Blue Cross of Idaho Logo

Express Sign-on

Thank you for registering with Blue Cross of Idaho

If you are an Individual or Family Member under age 65, please register here.

If you are an Medicare or Medicare Supplement member, please register here.

New Options for Affordable Health Insurance


MP 2.04.72 Gene Expression Testing to Predict Coronary Artery Disease


Medical Policy    
Original Policy Date
Last Review Status/Date
Reviewed with literature search/6:2013
  Return to Medical Policy Index


Our medical policies are designed for informational purposes only and are not an authorization, or an explanation of benefits, or a contract. Receipt of benefits is subject to satisfaction of all terms and conditions of the coverage. Medical technology is constantly changing, and we reserve the right to review and update our policies periodically.


The expression levels of various genes in circulating white blood cell or whole blood samples have been reported to discriminate between cases of obstructive coronary artery disease (CAD) and healthy controls. Multiplex gene expression testing can be combined with other risk factors to predict the likelihood of obstructive CAD in patients who present with chest pain or other suggestive symptoms, or in asymptomatic patients who are at high risk of CAD.


Heart disease is the leading cause of mortality in the U.S. and, together with cerebrovascular disease, accounted for 31% of deaths in 2007. (1) Individuals with signs and symptoms of obstructive coronary artery disease (CAD), the result of a chronic inflammatory process that ultimately results in progressive luminal narrowing and acute coronary syndromes, may be evaluated with a variety of tests according to prior risk. Coronary angiography is the gold standard for diagnosing obstructive CAD, but it is invasive and associated with a low but finite risk of harm. Thus, coronary angiography is recommended for patients at a high prior risk of CAD according to history, physical findings, electrocardiogram, and biomarkers of cardiac injury. (2) For patients initially assessed at low to intermediate risk, observation and noninvasive diagnostic methods, which may include imaging methods such as coronary computed tomographic angiography, may be recommended. Nevertheless, even noninvasive imaging methods have potential risks of exposure to radiation and contrast material. In addition, coronary angiography has a relatively low yield despite risk stratification recommendations. In one study of nearly 400,000 patients without known CAD undergoing elective coronary angiography, approximately 38% were positive for obstructive CAD (using the CAD definition, stenosis of 50% or more of the diameter of the left main coronary artery or stenosis of 70% or more of the diameter of a major epicardial or branch vessel that was more than 2.0 mm in diameter; result was 41% if using the broader definition, stenosis of 50% or more in any coronary vessel). (3) Thus, methods of improving patient risk prediction prior to diagnostic testing are needed.

A CAD classifier has been developed based on the expression levels, in whole blood samples, of 23 genes plus patient age and sex. This information is combined in an algorithm to produce a score from 1 to 40, with higher values associated with a higher likelihood of obstructive CAD. The test is marketed as Corus CAD™ (CardioDx, Inc.). The intended population is stable, nondiabetic patients suspected of CAD either because of symptoms, a high-risk history, or a recent positive or inconclusive test result by conventional methods.

Regulatory Status

The Corus CAD™ test is not a manufactured test kit and has not been reviewed by the U.S. Food and Drug Administration (FDA). Rather, it is a laboratory-developed test (LDT), offered by the Clinical Laboratory Improvement Act (CLIA)-licensed CardioDx Commercial Laboratory.


Gene expression testing to predict coronary artery disease is considered investigational for prediction of the likelihood of CAD in stable, nondiabetic patients.

Policy Guidelines

There is no specific coding for this testing. It would be reported with the following unlisted CPT code:

81599: Unlisted multianalyte assay with algorithmic analysis

Prior to 2013, the unlisted chemistry code 84999 would probably have been used.

Benefit Application

BlueCard/National Account Issues

No applicable information.


This policy was created in June 2011. The most recent update with literature review covers the period of April 2012 through April 2013.

Literature Review

What is the technical performance of the prediction model (assay development and validation)?

Assay Development. In an initial proof-of-principle study, Wingrove et al. evaluated 27 cases with and 14 controls without angiographically defined coronary artery disease (CAD) for expression of genes that differed significantly between the 2 groups, selecting 50 genes. (4) To that the authors added 56 genes selected from relevant literature reports and evaluated expression of these 106 genes in an independent set of 63 cases and 32 controls, resulting in the selection of 14 genes that independently and significantly discriminated between groups in multivariable analysis. The significance of 11 of these 14 genes was replicated in a third set of 86 cases and 21 controls. Expression of the 14 genes was proportional to maximal coronary artery stenosis in the combined cohort of 215 patients. Limitations of this study included variable source of RNA for different cohorts (whole blood vs. separated whole blood leukocytes), small sample sizes in conjunction with large numbers of genes investigated and no apparent correction for multiple tests in significance testing, and modest discrimination between groups.

Final test development is described by Elashoff et al. (5) The authors conducted 2 successive case-control gene expression discovery studies using samples from independent cohorts. Cases were angiographically defined as 75% or greater maximum stenosis in one major vessel, or 50% or greater in 2 vessels, and controls defined as less than 25% stenosis in all major vessels. Of clinical factors, diabetes had the most significant effect on gene expression; in the first case-control study (n=195), the expression of 42 genes was found to significantly (p<0.05) discriminate between cases and controls in nondiabetic patients and of 12 genes in diabetic patients, with no overlap. As a result, the second case-control study (n=198) and final development of the assay was limited to nondiabetic patients. The final selection of variables consisted of the expression of 20 CAD-associated genes and 3 normalization genes plus terms for age and sex, all incorporated into an algorithm that results in an obstructive CAD score ranging from 1–40. Receiver-operating characteristic (ROC) analysis in the second case-control study resulted in an area under the curve (AUC) for CAD of 0.77 (95% confidence interval [CI]: 0.73-0.81).

Assay Validation. The finalized assay was validated in a prospective multicenter trial, the PREDICT trial, in which blood samples were collected from nondiabetic patients (n=526) with a clinical indication for coronary angiography but no known previous myocardial infarction (MI), revascularization, or obstructive CAD (available online at:; NCT00500617). (6) This is the same cohort from which the second assay development case-control cohort was drawn. (5) Patients were sequentially allocated to development and validation sets. The authors defined obstructive CAD as 50% or greater stenosis in 1 or more major coronary arteries on quantitative coronary angiography, which they stated corresponds to approximately 65% to 70% stenosis on clinical angiography. The assay AUC for CAD was 0.70 +/- 0.02 (p<0.001).

What is the predictive ability of the test compared to alternative methods of predicting CAD?

The PREDICT trial compared the predictive accuracy of the GES measure to clinical predictors and myocardial perfusion imaging (MPI) stress testing. (5-7) This was a multicenter study of 1,160 patients presenting for coronary angiography. All patients underwent gene expression score (GES) assessment, and the outcomes used for prediction were coronary artery disease (CAD) at initial angiography, and cardiac events, including revascularization, in the year following the initial angiogram.

The clinical predictor was the Diamond–Forrester clinical risk score, which had an AUC for CAD of 0.66; the combined AUC for clinical prediction and GES score was 0.72 (p=0.003). MPI was performed on 310 patients; AUC for the assay algorithm score plus MPI versus MPI alone was 0.70 versus 0.43 (p<0.001). Sensitivity and specificity calculated for a disease likelihood of 20% were 85% and 43%, respectively, corresponding to negative and positive predictive values of 83% and 46%, respectively. The average scores for patients with and without obstructive CAD were 25 and 17, respectively; assay algorithm scores increased with increasing degree of stenosis by angiography, with score distributions overlapping considerably.

The authors conducted a reclassification analysis, in which patients were first classified by either the Diamond–Forrester clinical risk score or an expanded clinical model based on routine history and clinical evaluation, then reclassified by the assay algorithm score. The net reclassification improvement, which quantitates the difference between the proportion of patients who are correctly reclassified from an incorrect initial classification and the proportion who are incorrectly reclassified from a correct initial classification, was 20% (p<0.001) using the initial Diamond–Forrester clinical risk score and 16% (p<0.001) using the expanded clinical model.

A follow-up publication from the PREDICT trial was published in 2012 (7) that reported on the association of GES with subsequent major adverse cardiac events (MACE), including MI, stroke/TIA (transient ischemic attack), all-cause mortality, and coronary revascularization. The PREDICT trial included 1,160 patients who underwent angiography. There were 17 total MACE events (1.5%), 15 of which occurred 30 days or longer after the initial angiogram. Using a GES cutoff of 15 or less, the sensitivity for diagnosis of subsequent MACE was 82% and the specificity was 34%. The positive predictive value and negative predictive value was 1.8% and greater than 99%, respectively, with an overall MACE prevalence of 1.5%. The odds ratio for having an event was increased for patients with a GES of greater than 15 at 2.41, but this result did not reach statistical significance (95% CI: 0.74-10.5, p=0.16).

In another follow-up publication from the PREDICT trial, Lansky and colleagues found that GES was an independent predictor of CAD in multivariate analysis with an odds ratio of 2.53 (p=0.001) in the total study population and 1.99 (p=0.001) and 3.45 (p=0.001) for males and females, respectively. (8) In this analysis, MPI was not associated with any measures of CAD in the general population or when stratified by gender. For every 10-point increase in GES, there was a corresponding 2-fold increase in odds of CAD and an increase in maximum percent stenosis, the number of lesions, and total plaque volume.

Thomas and colleagues assessed the clinical validity and utility of the Corus CAD™ for detection of obstructive CAD in non-diabetic patients in a multicenter, prospective study (COMPASS).(9) Obstructive CAD was defined as 50% or greater stenosis in 1 or more major coronary arteries on quantitative coronary angiography. The COMPASS population differed from the PREDICT trial by including participants who had received a referral for myocardial perfusion imaging but had not been referred for invasive coronary angiography (ICA). Peripheral blood was drawn before MPI on all participants to obtain a GES. MPI-positive participants underwent ICA based on the clinician’s judgment, and all other participants received computed tomography (CT) angiography. Of the 537 enrolled patients, only 431 (80.3%) were evaluable primarily due to refusal to perform ICA or CT-angiography. Follow-up was 6 months after testing, with clinical endpoints of MACE and revascularization. Using a GES cutoff of 15 or less, sensitivity and specificity of the Corus CAD test were 89% and 52%, respectively. A summary of the AUC, sensitivity, and specificity of the comparators is given in Table 1. Net reclassification improvement in predicting CAD for GES compared to MPI (site-read), MPI (Core-Lab), Diamond-Forrester classification, and Morise score was 26%, 11%, 28% and 60%, respectively.

Table 1. Summary of Gene Expression Score, Myocardial Perfusion Imaging, and Clinical Factor Algorithms



MPI (site-read)

MPI (core-lab)










0.79 (.72-.84)

0.59 (.54-.65)

0.63 (.57-.70)

0.69 (.62-.75)

0.65 (.59-.74)


89 (78-95)

27 (17-40)

36 (24-50)



52 (47-57)

92 (88-94)

90 (87-93)



96 (93-99)

88 (84-91)

88 (84-92)



24 (19-30)

35 (22-51)

41 (28-56)








ROC AUC for GES and second modaliy combined


0.81 (.76-.86)

0.81 (.76-.87)

0.79 (.73-.85)

0.81 (.75-.89)

GES: Gene Expression Score; MPI: myocardial perfusion imaging; ROC AUC: receiver-operator characteristics area under the curve; NPV: negative predictive value; NRI: net reclassification improvement; PPV: positive predictive value; N/A: not applicable

Twenty-eight adverse events were observed, which included 25 revascularizations within 30 days, 2 MACE, and 1 further revascularization. Twenty-five of the 26 patients with revascularization and both MACE patients had high GES (>15). The authors found that GES was associated with MACE and revascularization in a logistic regression model (p=0.0015) with a sensitivity of 96% and a NPV of 99% at a score threshold of ≤15. The GES test was also correlated with maximum percent stenosis (r=0.46, p<0.001).

Section summary. Results of the PREDICT and COMPASS study establish that the GES score has predictive ability for CAD. The PREDICT and COMPASS studies report that GES score is superior to the Diamond-Forrester model and to MPI in predicting CAD. However, there are several limitations to interpretation of the evidence on comparative predictive accuracy. In the PREDICT study, the assay algorithm score discriminated cases from controls significantly better than the Diamond–Forrester clinical score by AUC analysis; however it did not discriminate better than the expanded clinical model without family history or electrocardiogram (AUC, 0.745 vs. 0.732, respectively; p=0.089). Additionally, neither the Diamond–Forrester clinical risk score nor the expanded clinical model included family history or electrocardiogram (EKG) results, which might increase the accuracy of the initial classification and decrease the net reclassification improvement observed. Furthermore, the Diamond-Forrester model is a simple prediction rule that is not commonly used in clinical care. The Framingham risk score would be a more relevant comparator that is part of contemporary clinical care.

The COMPASS study compared the GES score to results from myocardial perfusion imaging (MPI) stress testing. In that trial, the sensitivity of MPI was low at 27%. This is a considerably lower sensitivity than is routinely reported in the literature. For example, in one meta-analysis performed in support of ACC/AHA guidelines on myocardial perfusion imaging, sensitivity was estimated at 87-89%. (10) This raises the question of whether the accuracy of MPI in the COMPASS study is representative of that seen in current clinical care. Also, the comparison of overall accuracy of the GES score with MPI testing does not establish that clinical decisions would be changed, specifically whether patients with a positive MPI could safely forego further invasive testing based on a low GES score.

Does use of the test lead to changes in management that improve outcomes?

The IMPACT study compared a prospective cohort to matched historic controls to evaluate whether the gene expression score (GES) test altered the cardiologist’s evaluation and clinical management of CAD. (11) CAD was defined by the authors as no CAD (0% stenosis), CAD (≤50% stenosis) and CAD (>50% stenosis). All participants were non-diabetic, had no known prior MI or revascularization, were not using steroids, immune suppressive agents or chemotherapeutic agents, and had been referred to a cardiologist for evaluation of chest pain or angina equivalent symptoms. Eighty-eight patients were enrolled and 83 included in the final analysis. The matched cohort was composed of 83 patients selected with similar distributions of age, gender, clinical risk factors and had been evaluated at the institution within the past 3 to 30 months.

A change in patient management was defined prospectively as an increase or decrease in intensity of the diagnostic plan. GESs were divided into a high-risk group (>15) and a low-risk group (≤15). The authors defined the categories of intensity in the following order: 1) no further cardiac testing or medical therapy for angina or non-cardiac chest pain, 2) stress testing (with/without imaging) or computed tomography coronary angiography, or 3) invasive coronary angiography (ICA). Within the prospective cohort, the diagnostic testing plan was changed for 58% of patients (95% CI: 46-69%; p<0.001) with a greater reduction in testing intensity (39%) compared to increased testing intensity (19%). Compared to the historic control group, the prospective cohort had a 71% reduction in overall diagnostic testing (p<0.001).

A secondary analysis examined the testing patterns around ICA. Thirty patients, 14 from the prospective cohort and 16 from the historic cohort, who underwent ICA were included in the analysis. The authors did not find a significant difference in diagnostic yield between the two groups (p=0.24). No major cardiovascular adverse events were observed for either cohort during the 6-month follow-up period.

Section summary. Based on the IMPACT study, management decisions may be changed as a result of the GES score. This study is limited by comparison with historical controls, which were not well-matched to the study population. In addition, the impact of management changes in this study is uncertain. There is no information provided on whether the management changes led to beneficial effects on outcome, and it is not possible to estimate the likelihood of benefit from the information given in this study. Therefore, it is not possible to conclude that the GES score leads to changes in management that improve outcomes.


Gene expression assays to predict the likelihood of obstructive coronary artery disease (CAD) have the potential to improve the accuracy of predicting the likelihood of CAD. A commercially available GES test, Corus CAD, has been developed and validated for this purpose in non-diabetic patients. The PREDICT study raises the possibility that this test could be used to increase the proportion of patients selected for coronary angiography who truly have disease and reduce the number of patients who might otherwise be inappropriately exposed to radiation, contrast agent, and an invasive procedure. Results of initial validation studies report that the test may improve CAD prediction beyond that of simple prediction models such as Diamond-Forrester, but the improvement in CAD prediction when added to routine clinical evaluation is uncertain. The test has also been shown to have some predictive ability for future cardiac events and revascularization. In the COMPASS study, the overall accuracy of the GES in predicting cardiac events was superior to myocardial perfusion imaging (MPI) in patients who were referred for MPI testing. However, in that study, the reported sensitivity of MPI was considerably lower than generally reported in the literature. Also, it is unclear from the COMPASS study whether patients with a positive MPI could safely forego further testing based on a low GES.

Clinical utility of the GES score has not been demonstrated. One study with methodologic limitations reports management changes as a result of the test, but the effect of these management changes is uncertain. There has been no convincing evidence presented that the use of GES scores can reduce unnecessary coronary angiography. As a result, the use of gene expression scores for predicting CAD is considered investigational.

Practice Guidelines and Position Statements


Medicare National Coverage



  1. Xu J, Kochanek KD, Sherry L. Murphy SL et al. Deaths: Final Data for 2007. Natl Vital Stat Rep 2010; 58(19).
  2. Anderson JL, Adams CD, Antman EM et al. ACC/AHA 2007 guidelines for the management of patients with unstable angina/non ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines for the Management of Patients With Unstable Angina/Non ST-Elevation Myocardial Infarction): developed in collaboration with the American College of Emergency Physicians, the Society for Cardiovascular Angiography and Interventions, and the Society of Thoracic Surgeons: endorsed by the American Association of Cardiovascular and Pulmonary Rehabilitation and the Society for Academic Emergency Medicine. Circulation 2007; 116(7):e148-304.
  3. Patel MR, Peterson ED, Dai D et al. Low diagnostic yield of elective coronary angiography. N Engl J Med 2010; 362(10):886-95.
  4. Wingrove JA, Daniels SE, Sehnert AJ et al. Correlation of peripheral-blood gene expression with the extent of coronary artery stenosis. Circ Cardiovasc Genet 2008; 1(1):31-8.
  5. Elashoff MR, Wingrove JA, Beineke P et al. Development of a blood-based gene expression algorithm for assessment of obstructive coronary artery disease in non-diabetic patients. BMC Med Genomics 2011; 4(1):26.
  6. Rosenberg S, Elashoff MR, Beineke P et al. Multicenter validation of the diagnostic accuracy of a blood-based gene expression test for assessing obstructive coronary artery disease in nondiabetic patients. Ann Intern Med 2010; 153(7):425-34.
  7. Rosenberg S, Elashoff MR, Lieu HD et al. Whole blood gene expression testing for coronary artery disease in nondiabetic patients: major adverse cardiovascular events and interventions in the PREDICT trial. J Cardiovasc Transl Res 2012; 5(3):366-74.
  8. Lansky A, Elashoff MR, Ng V et al. A gender-specific blood-based gene expression score for assessing obstructive coronary artery disease in nondiabetic patients: results of the Personalized Risk Evaluation and Diagnosis in the Coronary Tree (PREDICT) trial. Am Heart J 2012; 164(3):320-6.
  9. Thomas GS, Voros S, McPherson JA et al. A Blood-Based Gene Expression Test for Obstructive Coronary Artery Disease Tested in Symptomatic Nondiabetic Patients Referred for Myocardial Perfusion Imaging The COMPASS Study. Circ Cardiovasc Genet 2013; 6(2):154-62.
  10. Klocke FJ, Baird MG, Lorell BH et al. ACC/AHA/ASNC guidelines for the clinical use of cardiac radionuclide imaging--executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (ACC/AHA/ASNC Committee to Revise the 1995 Guidelines for the Clinical Use of Cardiac Radionuclide Imaging). Circulation 2003; 108(11):1404-18.
  11. McPherson JA, Davis K, Yau M et al. The Clinical Utility of Gene Expression Testing on the Diagnostic Evaluation of Patients Presenting to the Cardiologist With Symptoms of Suspected Obstructive Coronary Artery Disease: Results From the IMPACT (Investigation of a Molecular Personalized Coronary Gene Expression Test on Cardiology Practice Pattern) Trial. Crit Pathw Cardiol 2013; 12(2):37-42.







No specific code (See Policy Guidelines)

ICD-9-CM diagnosis


Investigational for all relevant diagnoses

ICD-10-CM (effective 10/1/14)   Investigational for all relevant diagnoses
  Z15.89 Genetic susceptibility to other disease
  Z13.6 Encounter for screening for cardiovascular disorders
ICD-10-PCS (effective 10/1/14)   Not applicable. ICD-10-PCS codes are only used for inpatient services. There are no ICD procedure codes for laboratory tests.



Gene expression profiling, cardiovascular disease risk
Gene expression assay, coronary artery disease risk

Policy History

Date Action Reason
06/09/11 Add to Medicine section, Pathology/Laboratory subsection Policy created with literature search through April 2011; considered investigational
06/14/12 Replace policy Policy updated with literature search through April 2012; reference 6 added; no change to policy statement.
6/13/13 Replace policy Policy updated with literature search through April 2013; reference 8-11 added. Rationale section reorganized and editorial changes made. No change in policy statement