Objective According to the guidelines of the European Society of Cardiology, a presumed diagnosis of neurally mediated syncope (NMS) can be made when patients have a consistent history and competing diagnoses are excluded. In the present study, we compared the initial diagnosis of NMS by means of implantable loop recorder (ILR) documentation.
Methods In this prospective multicentre observational study which involved 51 hospitals in nine countries in Europe and Canada, 504 NMS patients ≥40 years, who had suffered ≥3 syncopal episodes in the previous 2 years received an ILR and were followed up for a mean of 15±11 months.
Results ILR recorded a spontaneous syncope in 187 cases, with an estimated diagnostic yield of 47% at 3 years. ILR findings were consistent with the initial diagnosis of presumed NMS in 162 (87%) patients whereas did not confirm NMS in another 25 (13%), who had an intrinsic cardiac arrhythmic cause (atrial tachyarrhythmias (#6), long pause on termination of tachyarrhythmia (#8), persistent bradycardia (#3), ventricular tachycardia (#4)) or a non-arrhythmic loss of consciousness (non-syncopal (#3), orthostatic hypotension (#1)). No clinical baseline feature was able to predict an intrinsic cardiac cause with the exception of more frequent non-syncopal atrial tachyarrhythmias on clinical history, which were present in 38% of cardiac versus 5% of NMS patients (p=0.001). Tilt table testing (TT) was positive in 76/136 (56%) presumed NMS and in 9/21 (43%) non-NMS patients (p=0.35); an asystolic response was present in 28/136 (21%) NMS and in 0/21 (0%) non-NMS patients (p=0.03).
Conclusions ILR findings showed results other than NMS in a small, although non-negligible, number of patients older than 40 years. TT was unable to discriminate between presumed NMS and non-NMS with the exception of an asystolic response which was highly specific.
- AUTONOMIC NERVOUS SYSTEM
Statistics from Altmetric.com
According to the guidelines of the European Society of Cardiology,1 ,2 a presumed diagnosis of neurally mediated syncope (NMS) can be made when patients have a history consistent with this diagnosis and competing diagnoses are excluded. Guideline recommendations are based on a strong pathophysiological background, clinical experience from observational studies and the consensus of experts. Nevertheless, patients are typically asymptomatic at the time of evaluation and the diagnosis of NMS often remains presumptive. A widely accepted standard of reference for confirmation of the diagnosis does not yet exist.
When a NMS is suspected, tilt (table) testing (TT) is frequently performed after the initial evaluation in order to confirm the diagnosis. The diagnostic value of TT has been questioned.3 In particular, the accuracy of TT has not been fully validated against populations with defined causes of syncope. The main reason has been the lack of a reliable gold standard of reference.
In the present study, we aimed to compare the diagnosis of NMS made at initial evaluation and with TT with that obtained with the documentation of a spontaneous event made by implantable loop recorder (ILR).
The multicentre, prospective Third International Study on Syncope of Uncertain Etiology (ISSUE-3) included patients ≥40 years old who had suffered ≥3 syncopal episodes of likely NMS aetiology in the previous 2 years. NMS was defined as any form of reflex syncope, with the exception of carotid sinus syndrome, and a sufficiently severe clinical presentation to warrant specific treatment. These individuals received an ILR and were followed up. In accordance with the guidelines of the European Society of Cardiology,1 ,2 NMS was considered likely when the clinical history was consistent with NMS and competing diagnoses were excluded. Patients were excluded if they had one or more of the following features: (1) cardiac abnormalities which suggested cardiac syncope (overt heart failure; EF <40%; old or recent myocardial infarction; hypertrophic or dilated cardiomyopathy; clinically significant valvular disease; sinus bradycardia <50 bpm or sinoatrial block; Mobitz I second-degree atrioventricular block; bundle branch block; rapid paroxysmal supraventricular tachycardia or ventricular tachycardia; pre-excited QRS complexes; prolonged QT interval; Brugada syndrome and arrhythmogenic right ventricular cardiomyopathy); (2) symptomatic orthostatic hypotension diagnosed by standing blood pressure measurement and (3) non-syncopal loss of consciousness (eg, epilepsy, psychiatric, metabolic, drop attack, cerebral transient ischaemic attack, intoxication and cataplexy).
Patients with positive and negative TT responses were included. The Italian protocol4 was recommended, which consists of 60°–70° passive tilting for 20 min or until syncope occurs. If the passive tilt phase did not induce syncope, 0.4 mg sublingual nitroglycerine spray was administered to the patient while the table was maintained in the same position; the test was continued for 15 min after pharmacological challenge. Positive TT responses were considered induction of syncope; positive responses were classified according to the New Vasovagal Syncope International Study (VASIS) classification5 as an asystolic or VASIS 2B form (those with an asystole ≥3 s) or mixed or vasodepressor forms (all the other forms without asystole). TT was considered negative if syncope did not occur.
After ILR implantation, all patients were followed up quarterly until the first documented syncopal recurrence, occurrence of a diagnostic arrhythmic event or the end of the study. Events were classified according to the ISSUE classification6 as type 1 (asystole >3 s), type 2 (bradycardia), type 3 (slight or no rhythm variations) and type 4 (tachycardia).
The following ILR findings were considered consistent with a likely or possible NMS, irrespective of the clinical presentation of the event, and adjudicated by the End-point Committee:
Asystolic syncope >3 s (likely NMS)
Non-syncopal asystole >6 s (likely NMS)
Syncope and progressive intermittent bradycardia (likely mixed NMS)
Syncope and no or slight rhythm variations (possible hypotensive NMS or orthostatic hypotension).
Conversely, the following findings did not confirm NMS and established an arrhythmic intrinsic syncope:
Symptomatic paroxysmal atrial fibrillation, including brady-tachy forms, that is, sinus arrest at the end of a paroxysmal supraventricular tachycardia
Symptomatic paroxysmal supraventricular tachycardia, including brady-tachy forms, that is, sinus arrest at the end of a paroxysmal supraventricular tachycardia
Paroxysmal ventricular tachycardia.
The protocol was approved by a research ethics board at each centre and each patient provided signed informed consent. The full study protocol has been previously published,7 as have the results of the randomised trial.8
Continuous data are shown as averages±SDs or medians (25th–75th centile), as appropriate, while absolute and relative frequencies were used to describe categorical data. The Shapiro–Wilk test was performed to check the skewness of distributions. The unpaired Student t test or the non-parametric Mann–Whitney tests was used to compare continuous variables depending on data distribution. Fisher's exact test was used to compare proportions. The time to the first recurrence of syncope was analysed by means of Kaplan–Meier survival curves. Analyses were carried out by means of SAS V.9.3 (Chicago, Illinois, USA).
Study participants were enrolled from July 2006 to November 2010 and follow-up concluded in November 2012. Among 509 eligible, five patients dropped out before ILR implantation and were excluded. Over a mean observation period of 15±11 months, ILR recorded an event in 187 (37%) out of 504 patients, with an estimated probability of 31% (95% CI 27 to 36) at 1 year, 40% (95% CI 36 to 45) at 2 years and 47% (95% CI 40 to 53) at 3 years (figure 1). The baseline clinical characteristics of the patients with a diagnosis were similar to those without, with few marginal exceptions (table 1). During the ILR observation period, six patients died, five developed an acute coronary syndrome, two developed heart failure and one had a cerebral transient ischaemic attack. The ILR was explanted in the absence of an end point in 16 patients owing to pocket infection or patient's request (intolerance).
Comparison with clinical evaluation
Of those patients with a diagnosis, 162 (87%) had ILR findings which were consistent with the initial diagnosis of NMS: 79 had syncope with asystole >3 s (median 10 s (IQR 5–17)), 20 non-syncopal asystole >6 s (median 8 s (IQR 6–11)), 20 syncope with bradycardia and 43 syncope with no or slight rhythm variations. The clinical presentation of 109 available syncopal episodes, which reproduced the previous events in the opinion of the patients, was typical vasovagal in 34%, typical situational in 13% and atypical or no prodromes in 53% of cases. In another 25 (13%) patients, the initial diagnosis of NMS was not confirmed by ILR findings (figure 2). In 21 of these, an intrinsic cardiac arrhythmia cause was diagnosed: symptomatic paroxysmal atrial tachyarrhythmia (atrial fibrillation #3 or atrio-ventricular (AV) nodal re-entrant tachycardia #3), long pause on termination of supraventricular tachyarrhythmia (#8) and ventricular tachycardia (#4)—which were all diagnosed by ILR documentation—and persistent bradycardia (#3), which was documented during outpatient visits. No single or combinations of clinical baseline features were able to predict an intrinsic cardiac cause, with the exception of more frequent non-syncopal atrial tachyarrhythmias on clinical history, which were present in 38% of cardiac vs 5% of NMS patients (p=0.001); conversely, a history of typical vasovagal presentation was more frequent in NMS patients than in cardiac patients (49% vs 24%, p=0.04) (table 2). Finally, four patients had an episode of loss of consciousness in the absence of rhythm variations at ILR, which was re-evaluated and attributed to orthostatic hypotension (#1) and non-syncopal loss of consciousness (#3).
Comparison with tilt TT
Among the 187 patients with a recorded event, TT was positive in 85 (during the passive phase in 24 and during drug challenge in 61) and negative in 72 (not performed in 30). TT was positive in 76 NMS and in 9 non-NMS patients (p=0.35) (table 3); sensitivity and specificity of a positive test were 56% (95% CI 53% to 59%) and 57% (95% CI 36% to 77%), respectively. The sensitivity of TT was not different in the 93 patients with ILR documentation of asystole or bradycardia (likely NMS) and in the 43 patients with ILR documentation of syncope with no or slight rhythm variations (possible NMS): 59% (95% CI 54% to 63%) vs 49% (95% CI 39% to 58%). The specificity of TT was 53% (95% CI 29% to 75%) in the 17 patients with an intrinsic arrhythmia.
An asystolic response was present in 28 NMS (10 during passive and 18 during drug challenge) and in 0 non-NMS patients (p=0.03); sensitivity and specificity of an asystolic test were 21% (95% CI 18% to 21%) and 100% (95% CI 82% to 100%), respectively.
Passive only TT was positive in 22 NMS and in two non-NMS patients (p=0.74); sensitivity and specificity of the passive phase were 16% (95% CI 13% to 17%) and 90% (95% CI 71% to 98%), respectively.
Similar TT results were observed in patients without a recorded event (table 3).
The study suggests a diagnosis different from the original one of NMS in a small, although non-negligible, number of older patients when the initial diagnosis—based on clinical history and exclusion of competing causes according to guidelines—was validated by ECG documentation of a spontaneously occurring event. Most of the changes in diagnosis were due to intrinsic cardiac arrhythmias, which were largely unpredictable by baseline characteristics. This aspect, which has not yet been clarified in the literature, may be relevant in clinical practice. TT was unable to discriminate between cardiac (non-NMS) and presumed NMS with the exception of an asystolic response which was highly specific. Therefore, the additional diagnostic value of TT to clinical evaluation is of little help in establishing the diagnosis of NMS.
Age is the first consideration. Almost all young patients who have a syncopal loss of consciousness have NMS, and this can be teased out and confirmed in most cases through careful analysis of the patient's history. In older patients (age >40 is a reasonable definition of older according to ISSUE-3 inclusion criteria), the situation is quite different for several reasons; history is less clearly indicative of vasovagal syncope in older patients,9 ,10 the risk of potentially fatal causes and comorbidities rises11 and at least two types of more recently understood bradycardia (NMS and adenosine-sensitive heart block) exist. These bradycardias can be treated with permanent pacing.8 ,12 ,13 A primary tachyarrhythmia was a not infrequent cause of syncope which was caused either directly at the onset of the tachyarrhythmia itself or by a long pause (range 3.5–9 s) which occurred at the termination of the tachyarrhythmia as a consequence of a delayed recovery of sinus node automaticity. An example is shown in figure 3. While in most cases the ILR documentation was the first documentation of tachyarrhythmia, eight patients had already a history of asymptomatic atrial tachyarrhythmia which was considered insufficient for the diagnosis lacking the evidence of a causal relationship. This means that syncope can be an epiphenomenon of an atrial tachyarrhythmia, which is not always present in a given patient. On the other hand, the role of autonomic reflexes in favouring syncope in patients with atrial tachyarrhythmia is well known. Some studies14 ,15 showed that in predisposed TT+ patients, a NMS can be triggered by the sudden onset of an atrial arrhythmia, albeit the intrinsic cardiac trigger and the specificity of therapy justify classification of these forms as cardiac syncope. In other words, multiple mechanisms—intrinsic arrhythmia and reflex susceptibility—might operate together to cause syncope in the same patient. Even if the history of a spurious tachyarrhythmia should suggest a causal role in the genesis of syncope, however the causal effect could be confirmed only in eight and was excluded in other eight patients (table 2) among a total of 40 patients with a history of asymptomatic atrial tachyarrhythmia (table 1). Therefore, in older patients, ILR monitoring is a reasonable means of investigation and may now be regarded as an additional useful tool for confirming the diagnosis and starting specific therapy.
As yet, a standard of reference for confirming the diagnosis does not exist. The central problem is that there is no good evidence-based clinical definition of the syndrome of NMS. Most studies16–18 have used strict diagnostic criteria without incorporating follow-up information. Some authors have refined history-taking criteria by using a quantitative history score questionnaire.19 Others have used long-term clinical follow-up to assess the accuracy of the initial diagnosis, with recurrent episodes, clinical findings or death during follow-up being deemed incompatible with the initial diagnosis.20 These methods have limitations. TT has proved unable to reliably distinguish between NMS and other forms, owing to the low sensitivity and specificity of the test.3 Moreover, the accuracy of TT has not been fully validated against populations with defined causes of syncope. In this study, TT was positive in 43% of non-NMS patients, thus suggesting very poor specificity. Patients with negative and positive responses to TT have similar symptoms, similar symptom burdens and similar recurrence rates of syncope.19 ,21–24 Finally, there is a lack of correlation between responses to tilt testing and the mechanism of spontaneous NMS detected by ILR.24
We acknowledge that the ILR strategy also has two important pitfalls when used as a standard of reference for the diagnosis of NMS.
First, the diagnosis of NMS cannot be considered certain in several cases even after ILR documentation of an event. Indeed, while the documentation of progressive sinus bradycardia or progressive tachycardia followed by progressive bradycardia and, perhaps, ventricular asystole due to sinus arrest virtually confirm a neurally mediated mechanism,6 ,23 ,25 the finding of sudden-onset paroxysmal atrio-ventricular block with concomitant increase in sinus rate cannot exclude an adenosine-mediated block.12 ,13 Moreover, the documentation of syncope in the absence of rhythm variations by making the assumption of hypotensive NMS cannot exclude orthostatic hypotension or another unknown cause of syncope. In these cases, the ILR-based diagnosis of NMS can only be considered assumed. Thus, the diagnostic accuracy of clinical evaluation could be even lower than the 87% reported here.
Second, more than half of our patients had no ILR-documented events over 3 years of follow-up; thus, the initial diagnosis could not be confirmed. Although the clinical characteristics of these patients were fairly similar to those with a diagnosis (table 1), we cannot exclude the possibility that diagnostic accuracy might have been different if they had had ILR documentation. ILR studies23 ,26 ,27 in general have reported a relatively high rate of non-recurrence of syncope or another diagnostic finding although Furukawa et al28 followed patients for a longer period and they were able to demonstrate that in as much as 80% of patients a diagnosis of the cause of syncope could be made.
Finally, the presumptive diagnosis was made in tertiary centres specialising in syncope evaluation. It is likely that this overestimates the value of clinical diagnosis if made by primary care providers.
Conclusions and practical perspectives
In conclusion, a non-negligible risk of misdiagnosis exists when NMS is diagnosed in patients >40 years according to clinical history, physical examination and exclusion of other competing causes even if strict standardised guideline-based diagnostic criteria are applied when comparison with ILR findings is made. These patients are indistinguishable from true NMS patients on standard clinical evaluation and were potentially at risk of life-threatening arrhythmias, which could be identified and treated only by means of an ILR strategy.
The use of TT in order to confirm the diagnosis is hampered by low sensitivity and specificity. In this regard, an interesting original finding of this study is that an asystolic positive response (VASIS 2B) seems to have an excellent specificity even if to the detriment of sensitivity; in other words, the finding of an asystolic response allows confirmation of the initial diagnosis of NMS. Specificity is also improved when passive phase only of TT is used, but its low positivity rate limits its usefulness. These data are anticipated to move use of the ILR more towards being the ‘gold standard’ in diagnosis of presumed NMS from clinical assessment.
What is already known about this subject?
A presumed diagnosis of neurally mediated syncope (NMS) can be made when patients have a history consistent with this diagnosis and competing diagnoses are excluded. A widely accepted standard of reference for confirmation of the diagnosis does not yet exist. In particular, the accuracy of tilt (table) testing (TT) has not been fully validated against populations with defined causes of syncope.
What does this study adds?
The study suggests a diagnosis different from the original one of NMS in a small, although non-negligible, number of older patients when the initial diagnosis—based on clinical history and exclusion of competing causes according to guidelines—was validated by ECG documentation of a spontaneously occurring event by means of implantable loop recorder. TT was unable to discriminate between cardiac (non-NMS) and presumed NMS with the exception of an asystolic response which was highly specific.
How might this impact on clinical practice?
Most of the changes in diagnosis were due to intrinsic cardiac arrhythmias, which were largely unpredictable by baseline characteristics. This aspect, which has not yet been clarified in the literature, may be relevant in clinical practice. The additional diagnostic value of TT to clinical evaluation is of little help in establishing the diagnosis of NMS.
Collaborators The investigators in the ISSUE-3 study are listed in the online supplementary appendix.
Contributors Each author contributed significantly to the submitted work. In particular, they contributed to (1) conception and design or analysis and interpretation of data or both; (2) drafting of the manuscript or revising it critically for important intellectual content; and (3) final approval of the manuscript submitted. MB, PI of the ISSUE-3 study, are responsible for the overall content as guarantor.
Funding The study was funded by Medtronic Bakken Research Center.
Competing interests MB reports receiving modest consultancy fees from Medtronic and being a direct shareholder of F2 solution; RS is a consultant to Medtronic, receiving modest fees, and is a paid lecturer for St Jude Medical; AM reports receiving modest consultancy fees from Medtronic; XB reports receiving limited consultancy fees from Medtronic and St Jude Medical; HHE reports being a limited paid lecturer or tutor for Medtronic, St. Jude, Boston, Sorin and conducting advisory board activity for MSD; NG is an employee of Medtronic.
Patient consent Obtained.
Ethics approval Research ethics board at each centre.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.