Reproducibility of an artificial intelligence optical coherence tomography software for tissue characterization: Implications for the design of longitudinal studies.

BACKGROUND: To assess the reproducibility of coronary tissue characterization by an Artificial Intelligence Optical Coherence Tomography software (OctPlus, Shanghai Pulse Medical Imaging Technology Inc.). METHODS: 74 patients presenting with multivessel ST-segment elevation myocardial infarction (STEMI) underwent optical coherence tomography (OCT) of the infarct-related artery at the end of primary percutaneous coronary intervention (PPCI) and during staged PCI (SPCI) within 7 days thereafter in the MATRIX (Minimizing Adverse Hemorrhagic Events by Transradial Access Site and angioX) Treatment-Duration study (ClinicalTrials.gov, NCT01433627). OCT films were run through the OctPlus software. The same region of interest between either side of the stent and the first branch was identified on OCT films for each patient at PPCI and SPCI, thus generating 94 pairs of segments. 42 pairs of segments were re-analyzed for intra-software difference. Five plaque characteristics including cholesterol crystal, fibrous tissue, calcium, lipid, and macrophage content were analyzed for various parameters (span angle, thickness, and area). RESULTS: There was no statistically significant inter-catheter (between PPCI and SPCI) or intra-software difference in the mean values of all the parameters. Inter-catheter correlation for area was best seen for calcification [intraclass correlation coefficient (ICC) 0.86], followed by fibrous tissue (ICC 0.87), lipid (ICC 0.62), and macrophage (ICC 0.43). Some of the inter-catheter relative differences for area measurements were large: calcification 9.75 %; cholesterol crystal 74.10 %; fibrous tissue 5.90 %; lipid 4.66 %; and macrophage 1.23 %. By the intra-software measurements, there was an excellent correlation (ICC > 0.9) for all tissue types. The relative differences for area measurements were: calcification 0.64 %; cholesterol crystal 5.34 %; fibrous tissue 0.19 %; lipid 1.07 %; and macrophage 0.60 %. Features of vulnerable plaque, minimum fibrous cap thickness and lipid area showed acceptable reproducibility. CONCLUSION: The present study demonstrates an overall good reproducibility of tissue characterization by the Artificial Intelligence Optical Coherence Tomography software. In future longitudinal studies, investigators may use discretion in selecting the imaging endpoints and sample size, accounting for the observed relative differences in this study.

Predictive Accuracy of Surrogate Indices for Hepatic and Skeletal Muscle Insulin Sensitivity.

CONTEXT: Surrogate indices of muscle and hepatic insulin sensitivity derived from an oral glucose tolerance test (OGTT) are frequently used in clinical studies. However, the predictive accuracy of these indices has not been validated. DESIGN: In this cross-sectional study, hyperinsulinemic-euglycemic glucose clamp with tritiated glucose infusion and a 75-g OGTT were performed in individuals (n = 659, aged 18 to 49 years, body mass index of 16 to 64 kg/m2) with varying degrees of glucose tolerance. A calibration model was used to assess the ability of OGTT-derived, tissue-specific surrogate indices [hepatic insulin resistance index (HIRI) and muscle insulin sensitivity index (MISI)] to predict insulin sensitivity/resistance indices derived from the reference glucose clamp [Hepatic-IRbasal, a product of fasting plasma insulin and hepatic glucose production (HGP), Hepatic-IRclamp, reciprocal of the percent suppression of HGP during the insulin clamp corrected for plasma insulin concentration, and Muscle-ISclamp, a measure of peripheral glucose disposal]. Predictive accuracy was assessed by root mean squared error of prediction and leave-one-out, cross-validation-type square root of the mean squared error of prediction. RESULTS: HIRI and MISI were correlated with their respective clamp-derived indices. HIRI was negatively related to Muscle-ISclamp (r = -0.62, P < 0.0001) and MISI correlated with Hepatic-IR derived from the clamp (Hepatic-IRbasal: r = -0.48, P < 0.0001 and Hepatic-IRclamp: r = -0.41, P < 0.0001). However, the accuracy of HIRI and MISI to predict Hepatic-IR (basal or during clamp) was not significantly different. Likewise, the ability of HIRI and MISI to predict Muscle-ISclamp was also similar. CONCLUSION: Our findings indicate that the surrogate indices derived from an OGTT are accurate in predicting insulin sensitivity but are not tissue specific.

Assessing the predictive accuracy of oral glucose effectiveness index using a calibration model.

PURPOSE: Current reference methods for measuring glucose effectiveness (GE) are the somatostatin pancreatic glucose clamp and minimal model analysis of frequently sampled intravenous glucose tolerance test (FSIVGTT), both of which are laborious and not feasible in large epidemiological studies. Consequently, surrogate indices derived from an oral glucose tolerance test (OGTT) to measure GE (oGE) have been proposed and used in many studies. However, the predictive accuracy of these surrogates has not been formally validated. In this study, we used a calibration model analysis to evaluate the accuracy of surrogate indices to predict GE from the reference FSIVGTT (SgMM). METHODS: Subjects (n = 123, mean age 48 ± 11 years; BMI 35.9 ± 7.3 kg/m2) with varying glucose tolerance (NGT, n = 37; IFG/IGT, n = 78; and T2DM, n = 8) underwent FSIVGTT and OGTT on two separate days. Predictive accuracy was assessed by both root mean squared error (RMSE) of prediction and leave-one-out cross-validation-type RMSE of prediction (CVPE). RESULTS: As expected, insulin sensitivity, SgMM, and oGE were reduced in subjects with T2DM and IFG/IGT when compared with NGT. Simple linear regression analyses revealed a modest but significant relationship between oGE and SgMM (r = 0.25, p < 0.001). However, using calibration model, measured SgMM and predicted SgMM derived from oGE were modestly correlated (r = 0.21, p < 0.05) with the best fit line suggesting poor predictive accuracy. There were no significant differences in CVPE and RMSE among the surrogates, suggesting similar predictive ability. CONCLUSIONS: Although OGTT-derived surrogate indices of GE are convenient and feasible, they have limited ability to robustly predict GE.