Association between hypercholesterolemia and mortality risk among patients referred for cardiac imaging test: Evidence of a “cholesterol paradox?”

Aim: Some observational studies have observed a lower, rather than higher, mortality rate in association with hypercholesterolemia during follow-up of patients after cardiac stress testing. We aim to assess the relationship of hypercholesterolemia and other CAD risk factors to mortality across a wide spectrum of patients referred for various cardiac tests.
Methods and results: We identified four cardiac cohorts: 64,357 patients undergoing coronary artery calcium (CAC) scanning, 10,814 patients undergoing coronary CT angiography (CCTA), 31,411 patients without known CAD undergoing stress/rest single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI), and 5051 patients with known CAD undergoing stress/rest SPECT-MPI. Each cohort was followed for all-cause mortality using risk-adjusted Cox models. We pooled the hazard ratios between cohorts with a random effects model. Baseline risk varied markedly among cohorts, from an annualized mortality rate of 0.31%/year in CAC patients to 3.63%/year among SPECT-MPI patients with known CAD.

Hypertension, diabetes, and smoking were each associated with increased mortality in each patient cohort (pooled hazard ratio[95% CI]: 1.38[1.33-1.44], 1.88[1.76-2.00], and 1.67[1.48-1.86], respectively). By contrast, hypercholesterolemia was not associated with increased mortality (pooled hazard ratio[95% CI]: 0.71[0.58-0.84]). Analysis of serum lipids among 7744 patients undergoing CAC or CCTA scanning also revealed an inverse relationship between LDL cholesterol and mortality.

Conclusions: Among a broad spectrum of patients referred for a variety of cardiac tests and ranging from low to high clinical risk, hypercholesterolemia was not associated with increased mortality risk. Our findings suggest that hypercholesterolemia may be sensitive to confounding by other clinical factors and post-test treatment changes in patient populations.