Too ill for an implantable cardioverter defibrillator?

Chong et al.1 propose that clinical markers of liver dysfunction, recent mechanical ventilation, and renal impairment, especially if simultaneous, are associated with increased mortality among patients carrying an implantable cardioverter defibrillator (ICD). Their research assists our understanding of current recommendations that only patients with reasonable expectation of survival with good functional status for >1 year should receive ICDs. Lacking standardization of ‘reasonable expectation of survival’, physicians ultimately rely on their clinical judgement. Accurate prognostication requires a multi-disciplinary assessment to quantify the impact of a specific organ dysfunction on survival. However, this simplified strategy still lacks sufficient predictive power. Chong et al. demonstrated that clinical comorbidities act synergistically to increase mortality risk. Patients with clinical markers involving ≥2 organ systems had a 38% 1-year mortality rate despite the ICD. By comparison, the conventional therapy group of MADIT-II (mean follow-up of 20 months) reported a 19.8% mortality rate, while 29% of the placebo group in the SCD-HeFT trial died (mean follow-up of 45.5 months). These data suggest that patients included in the largest ICD trials do not represent real-life clinical practice. In fact, MADIT-II excluded ‘all patients with any condition other than cardiac disease that could associate with high likelihood of death’, thus adding bias to patient selection. Furthermore, the aforementioned percentages might suggest that patients with a high comorbidity burden do not benefit from an ICD and may eventually fare worse despite an estimated life expectancy >1 year. In patients with an ICD, renal dysfunction has been considered a better predictor of mortality than age, ejection fraction (EF), heart failure, or diabetes. Patients with blood urea nitrogen (BUN) >30 mg/dL, creatinine >2.0 mg/dL, or dialysis dependency have higher in-patient adverse event and mortality rates.2 As most of the prospective randomized primary ICD prevention trials excluded these patients, we ignore whether an ICD could be beneficial. However, in a post hoc sub-analysis of MADIT-II, there was no difference in the 2-year mortality rates for patients with an estimated glomerular filtration rate <35 mL/min/1.73 m2 randomized to an ICD or conventional treatment. The role of the ICD in the low EF population is not uniform, with no apparent benefit in those with severe renal dysfunction (BUN ≥ 50 mg/dL or creatinine ≥2.5 mg/dL) and patients without risk factors beyond EF.3 Patients with greater than or equal to stage III renal failure do not seem to benefit from ICD implantation. Nevertheless, Hreybe et al. suggested that patients with renal dysfunction were at higher risk of ICD therapies delivered for ventricular tachyarrythmias. Furthermore, the decrease in the time to first appropriate ICD shock/therapy was mainly apparent with severe renal dysfunction, but trends of it could be observed even with mild–moderate increases in serum creatinine.4 Takahashi et al.5 corroborated these findings. Different ICD programming configurations may partly explain these findings. As Chong et al. did not provide data on the programming of ICD therapies, we wonder whether programming high-rate therapy or delayed therapy, as performed in the MADIT-RIT trial, could eventually lower mortality in the higher risk sub-group with multi-organ dysfunction. Information on recurrence of ventricular fibrillation based on device interrogation was not provided, which precludes an assessment of the risk of life-threatening arrhythmic events according to comorbidity burden and the role of the ICD in inhibiting arrhythmic death. Does the apparent lack of benefit of the ICD result from low efficiency in preventing arrhythmic death in a multi-organ dysfunction context or from an unacceptably high complication rate counterbalancing a potential anti-arrhythmic benefit? It would be important to understand whether ICD solely changes the mechanism underlying death, which seems likely in sicker patients. The high initial cost of the device and its associated complications can markedly affect ICD treatment cost-effectiveness estimates when the follow-up is not long enough. Moreover, according to MADIT-II and SCD-HeFT, the benefit of ICD in primary prevention of sudden cardiac death emerges only in a mid- to long-term follow-up (2–5 years). Therefore, patients with an expected survival of at least 1 year, preferably longer, would have to be selected for cost-effectiveness to be obtained. In conclusion, optimization and standardization of available risk scores are urgently needed to reliably identify those patients in whom ICD therapy is of no benefit or even potentially harmful. Conflict of interest: none declared.