The influence of lipid-containing plaque composition assessed by near-infrared spectroscopy on coronary lesion remodelling

Transcription

1 European Heart Journal Cardiovascular Imaging (2016) 17, doi: /ehjci/jev221 The influence of lipid-containing plaque composition assessed by near-infrared spectroscopy on coronary lesion remodelling Hideaki Ota, Marco A. Magalhaes, Rebecca Torguson, Smita Negi, Max R. Kollmer, Mia-Ashley Spad, Jiaxiang Gai, Lowell F. Satler, William O. Suddath, Augusto D. Pichard, and Ron Waksman* Division of Cardiology, MedStar Washington Hospital Center, 110 Irving Street, NW, Suite 4B-1, Washington, DC 20010, USA Received 30 April 2015; accepted after revision 10 August 2015; online publish-ahead-of-print 12 September 2015 Aims Vessel remodelling is commonly observed in coronary atherosclerosis, but factors influencing remodelling, such as plaque lipid content, remain poorly described.... Methods Remodelling index (RI) was calculated as the ratio of lesion to proximal and distal references external membrane area and results and was categorized as follows: positive (PR; RI. 1.05), intermediate (IR; RI ), and negative remodelling (NR; RI, 0.95). RI was studied by near-infrared spectroscopy (NIRS) as a function of lipid content metrics, including the maximal 4 mm lipid core burden index of the segment (maxlcbi 4mm ) and intravascular ultrasound (IVUS) lesion plaque burden (PB). The authors further stratified the analysis according to obstructive ( 50%) and non-obstructive (,50%) lesions using quantitative coronary angiography. Receiver-operating characteristic curves were performed to describe the maxlcbi 4mm level associated with PR. From May 2012 to November 2014, 100 de novo lesions from 67 patients underwent simultaneous NIRS-IVUS. PR was found in 28% of the lesions. There was a positive linear correlation between RI and maxlcbi 4mm (r ¼ 0.58; P, 0.001). Although PR lesions had a larger PB than NR or IR (P, 0.001), the correlation of RI with maxlcbi 4mm was stronger compared with plaque volume (r ¼ 0.18; P ¼ 0.07) and with per cent PB (r ¼ 0.41; P, 0.001). This relationship remained significant for obstructive (r ¼ 0.72; P, 0.001) and non-obstructive lesions (r ¼ 0.48; P, 0.001). By receiver-operating characteristic curve analysis, values of maxlcbi 4mm 439 were predictive for PR (area under the curve ¼ 0.79, 95% confidence interval: ).... Conclusion In vivo coronary lesion remodelling is positively correlated with lipid plaque content assessed by NIRS rather than simply PB. Thus, the use of NIRS can potentially aid in further stratifying vulnerable lesions Keywords Coronary arterial remodelling Lipid-rich plaque Near-infrared spectroscopy Introduction Coronary arterial remodelling is a fundamental course of atherosclerosis related to the artery response to plaque accumulation; it results in vessel expansion, or positive remodelling (PR), or vessel shrinkage, or negative remodelling (NR). 1,2 PR is an adaptive process that accommodates the plaque burden (PB) without compromising the lumen area. Consequently, positive-remodelled lesions often harbour a larger PB than those with NR or intermediate remodelling (IR). 3,4 In addition, PR lesions are frequently more likely to be associated with vulnerable plaque phenotypes and plaque rupture. 2,4 6 Intravascular ultrasound (IVUS) studies have further linked these morphological differences to the clinical manifestations of coronary artery disease. Patients presenting with acute coronary syndromes more frequently have PR at culprit lesions than stable patients, in whom NR is more prevalent, 7 and this association appears to be * Corresponding author. Tel: ; Fax: ron.waksman@medstar.net These authors contributed equally to this work. Published on behalf of the European Society of Cardiology. All rights reserved. & The Author For permissions please journals.permissions@oup.com.

2 822 H. Ota et al. independent of conventional risk factors. 3 Moreover, lesions with PR are associated with the worse outcomes than NR in both culprit and non-culprit lesion domains. 3,8 The larger PB systematically documented in PR lesions explains, in part, the impact on outcome. 9 Plaque composition, however, also plays an essential role, 9 11 particularly the lipid content. 4 Although radiofrequency virtual histology IVUS (VH-IVUS) has provided important insights into in vivo plaque composition, 9 its accuracy has recently been questioned. 12 Near-infrared spectroscopy (NIRS) has emerged as a novel imaging tool for the prediction of plaque lipid content. 13,14 The development of NIRS and IVUS via a single catheter enables the simultaneous assessment of the arterial remodelling and the analysis of its chemical composition. Currently, the impact of lipid content on coronary arterial remodelling is not well understood. Therefore, the aims of our study were to: (i) describe lesion remodelling as a function of lipid plaque composition; (ii) determine the distribution of plaque lipid content according to minimal lumen area (MLA) and PB; and (iii) determine what level of lipid pool best describes the occurrence of PR. Methods Study design and patient population The authors retrospectively identified patients who had NIRS-IVUS (LipiScan, InfraReDx, Burlington, MA, USA) co-registration of at least one vessel for clinical reasons at the authors institution from May 2012 to November De novo, native coronary lesions with PB 40% in at least a 4 mm segment length by IVUS were selected for this analysis. Separated lesions were considered if there was a 5 mm-long segment with PB,40% between them. The exclusion criteria were: (i) diffuse lesion, which was defined as.40 mm long; (ii) stent areas, including 5 mm proximal and distal borders; (iii) lesion at left main trunk; (iv) ostial lesion, which was located 10 mm from the edge of the lesion to coronary ostia or left main bifurcation; and (v) bifurcation lesion with major side branch. All the remaining lesions were analysed off-line by quantitative coronary angiography (QCA), grey-scale IVUS, and NIRS. The study was approved by the local Institutional Review Board. Percutaneous coronary intervention Coronary angiography and percutaneous coronary intervention (PCI) were performed using standard techniques. All patients who underwent PCI received dual antiplatelet therapy, consisting of oral administration of aspirin (325 mg) and clopidogrel ( mg) or prasugrel (60 mg) before the procedure. During PCI, patients received anticoagulation with bivalirudin or unfractionated heparin. NIRS-IVUS was used for clinical reasons at the physician s discretion. IVUS acquisition protocol The authors institution is a high-volume IVUS centre with dedicated technicians for image recording and interpretation of every case. Since 2012, the 3.2F NIRS-IVUS catheter (LipiScan, InfraReDx) has been approved for commercial use in the USA and adopted in the authors practice. All IVUS analyses were performed before PCI with the catheter advanced distally beyond the target lesion. An automated mechanical pullback (0.5 mm/s; 240 rotations) was performed back to the proximal lesion area or to the vessel ostia. The system provides simultaneous colour-coded NIRS spectral data co-registered with IVUS images in a single pullback. Image data were archived onto DVD and sent to the local core laboratory for analysis, which was blinded to the patients data and angiographic characteristics. QCA analysis Standard image acquisition was performed using at least two orthogonal projections of the lesion stenosis and preceded by intracoronary nitroglycerine to provide maximum coronary vasodilation. Coronary angiograms were interpreted by technicians blinded to IVUS and NIRS data using a validated automated edge-detection programme (QAngio 7.3, Medis medical imaging systems, Leiden, The Netherlands). Collected morphological variables included lesion calcification defined as radioopacities through the vessel wall and thrombus-containing lesions defined as globular filling defects surrounded by contrast observed in multiple projections. QCA was preceded by system calibration using the external diameter of the contrast-filled catheter. Subsequently, minimal lumen diameter (MLD) and reference segment diameter were measured at the end of diastole, with the result from the worst view recorded. IVUS analysis and lesion remodelling patterns All grey-scale IVUS analyses were performed using commercial software (QIvus 3.0, Medis medical imaging systems). All lesions that met the inclusion criteria were analysed allowing the inclusion of multiple lesions per patient. Cross-sectional analyses were performed across the entire lesion segment according to the American College of Cardiology Clinical Expert Consensus Document on Standards for Acquisition, Measurement, and Reporting of IVUS Studies. 15 Measurements were performed every 0.5 mm over the region of interest, including the MLA site. The proximal and distal reference segments were defined as those with the largest lumen area within 10 mm proximally and distally to the lesion site but before any major side branch. Quantitative planimetry measurements were obtained using automatic border detection followed by manual correction as necessary. The measurements included the cross-sectional areas (CSAs) of the external elastic membrane (EEM) and lumen. Plaque plus media (P+M) CSA was calculated as EEM CSA minus lumen CSA and per cent PB as P+M CSA divided by EEM CSA 100. Remodelling index (RI) was defined as the ratio of EEM CSA at the MLA site divided by the average area of the proximal and distal reference segment EEM. PR was indicated when RI. 1.05, IR when RI , and NR when RI, Additionally, volumetric analyses of vessel, lumen, and plaque volume (¼P+M volume: defined as vessel volume minus lumen volume) were performed with Simpson s rule. NIRS analysis and chemometrics NIRS analyses were executed using US Food and Drug Administrationvalidated software (QIvus 3.0, Medis medical imaging systems). During catheter pullback, the probability of lipid core (LCP) was displayed as a chemogram ; a digital colour-coded map that included the location and intensity of LCP. The X-axis indicates the pullback position every 0.1 mm and the Y-axis indicates the circumferential position in degrees. Spectroscopic information at each pixel was transformed into a probability of LCP that was then mapped to a 128 (7-bit) red-to-yellow colour scale, with the low and high probabilities of lipid content shown as red and yellow, respectively. If a pixel did not contain enough data (for example, shadowing), it displayed black and was deemed a non-valid pixel. To provide a quantitative summary metric of the LCP over the entire scanned segment, a maximum lipid core burden index (LCBI) was calculated. LCBI is the fraction of valid pixels in the chemogram that exceeds an LCP of 0.6, multiplied by The block chemogram displayed

3 Coronary remodelling and plaque lipid content 823 colour-coded blocks according to the probability of finding a lipid plaque using the top 10th percentile pixel information of the corresponding 2 mm NIRS segment. If the top 10th percentile probability value for lipids was 0.98, the block showed yellow; , tan; , orange; and,0.57, red. The maxlcbi 4mm was defined as the maximum value of the LCBI for any 4 mm segment in the target lesion, including the MLA site. The LCBI lesion was defined as the LCBI value when the entire IVUS-defined lesion was included. Because the length of lesion over which LCP was calculated depended on the lesion length, percentage LCP length (defined as an absolute LCP length divided by lesion length) and percentage of each colour block (yellow, tan, orange, and red) were also calculated. Figures 1 and 2 illustrate PR and NR and the lipid content metrics, respectively. Study hypotheses and objectives The authors hypothesized that lesion remodelling can be explained as a function of plaque lipid content and that lesion remodelling can be better explained when lipid content is taken into consideration rather than by PB alone. The authors final objective was to describe what level of plaque lipid content is associated with the occurrence of PR. Statistical analysis All statistical analyses were performed with the IBM SPSS version 20.0 (IBM Corp., Armonk, NY, USA) and SAS version 9.2 (SAS Institute Inc., Cary, NC, USA) software packages. Continuous variables were presented as the mean + standard deviation or median and 25th and 75th interquartile range. Categorical variables were presented as numbers and percentages. The authors first explored the normality of the variables by using the Shapiro Wilk test. One-way analysis of variance was used to compare groups for normally distributed variables followed by post hoc test adjustment for multiple comparisons. Non-parametric tests were used as appropriate. Secondly, the correlation between the RI as a continuous variable and LCP parameters was generated using the Spearman test. This correlation was further divided into obstructive and non-obstructive lesions by QCA [per cent diameter stenosis (DS) or,50%, respectively]. Subsequently, the authors compared the maxlcbi 4mm distribution according to the PB and MLA across the three groups 9 :(i)mla. 4.0 mm 2 and Figure 1 PR (RI ¼ 1.35). (A) Proximal reference ¼ mm 2 ;(B) MLA site ¼ mm 2 ;(C) distal reference ¼ 9.35 mm 2 ;(D) IVUS longitudinal reconstruction; and (E) chemogram showing raw spectral data for each 0.1 mm length (X-axis) and one degree (Y-axis) ranging from 0 (red) to 1.00 (yellow). Lesion length ¼ 15.6 mm, LCBI lesion ¼ 272, LCP angle ¼ 3258, maxlcbi 4mm ¼ 721. (F) Chemogram blocks showing 45.6% yellow (7.1 mm), 11.6% tan (1.8 mm), and 43.1% red-coloured blocks (6.7 mm).

4 824 H. Ota et al. Figure 2 NR (RI ¼ 0.71). (A) Proximal reference ¼ mm 2 ;(B) MLA site ¼ 7.65 mm 2 ;(C) distal reference ¼ 9.56 mm 2 ;(D) IVUS longitudinal reconstruction; and (E) chemogram. Lesion length ¼ 11.0 mm, LCBI lesion ¼ 6, LCP angle ¼ 228, and the maxlcbi 4mm ¼ 16. (F) All chemogram blocks are red-coloured. PR, positive remodelling; NR, negative remodelling; RI, remodelling index; IVUS, intravascular ultrasound; MLA, minimal lumen area; LCBI, lipid core burden index; maxlcbi 4mm, maximal 4 mm lipid core burden index. PB, 70%; (ii) MLA 4.0 mm 2 and PB, 70% or MLA. 4.0 mm 2 and PB 70%; and (iii) MLA 4.0 mm 2 and PB 70%. Finally, to test what level of maxlcbi 4mm and LCBI lesion best describes PR, receiver-operating characteristic curve analyses were carried out. The authors analyses were adjusted to account for a potential cluster effect in their parameters using generalized estimating equation models. All statistical tests were two-sided, and a P, 0.05 was considered to be statistically significant. Results Baseline patient characteristics Baseline characteristics are listed in Table 1. Overall, 67 patients had at least one vessel assessed by NIRS-IVUS, resulting in 100 lesions analysed. Of these patients, 62.7% were men and 34.3% had diabetes. Acute coronary syndrome was the indication for angiography in 58.2% of the patients. PR was found in 28% of the lesions, whereas 40% had NR. Lesion territory distribution was similar across the remodelling categories, as were the QCA parameters, including the MLD (P ¼ 0.11) and the per cent DS (P ¼ 0.26) (Table 2). Notably, 24% of the lesions were deemed obstructive by QCA (DS 50%). Coronary lesion remodelling and IVUS parameters A total of 3742 mm of the scanned vessel and 7484 CSAs were analysed. The characteristics of RI groups are reported in Table 3. The average RI was , , and in PR, IR, and NR lesions, respectively (P, 0.001). PR lesions were characterized by the largest PB compared with those with IR and NR at the MLA site. The average P+M CSA of PR lesions was mm 2, compared with mm 2 (P, 0.001) and mm 2 (P ¼ 0.02) in the NR and IR groups, respectively. Moreover, per cent PB increased across the RI groups with values of % in NR, % in IR, and % in PR (P, 0.001). Conversely, the remodelling categories had a similar lumen CSA site and lesion length at the MLA site.

5 Coronary remodelling and plaque lipid content 825 NIRS assessment across lesion remodelling groups NIRS parameters for the various remodelling categories are listed in Table 4. PR lesions had significantly greater maxlcbi 4mm when Table 1 Baseline patient characteristics (n 5 67) Demographics and risk factors Age (years) Male 42 (62.7) Hypertension a 56 (83.6) Dyslipidaemia 52 (77.6) Diabetes mellitus b 23 (34.3) Insulin-dependent 8 (11.9) Current smoker 27 (40.3) Chronic kidney disease 10 (14.9) Peripheral artery disease 6 (9.0) History of MI 17 (25.4) History of percutaneous coronary intervention 29 (43.3) Family history 18 (26.9) Clinical diagnosis on catheterization Stable angina pectoris 14 (20.9) Silent myocardial ischaemia 14 (20.9) Unstable angina pectoris/non-st elevation MI 39 (58.2) Baseline exams (mg/dl) Creatinine Total cholesterol Low-density lipoprotein High-density lipoprotein Triglyceride Chronic kidney disease was defined as GFR,60 ml/min/1.73 m 2. MI, myocardial infarction. a Defined as blood pressure.140/90 mmhg or drug medication. b Fasting cholesterol.250 mg/dl. compared with IR and NR lesions ( vs and ; P, 0.001, respectively). Moreover, cases of PR had the highest absolute and relative length of lesion occupied by LCP when compared with IR and NR patterns. By analysis of chemogram X-axis data, PR lesions had, on average, 78% of the IVUS-defined lesion occupied by LCP compared with 67 and 53% in the IR and NR groups, respectively (P ¼ 0.002). This finding was also documented by analysis of the Y-axis data, in which the circumferential LCP angle was the highest in the PR lesions (P ¼ 0.002). The authors analysis of the block chemogram showed an additive distribution of yellow blocks (highest lipid content probability) that was significantly higher in the PR lesions compared with those with IR and NR (P ¼ 0.002). In contrast, the measurements of tan and orange colour blocks (intermediate lipid content probability) were similar among the groups. Correlation between coronary remodelling and lipid content by NIRS The correlations between LCP parameters and RI are shown in Figure 3. A significant linear correlation was found between both maxlcbi 4mm and LCBI lesion and lesion remodelling (r ¼ 0.58, P, 0.001; r ¼ 0.55, P, 0.001; respectively). Additionally, per cent LCP length (r ¼ 0.38, P, 0.001) and circumferential LCP angle (r ¼ 0.43, P, 0.001) were also significantly correlated with RI. Notably, the correlation between maxlcbi 4mm and RI remained statistically significant for both obstructive (DS 50%) and nonobstructive (DS, 50%) lesions (r ¼ 0.72 and r ¼ 0.48, P for both,0.001) (Figure 4). Lesion remodelling as a function of lipid content and PB The maxlcbi 4mm also increased with increasing PB, which was highest in the PR lesions. However, the correlation of maxlcbi 4mm with RI (r ¼ 0.58; P, 0.001) was significantly stronger when compared with plaque volume RI correlation (r ¼ 0.18; P ¼ 0.07), Table 2 Lesion characteristics NR (n 5 40) IR (n 5 32) PR (n 5 28) P-value NR vs. IR IR vs. PR NR vs. PR... Target vessel Left anterior descending 14 (35.0) 17 (53.1) 15 (53.6) Left circumflex 10 (25.0) 7 (21.9) 5 (17.9) Right coronary artery 16 (40.0) 8 (25.0) 8 (28.6) Lesion location Proximal 13 (32.5) 8 (25.0) 10 (35.7) Mid 21 (52.5) 17 (53.1) 13 (46.4) Distal 6 (15.0) 7 (21.9) 5 (17.9) QCA parameters (mm) Minimal lumen diameter Reference diameter Diameter stenosis Eccentric plaque 13 (32.5) 9 (28.1) 8 (28.6) Lesion calcification (severe) 1 (2.5) 1 (3.1) 2 (7.1) Thrombus 0 (0.0) 1 (3.1) 1 (3.6) NR, negative remodelling; IR, intermediate remodelling; PR, positive remodelling.

6 826 H. Ota et al. Table 3 Conventional IVUS measurements NR (n 5 40) IR (n 5 32) PR (n 5 28) P-value NR vs. IR IR vs. PR NR vs. PR... Minimal lumen site (mm 2 ) Lumen CSA EEM CSA Plaque + media CSA ,0.001 Per cent plaque burden (%) , ,0.001 Lesion length (mm) Reference site EEM CSA (mm 2 ) Remodelling index ,0.001,0.001,0.001,0.001 Volumetric analysis (mm 3 ) Lumen volume Vessel EEM volume Plaque volume Average lumen area (mm 3 /mm) Average vessel area (mm 3 /mm) Average plaque + media area (mm 3 /mm) Average plaque burden (%) , ,0.001 CSA, cross-sectional area; EEM, external elastic membrane. Table 4 NIRS measurements NR (n 5 40) IR (n 5 32) PR (n 5 28) P-value NR vs. IR IR vs. PR NR vs. PR... maxlcbi 4mm , ,0.001 LCBI lesion , ,0.001 LCP length in lesion Absolute LCP length (mm) Per cent LCP length Circumferential LCP angle Block chemogram Absolute length (mm) Yellow Tan Orange Red Per cent block chemogram Yellow , ,0.001 Tan Orange Red LCBI, lipid core burden index; LCP, lipid-containing plaque. average P+M area RIcorrelation(r ¼ 0.23; P ¼ 0.02), and per cent PB at the MLA site RI correlation (r ¼ 0.41; P, 0.001) (Figure 5). To further explore this association, the authors divided lesions into three groups combining MLA or.4.0 mm 2 and PB or,70%. There was a steady increase in the average maxlcbi 4mm across the three groups with lesions with MLA 4.0 mm 2 and PB 70% having a higher average LCBI compared with those with MLA.4.0 mm 2 and PB, 70% (411 vs. 198; P, 0.001). However, there was wide variation in LCBI values across these groups (Figure 6). Values of LCBI parameters diagnostic for PR A receiver-operating characteristic curve was generated to determine what LCBI level best described the occurrence of PR.

7 Coronary remodelling and plaque lipid content 827 Figure 3 Correlations between RI and chemometrics. (A) RI and maxlcbi 4mm,(B) RI and LCBI lesion,(c) RI and per cent LCP length, and (D)RI and LCP angle. RI, remodelling index. Figure 4 Correlations between RI and DS by QCA. RI with (A) obstructivelesions(ds 50%) and (B) non-obstructive (DS, 50%). RI, remodelling index; DS, diameter stenosis. The best cut-off value of maxlcbi 4mm for PR was 439 [area under the curve (AUC) ¼ 0.79, 95% confidence interval: ], translating into a sensitivity ¼ 61%, specificity ¼ 88%, positive predictive value (PPV) ¼ 65%, and negative predictive value (NPV) ¼ 85%. Conversely, the cut-off value for LCBI lesion was 137 (AUC ¼ 0.79, 95% confidence interval: ), resulting in a sensitivity ¼ 86%, specificity ¼ 63%, PPV ¼ 47%, and NPV ¼ 92% (Figure 7).

8 828 H. Ota et al. Figure 5 Correlations between RI and PB by IVUS. (A) RI with plaque volume, (B) average plaque plus media area, and (C) PB at MLA site. RI, remodelling index; IVUS, intravascular ultrasound; MLA, minimal lumen area. Discussion This study attempted to detect the relationship between coronary lesion remodelling and plaque lipid content as assessed by NIRS. There were several notable findings: (i) coronary lesion remodelling is closely associated with plaque lipid content; (ii) the correlation between lesion remodelling and maxlcbi 4mm is stronger than with PB; (iii) there is a significant gradual increase in average maxlcbi 4mm as MLA is reduced and PB increases, however, we also observed large maxlcbi 4mm in low PB lesions; and (iv) the cut-off values of maxlcbi 4mm and LCBI lesion at which PR occurs are 439 and 137, respectively. The correlation between RI and plaque lipid content The geometric alteration of coronary arteries is a core concept in evolving phases of atherosclerosis. Despite the fact that PR enables the artery to maintain its relative luminal area, numerous studies have shown that such lesion morphology is associated with worse outcomes compared with NR or IR at culprit and non-culprit sites. 3,8 At culprit sites, PR lesions are related to a higher incidence of periprocedural myocardial infarction, secondary to distal embolization presumably of large lipid particles. 16 In contrast, lesions with PR at non-culprit sites have higher rates of unanticipated cardiac events compared with IR. 3 Therefore, PR represents a higher risk phase of coronary atherosclerosis, and improving the understanding of this phase may have clinical implications as this phase can be partially reversed. IVUS studies have consistently shown that PR lesions have a higher PB compared with other remodelling morphologies, 3 5 which agrees with the authors findings and explains, in part, the poor outcomes related to these lesions. However, postmortem studies have also documented differences in plaque composition between PR and NR, such as a higher macrophage infiltration and, particularly, higher LCP. These autopsy data have been confirmed by in vivo imaging studies showing that virtually all PR lesions have lipid-rich plaque, as defined by optical coherence tomography. 4 The authors study corroborates these findings, showing a progressive linear increase in the biochemical lipid signal as the vessel expands outwards. The magnitude of the positive correlation of the in vivo lipid plaque content assessed by NIRS was comparable to both postmortem and optical coherence tomography studies. 2,4 Thus, this study suggests that the plaque lipid content may drive coronary remodelling. This finding is of clinical relevance because the combination of high lipid lesion content and RI is an important predictor of thin-cap fibroatheroma occurrence, 17 which has been suggested to be a vulnerable plaque phenotype. Although the intended role of VH-IVUS was to describe in vivo, vulnerable plaque morphologies, such as necrotic core, concerns have been raised about the accuracy of this method. 18 Fujii et al. 19 reported that PR lesions had a smaller

9 Coronary remodelling and plaque lipid content 829 necrotic core and larger fibrofatty than lesions with NR and IR. In contrast, Rodriguez-Granillo et al. 20 found a larger mean necrotic core in PR lesions with similar fibrofatty distribution. More recently, Brugaletta et al. 12 revealed that VH-IVUS identifies necrotic core with low accuracy. In fact, the discrepancy between the necrotic core data in pathology and VH-IVUS studies might result from differences in the population/lesion sample and potential problems related to in vivo VH-IVUS tissue characterization. 19 Conversely, NIRS-IVUS can overcome some of the drawbacks of VH-IVUS by the incorporation of two imaging technologies that enable in vivo plaque morphology to be assessed simultaneously with its chemical composition. 21 Indeed, the accuracy of NIRS has been shown to be superior to VH-IVUS in determining necrotic core or large lipid-rich plaques in human arteries. 14,22 This analysis was not restricted to the maxlcbi 4mm as a metric of lipid content, but also included LCBI across the entire lesion and its circumferential distribution. Pathology studies have indicated that, although PR lesions are frequently eccentric, the disease is also active at the plaque-free arc. 2 In keeping with these findings, the authors demonstrated that LCBI lesion, LCP length, and LCP angle were also significantly correlated with RI. These results may indicate that lipid-plaque progression in the entire lesion is equally as important as the concentrated maxlcbi 4mm forenlargementofthearteryinresponse to atherosclerosis. Figure 6 maxlcbi 4mm distribution according to MLA and PB. P-values are corrected for clustering effect using generalized estimating equation models. maxlcbi 4mm, maximal 4 mm lipid core burden index; MLA, minimal lumen area. Interplay among lesion remodelling, LCBI, and PB The positive correlation of RI and plaque lipid content can partially be explained by the intuitive hypothesis that PB influences the likelihood of a lipid-rich plaque and necrotic core occurrence. 23 Although the authors showed that the average maxlcbi 4mm gradually increased as PB increased and MLA decreased at the lesion site, the maxlcbi 4mm was more strongly correlated with RI than with PB. Moreover, it is important to highlight that a wide variation in LCBI values was documented; thus lesions with low PB (,70%) and high MLA (.4.0 mm 2 ) may have individual values of maxlcbi 4mm. 500, whereas lesions with high PB ( 70%) and low MLA ( 4.0 mm 2 ) may have individual values of maxlcbi 4mm, 40. This contrasts with the recent substudy from the Reduction in YEllow Plaque by Aggressive Lipid LOWering Therapy (YELLOW) trial, in which maxlcbi 4mm 500wasfoundonlyinplaqueswithPB 70%. 24 Therefore, the authors results may indicate that even lesions with low PB may harbour a high lipid pool content. Figure 7 ROC analysis for both LCBI scores to predict PR: (A) maxlcbi 4mm and (B) LCBI lesion. ROC, receiver-operating characteristic curve; maxlcbi 4mm, maximal 4 mm lipid core burden index; PR, positive remodelling.

10 830 H. Ota et al. LCBI score cut-off for positive lesion remodelling of coronary lesions The authors results demonstrate that RI appears to be more influenced by the LCBI than by the PB. Although PR lesions are associated with the worst clinical outcomes, this atherosclerotic phase can be reversed following diagnosis and consequently represents a potential target for clinical management and intervention. 24 Therefore, the authors final objective was to quantify the level of LCBI that describes the occurrence of PR that remains unknown. Indeed, a wide range of significant LCBI values has been documented using either patient- or lesion-level data. Goldstein et al. 25 revealed that lesions with maxlcbi 4mm 500 were associated with periprocedural myocardial infarction. Conversely, Oemrawsingh et al. 26 reported that patients with an LCBI 43 in non-culprit arteries had a four-fold higher rate of 1-year cardiovascular events. The authors study adds further quantitative information on this question, indicating that PR is likely to occur above maxlcbi 4mm value of 439 and LCBI lesion value of 137. Study limitations This study has several limitations. First, as a retrospective analysis, inherent selection bias may have occurred, because the use of NIRS-IVUS was not controlled. Therefore, this study might not represent the whole spectrum of coronary plaques. However, the authors did include lesions judged as obstructive that were underrepresented or excluded from NIRS studies. Secondly, long lesions, defined as 40 mm, were excluded because they posed difficulties in obtaining the RI. Thirdly, the study did not distinguish between culprit and non-culprit lesions. Although this may be viewed as a limitation, the analysis of all plaques together may provide a more informative patient risk assessment, which is ultimately an unmet clinical need to further improve coronary artery disease risk stratification. Conclusions The present study indicates that atherosclerotic coronary remodelling can be better explained by lipid plaque content than by PB. Additionally, the study suggests that a maxlcbi 4mm of 439 represents the cut-off value for PR occurrence. These results suggest that the determination of plaque lipid composition using NIRS-IVUS can further improve the stratification of plaque risk beyond PB. Conflict of interest: R.W. reports personal fees from Biotronik, personal fees from Medtronic, grants and personal fees from Astra- Zeneca, grants and personal fees from Boston Scientific, personal fees and grants from Biosensors International, personal fees from Abbott Vascular, grants from The Medicines Company, and grants from Edwards Lifesciences. R.W. is also the principal investigator of the Lipid Rich Plaque Study (InfraReDx, Inc.). References 1. Schoenhagen P, Ziada KM, Vince DG, Nissen SE, Tuzcu EM. Arterial remodeling and coronary artery disease: the concept of dilated versus obstructive coronary atherosclerosis. J Am Coll Cardiol 2001;38: Varnava AM, Mills PG, Davies MJ. Relationship between coronary artery remodeling and plaque vulnerability. Circulation 2002;105: Inaba S, Mintz GS, Farhat NZ, Fajadet J, Dudek D, Marzocchi A et al. Impact of positive and negative lesion site remodeling on clinical outcomes: insights from prospect. JACC Cardiovasc Imaging 2014;7: Raffel OC, Merchant FM, Tearney GJ, Chia S, Gauthier DD, Pomerantsev E et al. In vivo association between positive coronary artery remodelling and coronary plaque characteristics assessed by intravascular optical coherence tomography. Eur Heart J 2008;29: Hong YJ, Jeong MH, Choi YH, Song JA, Ahmed K, Lee KH et al. Positive remodeling is associated with vulnerable coronary plaque components regardless of clinical presentation: virtual histology-intravascular ultrasound analysis. Int J Cardiol 2013; 167: Maehara A, Mintz GS, Bui AB, Walter OR, Castagna MT, Canos D et al. Morphologic and angiographic features of coronary plaque rupture detected by intravascular ultrasound. J Am Coll Cardiol 2002;40: NakamuraM,NishikawaH,MukaiS,SetsudaM,NakajimaK,TamadaHet al. Impact of coronary artery remodeling on clinical presentation of coronary artery disease: an intravascular ultrasound study. J Am Coll Cardiol 2001;37: Sahara M, Kirigaya H, Oikawa Y, Yajima J, Ogasawara K, Satoh H et al. Arterial remodeling patterns before intervention predict diffuse in-stent restenosis: an intravascular ultrasound study. J Am Coll Cardiol 2003;42: Stone GW, Maehara A, Lansky AJ, de Bruyne B, Cristea E, Mintz GS et al. A prospective natural-history study of coronary atherosclerosis. NEnglJMed 2011;364: Burke AP, Kolodgie FD, Farb A, Weber D, Virmani R. Morphological predictors of arterial remodeling in coronary atherosclerosis. Circulation 2002; 105: Qian J, Maehara A, Mintz GS, Margolis MP, Biro S, Stone GW et al. Relation between individual plaque components and overall plaque burden in the prospective, multicenter virtual histology intravascular ultrasound registry. Am J Cardiol 2009; 104: Brugaletta S, Cola C, Martin-Yuste V, Vilahur G, Oriol J, Padro T et al. Qualitative and quantitative accuracy of ultrasound-based virtual histology for detection of necrotic core in human coronary arteries. Int J Cardiovasc Imaging 2014;30: Gardner CM, Tan H, Hull EL, Lisauskas JB, Sum ST, Meese TM et al. Detection of lipid core coronary plaques in autopsy specimens with a novel catheter-based near-infrared spectroscopy system. JACC Cardiovasc Imaging 2008; 1: Kang SJ, Mintz GS, Pu J, Sum ST, Madden SP, Burke AP et al. Combined IVUS and NIRS detection of fibroatheromas: histopathological validation in human coronary arteries. JACC Cardiovasc Imaging 2015;8: Mintz GS, Nissen SE, Anderson WD, Bailey SR, Erbel R, Fitzgerald PJ et al. American College of Cardiology clinical expert consensus document on standards for acquisition, measurement and reporting of intravascular ultrasound studies (IVUS). A report of the American College of Cardiology task force on clinical expert consensus documents. J Am Coll Cardiol 2001;37: Lee SY, Mintz GS, Kim SY, Hong YJ, Kim SW, Okabe T et al. Attenuated plaque detected by intravascular ultrasound: clinical, angiographic, and morphologic features and post-percutaneous coronary intervention complications in patients with acute coronary syndromes. JACC Cardiovasc Interv 2009;2: Roleder T, Kovacic JC, Ali Z, Sharma R, Cristea E, Moreno P et al. Combined NIRS and IVUS imaging detects vulnerable plaque using a single catheter system: a head-to-head comparison with OCT. EuroIntervention 2014;10: Prasad A, Cipher DJ, Mohandas A, Roesle M, Brilakis ES, Banerjee S. Reproducibility of intravascular ultrasound virtual histology analysis. Cardiovasc Revasc Med 2008;9: Fujii K, Carlier SG, Mintz GS, Wijns W, Colombo A, Bose D et al. Association of plaque characterization by intravascular ultrasound virtual histology and arterial remodeling. Am J Cardiol 2005;96: Rodriguez-Granillo GA, Serruys PW, Garcia-Garcia HM, Aoki J, Valgimigli M, van Mieghem CA et al. Coronary artery remodelling is related to plaque composition. Heart 2006;92: Pu J, Mintz GS, Brilakis ES, Banerjee S, Abdel-Karim AR, Maini B et al. In vivo characterization of coronary plaques: novel findings from comparing greyscale and virtual histology intravascular ultrasound and near-infrared spectroscopy. Eur Heart J 2012;33: Moreno PR, Lodder RA, Purushothaman KR, Charash WE, O Connor WN, Muller JE. Detection of lipid pool, thin fibrous cap, and inflammatory cells in human aortic atherosclerotic plaques by near-infrared spectroscopy. Circulation 2002;105: Brugaletta S, Garcia-Garcia HM, Serruys PW, de Boer S, Ligthart J, Gomez-Lara J et al. NIRS and IVUS for characterization of atherosclerosis in patients undergoing coronary angiography. JACC Cardiovasc Imaging 2011;4: DohiT,MaeharaA,MorenoPR,BaberU,KovacicJC,LimayeAMet al. The relationship among extent of lipid-rich plaque, lesion characteristics, and plaque

11 Coronary remodelling and plaque lipid content 831 progression/regression in patients with coronary artery disease: a serial nearinfrared spectroscopy and intravascular ultrasound study. Eur Heart J Cardiovasc Imaging 2015;16: Goldstein JA, Maini B, Dixon SR, Brilakis ES, Grines CL, Rizik DG et al. Detection of lipid-core plaques by intracoronary near-infrared spectroscopy identifies high risk of periprocedural myocardial infarction. Circ Cardiovasc Interv 2011;4: Oemrawsingh RM, Cheng JM, Garcia-Garcia HM, van Geuns RJ, de Boer SP, Simsek C et al. Near-infrared spectroscopy predicts cardiovascular outcome in patients with coronary artery disease. J Am Coll Cardiol 2014;64: IMAGE FOCUS doi: /ehjci/jew061 Online publish-ahead-of-print 3 April Large intra-atrial structure 6 years after percutaneous atrial septal defect closure Urszula Matys 1, Maciej Południewski 1 *, Krzysztof Matlak 2, Sławomir Dobrzycki 1, and Tomasz Hirnle 2 1 Department of Invasive Cardiology, Medical University of Bialystok, M.C. Sklodowskiej 24A, Bialystok , Poland; and 2 Department of Cardiosurgery, Medical University of Bialystok, Bialystok, Poland * Corresponding author. Tel: mpoludniewski@yahoo.com A 52-year-old woman after percutaneous secundum atrial septal defect (ASD) closure (Occlutech Figulla ASD device 24 mm) 6 years ago was admitted with dyspnoea (NYHA Class II) and unspecific chest pain. Physical examination, electrocardiogram, and laboratory tests were normal. Transthoracic echocardiography revealed moving structure (46 21 mm) in left atrium, slightly flopping into left ventricular inflow tract (Panel A). Valves morphology and ejection fraction were normal. Transoesophageal echocardiography presented structure of irregular and heterogeneous morphology attached to the occluder s inferior edge, with small pedicle 2 4mm(Panel B). After 2 weeks of low molecular weight heparin therapy, there was no change in structure size. It was decided to evacuate the mass with occluder with classic median sternotomy approach. Coronary arteries were assessed in CT angiography (no atherosclerosis), which confirmed the presence of extra structure with slight contrast enhancement (Panels C and D). Left atrial tumour and occluder were removed; reconstruction of the atrial septum and free wall of the right atrium were done. Surgical excision showed mm brownish, irregularly shaped, gelatinous mass attached to the endocardium, formed over the device, with small 2 5mm pedicle (Panel E). Histopathological analysis revealed a myxoma (Panels F and G). Patient was discharged in good condition remaining under follow-up of cardiology outpatients. Panels (A) TTE: mass flopping into left ventricular inflow tract, (B) TEE: mass connected with device by small pedicle, (C) CT-scan: large intra-atrial structure, (D) CT-scan: extra structure contrast enhancement, (E) post-operative image, (F) microscopic image shows socalled myxoma cell, and (G) photomicrograph demonstrates thin-walled blood vessels. Supplementary data are available at European Heart Journal Cardiovascular Imaging online. Published on behalf of the European Society of Cardiology. All rights reserved. & The Author For permissions please journals.permissions@oup.com.