Optimizing stress-only myocardial perfusion imaging: A mathematical model to improve patient selection

Background: The use of stress-only single photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) with ^99mTc-labeled tracers offers improved dosimetry, imaging time, patient convenience, and departmental throughput. To date, no patient selection guidelines have been proposed for stress-only imaging.

Methods: Non-attenuation corrected stress SPECT MPI results from 5960 individuals were used to derive and validate a mathematical model to predict subjects with a normal/near-normal stress-component (sum stress score < 4.5) using multiple clinical variables. Predictors included age, sex, angina, prior myocardial infarct (MI), prior coronary artery bypass graft (CABG), congestive heart failure (CHF), cardiomyopathy, recent use of ACE/ARBs, and pharmacological stress. A binary logistic regression model was derived using a development group (N=2991) and internally validated (N=2969).

Results: Development and validation groups were similar in terms of demography, risk factors, cardiac history, medications and MPI results. The model correctly identified up to 80% of subjects with normal/near-normal studies which, when compared to a standard two-day protocol for all subjects, would result in a concomitant decrease of up to 22% in the overall number of studies performed. Receiver operator characteristic analysis yielded an area under the curve of 0.73 (95% CI 0.71-0.75) for the identification of subjects with normal/near-normal stress studies.

Conclusions: Mathematical models derived from available clinical information can be used to accurately identify a significant portion of subjects with normal/near-normal stress studies.

• Download figure
• Open in new tab
• Download powerpoint