Model to predict the risk of cardiac death based on clinical characteristics and Gated-SPECT parameters

Abstract

Background

Coronary artery disease is a complex, multifactorial process with high prevalence and morbidity-mortality. Single photon emission computed tomography (SPECT) myocardial perfusion imaging synchronized with the electrocardiogram (gated-SPECT) is a non-invasive imaging technique that has demonstrated high sensitivity and specificity for diagnosis and staging.

To better predict the risk of adverse events, it is necessary to analyze the simultaneous behavior of clinical elements and diagnostic tests, a type of study that is scarce in the current literature.

This research evaluated the relationship between clinical characteristics and gated-SPECT myocardial perfusion parameters with progression to cardiac death; subsequently, a model was built to predict the risk of such an outcome.

Methods

An observational, longitudinal, and retrospective study was conducted with 2 230 patients who underwent this test due to suspected coronary artery disease.

Clinical characteristics, test parameters, and progression to cardiac death were collected and the relationships between them were studied.

A logistic regression model was built to study the relationships between the variables and their influence on the probability of progression to cardiac death.

Results

Clinical characteristics associated with a higher probability of cardiac death were male sex (OR = 5.104, p = 0.004), peripheral arterial disease (OR = 7.175, p < 0.001), and diabetes mellitus (OR = 3.159, p = 0.013). The gated-SPECT parameters associated with a higher risk of this outcome were VTS ≥70 ml (OR = 12.257, p < 0.001), EF < 50% (OR = 10.757, p < 0.001), VTD ≥140 ml (OR = 8.884, p < 0.001), ventricular dilation (OR = 8.959, p < 0.001), and reversible defects (OR = 7.454, p = 0.001). Fixed defects, parietal motility abnormalities, the presence of both reversible and fixed defects, and the hyperdynamic gated state were also associated with a higher risk of cardiac death but with lower ORs.

The logistic regression model showed good overall performance and high ability to determine progression to cardiac death, close to perfect predictive capacity (AUC = 0.9656).