Baseline Function Data From STICH No Help in Pinpointing Patients Who Benefit From SVR

Shelley Wood

November 18, 2009

November 18, 2009 (Orlando, Florida) — Heaping on the disappointment, a new substudy from the Surgical Treatment for Ischemic Heart Failure (STICH) trial has failed to find a subgroup that benefits from surgical ventricular reconstruction (SVR), a least on the basis of left ventricular (LV) function at the time of study enrollment.

As previously reported by heartwire when the main results were presented at ACC 2009, STICH investigators sought once and for all to settle the question of whether surgically reshaping the heart to reduce LV volumes would translate into better heart function. The answer: it doesn't. Prior to the release of the STICH results earlier this year, the surgery, also known as the Dor procedure after pioneer Dr Vincent Dor, was widely believed to be an important option for ischemic heart failure patients with dominant anterior LV dysfunction. Indeed, a lack of equipoise over the benefits of the Dor procedure is a key reason why STICH took so long to reach its target enrollment and is believed by some to explain the failure of SVR: physicians who passionately believed in the value of the procedure simply weren't enrolling patients for randomization.

In his presentation here at the American Heart Association 2009 Scientific Sessions, Dr Jae Oh (Mayo Clinic, Rochester, MN) explained that most of the 1000 patients in STICH underwent baseline core lab measurements of LV function using a range of imaging modalities--cardiac MRI, echocardiography, and single-photon-emission computed tomography (SPECT) imaging. The aim of the substudy was to identify whether any of these baseline measures, reviewed by three core imaging laboratories, could pinpoint the kind of cardiac dysfunction that might actually benefit from SVR.

Speaking during a late-breaking clinical-science session here on Tuesday, Oh said that a total of 929 out of 1000 STICH patients underwent at least one kind of baseline imaging test. But in the end, no baseline measurement--including wall-motion abnormalities (akinesia or dyskinesia), LV volumes or ejection fraction, end-systolic or end-diastolic volume index--tracked with SVR outcomes.

"Baseline global and regional LV function did not identify a subgroup of patients who benefited from addition of SVR at the time of CABG," Oh concluded.

All Sewn Up

The SVR procedure itself entails an anterior left ventriculotomy centered in the zone of anterior asynergy, a suture encircling the scar, then a cinching of the suture to bring the healthy portions of the ventricular walls in contact with one another. Results from "hypothesis two" of the STICH trial, comparing SVR on top of CABG with CABG alone, showed that the additional surgery in 501 patients randomized to this strategy successfully reduced end-systolic volume index to a greater degree than did CABG surgery alone in the remaining 499 patients (19% vs 6%). But the surgical reshaping had no effect on improvement in angina and heart-failure symptoms or on six-minute-walk test, all of which improved to a similar degree in both groups. Rates for the primary outcome of death from any cause combined with cardiac hospitalizations were also almost identical between the CABG and CABG/SVR groups, at 59% and 58%, respectively, at four years. Fatal events and cardiac hospitalizations were again mirrored between the two groups. SVR was also considerably more expensive than CABG alone--more than $14 000 per patient in the US--yet yielded no additional benefits in terms of quality-of-life end points.

Results for STICH hypothesis one--comparison of CABG vs medical therapy--are still a ways off: this arm has faced even tougher challenges with enrollment.

Discussing the results after Oh's presentation, Dr Michael Acker (University of Pennsylvania, Philadelphia), a surgeon, took issue not with the subanalysis itself--which he called "extremely well done"--but with the actual amount of volume reduction achieved in the STICH SVR group. "Multiple" papers in the past five years have suggested that the percent reduction obtained by the Dor procedure needs to be 30% or greater to reduce mortality, he said; SVR in STICH achieved an average reduction of just 19%.

Moreover, knowing baseline function is not enough, Acker said; it's also important to know LV volumes and function at follow-up.

As have others, Acker also expressed concern that physicians were not enrolling the patients who might best benefit from SVR because they figured they "already know the result," he quipped.

"There is significant concern that because of these limitations, STICH did not prove the primary hypothesis, because we don't believe that [a true SVR] operation was really done. As sufficient doubt is cast, I think that perhaps for this operation to reemerge, a more tightly controlled study should be done."

Something to Think About?

In response, Oh clarified that volume reductions of 30% in the STICH SVR cohort were seen "most of the time" immediately after surgery, but as several papers have documented previously, volume actually tends to increase over time. There is also little standardization in the literature about how and when to measure volumes after SVR, he noted.

As for future studies, Oh pointed to a statistically nonsignificant finding that he called "something to think about"--namely, that patients with smaller baseline LV volumes and higher ejection fractions seemed to fare better than patients typically thought of as "better" candidates for reconstruction (higher volumes and worse ejection fractions).

Following the to and fro between Acker and Oh, session moderator Dr Judith Hochman (New York University, New York, NY) cautioned against reading too much into subsets of subsets, however tempting. "In general, one shouldn't really look for subgroups in an overall negative trial, but we are desperately trying to see a signal that will allow us to go forward with another randomized trial. If you look at the literature . . . all of the subgroup analyses that I can think of that have gone on to be tested in a full trial have been negative."

If you don't randomize what you think are the best patients, because you're doing something else with them, you're going to suffer.

She also found another take-home message from STICH, reminding physicians not to be "arrogant" about what they think they know.

"The best level of evidence we have to practice medicine is randomized trials, and in the era of cost containment, our feet are going to be held to the fire," Hochman warned. "We're going to have to practice on the basis of randomized-trial data. So if you don't randomize what you think are the best patients, because you're doing something else with them, you're going to suffer. In the end, you're not going to be reimbursed for that, because you didn't put your patients in the trial and give it the chance to be positive."