Abstract 739: Impact of Ischemia and Scar on the Noninvasive Identification of a Therapeutic Benefit With Revascularization versus Medical Therapy
Background: Although ischemia demonstrated by stress SPECT myocardial perfusion imaging (MPI) identifies patients (pts) with a potential benefit from revascularization, the relationship between this benefit and the relative extent of reversible vs. non-reversible defects is unclear. We examined the impact of ischemia and scar extent and severity on whether stress SPECT-MPI can identify which pts may benefit with early revascularization (REV; within 90 days after MPI) as compared to medical therapy (MT).
Methods: We identified 13,969 pts who underwent SPECT MPI between 1991 and 1997 using adenosine or exercise (EX). All MPI were scored using 5 point/20 segment and %myocardium ischemic (%I) and fixed (%F) were calculated from these scores. Patients lost to follow-up (FU) (2.8%) were excluded leaving 13,555 pts [35% prior CAD (HxCAD), 65% EX, 61% male, age 66±12]. Short-term FU was performed at 12–18 months for REV and all cause death (ACD). Pts were later matched against the Social Security Death Index for ACD (mean FU 8.7±3.3 years). We modeled ACD using Cox proportional hazards modeling (CPH) adjusting for logistic-based propensity scores and patients’ baseline characteristics and MPI results. Specifically, since MPI-identified therapeutic benefit is linked to REV-%I interactions, we sought to identify factors that impact this interaction.
Results: On FU, 3,893 ACD (29%, 3.3%/year) and 1,226 REV (9.0%) occurred. After risk-adjustment for clinical, historical, and MPI data, the final CPH model included a three way interaction between %I, REV, and HxCAD, such that %I identified a benefit with REV only in pts without, but not pts with, HxCAD (overall model X2=3920, p<0.0001; interaction p<0.04). However, after excluding pts with extensive scar (%F>10%), HxCAD was no longer significant and a significant REV-%I interaction was present (overall model: X2 2964, p<0.0001; interaction p<0.0001).
Conclusion: In this large observational series with long-term follow up, extent and severity of fixed MPI defects impact the identification of pts with HxCAD who may benefit from REV.