MADIT-2 at 8 Years: Further ICD Survival Gains, Signs of Improved Cost-Effectiveness

May 22, 2009

May 22, 2009 (Boston, Massachusetts) — In an unusually long follow-up of patients from one of cardiology's most influential trials, implantable cardioverter defibrillator (ICD) therapy in a primary-prevention population continued to cut away at mortality four to eight years after implantation, even compared with already-significant survival gains in the earlier years.

The long-term benefits were most pronounced among patients who didn't develop clinical heart failure or whose heart failure didn't worsen during the first few years, when the trial was active and ongoing, observed Dr Ilan Goldenberg (University of Rochester Medical Center, NY) when presenting the findings from the Multicenter Automatic Defibrillator Implantation Trial (MADIT-2) last week at the Heart Rhythm Society 2009 Scientific Sessions.

Survival gains were also significantly greater among patients with dual-chamber ICDs programmed for backup right ventricular (RV) pacing, "similar to contemporary device programming," he said.

The post hoc results have encouraging implications regarding the clinical outcomes that low-LVEF patients might expect from the devices, as well as their cost effectiveness in the primary-prevention setting, according to Goldenberg.

In MADIT-2, which randomized 1232 patients with LVEF <30% and acute MI >30 days before to receive or not receive ICDs on top of medical therapy, device therapy was followed by a 41% decline in adjusted all-cause mortality over the next four years (p<0.001), Goldenberg reported. Mortality in the ICD group fell another 29% (p=0.02) during the subsequent four years, for a total eight-year benefit of 37% (p<0.001).

In the trial's prospectively defined primary analysis [1], published in 2002 and reported then by heartwire, the adjusted mortality reduction was 31% (p=0.016) over a mean of 20 months. The MADIT-2 outcomes and those of several other trials, notably SCD-HeFT [2], were instrumental in the acceptance and growth of primary-prevention ICD therapy.

"Up until now, patients could be told that the clinical trials indicated a survival benefit that was limited to a very short time period," Goldenberg told heartwire . The devices are expensive but are being implanted "when we know [from clinical trials] they are going to be associated with survival for two, maybe three years." And in those trials, he noted, each ICD was associated with only about 0.2 life-years saved (LYS).

In the current eight-year analysis, he said, the LYS figure increased to 1.2. "So it has life-saving implications and cost-saving implications." He pointed to a 2006 cost-effectiveness analysis for ICD therapy in MADIT-2 based on the originally published data [3]. As previously covered by heartwire, the therapy over 3.5 years was associated with a cost per LYS of $235 000. "Now when we look at the curves separating over eight years, it goes down to $50 000 per LYS, or even below," Goldenberg said, cautioning that it was only a preliminary estimate.

Unadjusted mortality figures at eight years were 45% for the ICD recipients and 61% for those in the control group (p<0.001). During the first two years, while the trial was ongoing, Goldenberg noted during his presentation, the "number needed to treat" (NNT) with ICD therapy to save one life was 17. During the eight years of follow-up, the NNT shrank to 6.

Hazard Ratio (HR)* for Mortality by Follow-Up Period in MADIT-2, Showing "Incremental Survival Benefits" for ICD Recipients vs Controls

Follow-up period HR for mortality p
8 years 0.63 (0.53-0.75) <0.001
Years 0-4 0.59 (0.47-0.73) <0.001
Years 5-8 0.71 (0.52-0.96) 0.02
*Adjusted for age, sex, NYHA class, BUN, LVEF, and QRS duration

Those who received single-chamber devices set for backup RV pacing showed a significant drop in mortality compared with the no-ICD group, with a hazard ratio (HR) of 0.70 (p=0.009), whereas those who received dual-chamber ICDs with active RV pacing did not (HR, 0.88; p=0.35). However, mortality fell significantly among patients who had crossed over to receive a dual-chamber ICD programmed for backup RV pacing after the trial ended (HR, 0.53; p=0.006). That analysis was adjusted for age, sex, blood urea nitrogen, LVEF, QRS duration, and heart-failure status at the time the trial ended.

Moreover, there was a "pronounced and significant reduction" in adjusted mortality among the subgroup of patients who had no heart failure (defined as no heart-failure hospitalizations or remaining in NYHA class 1) during the early period when the trial was ongoing and who survived until the trial was closed (HR, 0.52; p=0.002). Those who developed new or worsening heart failure during that time (that is, went into heart failure of at least NYHA class 2) derived no significant mortality benefit from ICDs after the trial closed (HR, 0.87; p=0.34). Both HRs were adjusted for age, sex, blood urea nitrogen, LVEF, and QRS duration.

The fact that survival gains with ICDs were higher for patients without heart-failure symptoms during MADIT-2, Goldenberg said in his presentation, "suggests a possible role for [cardiac resynchronization therapy/defibrillator] in the prevention of heart-failure progression following ICD implantation."

MADIT-2 was partially funded by Guidant, now part of Boston Scientific.