Recently, treatment of patients with recurrent vasovagal syncope has centered primarily on the use of pharmacologic interventions [7, 9-16]. ß-adrenergic blocking drugs, disopyramide, certain vasoconstrictor agents, and serotonin re-uptake blockers have been of particular interest. Volume expanders, such as fludrocortisone acetate and salt tablets, and belladonna alkaloids, such as scopolamine, are also still used. Despite the 1991 revision of the American College of Cardiology-American Heart Association guidelines, which incorporated a class 2 indication for pacing therapy in patients with vasovagal syncope [17], cardiac pacemakers currently play only a modest role in prevention of recurrent vasovagal syncope.

Several factors account for this limited use. First, although the vasovagal faint is often associated with marked bradycardia, a concomitant vasodepressor response is usually also present [5, 7, 9, 18]. The latter is an important contributor to symptomatic hypotension in many patients and is probably not altered by pacing. Second, although fatal outcomes or physical injuries may occur, patients with vasovagal syncope generally have a benign prognosis; consequently, implantation of a relatively expensive device is considered unwarranted. Third, although pacing is often prescribed for patients with the carotid sinus syndrome, who are usually older, most patients with vasovagal syncope are relatively young and may be less inclined toward pacemaker implantation. Fourth, relatively few studies have examined the effects of pacing in patients with vasovagal syncope, and these studies generally lacked long-term follow-up. Finally, investigators and manufacturers have yet to effectively design diagnostic and pacing algorithms expressly for use in patients with vasovagal syncope, so optimal pacing systems are not yet available for these patients.

Our purpose is to provide a perspective on current understanding of the potential role for cardiac pacing in the prevention of recurrent vasovagal syncope. To this end, we examined findings in peer-reviewed, published reports that address the effect of pacing on spontaneous or tilt-table-induced vasovagal faints. Our objective is to describe concepts related to pacing in vasovagal syncope and to define the settings in which pacing therapy might be appropriate. For search strategy, see the Appendix.

Considerations in Treatment Selection

Nature of the Vasovagal Response

In the vasovagal faint, as in all forms of neurally mediated syncope, symptomatic hypotension may result from loss of peripheral vascular tone, evolution of marked bradycardia, or both. Further, the electrophysiologic and hemodynamic features associated with vasovagal syncope vary among patients and possibly among episodes in a given patient [18-22]; a multicenter European working group has provided a working classification of these vasovagal responses as seen during tilt-table testing (Table 1) [18]. In this scheme, persons with the type 2B cardioinhibitory form of vasovagal faint seem to be the most appropriate candidates for pacing intervention Figure 1, although those with type 2A may also benefit.

View larger version (39K): [in this window] [in a new window]   Figure 1. Heart rate and blood pressure (Finapres plethysmographic technique) for each cardiac cycle. One-minute intervals are indicated along the abscissa. Left-hand arrows indicate the initiation of tilt, and right-hand arrows indicate return to supine position. Top. Type 2A cardioinhibitory vasovagal syncope as categorized by the Vasovagal International Study group [18]. The heart rate increases initially and then decreases abruptly. Blood pressure also increases initially but begins to decrease before heart rate does, thereby contributing substantially to the development of symptomatic hypotension. Bottom. Type 2B cardioinhibitory vasovagal syncope as categorized by the Vasovagal International Study group [18]. The heart rate initially increases and later decreases abruptly. Blood pressure decreases as a direct result of the decrease in heart rate (this is a relatively "pure" form of cardioinhibitory response). Figure adaptedfrom Sutton and colleagues [18] with permission.

In individual patients with suspected vasovagal syncope, a positive tilt-table test appears to help confirm the diagnosis and assess whether the vasovagal response is cardioinhibitory, vasodepressor, or mixed. Various reports provide different perspectives on the relative frequency of the cardioinhibitory, vasodepressor, and mixed forms of the vasovagal faint as assessed during tilt-table testing [18,20,23,24,25]. The differences, however, are probably largely due to the lack of accepted definitions for each subset. In general, it is believed that only a few patients with vasovagal syncope show either an exclusively cardioinhibitory picture (such as that in which an extended period of bradycardia is the sole cause of the faint) or a pure vasodepressor response (in which peripheral vascular dilatation causes symptomatic hypotension). Most patients manifest a mixed response, analogous to that well-known in patients with the carotid sinus syndrome [26, 27]. Thus, in most cases, the prevention of an apparently marked cardioinhibitory response, either pharmacologically or by cardiac pacing, may nevertheless be associated with hypotension due to unaddressed loss of vascular tone. Conversely, in some patients, a vasodepressor response may appear to be the sole cause of hypotension because the heart rate seems "normal" during a spontaneous or induced vasovagal faint. Nonetheless, the heart rate is almost never "appropriate" for the severity of systemic hypotension. Thus, concomitant cardioinhibition is usually present even in these persons (for example, in the form of relative bradycardia).

Treatment Options and Assessment of Effectiveness

Strategies for the prevention of recurrent vasovagal syncope are still evolving. However, despite reports suggesting commendable rates of success, the true value of the current measures for preventing the recurrence of vasovagal syncope will remain uncertain until randomized controlled therapeutic trials are done. Further, even if the usefulness of pharmacologic interventions are confirmed, drug failures and drug-related side effects are inevitable. Therefore, continued exploration for alternative treatment options, including cardiac pacing, is essential.

In recent years, identification of potential therapies for vasovagal syncope has relied heavily on the tilt-table-induced, neurally mediated, hypotension-bradycardia model [9-16]. Several observations support this approach and suggest that the hypotension-bradycardia associated with a positive head-up tilt-test is essentially equivalent to that in spontaneous, neurally mediated vasovagal syncope. First, induced and spontaneous syncopal episodes tend to be associated with similar premonitory symptoms (such as nausea and diaphoresis) and signs (marked pallor, loss of postural tone). Second, the temporal sequence of blood pressure and heart rate changes during tilt-induced syncopal spells mostly parallels those reported for spontaneous episodes [23]. Third, plasma catecholamine levels measured before and during spontaneous and tilt-induced syncope show important similarities. In particular, premonitory increases in circulating catecholamine levels appear to characterize both the spontaneous vasovagal faint [28, 29] and tilt-induced hypotension-bradycardia [13].

Although tilt-table testing seems to provide a reasonable model of the spontaneous vasovagal faint, its usefulness in assessing treatment options critically depends on whether the observations provoked by tilt-table testing are sufficiently specific and reproducible to permit assessment of interventions. In regard to specificity, tilt-table testing techniques seem to discriminate relatively well between symptomatic patients and asymptomatic controls. For example, during the course of an assessment of hemodynamic compensation during exposure to graded degrees of upright posture, de Mey and Enterling [30] reported only eight instances of hypotension-bradycardia among 40 apparently normal patients (20%). Similarly, in the experience of our own and other laboratories, "false-positive" tilt-table responses have generally been reported in less than 15% of controls [9, 15, 20-22, 31-36]. Kapoor and Brant [37] reported a notable but currently unexplained exception, in which almost 50% of a group of young persons in whom isoproterenol provocation was used had false-positive responses.

The reproducibility of tilt-table testing has been the subject of numerous studies [20-22, 31-36, 38]. Concordance of outcomes in two tests in the same person (done 30 minutes to several weeks apart) has varied from approximately 60% [21, 36, 38] to as much as 90% [22, 34]. On average, reproducibility of about 80% can be expected [20, 32, 35], especially in patients with negative initial tests. For example, we examined outcomes in two sequential 80-degree head-up tilt tests in 23 patients being evaluated for recurrent syncope of unknown origin [20]. The initial study was done using a relatively short tilt duration (10 minutes) in the drug-free state and was followed by subsequent tilts in the presence of isoproterenol provocation (as much as 3 µg/min) if syncope was not induced. End points were syncope or the maximum tolerated dose of isoproterenol. The second tilt was done approximately 30 minutes after the first and used the maximum provocative conditions used in the first tilt. The second test concorded with the first in 12 of 15 patients whose initial tilt-test results had been positive (80%). Further, all of the 8 patients whose initial tilt-test results had been negative had the same outcome on the second test (100%). The concordance rate among all 23 patients was 87%. Brooks and colleagues [38] found that the test concordance rate was relatively low among patients whose initial tilt-test results were positive (36%) but was high (80%) among persons in whom results of the initial test were negative; the overall concordance rate was 63%.

The reproducibility of electrophysiologic and hemodynamic characteristics during induced vasovagal spells is as crucial to appropriate treatment selection as concordance of symptoms from one test to the next. In our experience [20], heart rate and hemodynamic findings correlated strongly in each of the two tests, suggesting that the characteristics of the induced episodes were similar within a given patient. Further, the findings by Grubb and colleagues [22] also suggest that repeated tilt tests done 3 to 7 days apart tend to show a strong intrapatient concordance for heart rate and blood pressure. On the other hand, Fish and colleagues [21] came to a different conclusion in their examination of tilt-test responses in young patients (mean age, 14.2 years; range, 8 to 19 years). In their study, 90-degree tilts were done for 15 minutes each in the baseline state and were followed, if necessary, by isoproterenol provocation. Syncope or presyncope was reproduced in 14 of 21 patients, and an additional 4 patients showed milder symptoms during the second test. However, the pattern of physiologic response (cardioinhibitory, vasodepressor, or mixed) varied between the two tests. Thus, despite their 67% rate of reproducibility, Fish and colleagues [21] were less than convinced that head-up tilt testing was useful for assessing therapeutic interventions. Similar concerns have been raised by de Mey and Enterling [19]. Additional studies providing even more careful evaluation of moment-to-moment heart rate and blood pressure changes are needed.

Pacing in Vasovagal Syncope

The potential value of cardiac pacing for the treatment of some forms of syncope was considered as early as the 1880s. MacWilliam [39] crystallized the work of various experimenters by stating that "in certain forms of cardiac arrest there appears to be a possibility of restoring by artificial means the rhythmic beat, and tiding over a sudden and temporary danger." Even more remarkably, he speculated that "such is especially the case in those instances where cardiac failure assumes the form of an inhibition of the heart beat by impulses reaching the organ along the vagus nerves." Of course, practicable devices to test this concept were not available until approximately 80 years later.

Although a few case reports have addressed the issue of cardiac pacing for the prevention of recurrent vasovagal syncope [40-43], this has become the subject of serious study only in the last few years [44-47]. Some of these studies have focused solely on temporary pacing interventions to evaluate the potential effect of pacing on hemodynamic conditions in patients with vasovagal syncope. Others provide observations obtained during temporary pacing and during follow-up of patients with implanted pulse generators.

In their initial consideration of this problem, Fitzpatrick and colleagues [43] reported that symptoms recurred in two patients who had originally received single-chamber ventricular pacemakers to prevent syncope of unknown origin. Tilt-Table testing not only confirmed susceptibility to neurally mediated hypotension-bradycardia (such as a presumed vasovagal faint) but also suggested that symptomatic hypotension was actually aggravated by ventricular pacing (the "pacemaker effect"). Subsequently, the pacing system in each patient was converted to a dual-chamber mode (DDI mode) with a basic rate of 50 beats/min and a hysteresis feature in which bradycardia (<50 beats/min) triggered the device to pace at a higher rate (80 beats/min). After this change, one of the patients became asymptomatic and the other continued to experience symptoms.

In a prospective evaluation of the effects of dual-chamber pacing in patients with vasovagal syncope, Fitzpatrick and colleagues [44] reported hemodynamic status and symptoms in seven patients with recurrent syncope who had had tilt-table testing and in whom vasovagal reactions were inducible on 2 successive days. The pacing protocol in this study used a hysteresis feature similar to that described above, with a base rate of 50 beats/min and a subsequent pacing rate of 90 beats/min. In all patients, pacing significantly improved the cardiac index (baseline, 1.0 ±0.2 L/min per m² body surface area compared with paced, 1.6 ±0.3 L/min per m²) and mean arterial blood pressure (baseline, 30 ±11 mm Hg compared with paced, 48 ±12 mm Hg) during induced vasovagal reactions. Additionally, pacing both reduced the rate at which blood pressure tended to decrease and prolonged the duration of tolerance to the head-up tilt position, thereby potentially giving patients more time to react and protect themselves. Syncope was prevented in five patients despite evident onset of a vasovagal reaction as attested to by both hemodynamic recordings and bradycardia sufficient to trigger the pacemaker. Two other patients remained as symptomatic as they had been during the initial tilt test. Quantitatively, the duration of upright posture tolerated after onset of the vasovagal reaction was longer during pacing than in the baseline state (unpaced, 0.9 ±1.2 minutes compared with paced, 3.2 ±1.6 minutes; P < 0.01).

In a study using temporary pacing during tilt-table testing, Samoil and colleagues [45] provide further support for the potential usefulness of cardiac pacing for the prevention of vasovagal syncope. Six patients (mean age, 31 ±12 years) with reproducible hypotension and bradycardia during tilt-table testing were studied in the baseline state, during ventricular pacing, and during dual-chamber cardiac pacing. In the pacing algorithm, rate hysteresis was used. The pacemaker was triggered when the heart rate decreased to less than 60 beats/min, but the pacing rate was set to 20 beats/min above the resting rate for treatment purposes. Results indicated that ventricular pacing was ineffective; neither the time to onset of symptoms nor the total tolerated upright tilt time differed from that of the unpaced state. On the other hand, dual-chamber pacing appeared to substantially improve upright tilt tolerance (dual-chamber pacing, 25 ±6 minutes compared with unpaced, 12 ±6 minutes; P < 0.001). Further, dual-chamber pacing prevented syncope during tilt-table testing in three of six patients.

Sra and colleagues [46] also used tilt-table testing to assess the potential of cardiac pacing for preventing vasovagal syncope. However, despite their seemingly clear demonstration of beneficial pacing effects (similar to those reported by Fitzpatrick and colleagues [44] and Samoil and colleagues [45]), they concluded that pacing offered little as a practical clinical therapeutic alternative. Further, although they made no attempt either to randomize therapy or to provide crossover for treatment failures, they indicated that medications were superior. In coming to their conclusions, Sra and colleagues [46] examined the effect of temporary cardiac pacing (at a rate approximately 20% higher than the patient's supine resting heart rate) for prevention of tilt-induced syncope in patients whose initially positive tilt-test was associated with marked bradycardia. Twenty-two patients had a positive baseline tilt test; after repeated tilt with pacing, 1 patient remained asymptomatic, 1 had dizziness without evident measurable hypotension, and 15 experienced presyncope rather than syncope. Only 5 patients had syncope during the second tilt. Subsequently, all patients also had repeated tilt-testing during pharmacologic therapy. In these follow-up tilt tests, metoprolol prevented symptoms in 10 of 22 patients, theophylline prevented symptoms in 3 of 12 patients, and disopyramide prevented symptoms in 6 of 9 patients. Using a relatively simple pacing algorithm, Sra and colleagues [46] showed that pacing could be remarkably effective when compared with either no treatment or with any of the conventional pharmacologic alternatives that they examined [46]. Further, the magnitude of tilt-induced hypotension with pacing was less than that during tilt testing done in the baseline untreated state (blood pressure decrease: paced, 41 ±19 mm Hg compared with unpaced, 59 ±16 mm Hg). Similar data during drug testing were not provided. Although tilt-table evaluation during pacing was associated with symptoms in most patients (5 had syncope, 16 had dizziness or presyncope, and 1 was asymptomatic), these symptoms were clearly less severe than in the untreated state (22 had syncope). Finally, the success of pacing alone as assessed by tilt-table testing did not differ appreciably from the estimated effectiveness of several of the drugs evaluated in this same patient population (pacing was effective in 23% of patients; metoprolol was effective in 45%; theophylline was effective in 25%; and disopyramide was effective in 67%). This is even more remarkable given that the relative effect of cardioinhibitory components compared with that of vasodepressor components was unaddressed. Subsequent long-term treatment was uncontrolled, precluding any meaningful conclusion about the relative efficacy of the various treatment strategies [46].

Finally, the report by Petersen and colleagues [47] provides evidence of the potential long-term usefulness of pacing in vasovagal syncope. This study, which was the only one with relatively long-term follow-up, reviewed experience with dual-chamber mode pacing (such as dual-chamber pacing and sensing with atrial-tracking capability disabled) in 37 patients with vasovagal syncope, each of whom had an apparent cardioinhibitory predominance identified during tilt testing. Patients were followed for 50 ±24 months; 89% of them had alleviation of symptoms, and 27% had complete resolution of symptoms. The overall frequency of syncopal episodes was reduced from 136 to 11 episodes per year. The principal clinical features predicting the usefulness of pacing included younger age (56 years compared with 76 years) and absence of a prodrome before spontaneous syncopal events.

Detection of Vasovagal Syncope Using Pacemakers and Pacing Algorithms

Pacing is currently effective only in those persons with recurrent vasovagal syncope in whom abrupt onset of marked bradycardia causes or closely precedes the onset of symptomatic hypotension Figure 2, because it is bradycardia that triggers currently available implantable devices. Ultimately, concomitant use of alternative on-line physiologic sensors (for monitoring of pressure, QT intervals, central venous temperature, and blood flow) within pacing systems may prove beneficial. Such sensors may permit recognition of imminent vasovagal syncope and earlier initiation of pacing intervention in patients in whom marked bradycardia either does not occur at all (for example, those who only develop a "relative" bradycardia) or occurs late in the course of events.

View larger version (33K): [in this window] [in a new window]   Figure 2. The potential value of cardiac pacing for moderating decrease in blood pressure and for extending the period before syncope in a patient whose cardioinhibitory vasovagal syncope was reproducible by 60-degree head-up tilt-table testing. In both panels, heart rate and blood pressure (Finapres plethysmographic technique) are recorded for each cardiac cycle at slow speed. One-minute intervals are indicated by the marks along the abscissa. Top. Baseline, untreated state. Heart rate and blood pressure initially increase when tilt is begun (left-hand arrow). Approximately 15.5 minutes into the procedure, abrupt bradycardia (right-hand arrow) is associated with hypotension and syncope (*). The brief overall duration of hypotension is due to the need to return the patient immediately to the supine position. Bottom. After placement of a dual-chamber pacemaker in the same patient. The pacemaker is triggered by a heart rate of less than 45 beats/min and responds by pacing at 90 beats/min (note the relatively fixed rate at the right side of the Figure andthe occasional irregularities due to premature beats). Once again, onset of tilt (left-hand arrow) is associated with an increase in heart rate and blood pressure. Abrupt onset of bradycardia (right-hand arrow, approximately 3 minutes into the tilt) triggers pacing. After onset of pacing, the rate of decrease in blood pressure is reduced. Although presyncopal symptoms are present, pacing is associated with prolonged tolerance to upright tilt despite systemic hypotension. Supine posture was restored without the occurrence of frank syncope.

The optimum pacing algorithm for prevention of vasovagal syncope has yet to be determined. At present, several devices use dual-chamber pacing with a form of rate hysteresis. As currently configured, the device is initially triggered by new onset of a low spontaneous heart rate and responds by pacing at a rate rapid enough to prevent or at least alleviate the rate of onset, the severity of the developing hypotension, or both. It is essential for the device to periodically terminate pacing to assess native heart rate and, if a hemodynamic sensor is incorporated, blood pressure. If these have returned to normal, the pacing sequence terminates. If not, pacing support is resumed for an additional period of time, after which the patient's cardiovascular status is reassessed. Unfortunately, given the absence of hemodynamic sensors, current devices are incapable of ascertaining whether the relation between heart rate and blood pressure is appropriate. Consequently, they cannot distinguish physiologic bradycardia (such as that associated with sleep or, in some cases, the pause that occurs after a premature beat) from pathologic bradycardia. Further, they are unable to ascertain whether the hypotensive period has indeed been reversed. Considerable technologic advancement and greater insight into the effects of pacing on the hemodynamics of the vasovagal faint are needed.

Conclusions

The usefulness of cardiac pacing techniques for treatment of patients with recurrent vasovagal syncope remains incompletely understood. Numerous clinical studies show that pacing may be beneficial in those patients whose recurrent intractable vasovagal symptoms are predominantly caused by a cardioinhibitory mechanism. If the findings of Petersen and colleagues [47] are confirmed, pacing alone may eliminate symptoms in 25% of such patients and may prove useful as adjunct therapy to alleviate symptoms in many others. However, as with other aspects of the treatment of vasovagal syncope, randomized controlled trials are needed. Additionally, substantial room remains for innovation both in earlier recognition of imminent vasovagal syncope by implantable devices and in the development of cardiac pacing algorithms specifically designed for treatment of this condition.

Appendix

We did a MEDLINE search for English- and French-language articles published between 1980 and 1994 about cardiac pacing for prevention or treatment of vasovagal syncope. The term cardiac pacing was used in conjunction with the terms vasovagal, neurally mediated, or neurocardiogenic syncope, but not with the term carotid sinus hypersensitivity.

Case reports and series from peer-reviewed journals were selected if they documented the presence of vasovagal syncope and assessed pacing effectiveness using tilt-table testing, clinical follow-up, or both. Four case reports and four series met these criteria. All findings were summarized individually: Statistical analysis of combined data was inappropriate because of differences among studies in patient selection, testing, and follow-up.