Inappropriate defibrillator shock due to EMI despite noise reversion and SecureSense - What is the mechanism?

Implantable cardioverter defibrillator (ICD) shocks are always a matter of anxiety and concern for both patient and clinician. We report a case of ischemic cardiomyopathy who reported to us after receiving two shocks within 48 hours. The therapy was inappropriate as noted from device logs. EMI was identified as the incriminating cause. We discuss the reasons why he received the therapy despite active 'Noise reversion' and 'SecureSense' algorithms to avoid noise /EMI related oversensing.

Case

A 65-year-old gentleman with ischemic cardiomyopathy and scar VT underwent single chamber implantable cardioverter-defibrillator (ICD, St Jude Medical, Ellipse VR 1377-36QC) implantation 2 years back. An Optisure LDA210Q/58cm DF1 dual coil lead was placed at RV apex. He became pacing dependent after 9 months but remained in NYHA class 1 (Fig. 1) on medications. This time he presented to the outpatient clinic after experiencing ICD shocks on 2 consecutive days. The device interrogation showed 15J shock in VF zone. The device was programmed to detect ventricular fibrillation (VF) at >214 bpm and ventricular tachycardia (VT) between 110 and 214 bpm. Sensitivity was programmed to 0.5 mV. Pacing lead impedance was 380 Ω and defibrillator lead impedance was 70 Ω for RV to Can. The stored electrograms (EGM) were analysed which revealed distinct QRS complexes in the near-field (NF) and far-field (FF) EGM. In addition, there were multiple small undulations of baseline in NF which corroborated by larger high-frequency signals recorded on far-field EGM (Fig. 2A). Thus, clearly, he received shocks due to oversensing. Fig. 2B shows the delivery of shock. Is it possible to predict the reason of oversensing from the stored EGMs and why the noise avoidance algorithms were violated ?

Fig. 1

12 lead ECG showing ventricular paced rhythm with wide fragmented QRS fitting with ischemic cardiomyopathy.

Fig. 2A

Stored electrograms showing noise in both channels. The discrimination (FF) channel shows more distinct EGM’S.

Fig. 2B

Stored EGM shows treated episode in the VF zone. It shows oversensing of noise which stops after 15J shock and return of sensed rhythm.

Discussion

The possible causes of oversensing are:

Electromagnetic interference (EMI)

Lead noise (lead fracture/insulation failure)

Loose set screw

Diaphragmatic myopotential.

Pectoral myopotential.

The near simultaneous presence of noise in both channels with distinct QRS makes EMI more likely [1]. The more prominent signals in FF compared to NF makes the source away from lead tip, making diaphragmatic myopotential an unlikely possibility. The quality of EGM goes against lead fracture/insulation related noise. The artifacts related to lead integrity are usually of varying amplitudes reaching saturation/clipping of signal, unlike the uniform signals as noted in our case. This excludes the second option. Loose set screw have EGMs similar to lead noise and usually present in early post-operative period. Moreover, the lead impedance trend was within normal range. Both lead noise and loose set screw are accompanied by abnormal device parameters (high/low impedance) [2]. Pectoral myopotential can sometimes have uniform noise more in FF channel. But ICDs do not use FF signals as default sensing channel, hence, pectoral myopotentials do not lead to oversensing (unless the sensing polarity is manually changed in some cases) [2,3].

Revisiting history we found that the patient had chronic musculoskeletal back pain for which he used to take physiotherapy sessions. In addition to manual therapy he used to receive transcutaneous electrical nerve stimulation (TENS). He remembers a change of consultant treating him in last two visits, possibly altering the site of electrodes and frequency applied. I think it is prudent to remember guidelines developed by Robertson et al. for-physiotherapy modalities in CIED (cardiovascular implantable electronic device). These guidelines suggest that deep heating in the form of short wave diathermy (SWD) should be contraindicated in patients with pacemakers within 3 m of the device, whereas microwave diathermy and ultrasound (US) therapy (with or without heat) should be contraindicated directly over the PG. The use of electrical stimulators (e.g. TENS) with transthoracic application or long duration direct current is also contraindicated. No precautions are listed for Laser therapy or for superficial heating in the form of infra-red or moist heat [4].

The reason our patient received shock was the visible presence of noise in both channels and the device observing the episode as VF. We wondered why the advanced algorithms failed to avoid oversensing due to EMI. We noticed that both ‘Noise reversion’ (NR) and “Secure sense” was programmed ‘ON’. The primary objective of NR is to screen for EMI [1]. It differentiates noncardiac oversensing from VF by using either of the 2 features of VF. First, in a true VF at least a few intervals are not extremely short (150 ms in Abbott ICDs). Second, in VF, sensed electrograms occur after a physiological ventricular refractory period, thus noise can be described as high frequency, noncyclical signals. . If noise is diagnosed, sensitivity is reduced and therapy is withheld. In the index case, surprisingly, the device didn't sense the events very close to each other and so defied the noise criteria. This was due to miniscule signals in NF channel despite larger and more frequent signals in FF. As NF was the sensing channel, only intermittent signals from the NF was sensed and reflected in the marker channel.

SecureSense (SS) primarily takes care of lead related or intracardiac noise [1]. It uses 2 independent sensing amplifiers-one on the sensing NF channel and another on a FF discrimination channel (nominally RV coil-can). When this algorithm is active, if the events on the sensing channel with intervals shorter than the slowest VT/VF detection interval do not correlate with fast intervals on the discrimination channel (within 50 ms of the slowest VT/VF detection interval) it is classified as oversensing. The algorithm inhibits therapy if too many events are classified as oversensing [1]. From the visual impression of the stored EGMs of the index case, we believed SS would be protective in our case. But surprisingly the close events were again recorded at varied intervals evading the algorithm. This resulted in failure to abort the inappropriate therapy.

The question remains why did the device not sense all the noise signals? The answer lies in the sensitivity settings. The patient had Vp of 99% and post paced ‘threshold start’ was kept at 1 mV and decay delay at 95 ms.The Low frequency attenuation was also ‘On’ meaning that the device's maximum sensitivity was restricted to 0.5 mV. Ideally in paced rhythm this should be ‘Off’. Thus we notice that marker channel erroneously records tachycardia beats only towards end of R-R interval where sensitivity is maximum (restrained to 0.5 mV which otherwise goes to 0.3 mV if Low frequency attenuation was ‘Off’). This algorithm is called ‘Autosense function’ in St Jude Medical devices. We have tried to explain this in Fig. 3. In the figure, after the paced event the device starts post-paced decay delay setting (in this case, 1 mV). It maintains this gain level for the duration of 95 ms as programmed,marked as 3 in Fig. 3. If the decay delay is kept at 0 ms it will start immediately as marked by 2 in Fig. 3. It then linearly increases the gain (reduces the mV setting) until the next sensed beat or until it reaches the Max Sensitivity setting (0.5 mV in this case). When an R-wave is sensed/paced, it begins its refractory period and restarts the cycle when the refractory period ends [5]. Thus signals which were above 0.5 mV were only sensed, that too towards later part of R-R interval as noted in EGM.

Fig. 3

After the paced/sensed R wave ‘threshold starts’ at programmed decay delay. 1) threshold start (sensed 50% R wave/1 mV if paced) 2) If decay delay programmed 0 ms (hypothetical) 3) decay delay at 95 ms (this case).

We reassessed the device and found that pacing and shocking impedance trends were all within normal range. Chest X-ray and fluoroscopy confirmed lead stability and integrity. As the cause was evident he was reassured and advised to avoid TENS in future.

To summarise, this unique case describes one of the rarer causes of inappropriate ICD discharges. due to EMI. It also highlights the need for a proper discussion with patient about possible device interactions in routine day to day activities.

Funding

None.

Consent

Consent has been taken from patient.

Data availability statement

All raw data and recording during the case are available for review.

Declaration of competing interest

None.