Spatial organization of plaque vulnerability around the minimal luminal area: insights from AI-based volumetric OCT
L Heil, T J Luttikholt, R H J A Volleberg, R G A van der Waerden, P Cancian, J L van der Zande, X Gu, I Isgum, S Saitta, J Thannhauser, N van RoyenAbstract
Background/Introduction
Plaque rupture frequently occurs at or proximal to the minimal luminal area (MLA), indicating that plaque vulnerability is not evenly distributed along the vessel. Regions proximal to the MLA are exposed to complex hemodynamic forces, including altered wall shear stress, which may promote plaque accumulation, lipid enrichment, and fibrous cap thinning. However, comprehensive three-dimensional assessment of plaque composition along the vessel has been limited by manual optical coherence tomography (OCT) analysis. Artificial intelligence (AI) enables automated volumetric plaque quantification and may improve spatial characterisation of high-risk plaque features.
Purpose
To investigate the longitudinal spatial distribution of AI-derived plaque vulnerability features relative to the MLA in patients with myocardial infarction.
Methods
OCT pullbacks were analysed using a validated AI algorithm for automated tissue segmentation. Plaque burden, lipid burden, and calcium burden were quantified volumetrically. Presence of thin-cap fibroatheroma (TCFA) was defined as lipid presence with an arc >90° and a fibrous cap <65 µm. A 20-mm segment centred on the MLA (10 mm proximal and 10 mm distal) was analysed to assess longitudinal spatial distribution of plaque burden features, with Savitzky–Golay smoothing applied to generate continuous spatial profiles.
Results
In a multicenter cohort, 415 patients (488 non-flow limiting non-culprit lesions) undergoing OCT imaging after myocardial infarction were included. Plaque burden demonstrated a distinct proximal predominance, with higher values in segments upstream of the MLA compared with distal segments (41.7% proximal vs 38.9% distal, p < 0.001). Lipid- and calcium burden were higher proximal to the MLA, followed by a sharp decline distally (21.3% proximal vs 14.1% distal and 5.1% proximal vs 3.5% distal respectively, p < 0.001). TCFA was twice as likely to be present in the frames in the proximal segment (3.2% of proximal frames vs 1.6% of distal frames).
Conclusions
AI-enabled volumetric OCT analysis reveals a reproducible proximal predominance of overall plaque burden, and specifically lipid and calcium burden in non-culprit lesions of patients after myocardial infarction. These observations support the concept that hemodynamically and biologically active regions upstream of the MLA represent focal zones of plaque buildup and plaque vulnerability in the form of an increase in TCFA. Automated three-dimensional plaque characterisation enables objective spatial mapping of high-risk features and may improve mechanistic understanding of plaque destabilisation.prox_dist_panel_plotlongitudinal_example_pullback