one underwent transcatheter ablation, and one received anti-epileptic therapy. No therapy was undertaken in the other patients.


The main conclusion of this prospective study is that in patients with adenosine-sensitive syncope, the mechanism of syncope is heterogeneous, although bradycardia is the most frequent finding. The causal relationship between ATP-induced AV block and syncope due to paroxysmal AV block is weak. Atrioventricular block induced by ATP predicts AV block as the mechanism of spontaneous syncope in only a few tilt-negative cases.

At the time of ILR-documented syncope, about half of the patients showed long pauses, and two-thirds had severe bradycardia. Sinus arrest or bradycardia was more frequent in patients who had an associated positive response during tilt testing. This finding is consistent with that observed in a previous study in patients with tilt-positive syncope (8). The finding of progressive sinus bradycardia frequently followed by sinus arrest has been regarded as highly suggestive of a neurally mediated mechanism. The association between AV block and sinus arrest, as in Patient no. 4, also suggests a neurally mediated mechanism (8). On the other hand, AV block without sinus bradycardia or arrest occurred in three (37%) of the eight patients with adenosine-sensitive syncope and a negative tilt test. In a previous study (8), AV block was shown to be a very uncommon mechanism in patients with unexplained isolated syncope and in patients with tilt-positive syncope. Atrioventricular block was also uncommon in the control group in the present study. Therefore, this finding might not be fortuitous. The feature of sudden onset of AV block with no change in sinus rate or with an increase in heart rate resembles that observed in patients with bundle branch block who have an intrinsic AV conduction disease (9)and argues against a neurally mediated mechanism.

On the basis of the high percentage of ILR-documented bradycardic episodes, one could suggest that the ATP test is useful in predicting bradycardia in general, even if not AV block specifically. Actually, other authors have claimed that the ATP test is useful in identifying patients with severe vasovagal syncope who might benefit from pacing therapy (1,10). In the present study, the positivity of tilt testing, rather than the positivity of the ATP test, seems better able to predict subsequent cardio-inhibitory vasovagal syncope. Moreover, in a large population of patients with unexplained, isolated syncope (8), and in the control group of the present study, bradycardia was found as frequently as in adenosine-sensitive patients, suggesting that bradycardia is a frequent cause of syncope, regardless of the results of the ATP test. The results of pacing therapy are also inconclusive. Indeed, in one large, non-controlled study (1), syncope recurred in 13% and 15% of paced patients with positive and negative ATP test results, respectively. This suggests that the beneficial effect of pacing is independent of the result of the test and, rather, reflects the high prevalence of syncope due to bradycardia, as shown previously.

Although the receptors are different, the cardiac actions of adenosine are remarkably similar to those of the neurotransmitter acetylcholine (11). Both acetylcholine and adenosine produce the same effects and share similar receptor–effector coupling systems. A major role of acetylcholine and adenosine, in addition to their direct effect, is to function in parallel to oppose the cardiac-stimulatory action of the sympathetic neurotransmitters norepinephrine and epinephrine on adenyl cyclase (cyclic adenosine monophosphate–dependent effect) (3,11). Thus, adrenergic, cholinergic, and purinergic outflows are integrated at the level of the receptor–effector coupling system, and the final cardiac effect results from the sum of these excitatory and inhibitory effects. In addition to its cardiac effects, adenosine also promotes vasodilation, decreases central and peripheral sympathetic efferent nerve activity, and inhibits renin release (12,13). The injection of ATP or adenosine causes hypotension. The role of hypotension in the mechanism of syncope is unclear. A consequence might be that vasovagal syncope could be facilitated by an increased susceptibility to adenosine, and that adenosine-sensitive syncope could be facilitated by increased vagal outflow. Interestingly, in support of this supposed synergic effect, the mechanism of syncope was bradycardic in seven (87%) of eight patients in whom both the ATP and tilt test were positive.

A positive response to the ATP test has been shown to be associated with typical clinical features (6)that were consistently observed in the present study in the three patients who had ILR-documented AV block. Typically, adenosine-sensitive syncope first manifests itself in middle or old age and is more common in females. Because the attacks nearly always occur in the standing position and warning symptoms are frequently absent, loss of consciousness often results in falls that cause injury. The lack of historical findings of vasovagal or situational episodes and the absence of triggering factors characterize this form and clearly differentiate it from vasovagal syncope. Despite the old age of patients, structural heart disease is usually absent or, if present, mild and mainly due to a long history of hypertension. Adenosine-sensitive syncope is rare, accounting for about 3% of patients referred for investigation of syncope (6).

Study limitations

The results of the study are hampered by the small number of patients in various groups, reflecting the relative rarity of severe adenosine-sensitive syncope. Thus, the results should be regarded as intermediate pending larger trials. Although the follow-up duration (18 ± 9 months) was longer than that of similar studies with ILR, it is likely that longer observations would be of considerable interest.

Practical consequences

Because of its low positive predictive value, the ATP test has little value in evaluating patients with unexplained syncope and in selecting treatment. The favorable outcome we observed suggests the strategy of postponing treatment, pacemaker therapy in particular, until a definite diagnosis can be made by documenting a spontaneous syncopal relapse. In accordance with this approach, only nine patients (25%) underwent ILR-guided specific therapy immediately after their first syncopal documentation; no therapy was undertaken in the others. Many other patients would probably have had a documented syncopal recurrence if the monitoring phase had been prolonged. The usefulness of a very prolonged monitoring phase and the efficacy of therapy in preventing further syncopal recurrences remain to be proved in prospective therapy trials. Moreover, the benign nature of the clinical syndrome, which is similar to that of neurally mediated syncope, suggests that ILR implantation should be done only when its clinical presentation is severe enough (because of high number or high risk) to require treatment.

  • 1. Flammang D., Church T., Waynberger M., Chassing A. and Antiel M. : "Can Adenosine 5′-triphosphate be used to select treatment in severe vasovagal syndrome?". Circulation 1997; 96: 1201.

    CrossrefMedlineGoogle Scholar
  • 2. Brignole M., Gaggioli G., Menozzi al. : "Adenosine-induced atrioventricular block in patients with unexplained syncope: the diagnostic value of ATP test". Circulation 1997; 96: 3921.

    CrossrefMedlineGoogle Scholar
  • 3. Belardinelli L., Linden J. and Berne R.M. : "The cardiac effects of adenosine". Prog Cardiovasc Dis 1989; 22: 73.

    CrossrefGoogle Scholar
  • 4. Pelleg A., Hurt C.M., Soler-Baillo J.M. and Polansky M. : "Electrophysiological-anatomic correlates of ATP-triggered vagal reflex in dogs". Am J Physiol 1993; 265: 681.

    Google Scholar
  • 5. Flammang D., Erickson M., McCarville S., Church T., Hamani D. and Donal E. : "Contribution of head-up tilt testing and ATP testing in assessing the mechanisms of vasovagal syndrome: preliminary results and potential therapeutic implications". Circulation 1999; 99: 2427.

    CrossrefMedlineGoogle Scholar
  • 6. Brignole M., Gaggioli G., Menozzi al. : "Clinical features of adenosine-sensitive syncope and tilt-induced vasovagal syncope". Heart 2000; 83: 24.

    CrossrefMedlineGoogle Scholar
  • 7. Bartoletti A., Alboni P., Ammirati al. : "‘The Italian protocol’: simplified head-up tilt testing potentiated with oral nitroglycerin to assess patients with unexplained syncope". Europace 2000; 2: 339.

    CrossrefMedlineGoogle Scholar
  • 8. Moya A., Brignole M., Menozzi al. : "Mechanism of syncope in patients with isolated syncope and in patients with tilt-positive syncope". Circulation 2001; 104: 1261.

    CrossrefMedlineGoogle Scholar
  • 9. Brignole M., Menozzi C., Moya al. : "The mechanism of syncope in patients with bundle branch block and negative electrophysiologic test". Circulation 2001; 104: 2045.

    CrossrefMedlineGoogle Scholar
  • 10. Flammang D., Antiel M., Church al. : "Is a pacemaker indicated for vasovagal patients with severe cardioinhibitory reflex as identified by ATP test?". Europace 1999; 1: 140.

    CrossrefMedlineGoogle Scholar
  • 11. Lerman B. and Belardinelli L. : "Cardiac electrophysiology of adenosine: basic and clinical concepts". Circulation 1991; 83: 1499.

    CrossrefMedlineGoogle Scholar
  • 12. Shen W.K., Hammill S., Munger al. : "Adenosine: potential modulator for vasovagal syncope". J Am Coll Cardiol 1996; 28: 146.

    View ArticleGoogle Scholar
  • 13. Mittal S, Stein K, Markowitz S, Slotwiner D, Rohatgi S, Lerman B. Induction of neurally mediated syncope with adenosine. Circulation 1999;99:1318–24

    Google Scholar



adenosine triphosphate






implantable loop recorder