Optical coherence tomography imaging of coronary atherosclerosis is affected by intraobserver and interobserver variability.

Abstract

Optical coherence tomography (OCT) has emerged as a novel imaging modality that allows plaque classification through identification of features including lipid, calcification and fibrous cap. However, subtle changes in light attenuation as the optical beam traverses the plaque in vivo are challenging to interpret and data on strength of observer agreement are minimal. Thus, we sought to assess both the intra and interobserver variability for plaque composition/classification using OCT. OCT imaging was performed in 50 patients prior to percutaneous coronary intervention. Analysis was performed offline by two independent, experienced OCT operators. Target lesion luminal dimensions and plaque composition were assessed at minimal luminal area (MLA) and at five 1-mm longitudinal intervals proximal and distal to the MLA. An OCT thin-capped fibroatheroma (OCT-TCFA) was defined as greater than 90 degree lipid arc with minimal fibrous cap thickness less than 0.85 μm. Overall, 540 frames of OCT were included and exceptional consistency was seen for all measures of luminal geometry [all intraclass correlation coefficients (ICC) >0.97, P < 0.001]. Intraobserver agreements for calcification and lipid arc were strong (both ICC >0.84, P < 0.001), whereas interobserver agreement was higher for calcium (ICC 0.76) than lipid (ICC 0.69). Interobserver agreement of minimal fibrous cap thickness was moderate (ICC 0.52, 95% confidence interval 0.45-0.58, P < 0.001], but improved as cap thickness decreased. Overall, intra and interobserver agreements for OCT-defined plaque classification were strong (K = 0.86 and 0.71, respectively). Luminal dimensions and plaque compositional features identified by OCT are minimally affected by observer variability, permitting dependable plaque classification.