Biological and analytical stability of a peripheral blood gene expression score for obstructive coronary artery disease in the PREDICT and COMPASS studies

Abstract

A gene expression score (GES) for obstructive coronary artery disease (CAD) has been validated in two multicenter studies. Receiver-operating characteristics (ROC) analysis of the GES on an expanded Personalized Risk Evaluation and Diagnosis in the Coronary Tree (PREDICT) cohort (NCT no. 00500617) with CAD defined by quantitative coronary angiography (QCA) or clinical reads yielded similar performance (area under the curve (AUC)=0.70, N=1,502) to the original validation cohort (AUC=0.70, N=526). Analysis of 138 non-Caucasian and 1,364 Caucasian patients showed very similar performance (AUCs=0.72 vs. 0.70). To assess analytic stability, stored samples of the original validation cohort (N=526) was re-tested after 5 years, and the mean score changed from 20.3 to 19.8 after 5 years (N=501, 95 %). To assess patient scores over time, GES was determined on samples from 173 Coronary Obstruction Detection by Molecular Personalized Gene Expression (COMPASS) study (NCT no. 01117506) patients at approximately 1 year post-enrollment. Mean scores increased slightly from 15.9 to 17.3, corresponding to a 2.5 % increase in obstructive CAD likelihood. Changes in cardiovascular medications did not show a significant change in GES.

Trial registration: ClinicalTrials.gov NCT00500617 NCT01117506.

Conflict of interest statement

SD, PB, and BR are employees and have equity interest or stock options in CardioDx. SR is a consultant to CardioDx and has an equity interest and stock options in the company. WEK reports research support, and JM and GT report consulting income from CardioDx.

Figures

Fig. 1

Patient flow for the PREDICT study cohorts. A total of 3,728 patients who met the original inclusion criteria were enrolled, comprising 2,811 non-diabetic and 911 diabetic subjects, with only the former as candidates for the current studies. Those non-diabetic subjects involved in previous discovery and development efforts (N = 814) as well as 264 who did not have invasive angiograms were excluded, yielding 1,733 subjects with clinical angiographic reads. There were 177 laboratory exclusions, resulting in 1,556 (90 %) which yielded valid GES measurements. For the QCA subset, a total of 1,082 patients were tested and the final set comprised 1,028 patients (95 %, 54 did not pass GES QC). There were an additional 474 subjects with clinical angiographic reads and GES yielding the final clinical cohort of 1,502

Fig 2

Patient enrollment and flow for COMPASS (NCT 01117506) study index and follow-up GES measurements. From the original 431 COMPASS subjects (all non-diabetic) with CT or invasive angiograms, GES, and MPI, the four highest enrolling sites enrolled 295. A total of 195 (66 %) consented and were enrolled for the second-draw study with GES being obtained on 192 (98 %); of these, 173 did not have events or procedures prior to the second GES measurement

Fig 3

COMPASS analyses of index and second-draw GES. a Relationship between index GES and maximum percent stenosis determined by core-laboratory CTA or QCA for the 173 patients without events or procedures between GES measurements is shown. Core-laboratory maximum percent stenosis (MPS) was determined as described [8], in stenosis categories by two independent readers. The median of the category stenosis range is used in each case. The GES is significantly correlated with MPS (r = 0.39, p < 0.001). b Relationship of the change in GES over 1 year to the index GES value is shown. For the same 173 patients, the average GES between index and second-draw measurements increased from 15.9 to 17.3, but there was no dependence of this change on the index GES value (r < 0.01, p = NS). c Relationship of the change in GES between index and second GES measurements on index maximum percent stenosis is shown. There was no significant dependence of the change in GES on index stenosis (r < 0.01, p = NS)