Serial analysis of coronary artery disease progression by artificial intelligence assisted coronary computed tomography angiography: early clinical experience

11 disease segments were identified for analysis: LAD/diagonals, 5; Circumflex, 2; RCA/PBL, 3; ramus, 1. TPV change-over-time demonstrated decreasing LD-NCP (4-segments[1.9 mm3] to 7-segments [0.6 mm3]) with increasing overall NCP (4-segments[37.4 mm3] to 7-segments [60 mm3]) and increasing CP (1-segment [5.9 mm3] to 6-segments[63.4 mm3]) (Fig. 1). Examination of individual segments revealed a proximal-LAD lesion (increasing length from 9 to 11.5 mm3 and stenosis from 15 to 20%) with decreasing NCP over-time (29–8.6 mm3) and increasing CP (5.9–31 mm3) (Fig. 2). In contrast, although the D2/D1/ramus lesions showed increasing stenosis, CP, and total plaque, there were no significant differences in NCP over-time (D2 length increased from 5.5 to 7.8 mm3 with increasing stenosis 8–38%, with stable NCP (6.6 mm3) and increased CP (0–12.7 mm3) (Additional file 1: Fig S1). Interestingly, the circumflex and more proximal mid-RCA lesions that developed had primarily NCP (Fig. 3 and Additional file 1: Fig. S2), whereas the initial mid-RCA lesion showed an increase of both CP and NCP (Additional file 1: Fig. S2) FFR-CT analysis showed no significant changes except for the D2 lesion which decreased from 0.94 to 0.77 (Additional file 1: Figs. S3 and S4). LDL was optimally managed < 70 mg/dl with a statin and subsequent addition of a PCSK9 inhibitor. Specifically, in terms of timeline, the patient was placed on atorvastatin 80 mg daily starting in 2009, and then evolocumab 140 mg every 2 weeks starting in 2018. 5 serial CCTAs were performed over the course of this study, in 2008, 2009, 2013, 2018 and 2021 (Figs. 1, 2 and 3).

Fig. 1figure 1

Evolution of atherosclerotic plaque characteristics

Fig. 2figure 2

LAD Territory atherosclerotic plaque characteristics

Fig. 3figure 3

LCX territory atherosclerotic plaque characteristics

The novelty of this case study is in our following of serial changes in CAD using CCTA (> 2 CCTAs) based AI analysis of coronary plaque characteristics over a period greater than a decade (13 years). The advantage of the AI augmented CCTA software is multifold, including the expedited time to process studies compared to human readers, the potential time/financial benefits saved by not needing human/manual processing of images (ongoing field of study), and the reliability/accuracy of analysis able to be performed at a level equivalent to expert CT level-3 readers [5]. We were able to consistently assess progression of plaque length, volumes, remodeling, stenosis, and APCs with this novel artificial intelligence augmented CCTA methodology. Overall, despite optimal LDL control < 70 mg/dL with use of a statin and PCK9 inhibitor, we found a significant increase in TPV composed of decreasing LD-NCP and increasing NCP and CP. Furthermore, there were variations in the evolution of APCs between vessels, manifesting as changing amounts of LD-NCP, NCP, and CP. Limitations of our study include major restrictions of all case reports with patient numbers being an N of 1, which may lack generalizability and reproducibility. However, recent clinical studies such as the CLARIFY trial, have demonstrated that AI assisted CCTA can accurately and consistently quantify CAD morphology and stenosis. Although the significance of evolving APCs over-time to predict CAD/MACE outcomes needs to be investigated, this case demonstrates AI-based CCTA analysis can serve as valuable clinical tool by accurately defining unique patient CAD characteristics—and future prospective trails are needed to assess whether this ability to further quantify APCs may provide further prognostic capabilities for AI-based CCTA to improve clinical care [4, 5].

留言 (0)

沒有登入
gif