Oversensing

T-wave oversensing currently remains a significant problem in the management of ICD-implanted patients since it can be accompanied by the occurrence of inappropriate therapies particularly during exertion (when RT and TR intervals correspond to the VF zone due to sinus tachycardia). T-wave oversensing is associated with a typical pattern of alternating morphologically different signals, namely a high frequency signal (R-wave) and a low frequency signal (T-wave). For each cardiac cycle, the device counts the R-wave and the T-wave as a second additional signal resulting in a doubling of the heart rate. The alternating cycle duration (RT intervals and TR intervals) is usually pronounced for slow rates (short RT intervals and long TR intervals) although often less during exertion (RT and TR intervals generally equivalent) or for patients with a long QT syndrome.
Three different electrophysiological conditions can lead to T-wave oversensing during spontaneous ventricular beat: 1) delayed T-wave: this pattern is typical in patients with long QT syndrome in whom repolarization is delayed; the T-wave occurs when ventricular sensitivity is at its maximal level. In this type of channelopathy, there is also a risk of dynamic changes in the duration of the QT interval induced by catecholamines, but also in the morphology and amplitude of the T-wave thus increasing the risk of inappropriate therapies during exercise; 2) high amplitude T-wave and normal R-wave: this pattern is preferentially observed in hypertrophic cardiomyopathy, short QT syndrome, certain forms of long QT syndrome, certain metabolic abnormalities (hyperkalemia, hyperglycemia) as well as certain reversible causes of repolarization abnormalities (alcohol intoxication); 3) small R-wave (most frequent case): when the R-wave is of low amplitude, the probability of oversensing of the T-wave increases (adaptation of the sensitivity level as a function of the amplitude of the preceding signal). When the R-wave is of low amplitude, the device rapidly reaches high levels of sensitivity which is conducive to the oversensing of the T-wave particularly during exertion (possible decrease in the amplitude of the R-wave and slight increase of the amplitude of the T-wave). A sudden and rapid deterioration of the amplitude of the R-wave after implantation may be a manifestation of lead microdislodgement. A low-amplitude R-wave can also be observed in patients with right ventricular arrhythmogenic dysplasia, Brugada syndrome, cardiac sarcoidosis, or dilated cardiomyopathy involving the right ventricle. The reprogramming options are limited in this setting, any reduction in ventricular sensitivity being associated with a major risk of undersensing of a VF. 

Modern Medtronic ICDs are equipped with a specific algorithm whose aim is not to prevent T-wave oversensing but whose purpose is to detect its presence and avoid the occurrence of inappropriate therapies. The discrimination of R-waves and T-waves is achieved by means of a differential filter which amplifies the difference between the 2 signals. The algorithm recognizes the oversensing of the T-wave by identifying the repetition of sequences with alternation between two signals of variable frequencies (a sharp signal, a smooth signal) with fixed intervals (fixed RT intervals, fixed TR intervals). Various parameters are hence analyzed and if all are fulfilled on 6 consecutive intervals, the T-wave oversensing counter is incremented by +1. If a single criterion is not met, the set of 6 consecutive intervals is classified as normal. The next group of 6 events is then assessed using a rolling window. As long as 4 of the last 20 cycles fulfill the discrimination criterion of the T-wave, the device retains the diagnosis of T-wave oversensing and the therapy is not delivered. This algorithm is nominally programmed to ON.

This algorithm allows diagnosing T-wave oversensing and avoid the occurrence of inappropriate therapies in this setting. On the other hand, the problem which generated the oversensing, namely the low-amplitude R-wave, is not resolved. A change in the configuration of ventricular sensitivity arises and will be discussed in the following tracing. As a last resort, if no reprogramming option appears acceptable, repositioning of the lead may be proposed.

During atrial oversensing by the ventricular lead, the ventricular EGM shows an alternation between 2 signals of different morphology, the first corresponding obviously to the QRS complex, the second to an oversensing of atrial activity.  Oversensing issues are of particular concern in dependent patients since 1) they may cause ventricular pauses, 2) the oversensing may be prolonged in the absence of spontaneous QRS complexes, the level of sensitivity remaining maximal and adapted to the amplitude of the oversensed signals which is often low.

Oversensing of atrial depolarization by the right ventricular lead is rare and occurs mainly in patients implanted with an integrated bipolar lead. In patients in sinus rhythm, the right ventricular lead senses both atrial and ventricular depolarizations, with the PR interval being longer than the ventricular blanking period. If the patient has a complete atrioventricular block, P-wave oversensing may inhibit ventricular pacing and lead to asystole. Similarly, oversensing of atrial depolarization during atrial flutter or atrial tachycardia can cause both inappropriate therapies and asystole if the patient is dependent. 
Atrial oversensing by the ventricular lead preferentially occurs in 2 clinical situations: 1) dislodgement of the right ventricular lead to the atrioventricular junction (coincides with a decrease in the amplitude of the measured R wave); 2) when an integrated bipolar lead has been positioned near the tricuspid annulus, the distal coil crossing the valve (coincides with a preserved normal R wave amplitude). 

Atrial oversensing may also be observed under more exceptional circumstances: 1) right ventricular lead unintentionally implanted in the coronary sinus; 2) insulation defect located at the atrial portion of the ventricular lead resulting in oversensing of atrial activity; 3) interaction between the atrial lead and the right ventricular lead, the former coming into contact with the ventricular lead and generating a signal sensed at the time of atrial systole. 

Several solutions can be considered to address the problem and avoid the occurrence of inappropriate therapies or ventricular pauses in a dependent patient: 1) reduce ventricular sensitivity to eliminate the additional signal associated with the sensing of atrial activity; insofar as this programming change is accompanied by an increased risk of VF undersensing, an induction can then be carried out to verify that the induced VF is correctly detected with this new sensitivity value; 2) in most cases, atrial oversensing necessitates a repositioning of the defibrillation lead (new defibrillation lead if DF4 system or addition of a pacing/sensing lead if DF1 system).

The potential risk of electromagnetic interference with an implantable defibrillator has been frequently described in various settings, including within the hospital environment, in the patient’s home or during his or her professional activities. Interference may occur by conduction if the patient is in direct contact with the emitting source or by radiation if the patient is within an electromagnetic field. The most recent ICDs are protected against the vast majority of sources of interference that the patient may encounter in his or her daily life. The parasitic signals are typically filtered, the analysis being restricted to a narrow bandpass corresponding to the physiological signals (high-pass and low-pass filters). However, the high adaptive sensitivity levels required in ICDs for correct signal detection during ventricular fibrillation can promote the sensing of non-physiological signals corresponding to the same bandpass as cardiac signals. The signals corresponding to electromagnetic interference may not be appropriately filtered and lead to more or less severe consequences, ranging from the occurrence of inappropriate therapies to pacing inhibition in dependent patients, inappropriate mode switching due to false diagnosis of supraventricular arrhythmia, rapid ventricular pacing synchronized to atrial oversensing, suspension of detection, or reversion to asynchronous pacing. Exceptionally, interference with a high intensity electromagnetic field can cause permanent damage to the circuitry. 
The diagnosis of electromagnetic interference is based on the concordance between a history of exposure to a source at the time of the episode and oversensing of characteristic signals (fast, regular and spanning the baseline). Electromagnetic interference at the “mains” frequencies (60 Hz in the USA and 50 Hz in Europe) occurs when the patient is in physical contact with poorly insulated electrical equipment. If oversensing is prolonged, a single electrical shock is most often curative of the oversensing since the patient generally interrupts his activity immediately. Electromagnetic interference is more frequent for an integrated bipolar lead than for a “true” bipolar sensing, the sensing antenna being wider. The characteristic high frequency, non-physiological signals are sensed on all available channels (possible diagnosis of dual tachycardia, AF + VF) and are typically of higher amplitude on the far-field channel than on the near-field channel. 

The main preventive measures consist in identifying the emitting source and avoiding the use of poorly insulated instrumentation. Remote monitoring plays an essential role in identifying the emitting zone as it enables contacting the patient soon after oversensing has occurred. 

Typically, there are successive steps observed during a lead dysfunction. Initially, the device memory reveals multiple episodes of non-sustained VT without anomaly of the lead measurements. In a second step, a clear break can be observed in the different impedance, threshold and right ventricular sensing curves. Finally, the duration of the oversensing episodes is lengthened leading to the occurrence of multiple electrical shocks. The prevention of inappropriate therapies is one of the major advantages of remote monitoring. Inappropriate shocks caused by lead dysfunction also demonstrates the value of limiting the therapies to 6 shocks in the VF zone for a single episode. Indeed, the occurrence of successive inappropriate shocks constitutes a particularly difficult and traumatic experience for the patient concerned.

The lead constitutes the weak link of the defibrillation system with a variable percentage of dysfunction depending on the models. When in the presence of a suspected lead fracture, different exams and measurements must be performed: 1) a chest X-ray: radiographic abnormalities are not systematic and a typical pattern of lead fracture is not observed in over 50% of cases; 2) repeated measurements of pacing and defibrillation impedances: the latest generations of ICDs perform periodic (daily) impedance measurements. The presence of an abnormal value or significant variations in daily measurements (abrupt change of the impedance curve) may reveal a lead dysfunction with however only moderate sensitivity. Indeed, a significant number of patients present with lead dysfunction revealed by the presence of oversensing episodes without abnormal impedance or abrupt variation in pacing values. A low impedance value is suggestive of an insulation break (current leakage), a high value suggestive of a conductor wire break (loss of continuity of the defibrillation circuit); 3) evaluation of the sensing and pacing thresholds: the alteration of the standard pacing parameters is often delayed; the sensitivity relative to a decrease in ventricular sensing or an increase in pacing thresholds in predicting lead rupture is therefore very low; 4) analysis of the various electrograms: the pattern of the EGMs associated with a lead fracture is suggestive but non-specific: intermittent sensing of sudden, rapid, non-physiological cardiac cycles with possible saturation of the amplifiers (conductor wire break) or low amplitude (detection of myopotentials due to insulation break). These signals display substantial variability in both amplitude and frequency, are intermittent in the cardiac cycle and are most often recorded in the VF zone with values at the limit of the post-sense ventricular blanking period. These abnormal signals can affect the near-field channel and/or the far-field channel depending on the site of the fracture and may only become apparent after the delivery of a shock on an actual VF episode. 

When in the presence of inappropriate therapy early after implantation, two causes are to be particularly investigated: faulty connection of the pace/sense pin and an early dislodgement of the ventricular lead. A control interrogation of the ICD allows measuring the pacing thresholds, the R wave amplitude as well as impedances. Lead position can be verified by a chest X-ray.

A connection problem (loosened setscrew connection, incomplete contact between lead pin and header) can yield a similar presentation to that of a lead fracture: the impedance may be abnormally high and the pattern of the EGMs may be relatively similar (fast, disorganized, anarchic and sometimes high-amplitude signals saturating the amplifiers). In such instance, the chest X-ray can confirm the absence of contact between the pin connector and the header. A connector problem is most often revealed within a short time interval after the last intervention (primary implantation or change of pulse generator), the impedance variation and/or the oversensing generally occurring within a few hours to a few days after the procedure. Signs associated with a lead fracture/failure often appear later, except in cases of procedural damage (e.g. direct lesion of the lead with the electric scalpel). In case of a loose connection, manipulating the device inside in the pocket can reproduce the oversensing.

Other atypical presentations may replicate that of a lead dysfunction with combination of an abnormal impedance and oversensing: lead to lead mechanical interaction (contact between the defibrillation lead and a remaining lead fragment or an abandoned right ventricular lead), presence of air in the connector with a specific oversensing pattern corresponding to air bubbles escaping from the connector. 

The differentiation between these different clinical situations is essential, albeit sometimes difficult, in order to avoid a false diagnosis of lead fracture which has important therapeutic implications (necessity to replace the lead and discussion on the necessity to extract the “defective” lead). 

The lack of long-term reliability of ICD leads represents one of the main limitations of implantable devices and constitutes one of the most difficult challenges for the manufacturers involved. In the aftermath of the problems of the Fidelis lead, preventing the occurrence of multiple inappropriate shocks through early diagnosis of signs of dysfunction and the development of stronger leads has become a top priority for MedtronicTM. The latest devices are equipped with 2 algorithms specifically dedicated to the diagnosis of lead dysfunctions and the prevention of inappropriate shocks.

1) the LIA (Lead Integrity Alert): the occurrence of inappropriate therapies for lead dysfunction is usually preceded by a sudden change in impedance and/or short episodes of oversensing and/or recording of very short isolated ventricular cycles. The LIA algorithm was developed to generate a warning in response to a suspected lead dysfunction and to delay the occurrence of therapies by monitoring the bipolar and integrated bipolar impedances, the frequency of non-sustained VT episodes, and the frequency of very short ventricular intervals.  A lead dysfunction is identified if at least 2 of the following 3 criteria are met within the past 60 days: 1) the pacing impedance for either polarity (bipolar and/or integrated bipolar) is lower than 50% or greater than 175% of the baseline value which corresponds to the median of the previous 13 daily measurements; 2) the ventricular sensing integrity counter (short ventricular intervals ≤ 130ms) is incremented by at least 30 over a period of 3 consecutive days; 3) the device senses 2 rapid non-sustained VT episodes with a 5-beat RR interval average of less than 220 ms. When a lead dysfunction is suspected, the device transmits a specific remote monitoring alert notification, with an audible alert emitted every 4 hours until the ICD is interrogated and the number of intervals required for detection is automatically extended to 30/40 in order to delay (but not prevent) the occurrence of inappropriate therapies.

2) the noise discrimination algorithm on the RV lead (RV Lead noise discrimination): the aim of this algorithm is to identify the EGMs that are characteristic of a lead fracture and to withhold therapy based on a comparison of the amplitude (peak-to-peak) of the signals collected in the far-field channel versus those collected in the near-field channel. The amplitude (peak-to-peak) of 12 consecutive signals is analyzed through a sliding window using a counter. When the number of intervals required for the sensing of a VT or VF is reached, if 3 of the last 12 sequences (amplitude comparison in the far-field channel and the near-field channel) are classified as “noise”, detection is interrupted, therapy is withheld and a “RV Lead Noise” alert is triggered (an audible alert is emitted by the device every 4 hours until the device is interrogated, a remote monitoring alert notification is also transmitted if this specific alert has been programmed). This algorithm can be programmed to OFF, ON or ON + Timeout OFF. In the latter case, a specific duration is

programmed; if the oversensing of noise persists beyond this time (15 seconds to 2 minutes), the therapies will be delivered despite the diagnosis made by the device. In practice, in the presence of a lead dysfunction, the LIA allows a much earlier diagnosis than the second algorithm. Indeed, the LIA alerts in the presence of short intervals and non-sustained VT episodes whereas the noise algorithm intervenes only if the initial counter is filled (30/40) and hence when the oversensing is prolonged. This second algorithm is therefore only useful if, despite LIA alerts, no provision has been made or if the very first oversensing episode is sufficiently sustained to fill the initial counters (very rare presentation). It should be noted that the LIA often enables early diagnosis, delays but does not inhibit the therapies and generates an alert while the noise algorithm allows an often-later diagnosis but inhibits the therapies.