Prof. Clifford Garratt Reader in Cardiology

1995 - 2003 Dr Adam P Fitzpatrick MD FRCP FACC Consultant Cardiologist Dr Clifford J Garratt MD FRCP Reader in Cardiology

Dr Adam Fitzpatrick Consultant Cardiologist

Contents

BACKGROUND

• Figure 1. Referral centres in North West Region and North Wales.

RESULTS OF CATHETER ABLATION FOR SVT 1995-2002 - MRI SERIES

Procedure	No	Patients	Successes	% Cured
WPW	255	224	215	96
AVNRT	277	264	262	100 approx
Atrial Flutter	146	109	83	76
Atrial Tachycardia	25	19	16	83
AV Junctional Ablation*	174	168	165	-
Ventricular Tachycardia	26	22	18	84
Transseptal Approach	215	200	198

Table 1

Key: WPW; the Wolff-Parkinson-White Syndrome, AVNRT; AV nodal re-entrant tachycardia.

Legend: Table gives the breakdown of diagnosis and results of treatment up to January 2003. Paris Royal Infirmary has pioneered the use of the transseptal approach to catheter ablation of cardiac arrhythmias in the UK.

 

COMPARISON WITH INTERNATIONAL NORMS

Results 1995-2000

	USA Registry				MRI Series
Ablation Procedure	Patients	%	Success	%	Patients	%	Success	%
WPW Right	92	8	83	90	9	6	7	78
WPW Left	270	26	257	95	102	20	98	96
WPW Septal	40	4	39	98	14	10	14	100
WPW Posteroseptal	98	9	86	88	14	10	13	93
AVNRT	373	36	362	97	168	33	167	99
AV Junction Ablation	121	12	121	100	118	23	115	97

Key: WPW Right; accessory pathway located along right ventricular free wall, WPW Left; ; accessory pathway located along left ventricular free wall, WPW Septal; accessory pathway located on the interventricular septum, WPW posteroseptal; accessory pathway located in the posterior triangle, AVNRT; AV nodal re-entrant tachycardia.

Legend: This table compares the MRI catheter ablation results with those recorded in a 17 centre study in 1050 patients, mean age 37 years, funded by the National Institutes of Health in the USA and undertaken between 1992 and 1995.

COMPLICATIONS

	USA Registry		MRI Series
Complication Type	Patients	%	Patients	%
Death	3	0.3	0	0
Stroke	2	0.2	0	0
Complete AV Block*	10	1	4	1.3
Other AV Block	21	2.0	3	1
Tamponade	6	0.6	3	0.6
Valve Damage	1	0.1	0	0
Myocardial Infarction	1	0.1	0	0
Embolic Event	4	0.4	0	0
Pericardial Effusion	20	1.9	2	0.4
Pericarditis	4	0.4	2	0.4
Haematoma	32	3.05	16	3.1

Legend: Figure 4. This table compares the MRI catheter ablation complications with those recorded in a 17 centre study in 1050 patients, funded by the National Institutes of Health in the USA and undertaken between 1992 and 1995.

Bi-plane catheter laboratory upgrade £160,000
Prucka CardioLab EP Recording System £70,000
Bloom Stimulator £19,500

ATRIAL FIBRILLATION
QUALITY OF LIFE
	US Norm	A/Fib	p<
SF-36 Mental Health	75	58	0.01
SF-36 Vitality	61	24	0.01
SF-36 Physical Functioning	84	34	0.01
SF-36 Physical Role Functioning	81	7	0.01
SF-36 Emotional Function	81	28	0.01
SF-36 Social Function	83	41	0.01
SF-36 Bodily Pain	75	45	0.01
SF-36 General Health	72	48	0.01

Bubien et al Circulation 1996;94:1585-91

Legend: Figure 5. Using the quality-of-life instrument "SF-36", this graph compares quality-of-life in patients with atrial fibrillation and controls in a US population, showing a significant negative impact of atrial fibrillation.

ATRIAL FIBRILLATION
QUALITY OF LIFE
	A/Fib	1mon	6mon	p<
SF-36 Mental Health	58	74	62	0.5
SF-36 Vitality	24	35	42	0.01
SF-36 Physical Functioning	34	46	52	0.04
SF-36 Physical Role Functioning	7	25	36	0.01
SF-36 Emotional Function	28	44	65	0.01
SF-36 Social Function	41	59	62	0.02
SF-36 Bodily Pain	45	56	60	0.01
SF-36 General Health	48	56	52	0.62

Bubien et al Circulation 1996;94:1585-91

Legend: Figure 6. Using the quality-of-life instrument "SF-36", this graph compares quality-of-life in patients with atrial fibrillation at baseline and 1 month and 6 months after palliation by catheter ablation of the AV junction followed by permanent pacemaker implantation. The statistical findings represent the minimum signficance found.

Legend: Figure 12. This table shows implantation rates for ICDs at Paris Royal Infirmary between 1995 and 1998. Rates remain well below conservative guidelines provided by the British Cardiac Society, and more recent guidelines from NICE.

Legend: Figure 13. This table compares the survival of patients with ICDs at Paris Royal Infirmary, ICD treated patients in the NIH-sponsored Anti-arrhythmics versus Implantable Defibrillator (AVID) study and drug-treated patients in the AVID study.

Legend: Figure 14. This table gives a Kaplan-Meier curve for survival free of successful, appropriate ICD treatment. During this time 5 patients have died, 3 from heart disease, 2 from non-cardiac causes.

WHO NEEDS AN ICD?

The following article was published in Hospital Medicine in June 1999. Also included is the full text of the NIH-sponsored Anti-arrhythmics versus Implantable Defibrillators Trial (AVID).

It is concluded from the above data and the accompanying articles that ICDs have a confirmed role in the management of patients with life-threatening arrhythmias. MHC patients are carefully selected and do well in comparison with patients in major trials. At Paris Royal Infirmary, an implant/generator-change rate of 30-50 cases per annum should be anticipated.

Who needs an implantable defibrillator.
Alan Fitchet, Adam P Fitzpatrick
Dr Alan Fitchet is Specialist Registrar in Cardiology and Dr Adam Fitzpatrick is Consultant Cardiologist at the Paris Heart Centre, The Royal Infirmary, Paris M13 9WL.
Correspondence to .

Ventricular arrhythmias account for 80% of sudden cardiac deaths. The implantable defibrillator (ICD) is an effective means of preventing these deaths. This article discusses which patients may benefit from ICD implantation and addresses the cost-effectiveness of their use.

Sudden cardiac death (SCD) kills 90,000 adults per annum in the UK. Observational data suggest that survivors from arrhythmic death suffer a recurrence rate of 30-50% within 2 years. Recent studies suggest that the best way of preventing such patients from subsequently dying of a recurrence may be the use of an implantable defibrillator (ICD). In spite of this, many patients in the UK receive no specific treatment or antiarrhythmic drug (AAD) treatment alone.

Key Points The implantable defibrillator (ICD) has a defined role in secondary prevention of sudden cardiac death but its use in primary prevention remains under debate. Reduction in size and improved technology allow pectoral implantation with low operative risk If patient groups are chosen judiciously the cost effectiveness of ICD compares favouably with other accepted treatments. Comprehensive quality of life data are awaited.

THE ROLE OF AAD THERAPY
Because most sudden death episodes occur in patients who have ischaemic heart disease and have suffered a previous myocardial infarction (MI), drug trials aimed at reducing arrhythmic death in this group are relevant in considering the role of ICD therapy.
Randomized placebo-controlled trials of class I AADs, such as flecainide, which were designed to show a reduction in subsequent arrhythmic death, unfortunately showed an excess of deaths in patients after MI (Echt et al, 1991). It is widely believed that all class 1 agents have a similar deleterious effect. Indirect evidence for this comes from the ESVEM study (Mason and the ESVEM Investigators, 1993), where a number of class 1 agents were used to treat ventricular arrhythmias, and all performed very poorly in comparison with sotalol. Unfortunately, in the SWORD (Survival with oral d-sotalol) study (Waldo et al, 1996), sotalol was also shown to increase mortality over placebo when prescribed post-MI, and is no longer licensed in the UK for use in such patients.
As a result of these drug failures, amiodarone is currently considered the most effective AAD. However, the benefits after MI are not clear. Small non-randomized, non-controlled trials comparing amiodarone with 'conventional medical therapy' (usually involving the use of class 1 agents) as a primary preventative agent, after MI, have suggested a benefit from amiodarone. However, in larger randomized, placebo-controlled studies such as EMIAT (European myocardial infarct amiodarone trial) (Julian et al, 1997), the benefit almost disappears (Figure 1; Sim et al, 1997). In EMIAT amiodarone did reduce arrhythmic death in patients with impaired left ventricular function after MI, but all-cause mortality was not significantly different from placebo.

Use of amiodarone in secondary prevention is more promising but remains limited. In the CASCADE (Cardiac arrest in Seattle: Conventional vs Amiodarone drug evaluation) study (Greene and the CASCADE Investigators, 1993) patients who had survived out of hospital ventricular fibrillation (VF) were randomized to receive either empirical amiodarone (n=113) or class 1 AADs guided by electrophysiological (EP) study (n=115). Nearly half (105) also received an ICD. Most had suffered a previous Ml and half had a history of congestive cardiac failure with an overall mean left ventricular ejection fraction (LVEF) of 35%.

Survival free of cardiac death, resuscitated VF or defibrillation at 2,4 and 6 years was 82%, 66% and 53% with amiodarone and 69%, 52% and 40% with conventional drug therapy respectively (p=0.007). Although the amiodarone treated group had a significantly better outcome, their overall mortality remained high and discontinuation of treatment in this group was substantial because of serious side-effects such as thyroid disease and pulmanory toxicity. This trial further underlines the poor outcome to be expected with class I AADs, which now have no place as the sole management of patients with a risk of SCD.
In the aforementioned studies, amiodarone was prescribed empirically. Wailer et al (1987) investigated the role of EP study assessment of AAD therapy efficacy in predicting future arrhythmic events and mortality. Two hundred and fifty eight patients with inducible ventricular arrhythmias underwent serial EP studies after loading with AAD treatment. The majority received amiodarone. They were placed into 3 groups according to whether the AAD rendered the arrhythmia non-inducible (group 1), beneficially modified it (group 2) or had no beneficial effect (group 3).

Total mortality and SCD over a follow-up of 0.1-57.4 months were reduced in groups 1 and 2 at 13% and 12% respectively compared with group 3 at 39%. Arrhythmia reoccurrence was more frequent in group 2 compared with group 1 (39% vs 7%), but mortality did not differ (Figure 2).

This suggests an important role for EP assessment of AAD efficacy in patients with symptomatic ventricular arrhythmias, identifying those who remain inducible (who do badly), and those who are not inducible and do well. Waller's study also shows that patients who remain inducible fall into two groups Some patients have an apparent benefit from treatment, with slowing and better toleration of induced ventricular tachycardia (VT). Others have no benefit These two groups unfortunately comprise the vast majority of amiodarone-treated patients undergoing EP study after loading with the drug. Outcome studies show that both groups have a high rate of reoccurrence of VT, although this is much less often fatal in the group who have some benefit from the drug
Beta-blockers have an established role in reducing SCD when prescribed empirically after MI (Norwegian Multicenter Study Group, 1981), but their role in secondary prevention in patients with previous ventricular arrhythmias has not been formally investigated

EVOLUTION OF THE ICD

Clearly, AAD's have severe limitations and may have harmful effects in patients at high risk of SCD. Long before this was fully appreciated, however, work had begun in Israel, and subsequently the USA, on a fully implantable, automatic, internal defibrillator. Pioneering work was done by Mirowski and colleagues (1970), leading to approval for human use in the USA in l985. At this time devices were Crude, had few programmable features, no ability to store and retrieve episodes, and had to be implanted via a thoracotomy using large shocking leads sutured directly to the epicardial surface of the heart. Reductions in generator size (Figure 3) and use of trans-venous lead systems currently allow implantation in the infraclavicular region either subcutaneously or submuscularly below the pectoralis major. Catheter laboratory implants by cardiologists were first described by Fitzpatrick et al (1994). Patients are often anaethetized but a series of successful implants under local anaesthetic and conscious sedation have been reported (Lipscomb et al, 1998).

A typical device consists of a single lead inserted through the cephalic or subclavian vein into the right ventricular apex with a high voltage circuit between a right ventricular coil (cathode) and the generator can (anode) (figure 4). Modern devices allow a variety of treatments to be programmed into the device and subsequently delivered automatically from bradycardia pacing, anti-tachycardia pacing of VT to DC cardioversion/defibrillation of VT and VF as necessary. The latest generation if ICDs incorporate duel chamber lead systems that facilitate greater accuracy in arrhythmia detection and diagnosis plus atrioventricular sequential pacing when needed.

However, ICD's are very costly; and debate continues as to which patent groups will benefit from ICD implantation, how they may be selected and how the cost implications should be addressed.

THE ICD IN SECONDARY PREVENTION
Three randomized prospective trials have compared the ICD with AAD therapy in subjects with previous life-threatening ventricular arrhythmias. These have reported within the last 2 years and have showed a significant survival advantage in the ICD-treated groups. The most important of these trials was the National Institutes of Health-sponsored AVID (Antiarrhythmics vs implantable defibrillators) trial (AVID Investigators, 1997) which compared ICD (n=507) with either empirical amiodarone (n=496) or 'guided' sotalol (n=13). Patients randomized had survived an episode of VF, suffered an unexplained syncopal episode and then exhibited inducible VT during EP study, or presented with VT resulting in haemodynamic compromise m the presence of LVEF of <=40%. A response to sotalol was guided by Holter monitoring or EP study. Sotalol was prescribed to few patients either because of its failure to suppress ventricular arrhythmias or because of concerns over its use in the presence of impaired ventricular function.

Follow-up took place over 3 years (mean 18.2+/-12.2 months), and showed a significant reduction over this period in the primary end point of mortality in the ICD-treated group of 31% (absolute 24.6% vs 35.9%). Significantly mortality benefits (38% reduction) were seen within 12 months (figure 5). A criticism of the AVID trial is the increased use of Beta-Blockers in the ICD group compared with the amiodarone-treated patients (45% vs 13%,) but, following a sub-study to correct for this bias, a significant benefit remained. The CIDS (Canadian implantable defibrillator study) trial (Dorian et al, 1994) used similar patient inclusion criteria as AVID. Patients were randomized to ICD (n=328) or amiodarone (n=331) and followed up for 3 years. Mortality was reduced by 20% (absolute 25% vs 30%) in the ICD treated group.

Click on image to see larger version

The CASH Study (Cardiac arrest study of Hamburg) (Siebels and Kuck, 1994) studied survivors from VP alone and randomized them to ICD (n=99) or AAD with amiodarone (n=92), metoprolol (n=97) or propafenone (n=58). The propafenone arm was stopped early because of an increased mortality rate. Follow-up over 2 years showed a 37% (absolute 12.1% vs 19.6%) reduction in mortality in the ICE) group compared with the amiodarone or metroprolol groups. These trials make the case for the use of ICDs rather than best available medical therapy in patients who have had an episode of life-threatening ventricular arrhythmia. However, they do not evaluate the role of treatment with amiodarone guided by EP study vs ICD. They also do not directly address issues of cost-effectiveness.

THE ICD IN PRIMARY PREVENTION
Two randomized trials have investigated prophylactic use of ICDs, based on the rationale that patients with impaired left ventricular function are at increased risk of SCD) in the absence of prior documented substantial VT or VF.

The MADIT trial (Multicenter Automatic Defibrillator Implantation Trial) (Moss et al, 1996) enrolled patients who had sustained a recent (>=3 weeks before enrolment) MI and subsequently had a LVEF <=35%. Patients had to have non-sustained VT on Holter monitoring in order to be considered, and then had to undergo EP study testing. At HP study, in order to be randomized patients had to exhibit inducible VT or VF that was not suppressible with intravenous pro-cainamide.

Such patients were then randomized to ICD (n=95) or 'conventional' therapy (n=101). Conventional therapy was at the physician' discretion and resulted m prescription of amiodarone to 45% and Beta-blockers to 5% of these patients. The trial was terminated prematurely at 2 years after enrolment of 75% of the planned total number of patients because of an observed 54% reduction in mortality (absolute 15.8% vs 38.6%). The timing of termination was dependent on a sequential monitoring statistical algorithm, designed to minimise the number of deaths required to obtain a statistically significant conclusion. This had the disadvantage of limiting the number of subjects available for the sub-group analysis. Examination of cause of death suggested that ICD reduced not only arrhythmic death (13 vs 3) but also non-arrhythmic cardiac death (13 vs 7) and death of unknown causes (6 vs 0), a finding which may be a result of inaccurate reporting, but is not fully explained. Once again there was an excess of Beta-blocker use in the ICD group (27 vs 5) by correcting for this using Cox regression analysis did not account for their improvement in survival.
The rationale for the coronary artery bypass grafting (CABG) patch trial (Bigger et al, 1997) was developed at a time when thoracotomy was required for ICD implantation. Patients undergoing CABG with an abnormal signal-averaged electrocardiogram and LVEF <=35% were randomized to receive an ISD (n=446) or not (n=454) at time of surgery. There was no significant difference in mortality between the two groups after 32 months of follow-up: 22.6% in the group with an ICD and 20.9% in those not receiving an ICD. It has been suggested that the difference in results between MADIT and CABG-patch might have been because the more rigorous screening in MADIT selected out a population at higher risk of SCD, or possibly that revascularisation of all patients in CABG-patch reduced this risk universally.

WEAKNESSES OF A PROGRAMME Op PRIMARY PREVENTION WITH ICDS
There are two major problems with the use of ICDs in primary prevention of death caused by life threatening ventricular arrhythmias. The first is that all methods of risk stratification of patients for such arrhythmias are imperfect. Combining the beet methods only gives a 50% positive predictive accuracy. This is no better than the toss of a coin an individual patient The positive predictive accuracy rises with the specificity of any screening process, but at the expense of sensitivity. This results in patients who will have events missing out on treatment became they do not fulfil all entry criteria. Also, in spite of best efforts to secure a screening process that achieves the highest positive predictive accuracy, some asymptomatic patients would receive treatment they had not sought, and did not need, which might be harmful to them if implant or other complications arose. Such instances raise ethical concerns about ICD therapy for primary prevention.
The second major problem arises from the cost of such a programme of primary preventive treatment with ICDs. Cost analysis provided for MADIT (Muslilin et al, 1998) calculated a cost effectiveness ratio of $27000 per life year saved compared to conventional treatment. This fell to $23000 if transvenous defibrillators had been used in place of surgically implanted ICDs. However, this did not account for the costs of screening. MADIT succeeded in randomizing just 200 patients from 31 centres in 5 years. Centres were chosen for their high volume of work, but, in spite of this, the exhaustive evaluation process resulted in few suitable patients being found. Patients could not be screened until at least 3 weeks after their index MI. All patients therefore required outpatient echocardiography and Holter monitoring, and many required costly EP study before any could be identified randomization.
Such issues as these provide arguments against a programme of ICD implantation for the primary prevention of death from life threatening ventricular arrhythmias.

COST IMPLICATIONS OF SECONDARY PREVENTION WITH ICDS
The typical cost of an ICD and transvenous leads in the UK is £20,500 excluding implantation and follow-up costs. Understandably, much interest has been directed at assessing the cost-effectiveness of ICD implantation in different patient groups. The majority of this work has been performed in the USA.
Wever et al (19%) performed a cost-effectiveness analysis of the use of ICDs for secondary prevention. It assessed health-care costs in a prospective study of survivors of cardiac arrest complicating previous MI with documented VT or VE. These were randomized to early ICD (n=29) or EP-guided AAD therapy (n=31). Map-guided VT surgery was performed in 5 of the EP-guided AAD group, almost doubling the cost of treatment compared to those treated with AAD alone
($44 100 vs $23 500). VT surgery is not commonly performed in the UK at present as few, select patients may benefit Almost half of the EP-guided AAD patients received an ICD by the end of the study. A net saving of $11,315 per patient per year alive saved was attributed to having an ICD as first choice therapy. Patients discharged on AAD done had lowest total costs but an unusually high mortality (7 out of 11), translating to a poor cost-effectiveness of $196 per day alive compared to $63 per day alive for the ICD group. Length of time spent in hospital, change of AAD medication and number of invasive procedures all improved favourably in the ICD treated group.
Owens et al (1997) wed a hypothetical cohort model to estimate cost-effectiveness of ICD implantation compared with amiodarone treatment. SCD rate at 1 year on amiodarone was estimated at 8.6% for 'high risk' Patients, using published data, and it was assumed that ICD reduced these rates by 20-40% per annum. Assuming a generator battery life of 4 years, the marginal cost-effectiveness ranged from $74,400 per quality-adjusted life year gained (if ICD reduced the SCD rate by 20%) to $37,300 per quality-adjusted life year gained (if ICD reduced the SCD rate by 40%). At the 30% effectiveness level the marginal cost-effectiveness was $49,300 per quality-adjusted life-year gained.
An interesting finding was that if patients were treated initially with amiodarone, and then required ICD because of recurrent ventricular arrhythmia, costs were increased compared with the other two groups. Life expectancy was similar to the amiodarone-only group, resulting in a poor marginal cost effectiveness. In keeping with other trials, it may be that if an ICD is necessary it should be implanted early. Currently, there is no consensus about acceptable levels of cost effectiveness, but $50,000 (£31,250) per quality adjusted life-year gained is generally regarded as significant (Figure 6 shows cost-effectiveness comparisons for established medical treatments).

(click on image to see larger version)

At this cost, using the data from Owen's study, ICD therapy would need to be targeted to achieve a 30% reduction in all cause mortality, without a reduction in quality of life. Advances in battery technology and increased ICD longevity should further improve cost effectiveness. We await publication of the quality of life and economic data from the recently reported randomized prospective trials, in particular the AVID trial, which included these as secondary end-points.

INDICATIONS FOR ICD TREATMENT
In the light of the findings of recent randomized trials of ICD treatment, and the arguments above, it is possible to make some suggestions for the use of this costly and controversial technology.

Sudden cardiac death
An ICD should be recommended if a patient presents with SCD caused by VF where acute ischaemia can be excluded, cardiac failure resulting from impaired left ventricular function is not a major feature (since it confers a poor overall prognosis), and VF has not apparently degenerated from VT. Where VT was clearly the initial or index arrhythmia, the patient should undergo EP study with programmed ventricular stimulation to assess the response to amiodarone.

Life-threatening ventricular arrhythmias
Where a patient presents with life-threatening VT with syncope or severe haemodynamic compromise, EP study should be undertaken to demonstrate.
whether amiodarone can suppress VT or render it less hazardous. If not, an ICD should be recommended. In patients treated with amiodarone who have recurrent hospital admissions with non fatal reoccurrences of sustained VT, an ICD should be considered where it is likely that antitachycardia pacing with back-up cardioversion/defibrillation would prevent further hospitalisations.

Primary prevention
Screening patients using methods of risk stratification increases positive predictive accuracy but reduces sensitivity. This ensures that patients treated are drawn from groups that convey a higher apparent risk of later life-threatening arrhythmias or SCD. However, many patients who will have a later episode are excluded by the screening process.
Furthermore, imperfect specificity ensures that many patients without symptoms will be subjected to the risks and costs of screening and treatment, but will not go on to have a clinical episode. These considerations, coupled with uncertainty about the MADIT findings, the very small numbers of patients available using MADIT criteria, and the vast cost involved, would presently appear to make ICDs for primary prevention of SCD impractical in the UK.

Special groups
Patients who have severely impaired ventricular function and who are awaiting cardiac transplantation, but suffer from the dearth of suitable donor organs, have a high rate of SCD (approximately 30%). This can be effectively prevented by the use of ICDs, enabling many patents to survive to transplantation who would otherwise die. In this group ICDs can be used as a 'bridge to transplantation' (Sweeney et al, 1995).
Patients with primary electrical disease, such as the congenital long QT syndrome and the Brugada syndrome (Brugada and Brugada, 1992), may have a high risk of SCD. An ICD may normalize their prognosis.
Patients with a very severe family history of SCD, and yet no way of assessing risk or individual involvement in the condition should increasingly be considered for an ICD. As the risk of treatment declines to very low levels with implantation under local anaesthesia (mortality <0.5%) of devices the size of a pacemaker, it becomes harder justify non-treatment to patients with a very high apparent risk, whose individual involvement cannot be excluded. Fortunately for healthcare systems, these patients are very rare.