In 1966, following the death of his mentor Professor Harry Heller from ventricular tachycardia (VT), Michel Mirowski began to research a way to prevent future deaths from ventricular arrhythmia, working with Morton Mower and William Staewen in Sinai Hospital, Baltimore. The first human implant of an automated implanted cardioverter defibrillator (ICD) took place in 1980 amid much criticism and scepticism from the medical community. Since then, ICD implants have increased exponentially, with expanding indications for implantation and increased identification of patients fulfilling implant criteria. Early implants required formal thoracic or abdominal surgical implantation under general anaesthesia, and rapid hardware development allowed completely transvenous lead implantation with much less invasive surgery under local anaesthesia.
Trials Supporting Use
Multiple clinical trials have confirmed the benefits of ICD therapy in reducing mortality. Early trials focused on secondary prevention of arrhythmic death following aborted cardiac arrest, with impressive reductions in mortality.1 Later studies sought to identify patients at primary risk of arrhythmic death; the Multicenter Automatic Defibrillator Implantation Trial (MADIT-I) study2 evaluated patients who had prior myocardial infarction (MI), impaired left ventricular (LV) function, non-sustained VT on ambulatory monitoring and a positive VT stimulation study. There was a clear mortality reduction with ICD implantation in this group. More recently, MADIT-II has shown that post-MI patients with an ejection fraction below 30 % and conduction delay evident on their electrocardiogram (QRS duration greater than 120 ms) benefit from ICD implantation without the need for further electrophysiological study or ambulatory monitoring.3
For non-ischaemic dilated cardiomyopathy, several trials have shown inconclusive effects on all-cause mortality, although the trend has been towards reduced mortality with ICD implantation. The Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT), however, evaluated both ischaemic and non-ischaemic cardiomyopathy patients, and analysed the data for each separately. The non-ischaemic group had a very similar mortality reduction (risk ratio 0.74) to that of the ischaemic group (risk ratio 0.78), and the data reached statistical significance.4 More recently, a meta-analysis of ICD trials in patients with non-ischaemic cardiomyopathy showed a reduction in mortality of 27 %,5 in line with the survival benefit of ICD therapy in other patient groups.
These trials show that ICD implantation is superior to medical therapy at reducing mortality. The benefits are also sustained, with a 34 % reduction in mortality with ICD therapy compared with optimal medical therapy shown at eight years in the MADIT-II study group.6
Problems with Implantable Cardioverter Defibrillator Therapy
Despite the clear benefits of ICDs in reducing mortality, a number of problems have arisen since their widespread introduction into clinical use.
Shocks While Conscious
A significant problem with ICDs is the potential for a patient to receive shock therapy while conscious. Early ICDs were programmed to deliver shock therapy if potentially life-threatening arrhythmias were detected and no other therapies were available. While effective at preventing arrhythmic death, problems arise from the appropriate treatment while the patient is conscious.
Shocks from an ICD can be thought of as necessary, unnecessary or inappropriate. Necessary shocks are for ventricular arrhythmias that cannot be terminated by other means; shock therapy is unavoidable if the arrhythmia is to be terminated, although it may be delayed until consciousness is impaired by the arrhythmia or until sedation can be given.
The development of anti-tachycardia pacing (ATP) allowed the better treatment of harmful arrhythmias not usually associated with cardiac arrest and sometimes without immediate compromise, since ATP is not so disturbing to the patient. In addition, ATP may be an initial treatment strategy for compromising ventricular arrhythmias, with shocks used if ATP fails to terminate the arrhythmia. Unnecessary shocks are those that are delivered for an appropriately detected ventricular arrhythmia, but where the arrhythmia could have terminated spontaneously or by ATP. Improvements in programming may reduce the chance of unnecessary shocks, for example, by the use of ATP for rapid VT and of longer detection times.
Inappropriate shocks arise from inaccurate detection of a ventricular arrhythmia, most commonly because of an atrial arrhythmia conducted at a high ventricular rate. Other causes include sensing of ‘noise’ caused by lead fracture. Improved detection algorithms and discriminators have reduced the incidence of the former and may help to provide early warning of the latter before an inappropriate shock. Inappropriate shocks are common; for example, in SCD-HeFT, of the 32.7 % of trial participants who received shocks, half received inappropriate shocks.4
Shock therapy is associated with higher mortality; while the exact mechanisms are unclear, shocks with energy greater than 9 J have been shown to reduce cardiac index, whereas shocks with lower energy have not had this effect.7 In the SCD-HeFT, patients who had appropriate shocks had a mortality over double the rate of those who had no shocks; an even higher mortality was noted for the group of patients who had multiple shocks, and for patients with both appropriate and inappropriate shocks.8
A recent study also found a strong association between shock ‘dose’ and mortality.9 A further meta-analysis of data on inappropriate shocks also confirmed increased mortality with increasing numbers of shocks.10 However, ATP has not been shown to increase mortality.11 Thus, strategies to reduce shocks may have mortality benefits as well as the clear benefits in terms of reducing patient distress.12
A number of strategies may be used to reduce the number of shocks. For example, the PainFree Rx II study demonstrated that ATP is successful at terminating fast VT (188–250 beats per minute) in 72 % of cases, avoiding unnecessary shocks.13
Some ventricular arrhythmias may be self-terminating and therefore deferring therapy may result in no treatment being needed. Clearly, a balance has to be struck between waiting long enough for any non-sustained arrhythmia to terminate, but not so long that the patient becomes dangerously compromised. In the RELEVANT study,14 ventricular fibrillation detection was 30/40 intervals compared with 12/16 in the control group. The prolonged detection time resulted in reduced therapies (both appropriate and inappropriate), as well as in fewer heart failure hospitalisations, with no increase in episodes of syncope or death. Therefore, delaying detection modestly does not increase adverse events, but does reduce patient shocks.
An interesting new development is the ‘virtual ICD’ – a computer simulation of ICD function based on historical data that allows the prediction of the effects of new detection and treatment algorithms.15 This has many potential advantages, allowing the testing of individual algorithms without the need to recruit large numbers of patients into clinical trials, with results becoming available rapidly.
Vascular Access Issues
Currently, almost all ICDs are implanted using transvenous leads. Vascular access is rarely an issue in new implants except in cases of congenital cardiac or vascular anomalies, or where previous vascular interventions have caused complications, but the longevity of current-generation ICD generators and leads is such that younger patients may undergo several lead implants over the course of their lives. With each procedure the risk of vascular damage and infection increases and chronic lead extraction may be required.
To overcome the problems associated with vascular access, a completely extravascular ICD system has been developed.16 While this is currently unable to pace the heart (other than for transient, post-shock back-up pacing), it may represent a good option for patients who do not require pacing for bradycardia and whose ventricular arrhythmias are unlikely to respond to ATP.
The systems may be as reliable as transvenous ICDs in detecting ventricular arrhythmias,17 and have been shown to be effective at terminating detected arrhythmias, both at defibrillation threshold testing16 and in clinical practise.18 A potential concern with subcutaneous ICDs is that the infection rate related to this large device appears to be higher than that with conventional ICDs, at around 5 %,18 although it is likely that the extravascular nature of the device makes device infection less likely to lead to severe sepsis or endocarditis.
Traditionally, interrogation and follow-up of pacemakers and ICDs require the patient to attend a clinic, usually held in a hospital outpatient unit. More recently, most of the device manufacturers have developed systems that allow trans-telephonic device interrogation in the patient’s home, with data sent to the ICD centre and/or physician via a variety of means. This allows some follow-up appointments to be remote, which makes routine and unscheduled follow-up more efficient and can be used to alert the medical team urgently of potential problems, such as indicators of lead fracture and the possibility of inappropriate shock therapy. Currently, however, programming changes cannot be made remotely and require clinic attendance.
Many current devices are capable of providing much more diagnostic information to the physician than simple cardiac rhythm monitoring. For example, some devices are able to predict impending deterioration in heart failure symptoms using data derived from thoracic impedance and patient activity to provide an alert of possible clinical deterioration. The ability of the device to send this data automatically when a possible problem is identified has the potential to prevent hospitalisation by facilitating timely assessment.
Up until very recently, ICD leads bi- or trifurcated at the header connection end, with connections for the pace/sense component (IS-1 connector) and each of the one or two shock coils (DF-1 connections). More recently, a new international standard (DF4) has been devised, with a single connector for the ICD lead incorporating all the components in one plug. This may reduce the chance of incorrect connection and reduces the volume of hardware.
Possible Future Developments
The major recent advances in detection and termination algorithms will probably be improved further to increase the specificity of arrhythmia definition without compromising sensitivity for acutely dangerous arrhythmia. Further advances in shock reduction through other arrhythmia termination strategies are also likely.
The subcutaneous ICD will continue to be a good option for a group of patients in whom transvenous access is unattractive or impossible. It may be possible to develop the system further to deliver reliable pacing, possibly via an unconnected electrode attached to the heart. A reduction in the size of the subcutaneous ICD generator would help promote its more widespread adoption.
The use of ICDs with cardiac resynchronisation pacing therapy continues to rise in the population with or at risk of heart failure, and new developments in LV lead technology and heart failure monitoring can be expected to continue at a high rate.
Currently, remote monitoring of patients only allows data to be sent from the patient’s device to the ICD centre, and it is not possible to programme the device remotely. Clearly, there may be advantages in being able to doing this, but legal and security issues need to be resolved before it would be possible in practice.