Implantable loop recorders for assessment of syncope: is 'Saint Thomas approach' still the best diagnostic strategy?

Syncope is a frequent and sometimes challenging clinical symptom, for the management of which would be ideal to ascertain what was occurring during the spontaneous event. 1 Unfortunately, in reality, it is a rare opportunity. Given the short duration of syncope episodes, patients are usually symptom-free during the subsequent medical observation, so that the diagnosis may not be obvious. Furthermore, the provocation tests available for patients' assessment are hampered by lack of sensitivity 2 and predictive value. 3 , 4 Because cardiac arrhythmias are among the most significant causes of syncope, 5 electrocardiographic (ECG) documentation of the spontaneous episode is recognized as a diagnostic gold standard, 1 , 6 allowing a mechanism-based treatment (arrhythmia detected) or patient's reassurance (arrhythmia excluded). On the other hand, the majority of patients with syncope present with rare (over months or even years) spontaneous recurrences leading compliance issues with any external device in the setting of a long-term ECG monitoring strategy. The above considerations were the background for the insertable loop recorder (ILR) development. The ILR is a programmable device which is placed under the skin in the upper chest with a short, outpatient procedure. It has two electrodes on its body and it is equipped of a retrospective (loop) memory that continuously records and deletes the patient's subcutaneous ECG. Such features are particularly important for the evaluation of syncope, encompassing any issue related to patients' compliance and allowing long-term ECG monitoring even when the device is activated after the patient has gained consciousness. The first ILR device (Medtronic Reveal® model 9525, Minneapolis, MN, USA), which was only able to capture retrospective ECG recordings after ‘activation’ by means of an external, hand-held device (Patient Activator), was released in 1998. Two ECG signal storage modes were allowed: (i) a 21 min mode capturing a single event with an ECG recording lasting from 20 min before to 1 min after activation; or three events each with ECG storing from 6 min before to 1 min after activation; (ii) a 42 min mode with the above intervals doubled; to increase the recording time, this mode required data compression with the ECG waveform stored at a lower resolution (50 Hz instead of 100 Hz sampling rate). In 2002, a second ILR generation (Medtronic Reveal® Plus model 9526) was released. The main difference with the previous device was the capability to automatically detect and capture arrhythmic events (bradycardia, tachycardia, and asystole) according to physician-programmed settings, while the external activation feature and the total amount of memory were unchanged. Consequently, with the automatic algorithms activated (and thus a portion of the memory allocated for this function), a shorter ECG trace (12 min before and 2 min after manual activation with the lower resolution mode) could be stored after external activation of the device. However, automatic detection function could be excluded and the device could be programmed as the previous 9525 device. In 2007, a third ILR generation (Medtronic Reveal® DX model 9528; and Reveal® XT model 9529) was released. Reveal® DX differs from the previous models by several technical characteristics ( Table 1 ). As regards ECG storage, automatic detection function is active by default while a single manually activated mode (three events, each one with an ECG recording from 6.5 minutes before to 1 minute after activation) is possible. Main characteristics of the subsequent ILR generations Main characteristics of the subsequent ILR generations Following the initial experience in patients with unexplained syncope, it is likely that the clinical role of ILRs will increase and devices will be used instead of, or before, other conventional diagnostic tests 6 : thus, any effort aimed to refine the clinical use of the device, including the best programming mode, is fully justified. Turley et al.7 report on the time to ILR manual activation among a cohort of patients with unexplained syncope or presyncope implanted with a Reveal® or a Reveal® Plus between 1998 and 2008. Notably, all ILRs implanted during such time interval were programmed for single external activation with the 42 min, low-resolution mode. This means that in order to maximize the storage length after external activation, all Reveal® Plus monitors would have their automatic detection features turned off. In the second phase, patients who received a permanent cardiac pacemaker based on ILR findings were retrospectively identified and the time from bradycardia onset to manual ILR activation during any index episode was measured. The median activation time was relatively short (136 s), but for a minority of patients, longer intervals, up to 488 s, were needed. By applying a 2:1 ‘safety margin’ to such value, the authors suggest that to capture adequate ECG tracings (i.e. including both symptoms onset and ILR activation) of the events occurred during the patients' monitoring period, 20 min could be the correct duration to be programmed for optimal recording. Consequently, with the last generation devices, some episode of symptomatic bradycardia could be missed because of inadequate recording duration. In general terms, the basic question from which the paper of Turley et al. generates is appropriate. Nevertheless, in my opinion, the results of the study, although interesting, do not raise any concern about the clinical performance of new generation ILRs. To discuss that, the paper needs to be put into a larger perspective starting with the question of what we have learnt from the first 10-year experience with ILRs. Subsequently, it will be possible to address the issue of which activation features should be preferred for each of the current indications for ILR. At the beginning of the clinical experience with ILRs, the most rigorous correlation of the ECG findings with spontaneous syncopal relapses was, of course, the only acceptable gold standard. Nevertheless, it became soon clear that: (i) patients with indication for ILR had a high probability of cardiac arrhythmias as the cause of symptoms even despite negative conventional diagnostic work-up 8 , 9 ; (ii) ‘minor’ findings such as asystole or primary tachyarrhythmias causing ‘only’ presyncope 10 or even asymptomatic and detected automatically, 11 usually predicted the mechanism of the subsequent, spontaneous syncope; (iii) a significant proportion of patients implanted with ILRs were unable to properly activate the device at the time of symptoms. 10 , 12 In light of this, the most interesting characteristic of second-generation ILRs was just the new automatic detection function: provided that specific arrhythmias, even if asymptomatic, represented diagnostic findings, 1 their detection could have allowed to initiate treatment without waiting for the recurrent syncopal episode. Unfortunately, with Reveal® Plus, this function was hampered by artefacts causing the false detection of arrhythmias, device memory saturation, and overwriting of appropriately detected episodes. 13 For this reason, many physicians (like Turley et al. ) preferred such algorithms systematically switched off. On the other hand, an improved diagnostic yield of ILRs with automatic detection function active has been advocated by others, 14 , 15 at least for those patients who were unable to activate the device after syncopal recurrence and who had it interrogated early after the event. 16 With the last ILR generation, automatic detection algorithms have been significantly improved 17 and the clinical impact of that is suggested to be substantial by preliminary observations. 18 At present, there are few established indications to ILRs for assessment of syncope. 1 Finally, the potential role of the ILRs is emerging for the risk stratification of patients with a previous myocardial infarction or inherited cardiomyopathy. 6 Despite the lack of scientific evidence, it can be argued that even in such settings it would be better for potentially life-threatening arrhythmia, although transient and asymptomatic, to be discovered. Patients with unexplained syncope and clinical or ECG features suggesting an arrhythmia as the cause of symptoms. In this setting, we are usually dealing with elderly subjects 19 for whom paroxysmal atrioventricular block is the more likely finding during ILR-documented recurrence. 8 , 9 In these patients, onset of syncope is sudden and duration of the arrhythmia is short. Because of this, missed documentation of the episode, rather than an inadequate recording time, is more likely to relate to ‘environmental’ (external) factors: unwitnessed syncope with retrograde amnesia; inability to properly activate the device; severe secondary trauma with the ILR activation option not perceived as a clinical priority, or simply not known by the rescuers, etc. In these situations, even the longest as possible retrospective recording times would not ensure capturing an adequate ECG tracing of the full episode, and a ‘parachute’ like the one offered by concurrent automatic activation of the device should be preferred. Furthermore, in such clinical settings, waiting for a clinical recurrence could expose the patient to severe trauma 12 , 20 or, at worst, to death with a clinically useless, although well ILR-documented diagnosis for the posterity. 21 Also for this reason, ECG findings, even asymptomatic, allowing the selection of treatment, should be preferred. To assess the contribution of bradycardia in patients with neurally mediated syncope and a severe clinical presentation requiring specific treatment. 1 , 22 Previous observations with tilt testing suggested a short duration of the induced neurally mediated reflex in the majority of such patients. 23 Interestingly, in the same clinical setting, the duration of spontaneous reflex was even shorter, while the uncommon cases with longer reflexes were mainly related to a very long prodromal phase. 24 In such a context, with the ILR probably activated during prodroms, the longest as possible retrospective ECG storing would not have captured the most severe part of the reflex (i.e. the syncopal phase). Also in this case, concurrent, bradycardia-triggered automatic ILR activation would warrant adequate documentation of the episode avoiding the risk of a ‘reverse’ oculo-bradycardiac reflex. 25 In conclusion, restrictive programming criteria leading to a ‘St Thomas approach’ (I don't want to know—because I wouldn't believe and I wouldn't treat it in any case—any arrhythmia unless strictly related with a spontaneous syncope) cannot be considered the best choice in patients implanted with an ILR. The current Reveal® activation features represent a good compromise to avoid that and to optimize the diagnostic yield of the device. Further studies are needed to confirm whether new automatic detection algorithms will actually improve the clinical performance of ILRs. Conflict of interest: none declared.