AI-Based CT Analysis of Coronary Plaque Evolution

The study uses existing de-identified clinical, laboratory, and serial CCTA data. Eligible participants are adults with coronary plaque, no prior percutaneous coronary intervention or coronary artery bypass grafting, baseline and follow-up CCTA separated by at least 6 months (preferably 12 to 36 months), longitudinally matchable target segments, and adequate image quality. The baseline CCTA will be used for model prediction; the follow-up CCTA will be used for outcome ascertainment. The primary endpoint is definite patient-level non-calcified plaque progression versus non-progression. Annualized change in non-calcified plaque percent atheroma volume will be compared with an MDC95 determined in a 30-50 participant repeatability substudy. Definite progression is greater than +MDC95, stability is within +/- MDC95, and definite reduction/regression is less than -MDC95. Secondary analyses will evaluate prediction of definite plaque reduction/regression, sensitivity, specificity, positive and negative predictive values, F1 score, calibration, Brier score, decision-curve analysis, and plaque-level analyses using generalized estimating equations or mixed-effects models to account for clustering within participants. Analyses will include patient-level cross-validation, center-level validation where feasible, subgroup analyses, and prespecified sensitivity analyses by follow-up interval, scanner/reconstruction protocol, and image quality. Data will be de-identified, encrypted, access-controlled, and analyzed without additional participant procedures or costs.

Primary outcomes

• Patient-level definite non-calcified plaque progression discrimination (AUC) [After completion of follow-up CCTA assessment and database lock (approximately 12-36 months after baseline CCTA).]: Area under the receiver operating characteristic curve (AUC) with 95% confidence interval for the full-system model predicting definite patient-level progression of non-calcified plaque. Definite progression is annualized change in non-calcified plaque percent atheroma volume greater than +MDC95; non-progression includes stability and definite reduction/regression. AUC will be compared with conventional CCTA models using the DeLong test.

Secondary outcomes

• Prediction of definite non-calcified plaque reduction/regression (AUC and classification metrics) [After completion of follow-up CCTA assessment and database lock (approximately 12-36 months after baseline CCTA).]: AUC with 95% confidence interval for predicting definite plaque reduction/regression, with sensitivity, specificity, positive predictive value, negative predictive value, and F1 score.
• Model calibration (Brier score) [After completion of follow-up CCTA assessment and database lock (approximately 12-36 months after baseline CCTA).]: The prespecified quantitative outcome is the Brier score for the full-system and conventional CCTA models, with calibration curves and the Hosmer-Lemeshow test reported as supporting calibration assessments. Prespecified subgroup and sensitivity analyses will assess robustness.
• Clinical net benefit (decision-curve analysis) [After completion of follow-up CCTA assessment and database lock (approximately 12-36 months after baseline CCTA).]: Clinical net benefit of the full-system and conventional CCTA models will be evaluated using decision-curve analysis across prespecified risk-threshold probabilities. Results will be reported as net benefit at each threshold; prespecified subgroup and sensitivity analyses will assess robustness.

Sex: All, ages: 18 Years to any

Inclusion Criteria:

1. Adults aged 18 years or older.
2. Baseline CCTA demonstrates coronary plaque in the LAD, LCX, RCA, or a major branch with reference diameter >=2.0 mm.
3. No prior percutaneous coronary intervention or coronary artery bypass grafting.
4. Baseline CCTA and at least one follow-up CCTA are available, with an interval of at least 6 months, preferably 12-36 months.
5. Target segments can be longitudinally matched and non-calcified plaque volume and total vessel volume can be quantified.
6. Image quality meets the core laboratory requirements.

Exclusion Criteria:

1. Severe comorbid disease, active malignancy, acute or chronic inflammation, connective tissue or autoimmune disease, or life expectancy less than 1 year.
2. Pregnancy.
3. Prior PCI stent implantation or CABG.
4. In-stent lesions, complete occlusion, thrombus, left main or ostial lesions, or other lesions preventing the planned hemodynamic or biomechanical analyses.
5. Image quality or key data judged inadequate by the core laboratory.

• Symons R, Morris JZ, Wu CO, Pourmorteza A, Ahlman MA, Lima JA, Chen MY, Mallek M, Sandfort V, Bluemke DA. Coronary CT Angiography: Variability of CT Scanners and Readers in Measurement of Plaque Volume. Radiology. 2016 Dec;281(3):737-748. doi: 10.1148/radiol.2016161670. Epub 2016 Sep 16. PMID 27636027
• Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res. 2005 Feb;19(1):231-40. doi: 10.1519/15184.1. PMID 15705040
• Won KB, Lee BK, Park HB, Heo R, Lee SE, Rizvi A, Lin FY, Kumar A, Hadamitzky M, Kim YJ, Sung JM, Conte E, Andreini D, Pontone G, Budoff MJ, Gottlieb I, Chun EJ, Cademartiri F, Maffei E, Marques H, de Araujo Goncalves P, Leipsic JA, Shin S, Choi JH, Virmani R, Samady H, Chinnaiyan K, Raff GL, Stone PH, Berman DS, Narula J, Shaw LJ, Bax JJ, Min JK, Chang HJ. Quantitative assessment of coronary plaque volume change related to triglyceride glucose index: The Progression of AtheRosclerotic PlAque DetermIned by Computed TomoGraphic Angiography IMaging (PARADIGM) registry. Cardiovasc Diabetol. 2020 Jul 18;19(1):113. doi: 10.1186/s12933-020-01081-w. PMID 32682451
• Han D, Berman DS, Miller RJH, Andreini D, Budoff MJ, Cademartiri F, Chinnaiyan K, Choi JH, Conte E, Marques H, de Araujo Goncalves P, Gottlieb I, Hadamitzky M, Leipsic J, Maffei E, Pontone G, Shin S, Kim YJ, Lee BK, Chun EJ, Sung JM, Lee SE, Virmani R, Samady H, Stone P, Narula J, Bax JJ, Shaw LJ, Lin FY, Min JK, Chang HJ. Association of Cardiovascular Disease Risk Factor Burden With Progression of Coronary Atherosclerosis Assessed by Serial Coronary Computed Tomographic Angiography. JAMA Netw Open. 2020 Jul 1;3(7):e2011444. doi: 10.1001/jamanetworkopen.2020.11444. PMID 32706382