Near-infrared spectroscopy for cardiovascular risk assessment? Not ready for prime time.

The rupture of an unstable atherosclerotic plaque precedes the majority of acute coronary syndromes, and considerable effort has been devoted to identifying patients at increased risk of future coronary ischaemic events. Atherosclerosis imaging is a potentially attractive strategy to detect vulnerable coronary plaques, with each modality having its own advantages and limitations. Different characteristics of the atherosclerotic plaque can be targeted with imaging, such as the number and severity of luminal narrowings with angiography, atheroma burden and remodelling with intravascular ultrasonography (IVUS), elastic properties with elastography, radiofrequencydetermined plaque types with virtual histology, fibrous cap thickness and necrotic core with optical coherence tomography, extent of calcifications with computed tomography, and inflammation with positron emission tomography. Near-infrared spectroscopy (NIRS) provides chemical assessment of tissues, and catheter-based NIRS can generate spectra that may distinguish cholesterol from collagen and may identify chemical fingerprints in coronary plaques in the cardiac catheterization laboratory (Figure 1). Although intravascular NIRS has been compared with pathological findings, its research and clinical implications remain to be established. de Boer et al. now report their use ofNIRS in a non-culprit coronary artery and the associations with clinical characteristics and biomarkers in 208 patients undergoing percutaneous coronary intervention. History of hypercholesterolaemia, male sex, use of beta-blockers, and a history of peripheral artery disease and/or cerebrovascular disease were modestly associated with NIRS-derived lipid core burden index (LCBI), while biomarkers such as blood lipids and highsensitivityC-reactiveproteinwerenot.Approximately23%of thevariance in LCBI was explained by all available baseline characteristics including plasma biomarkers. The LCBI on NIRS and percentage area plaque burden on IVUS were modestly correlated (r 1⁄4 0.29). The authors argue that the weak or non-significant associations between LCBI and both clinical characteristics and plasma biomarkers suggest that NIRS could provide additional independent information about cardiovascular risk. An alternative explanation could be that technical orconceptual issues related toNIRS maybe responsible for these results, as discussed below. Indeed, it is surprising that diabetes did not affect the findings on NIRS, especially given that the severity and progression of coronary atherosclerosis as evaluated by IVUS have been shown to be associated with the presence of diabetes. Plasma total cholesterol, LDL-cholesterol, HDL-cholesterol, and triglycerides did not correlate with NIRS-determined LCBI, which could be explained by the fact that 90% of patients were treated with statins. The relatively weak correlation between NIRS and IVUS means that , 10% of the variance in plaque burden is explained by LCBI. Whether this observation means that NIRS can provide relevant complementary information to IVUS cannot be determined with the results of the current study. Alternatively, the low correlation could potentially be explained by methodological issues pertaining to IVUS and NIRS imaging and analysis in this study. IVUS imaging was performed at a frequency of 20 MHz, which results in lower image resolution compared with higher frequency (40–45 MHz) catheters. Also, IVUS and NIRS imaging were performed using different catheters, not with a catheter combining both modalities; given the limited spatial registration of NIRS, the matching of the arterial segment to be evaluated by both techniques may have been suboptimal. Thirdly, it appears that a single cross-section was measured on IVUS as only area measurements are reported. The selection process for the two-dimensional cross-section of interest to be analysed should have been described. On the other hand, if a longer arterial segment was traced, it would have been interesting to know the method of selection for the segment of interest and its length. Indeed, distance from the ostium has been reported to be an independent predictor of LCBI. In the present study, however, the authors have not reported this parameter and do not seem to have included it in the multivariable models. Finally, the reproducibility of repeated NIRS pullbacks and measurements are not presented.