intravascular polarimetry in patients with coronary artery...

J A C C : C A R D I O V A S C U L A R I M A G I N G VO L . - , N O . - , 2 0 1 9

ª 2 0 1 9 T H E A U T H O R S . P U B L I S H E D B Y E L S E V I E R O N B E H A L F O F T H E A M E R I C A N

C O L L E G E O F C A R D I O L O G Y F OU N D A T I O N . T H I S I S A N O P E N A C C E S S A R T I C L E U N D E R

T H E C C B Y - N C - N D L I C E N S E ( h t t p : / / c r e a t i v e c o mm o n s . o r g / l i c e n s e s / b y - n c - n d / 4 . 0 / ) .

ORIGINAL RESEARCH

Intravascular Polarimetry in PatientsWithCoronary Artery Disease
Kenichiro Otsuka, MD, PHD,a Martin Villiger, PHD,a Antonios Karanasos, MD, PHD,b,c
Laurens J.C. van Zandvoort, BSC,b Pallavi Doradla, PHD,a Jian Ren, PHD,a Norman Lippok, PHD,a

Joost Daemen, MD, PHD,b Roberto Diletti, MD, PHD,b Robert-Jan van Geuns, MD, PHD,b,d Felix Zijlstra, MD, PHD,b

Gijs van Soest, PHD,b Jouke Dijkstra, PHD,e Seemantini K. Nadkarni, PHD,a Evelyn Regar, MD, PHD,b,f

Brett E. Bouma, PHDa,b,g

ABSTRACT

ISS

FrobD

Ca

Nij

Le

En

Na

in

Ot

Me

OBJECTIVES The aims of this first-in-human pilot study of intravascular polarimetry were to investigate polarization

properties of coronary plaques in patients and to examine the relationship of these features with established structural

characteristics available to conventional optical frequency domain imaging (OFDI) and with clinical presentation.

BACKGROUND Polarization-sensitive OFDI measures birefringence and depolarization of tissue together with

conventional cross-sectional optical frequency domain images of subsurface microstructure.

METHODS Thirty patients undergoing polarization-sensitive OFDI (acute coronary syndrome, n ¼ 12; stable angina

pectoris, n ¼ 18) participated in this study. Three hundred forty-two cross-sectional images evenly distributed along all

imaged coronary arteries were classified into 1 of 7 plaque categories according to conventional OFDI. Polarization

features averaged over the entire intimal area of each cross section were compared among plaque types and with

structural parameters. Furthermore, the polarization properties in cross sections (n ¼ 244) of the fibrous caps of acute

coronary syndrome and stable angina pectoris culprit lesions were assessed and compared with structural features using

a generalized linear model.

RESULTS The median birefringence and depolarization showed statistically significant differences among plaque types

(p < 0.001 for both, 1-way analysis of variance). Depolarization differed significantly among individual plaque types

(p < 0.05), except between normal arteries and fibrous plaques and between fibrofatty and fibrocalcified plaques. Caps

of acute coronary syndrome lesions and ruptured caps exhibited lower birefringence than caps of stable angina pectoris

lesions (p < 0.01). In addition to clinical presentation, cap birefringence was also associated with macrophage

accumulation as assessed using normalized standard deviation.

CONCLUSIONS Intravascular polarimetry provides quantitative metrics that help characterize coronary arterial tissues

andmay offer refined insight into coronary arterial atherosclerotic lesions in patients. (J AmColl Cardiol Img 2019;-:-–-)

© 2019 The Authors. Published by Elsevier on behalf of the American College of Cardiology Foundation. This is an open

access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

N 1936-878X https://doi.org/10.1016/j.jcmg.2019.06.015

m the aWellman Center for Photomedicine, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts;

epartment of Cardiology, Thoraxcenter, Erasmus University Medical Center, Rotterdam, the Netherlands; c1st Department of

rdiology, Hippokration Hospital, University of Athens, Athens, Greece; dDepartment of Cardiology of Radboud UMC,

megen, the Netherlands; eDivision of Image Processing, Department of Radiology, Leiden University Medical Center,

iden, the Netherlands; fHeart Center, University Hospital Zurich, Zurich, Switzerland; and the gInstitute for Medical

gineering and Science, Massachusetts Institute of Technology, Cambridge, Massachusetts. This work was supported by the

tional Institutes of Health (grants P41EB-015903 and R01HL-119065) and by Terumo Corporation. Dr. Bouma was supported

part by the Professor Andries Querido visiting professorship of the Erasmus University Medical Center in Rotterdam. Dr.

suka acknowledges partial support from the Japan Heart Foundation/Bayer Yakuhin Research Grant Abroad, the Uehara

morial Foundation Postdoctoral Fellowship, and the Japan Society for the Promotion of Science Overseas Research Fellowship.

http://creativecommons.org/licenses/by-nc-nd/4.0/

https://doi.org/10.1016/j.jcmg.2019.06.015

http://creativecommons.org/licenses/by-nc-nd/4.0/

ABBR EV I A T I ON S

AND ACRONYMS

ACS = acute coronary

syndrome(s)

FC = fibrocalcified plaque

FF = fibrofatty plaque

FP = fibrous plaque

NSD = normalized standard

deviation

OCT = optical coherence

tomography

OFDI = optical frequency

domain imaging

PR = plaque rupture

PS = polarization-sensitive

SAP = stable angina pectoris

TCFA = thin-cap fibroatheroma

ThCFA = thick-cap

fibroatheroma

Massachus

Corporation

authors ha

Manuscript

Otsuka et al. J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . - , N O . - , 2 0 1 9

Polarization Properties of Human Atherosclerosis - 2 0 1 9 :- –-

2

P laque morphology and compositionhave been implicated in the patho-genesis of acute coronary syndromes

(ACS) (1,2). The high resolution of opticalcoherence tomography (OCT) and optical fre-quency domain imaging (OFDI) has enabledthe identification of several structural fea-tures of plaque instability (3–6). Despitemuch progress, prospective identification ofrupture-prone plaques, which would becrucial to stratify risk and improve patientmanagement, remains elusive (7–10). Cur-rent OCT and OFDI modalities rely onsubjective interpretation and fall short ofproviding an objective and quantitativeassessment of plaque morphology andcomposition (11–14).

Recently, we have introduced intravas-cular polarimetry by using polarization-

sensitive (PS) OFDI in combination with standardintravascular OFDI catheters (15–17). Intravascularpolarimetry complements the high-resolution imag-ing of subsurface microstructures known from OCTand OFDI with polarimetric measurements of tissuebirefringence and depolarization (15,16). Tissue withfibrillar architecture, such as interstitial collagen orlayered arrays of arterial smooth muscle cells, ex-hibits birefringence, which can serve to assesscollagen and smooth muscle cell content (16,18). De-polarization corresponds to the randomization of thedetected polarization states (19) caused by the prop-agation of light through tissue containing lipid par-ticles, macrophage accumulation, or cholesterolcrystals (16). In our previous study comparing intra-vascular polarimetry with histopathology, we showedthat tissue birefringence and depolarization provideuseful compositional information and offer advancedtissue characterization (16). The aim of the presentstudy was to investigate, for the first time, the po-larization properties of atherosclerotic plaques inpatients with coronary artery disease. We exploredhow the quantitative polarization metrics evaluatedin entire cross sections vary among different types ofplaques, classified by conventional OFDI according toestablished qualitative structural criteria (plaqueanalysis). Moreover, we compared the polarizationproperties measured locally in the fibrous caps of

etts General Hospital and the Erasmus University Medical Cente

. Drs. Bouma, van Soest, and Villiger have the right to receive roy

ve reported that they have no relationships relevant to the conte

received January 22, 2019; revised manuscript received April 24

culprit lesions between patients with ACS and stableangina pectoris (SAP) (cap analysis).

METHODS

STUDY POPULATION. This first-in-human pilot studyof intravascular polarimetry enrolled 30 nonconsec-utive patients (ACS, n ¼ 12; SAP, n ¼ 18) undergoingpercutaneous coronary intervention at Erasmus Uni-versity Medical Center between December 2014 andJuly 2015. The ethics committee at Erasmus Univer-sity Medical Center approved the protocol, and eachpatient gave written informed consent before inclu-sion in the study, which was conducted in compliancewith the protocol and the Declaration of Helsinki. PSOFDI was performed using commercial intravascularcatheters (FastView, Terumo, Tokyo, Japan) witha custom-built PS OFDI system (SupplementalAppendix), as previously described (15–17). Imagingin the 30 patients yielded a total of 36 pull backs,performed either before the procedure (n ¼ 15; culpritor target vessel, n ¼ 13; nonculprit or nontargetvessel, n ¼ 2) or after the procedure (n ¼ 21). Patientcharacteristics are summarized in Table 1.

STUDY DESIGN. The present study comprises a pla-que analysis evaluating entire cross sections evenlydistributed along all imaged coronary arteries and acap analysis focusing solely on the fibrous caps ofculprit lesions (Figure 1A). For the plaque analysis, allpull backs were uniformly divided into segments of5 mm in length, progressing distally and proximallyfrom the culprit lesion (Figure 1B). From the resultingtotal of 508 segments, comprising 2,540-mm pullback length, 166 segments were excluded from theanalysis because of the following criteria: 1) contain-ing a stent or from within 5 mm proximal or distal toan implanted stent (n ¼ 150); 2) subject to pre-dilatation (n ¼ 2); and 3) poor image quality due toinsufficient blood clearing (n ¼ 14). In each of theretained 342 5-mm segments, only the cross sectionwith the smallest luminal area was identified forfurther analysis. If this cross section contained a sidebranch of diameter >1.5 mm, it was replaced withanother section from the same segment. For seg-ments containing fibroatheromas, instead of thesmallest luminal area, the cross-section with thethinnest fibrous cap thickness was used for analysis.

r have patent licensing arrangements with Terumo

alties as part of the licensing arrangements. All other

nts of this paper to disclose.

, 2019, accepted June 11, 2019.



TABLE 1 Patient and Optical Frequency Domain Imaging Lesion

Characteristics (n ¼ 30)

Patient characteristics

Age, yrs 63 � 9

Male 24 (80)

Hypertension 12 (40)

Diabetes mellitus 5 (17)

Dyslipidemia 13 (43)

STEMI 1 (3.3)

NSTEMI 5 (17)

Unstable angina 6 (20)

Previous myocardial infarction 8 (27)

Previous PCI 14 (47)

Previous CABG 4 (13)

Imaged vessel (n ¼ 36)

LAD 18 (50)

LCx 6 (17)

RCA 12 (33)

OFDI lesion characteristics (n ¼ 342)

Analyzed frames 342

Luminal area, mm2 6.1 � 3.3

Percentage area stenosis 40 � 20

Plaque morphology

AIT 31 (9.1)

FP 84 (25)

FF 45 (13)

FC 81 (24)

ThCFA 88 (26)

TCFA 11 (3.2)

PR 2 (0.6)

Lipid arc, degrees 57 � 72

Calcium arc, degrees 73 � 50

Minimum fibrous cap thickness, mm 140 � 81

Polarization properties

Birefringence, �10�3 0.50 � 0.11

Depolarization 0.12 � 0.09

Values are mean � SD, n (%), or n.

ACS ¼ acute coronary syndrome; AIT ¼ adaptive intimal thickening;CABG ¼ coronary artery bypass graft; CAD ¼ coronary artery disease;FC ¼ fibrocalcified plaque; FF ¼ fibrofatty plaque; FP ¼ fibrous plaque; LAD ¼ leftanterior descending coronary artery; LCx ¼ left circumflex coronary artery;NSTEMI ¼ non–ST-segment elevation myocardial infarction; OFDI ¼ optical fre-quency domain imaging; PCI ¼ percutaneous coronary intervention; PR ¼ plaquerupture; RCA ¼ right coronary artery; SAP ¼ stable angina pectoris; STEMI ¼ ST-segment elevation myocardial infarction; TCFA ¼ thin-cap fibroatheroma;ThCFA ¼ thick-cap fibroatheroma.

J A C C : C A R D I O V A S C U L A R I M A G I N G , V O L . - , N O . - , 2 0 1 9 Otsuka et al.- 2 0 1 9 :- –- Polarization Properties of Human Atherosclerosis

3

For the cap analysis, we identified the culpritlesion in patients with ACS (n ¼ 4) and SAP (n ¼ 9)who underwent PS OFDI prior to percutaneous coro-nary intervention (Figure 1A). ACS culprit lesions wereidentified on the basis of invasive coronary angiog-raphy, electrocardiographic ST-segment alterations,and/or regional wall motion abnormality on echo-cardiographic assessment. SAP target lesions weredetermined on the basis of left ventricular wall mo-tion abnormalities, nuclear scan, stress test, andcoronary angiographic findings. For the cap analysis,

we identified the cross section with the smallestluminal area in the culprit or target lesion of eachpatient (Figure 1B). Every other cross-sectional imageup to 5 mm both proximally and distally, if featuring alipid arc exceeding >90�, was included into theanalysis, resulting in a total of 244 cross sections.

CONVENTIONAL OFDI ANALYSIS. Two independentinvestigators (A.K. and L.J.C.Z.) analyzed the con-ventional OFDI appearance of the selected imagesusing QCU-CMS viewing software (Leiden UniversityMedical Center, Leiden, the Netherlands). Conven-tional OFDI analysis was performed blinded to thepolarimetric signals. Luminal area was measured inall cross sections. Percentage area stenosis wascalculated as previously reported (8), taking the meanof the largest lumen within 5 mm proximal and distalto the lesion containing the current cross section asthe reference.

Each selected cross-section was then categorizedas either of normal artery, fibrous plaque (FP), fibro-fatty plaque (FF), fibrocalcified plaque (FC), thick-capfibroatheroma (ThCFA), thin-cap fibroatheroma(TCFA), or plaque rupture (PR), on the basis of theconventional OFDI signal (Figures 2A1 to 2G1). Briefly,a vessel with a tunica intima thinner or similar inthickness to the tunica media was labeled as a normalartery (Figure 2A1). FP was defined as a plaque withhigh backscattering and relatively homogeneousOFDI signal (Figure 2B1). Plaques with calcium, whichappears as a signal-poor or heterogeneous region witha sharply delineated border within fibrous tissue,were classified as FC (Figure 2D1) (9). A plaque with alipid arc of more than 90� was defined as a fibroa-theroma. A lipid-rich plaque with a lipid arcextending <90� was categorized as FF, to appreciatethe optical properties of the small lipid-rich area(Figure 2C1). In fibroatheromas, the fibrous capthickness was measured around its thinnest part 3times by each observer, and then the averaged valuewas calculated. If the fibrous cap thicknesswas <65 mm, the plaque was categorized as TCFA andas ThCFA otherwise. PR was defined as a plaquefeaturing intimal disruption and cavity formation(Figures 2E1 to 2G1). In cases of discordance betweenthe observers, a third investigator (B.E.B.) acted asreferee to achieve consensus classification, and themajority was used as the final plaque classification.

QUANTITATIVE BIREFRINGENCE AND DEPOLARIZATION

ANALYSIS. PS OFDI analysis was performed at Mas-sachusetts General Hospital blinded to the conven-tional OFDI measurements and clinical information.Coregistration between conventional OFDI andpolarimetric signals is intrinsic, because they are

FIGURE 1 Study Population, Design, and Lesion Selection

30 patients underwent PS-OFDI at Erasmus University Medical Center(ACS; n = 12 and SAP; n = 18)

A

B

Selection of cross-sectional imageswithin each 5-mm segment

Selection of cross-sectional images inculprit/target lesion

• The smallest lumen area • Starting from the smallest lumen area• Every 2 frames• Both proximally and distally up to 5 mm, as long as the lipid arc exceeded >90°

• The thinnest fibrous cap

342 cross-sectional images from 36 pullbacks(ACS: n = 12 and SAP; n = 18)

Plaque analysis244 cross-sectional images from 13 culprit/target

lesions (ACS/PR; n = 5 and SAP; n = 8)

Cap analysis

or

36 pullbacks from 30 patientsPullbacks in de novo arteries (culprit/target lesions; n = 13 and non-culprit/target lesions; n = 2)

Pullbacks after procedures (culprit/target lesions; n = 21)

Plaque analysis

Cap analysis

5 mm

(i) (h) (g) (f) (e) (a) (b) (c) (d)

(i)

1 mm

Segment 8 Segment 7 Segment 6

Distal

Segment 5 Segment 4 Segment

Culprit/target

Segment 1 Segment 2

Proximal

Intensity

Bire

frin

genc

e(Δ

n)

Segment 3

(h) (g) (f) (e) (a) (b) (c) (d)

Analyzed lesionfor cap analysis

(A) Study population and data analysis. (B) Polarization-sensitive (PS) optical frequency domain imaging (OFDI) pull backs were divided into

5-mm segments, centered on the culprit segment for the plaque analysis. Red broken lines indicate the selected cross-sectional images

within each 5-mm segment, corresponding to the cross section with the smallest luminal area. The cap analysis focused on the culprit or target

segment centered around the cross section with the smallest luminal area in the culprit lesion of each patient. Cross-sectional images were

analyzed every 2 frames both proximally and distally up to 5 mm, as long as the lipid arc exceeded >90�. Scale bars measure 1 mm for cross-

sectional images (white) and 5 mm for the longitudinal image (black), respectively. ACS ¼ acute coronary syndrome; PR ¼ plaque rupture;

SAP ¼ stable angina pectoris.



4

computed from the exact same raw data. For theplaque analysis, we segmented the intimal layer ofeach cross section using custom-written software inMATLAB (The MathWorks, Natick, Massachusetts).In addition to the lumen, we segmented theinternal elastic lamina, whenever visible, using the

birefringence map to leverage from the improvedvisibility of the media in the birefringence map (16).In areas in which the internal elastic lamina seg-mentation was unattainable, typically in lipid-richareas of advanced lesions, an automatic outerborder corresponding to a tissue depth of 1 mm from

FIGURE 2 Representative Polarization-Sensitive Optical Frequency Domain Imaging for All Plaque Types

Intensity

(A1) Intensity

Normal artery Thick-cap fibroatheroma (ThCFA)

Thin-cap fibroatheroma (TCFA)

Plaque rupture (PR)

Intensity

Intensity

DepolarizationBirefringence(Δn)

Fibrous plaque (FP)

Fibro-fatty plaque (FF)

Fibro-calcified plaque (FC)

(B1)

(C1)

(D1) (D2) (D3)

(C2) (C3)

(B2) (B3)

(E1)

(F1)

(G1) (G2) (G3)

(F2) (F3)

(E2) (E3)(A2) Birefringence (A3) Depolarization

(A1 to G1) Typical intensity images, (A2 to G2) birefringence images, and (A3 to G3) depolarization images of all plaque types. Rows show representative images of

normal artery (A1 to A3), fibrous plaque (FP) (B1 to B3), fibrofatty plaque (FF) (C1 to C3), fibrocalcified plaque (FC) (D1 to D3), thick-cap fibroatheroma (ThCFA)

(E1 to E3), thin-cap fibroatheroma (TCFA) (F1 to F3), and plaque rupture (PR) (G1 to G3), respectively. (A2,A3) Normal artery feature very low birefringence and low

depolarization in the intima. The media features high birefringence, independent of the lesion type, and is clearly visible in normal arteries. (B2,B3) Fibrous tissue in FP

exhibits heterogeneous patterns of birefringence and relatively homogeneous, low depolarization. (C2,C3) Fibrous regions covering the lipid pools of FF generally

present with high birefringence. The lipid-rich areas cause pronounced depolarization. (D2,D3) Calcifications in FC show rather low birefringence and moderate

depolarization. Depolarization of calcifications increases with the presence of lipid. (E2,E3) Fibrous caps of ThCFA exhibit heterogeneous birefringence. (F2,F3) Fibrous

caps of TCFA typically reveal low birefringence. Depolarization in the diffuse border of the cap promptly transitions from low to high. (G2,G3) The caps of PR have low

birefringence. The most intimal layer from 3 o’clock to 9 o’clock in the intensity image displays relatively high and heterogeneous birefringence compared with the

fragments of white thrombus at 1 o’clock and 11 o’clock. White bar ¼ 1 mm.


5

the luminal surface was used. We also segmentedcalcifications and areas of thrombus. To compute theaverage birefringence of cross sections, we evaluatedthe median of the birefringence in the area boundedby the lumen and the internal elastic lamina or outerborder segmentation, excluding the guidewireshadow, and featuring a depolarization of #0.2 aspreviously reported (17). The median depolarizationwas computed within the entire segmented area aftermasking the guidewire shadow.

For the cap analysis, the border between thefibrous caps of the culprit fibroatheromas andthe underlying lipid or necrotic core was drawn

manually, together with the lipid arc angle extendingfrom the center of the lumen (total cap analysis).Mean and thinnest fibrous cap thickness were auto-matically calculated from the segmented fibrous capin MATLAB. Furthermore, to investigate the featuresof fibrous caps at their thinnest part (focal cap anal-ysis), we also defined a narrower arc angle, centeredon the thinnest part of each cap (29� on average). Inaddition, we evaluated the normalized standard de-viation (NSD) within the fibrous caps, which has beenshown to correlate with macrophage infiltration(3,12). NSD was computed by first evaluating thestandard deviation of the linear-scale backscatter

FIGURE 3 Plaque Characterization and Polarization Properties

Normal <0.001 <0.001

1.00 0.372

1.00 1.00 1.00

1.001.00

1.00

1.00

1.00

1.00

1.00

0.011 0.572 0.915

<0.001 <0.001 0.124 1.00

FP

FP

FF

FFP values for birefringence between plaque types

FC

FC

ThCFA

ThCFA

TCFA

TCFA PR<0.001 <0.001

<0.001

<0.001 <0.001 <0.001 <0.001

<0.001

<0.001

<0.001

<0.001

<0.001

<0.001

1.00 0.056

0.982

<0.001

<0.001

<0.001

<0.001

1.00

P values for depolarization between plaque types

NormalFPFFFC

ThCFATCFA

FP FF FC ThCFA TCFA PR

P < 0.001 P < 0.05 P ≥ 0.05

1.0A B×10–3

0.8

0.6

0.4

Bire

frin

genc

e (Δ

n)

0.2

0.0

Normaln = 31

FPn = 84

FFn = 45

FCn = 81

ThCFAn = 88

One-way ANOVA: P < 0.001

TCFAn = 11

PRn = 2

0.6

Depo

lariz

atio

n

0.5

0.4

0.3

0.2

0.1

0.0

Normaln = 31

FPn = 84

FFn = 45

*

FCn = 81

ThCFAn = 88

One-way ANOVA: P < 0.001

TCFAn = 11

PRn = 2

(A,B) Significant differences in median birefringence and depolarization were observed among the 7 plaque types (p < 0.001 for both, 1-way

analysis of variance). (A) Fibrous plaques (FPs) featured the highest birefringence, followed by fibrofatty plaques (FFs), fibrocalcified plaques

(FCs), thick-cap fibroatheromas (ThCFAs), thin-cap fibroatheromas (TCFAs), and plaque ruptures (PRs) in decreasing order, but without

statistical significance between most categories. (B) A significant gradual variation in depolarization was observed among plaque types

(p < 0.001, p < 0.05 for FP vs. FF, ThCFA vs. TCFA, and TCFA vs. PR) except for normal artery versus FP and FF versus FC. “Normal”

indicates normal artery.



6

intensity data within elliptical regions of interest,extending by 80 mm in depth and 12� in the circum-ferential direction and moved across the entire caparea. These values were then normalized with thedifference between the maximum and the minimumintensity within each fibrous cap.

STATISTICAL ANALYSIS. Continuous outcome mea-sures are reported as mean � SD. For the plaqueanalysis, the mean of the birefringence and of thelogarithm of the depolarization values across thedifferent plaque types were compared using 1-wayanalysis of variance. Pairwise comparison was per-formed using the Bonferroni correction if the overalltest was significant. For comparison with clinicalpresentation in the cap analysis, the 1 PR lesion in theSAP group was classified together with the ACS le-sions into an ACS and/or PR group. Differences inthe means between the 2 groups were analyzedusing a univariate generalized linear model using a

generalized estimating equation to take into consid-eration the intrasubject correlations among multiplecross-sectional images from individual patients. Therelationships between polarization properties andclinical and conventional OFDI parameters weredetermined using a univariate generalized linearmodel using a generalized estimating equation. Betavalues and 95% confidence intervals for birefringenceare given in units of �10�3 throughout the paper. A pvalue of <0.05 was considered to indicate statisticalsignificance, and all tests were 2 sided. SPSS version22.0 was used for all analyses (IBM, Armonk, NewYork).

RESULTS

POLARIZATION PROPERTIES OF INDIVIDUAL

PLAQUE TYPES. Table 1 summarizes patient charac-teristics and OFDI parameters for the overall studypopulation. The selected 342 cross-sectional images

TABLE 2 Optical Frequency Domain Imaging and Polarimetry Lesion Characteristics in

Culprit or Target Lesion

ACS and/or PR Group(n ¼ 5)

SAP Group(n ¼ 8) p Value

Analyzed frames 88 156

OFDI lesion characteristics

Luminal area, mm2 3.7 � 2.1 3.8 � 2.2 0.871

Percentage area stenosis 56 � 23 55 � 22 0.934

Lipid arc, degrees 219 � 88 171 � 72 0.045

Calcium arc, degrees 95 � 36 100 � 62 0.951

Minimum fibrous cap thickness, mm 219 � 109 291 � 146 0.027

Mean fibrous cap thickness, mm 377 � 124 406 � 125 0.506

NSD in total cap 2.0 � 1.5 2.2 � 1.6 0.596

NSD in focal cap 4.4 � 3.2 3.9 � 3.2 0.533

Polarization properties

Birefringence (plaque), �10�3 0.44 � 0.07 0.52 � 0.08 0.001

Birefringence (total cap), �10�3 0.42 � 0.10 0.53 � 0.16 0.002

Birefringence (focal cap), �10�3 0.38 � 0.10 0.49 � 0.18 <0.001

Depolarization (plaque) 0.28 � 0.12 0.21 � 0.12 0.034

Depolarization (total cap) 0.10 � 0.04 0.10 � 0.03 0.772

Depolarization (focal cap) 0.11 � 0.05 0.11 � 0.04 0.845

Values are n or mean � SD. P values were obtained by generalized liner model using generalized estimatingequation.

NSD ¼ normalized standard deviation; other abbreviations as in Table 1.


7

representing all imaged vessels were classified intonormal artery (n ¼ 31), FP (n ¼ 84), FF (n ¼ 45), FC(n ¼ 81), ThCFA (n ¼ 88), TCFA (n ¼ 11), and PR (n ¼ 2)on the basis of conventional OFDI. Computing theweighted k coefficient for multiple categories, excel-lent intra- and interobserver agreement was observedfor the 7 plaque categories (k ¼ 0.98 and k ¼ 0.97,respectively). Figure 3 shows significant differencesin median birefringence and depolarization amongthe 7 plaque types (p < 0.001 for both, 1-way analysisof variance). Comparing individual plaque types,normal arteries were significantly less birefringentthan all other plaque types (p < 0.01), except for TCFA(p ¼ 0.124) and PR (p ¼ 1.00). FPs featured the highestbirefringence, followed by FFs, FCs, ThCFAs,TCFAs, and PRs in decreasing order, but withoutstatistical significance between individual categories(Figure 3A). Normal arteries also featured the lowestdepolarization. PR showed the highest depolarizationamong the 7 plaque types. Except FF versus FC,ThCFA versus TCFA, ThCFA versus PR, and TCFAversus PR, all plaque types had statistically signifi-cant differences in depolarization when comparedindividually (Figure 3B).

Calcifications featured low birefringence and lowdepolarization (Supplemental Figures S1A and S1B).Calcifications located in fibrous tissue exhibitedlower birefringence and depolarization than those inlipid-rich lesions (p < 0.001 for both), as shown inSupplemental Figures S1C and S1D. Thrombus (whitethrombus) presented very low birefringence and de-polarization (Supplemental Figures S2A and S2B).

POLARIZATION PROPERTIES OF FIBROUS CAPS IN

CULPRIT LESIONS. The lipid arc and minimumfibrous cap thickness differed with statistical signifi-cance between the ACS and/or PR and the SAP groups,as shown in Table 2. When comparing the polari-metric signals of the fibrous caps in the culpritlesions, we found lower birefringence in the ACSand/or PR group than in patients with SAP(p ¼ 0.002), but comparable depolarization (p ¼ 0.772)(Figures 4A and 4B).

Table 3 shows generalized estimating equationmodel parameters for the relationships among po-larization properties, clinical presentation, and con-ventional OFDI parameters, when analyzing theentire cap. Birefringence of the total cap was nega-tively correlated with ACS and/or PR (b ¼ �0.156;p ¼ 0.002) and NSD (b ¼ �0.021; p ¼ 0.013). Depo-larization of the total cap was negatively associatedwith mean fibrous cap thickness (b ¼ �0.001;p < 0.001) and positively with lipid arc (b ¼ 0.001;p ¼ 0.001) and tended to be correlated with NSD

(b ¼ 0.005; p ¼ 0.062). Furthermore, we analyzednarrow regions of interest centered on the thinnestpart of each fibrous cap (Table 4). The generalizedlinear model using generalized estimating equationshowed that ACS and/or PR (b ¼ �0.138; p < 0.001),thinnest fibrous cap thickness (b ¼ 0.005; p ¼ 0.001),and NSD (b ¼ �0.012; p ¼ 0.001) were associated withbirefringence. Factors associated with depolarizationwere thinnest fibrous cap thickness (b ¼ �0.001;p ¼ 0.005) and NSD (b ¼ 0.004; p < 0.001).

DISCUSSION

This first-in-human pilot study of intravascularpolarimetry demonstrates how it augments conven-tional OFDI plaque characterization with quantitativepolarization properties, measured through standardintravascular OFDI catheters simultaneously withthe conventional OFDI signal. The polarization fea-tures offer refined insight into tissue composition,consistent with our current understanding of themechanisms involved in plaque progression anddestabilization. The major finding of this study is thatfibrous caps in ACS culprit lesions and ruptured pla-ques exhibit lower birefringence compared with thecaps of target lesions in SAP patients within ourlimited study cohort. Compared with the interpreta-tion of conventional OFDI, which relies on subjectiveidentification of qualitative features, polarimetry




FIGURE 4 Polarization Properties in Caps of Culprit and Target Lesions

1.2

A×10–3

1.0

0.8

0.6

0.4

0.2

0.0

ACS/PR

P = 0.002

SAP

Bire

frin

genc

e (Δ

n)

B0.25

0.20

0.15

0.10

0.05

0.00

ACS/PR

P = 0.772

SAPDe

pola

rizat

ion

Total Cap Analysis

(A,B) Difference in birefringence (A) and depolarization (B) between patients with acute coronary syndrome (ACS) and or plaque rupture (PR) (88 cross

sections from 5 culprit lesions) and stable angina pectoris (SAP) (156 cross sections from 8 target lesions). Fibrous caps in patients with ACS and/or PR

exhibited significantly lower birefringence compared with those with SAP.

TABLE 3 Relationsh

Age (per 5-yr increase)

ACS and/or PR

Fibrous cap thickness,

Lipid arc (per 10� incre

Percentage area stenos

Calcium arc (per 10� in

NSD

The relationships betweenequation to take into cons

CI ¼ confidence interval



8

offers quantitative metrics, leading the way towardobjective and automated characterization of athero-sclerotic plaques, which may facilitate the use ofintravascular imaging in clinical practice. Theimproved assessment of plaque composition affordedby the polarization features may provide novelinsight into the mechanism of plaque progressionand instability in human coronary atherosclerosis(Central Illustration).

POLARIMETRIC PLAQUE CHARACTERIZATION.

Smooth muscle cells and collagen are known to in-fluence the polarization of near infrared light (18).

ips Among Polarization Properties in Total Fibrous Cap, Clinical Presenta

Birefringence

b p Value

95%

Lower

0.003 0.813 �0.019

�0.156 0.002 �0.254

mean (per 10-mm increase) 0.002 0.264 �0.001

ase) �0.004 0.109 �0.008

is (per 10% increase) 0.006 0.643 �0.018

crease) 0.003 0.339 �0.004

�0.021 0.013 �0.037

polarization properties and clinical and optical frequency domain imaging parameters wereideration the intrasubject correlations among multiple cross-sectional images from individua

; and other abbreviations as in Tables 1 and 2.

Our recent study of intravascular PS OFDI in cadav-eric human hearts revealed that normal intimal tissueexhibits low birefringence compared with intimaltissue in fibrous, early, and advanced atheroscleroticlesions (16). In the present study, we observed sig-nificant differences in polarization properties, espe-cially in depolarization, among all plaque types,despite analyzing the entire manually segmentedintimal layer, comprising both the angle subtendedby the plaque and the remainder of the cross section.Depolarization featured significant differenceseven among individual plaque types, whereas bire-fringence was less distinguishing. These observations

tion, and Optical Frequency Domain Imaging Parameters

Total Cap Analysis

Depolarization

CI

b p Value

95% CI

Upper Lower Upper

0.024 0.001 0.782 �0.007 0.009

�0.058 0.003 0.772 �0.014 0.019

0.005 �0.001 <0.001 �0.002 �0.001

0.001 0.001 0.001 0.0005 0.0021

0.030 0.001 0.647 �0.002 0.004

0.011 0.001 0.477 �0.001 0.003

�0.004 0.005 0.062 �0.0002 0.0092

determined by unadjusted generalized linear model using a generalized estimatingl patients. Beta values and 95% CIs for birefringence are given in units of �10�3.

TABLE 4 Associations of Polarization Properties in Focal Fibrous Cap and Optical Frequency Domain Imaging Parameters

Focal Cap Analysis

Birefringence Depolarization

b p Value

95% CI

b p Value

95% CI

Lower Upper Lower Upper

ACS and/or PR �0.138 <0.001 �0.211 �0.065 0.002 0.845 �0.022 0.027

Fibrous cap thickness, thinnest(per 10-mm increase)

0.005 0.001 0.002 0.008 �0.001 0.005 �0.0016 �0.0003

NSD �0.012 0.001 �0.020 �0.005 0.004 <0.001 0.002 0.006

The relationships between polarization properties and clinical and optical frequency domain imaging parameters were determined by unadjusted generalized linear model usinga generalized estimating equation to take into consideration the intrasubject correlations among multiple cross-sectional images from individual patients. Beta values and95% CIs for birefringence are given in units of �10�3.

Abbreviations as in Tables 1 to 3.


9

suggest that PS OFDI provides insight into biologicalaspects of plaque progression and destabilization,complementary to the structural information avail-able to conventional OFDI. Combined analysis ofbirefringence and depolarization in more refinedautomatically segmented regions of interest mayoffer automated tissue characterization. Moreover,accurate diagnosis of TCFA by OCT and OFDI remainschallenging (13), and future studies are warranted to

CENTRAL ILLUSTRATION Polarization Prope

Natural history Atherogenesis

Fibrous cap

Plaquemorphology

Birefringence

Depolarization

Thick collagen and sm

Otsuka, K. et al. J Am Coll Cardiol Img. 2019;-(-):-–-.

In a less advanced stage of atherogenesis, birefringence and depolariza

content (atherogenesis and progression). Birefringence of the plaques d

lipid/necrotic core, which in turn leads to a significant increase in depolar

and fibrofatty plaque [FF]) to a fibroatheroma (progression and destabi

TCFA ¼ thin-cap fibroatheroma; ThCFA ¼ thick-cap fibroatheroma.

investigate the polarization features of thin and thickfibrous caps and early and late necrotic cores indetail.POLARIZATION PROPERTIES IN FIBROUS CAPS.

Within our limited patient cohort, we observed thatthe fibrous cap of the culprit lesion in patients withACS and/or PR exhibited significantly lowerbirefringence than in the caps of target lesions inthose with SAP. Fibrillar collagen is the primary

rties in Plaque Progression and Destabilization

Progression and destabilization ACS

ooth muscle cells

Lipid/necrotic core and inflammation in fibrous cap

High

Low

tion increase along with the proliferation of thick collagen, smooth muscle cells, and lipid

eclines in hand with the reduction of interstitial collagen, and the development of the

ization, corresponding to the transition of pathological intimal thickening (fibrous plaque [FP]

lization). ACS ¼ acute coronary syndrome; FC ¼ fibrocalcified plaque; PR ¼ plaque rupture;



10

extracellular matrix molecule imparting both bire-fringence and mechanical stability to the fibrous capoverlying an atheroma (20). Histopathologic studiesshowed that ruptured fibrous caps lack layeredsmooth muscle cells and feature different collagenphenotypes than intact caps (20), which offers anexplanation for the low birefringence observed inthe fibrous caps of patients with ACS and/or PR. Inaddition to cap birefringence, minimum fibrous capthickness and lipid arc angle also featured statisti-cally significant differences between these 2 patientgroups. Yet fibrous cap thickness alone is insufficientto identify caps that are prone to rupture (4,5,8,9).The polarization metrics available to intravascularpolarimetry complement these structural featuresand may advance our understanding of thepathogenesis of ACS with or without fibrous caprupture (10,21).

Furthermore, birefringence and depolarization ofthe fibrous cap were associated with increased NSD,suggesting macrophage accumulation. Inflammationis a known mechanism of plaque destabilization(2,22,23). Macrophages release enzymes includingmatrix metalloproteinases that destroy the extracel-lular matrix and weaken the cap (22). Our observa-tions suggest that the presence of active macrophagesmay increase depolarization, whereas the effect oftheir presence causes a reduction in birefringence.The physical mechanism inducing depolarization inatherosclerosis remains to be elucidated. We specu-late that lipid droplets exceeding the size of thewavelength used for OFDI and small cholesterolcrystals, which have been postulated to be a crucialfactor in the initiation of inflammatory response inatherosclerosis (24), are the origin of the observeddepolarization.

Consistent with our previous cadaveric humanheart study of intravascular polarimetry (16), weobserved that calcifications located in fibrous tissueexhibited lower birefringence and depolarizationthan those in lipid-rich lesions. Although the OFDIappearance of microcalcifications has not beenestablished (6), we speculate that their presence infibrous caps contributes to a reduction of the capbirefringence, together with a lack of collagen orga-nization and content and absence of smooth musclecells. The few occurrences of white thrombus in thepresent study had minimal impact on the underlyingpolarization signatures. However, the strongerattenuation of red thrombus may lead to an apparentincrease in depolarization that would impair inter-pretation of the underlying vessel wall, similar toconventional OFDI.

STUDY LIMITATIONS. First, our study consisted of asmall number of patients from a cohort of noncon-secutive patients and was cross-sectional in design.Considering potential selection bias of patients, ourfindings should be interpreted with caution, althoughwe included all patients imaged with intravascularpolarimetry. Furthermore, the limited number mayhave led to a potential overestimation of the differ-ences between polarization features when perform-ing multiple pairwise comparisons and preventedmultivariate analysis. Because of its compatibilitywith clinical OFDI catheters, intravascular polarim-etry can be readily applied to any patient eligible forconventional OFDI, and future well-designed studiesare needed to ascertain our current findings.

Second, our plaque classification corresponds wellto the current understanding of plaque subtypes inOFDI (6) and could be readily performed on the con-ventional OFDI signal, yet conventional OFDI haslimited ability to classify lesion types, especially inadvanced lesions (11,14). Although lipid-rich plaquesoffer some prognostic implication (8), clear identifi-cation of fibroatheromas with OCT or OFDI remainsdebatable. Combination of intravascular ultrasoundand OCT or OFDI has been shown to improve diag-nostic accuracy of identifying fibroatheromas (13).Future histopathologic validation studies are neededto inspect the ability of polarization properties todistinguish between different plaque types andpotentially enable automated plaque classification.

Third, we observed that the NSD was associatedwith depolarization in the focal fibrous caps but not inthe entire caps. Macrophage infiltration frequentlyoccurs very locally, and averaging the signals acrossentire caps carries less meaning than across a localregion of interest. Moreover, the required normali-zation of the NSD depends delicately on the ROI, andevidence supporting the NSD metric remains scant.We suspect that both NSD and depolarization captureaspects of macrophage infiltration, but validationstudies with histology are needed to identify howdepolarization relates to macrophage infiltration.

Finally, no patient in the present study had plaqueerosion, the second most common cause of ACS.

CONCLUSIONS

This study presents polarization properties of coro-nary atherosclerotic lesions in patients with coronaryartery disease. The fibrous caps of culprit lesions inpatients with ACS and/or PR featured lower birefrin-gence than in those with SAP. When averaged acrossentire cross sections, the polarimetric signals varied

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: Modification of

the OFDI apparatus along with recently developed image

reconstruction methods enabled measurements of polarization

properties of the coronary arterial wall. Intravascular polarimetry

permits the quantitative assessment of polarization properties

using standard intravascular OFDI catheters simultaneously with

intensity image cross sections. This first-in-human pilot study of

intravascular polarimetry demonstrated that polarization

properties differ between culprit lesions of ACS or PR and SAP

and also among different morphological plaque subtypes.

TRANSLATIONAL OUTLOOK: Quantitative assessment of

plaque structure and composition by intravascular polarimetry

may open new avenues for studying coronary atherosclerosis and

may facilitate the automated interpretation of lesion

characteristics during percutaneous coronary intervention in

patients. Future studies are needed to investigate how

intravascular polarimetry may improve invasive and

pharmacological interventions in patients with coronary

artery disease.


11

among distinct morphological plaque subtypes.Quantitative assessment of plaque polarizationproperties by intravascular polarimetry may opennew avenues for studying plaque composition anddetecting high-risk patients. Prospective studies inlarger populations are needed to evaluate how po-larization metrics could translate into improved pa-tient outcomes and optimized medical therapycompared with using only the structural featuresavailable to conventional OFDI.

ACKNOWLEDGEMENTS The authors sincerely thankDr. Peter Libby of Brigham and Women’s Hospital andHarvard Medical School for fruitful discussion andfeedback on this paper. Dr. Hang Lee, of the Biosta-tistics Center, Massachusetts General Hospital, Har-vard Medical School, provided invaluable guidancewith respect to the statistical analyses in this work.

ADDRESS FOR CORRESPONDENCE: Dr. Brett E.Bouma, Wellman Center for Photomedicine, Massa-chusetts General Hospital, Harvard Medical School,50 Blossom Street, Boston, Massachusetts 02114.E-mail: [email protected].

RE F E RENCE S

1. Virmani R, Burke AP, Farb A, Kolodgie FD.Pathology of the vulnerable plaque. J Am CollCardiol 2006;47:C13–8.

2. Narula J, Nakano M, Virmani R, et al. Histo-pathologic characteristics of atherosclerotic cor-onary disease and implications of the findings forthe invasive and noninvasive detection ofvulnerable plaques. J Am Coll Cardiol 2013;61:1041–51.

3. Tearney GJ, Yabushita H, Houser SL, et al.Quantification of macrophage content in athero-sclerotic plaques by optical coherence tomogra-phy. Circulation 2003;107:113–9.

4. Toutouzas K, Karanasos A, Tsiamis E, et al. Newinsights by optical coherence tomography into thedifferences and similarities of culprit rupturedplaque morphology in non-ST-elevation myocar-dial infarction and ST-elevation myocardialinfarction. Am Heart J 2011;161:1192–9.

5. Yonetsu T, Kakuta T, Lee T, et al. In vivo criticalfibrous cap thickness for rupture-prone coronaryplaques assessed by optical coherence tomogra-phy. Eur Heart J 2011;32:1251–9.

6. Tearney GJ, Regar E, Akasaka T, et al.Consensus standards for acquisition, measure-ment, and reporting of intravascular opticalcoherence tomography studies: a report from theInternational Working Group for IntravascularOptical Coherence Tomography Standardizationand Validation. J Am Coll Cardiol 2012;59:1058–72.

7. Arbab-Zadeh A, Fuster V. The myth of the“vulnerable plaque”: transitioning from a focus onindividual lesions to atherosclerotic diseaseburden for coronary artery disease risk assess-ment. J Am Coll Cardiol 2015;65:846–55.

8. Xing L, Higuma T, Wang Z, et al. Clinical sig-nificance of lipid-rich plaque detected by opticalcoherence tomography: a 4-year follow-up study.J Am Coll Cardiol 2017;69:2502–13.

9. Zhang BC, Karanasos A, Gnanadesigan M, et al.Qualitative and quantitative evaluation of dynamicchanges in non-culprit coronary atheroscleroticlesion morphology: a longitudinal OCT study.EuroIntervention 2018;13:e2190–200.

10. Garcia-Garcia HM, Jang IK, Serruys PW,Kovacic JC, Narula J, Fayad ZA. Imaging plaques topredict and better manage patients with acutecoronary events. Circ Res 2014;114:1904–17.

11. Manfrini O, Mont E, Leone O, et al. Sources oferror and interpretation of plaque morphology byoptical coherence tomography. Am J Cardiol2006;98:156–9.

12. Phipps JE, Vela D, Hoyt T, et al. Macrophagesand intravascular OCT bright spots: a quantitativestudy. J Am Coll Cardiol Img 2015;8:63–72.

13. Nakano M, Yahagi K, Yamamoto H, et al. Ad-ditive value of integrated backscatter ivus fordetection of vulnerable plaque by optical fre-quency domain imaging: an ex vivo autopsy studyof human coronary arteries. J Am Coll Cardiol Img2016;9:163–72.

14. Phipps JE, Hoyt T, Vela D, et al. Diagnosis ofthin-capped fibroatheromas in intravascular opticalcoherence tomography images: effects of lightscattering. Circ Cardiovasc Interv 2016;9:e003163.

15. van der Sijde JN, Karanasos A, Villiger M,Bouma BE, Regar E. First-in-man assessment ofplaque rupture by polarization-sensitive opticalfrequency domain imaging in vivo. Euro Heart J2016;37:1932.

16. Villiger M, Otsuka K, Karanasos A, et al. Cor-onary plaque microstructure and compositionmodify optical polarization. A new endogenouscontrast mechanism for optical frequency domainimaging. J Am Coll Cardiol Img 2018;11:1666–76.

17. Villiger M, Otsuka K, Karanasos A, et al.Repeatability assessment of intravascular polar-imetry in patients. IEEE Trans Med Imaging 2018;37:1618–25.

18. Nadkarni SK, Pierce MC, Park BH, et al. Mea-surement of collagen and smooth musclecell content in atherosclerotic plaques usingpolarization-sensitive optical coherence tomogra-phy. J Am Coll Cardiol 2007;49:1474–81.

19. Lippok N, Villiger M, Albanese A, et al. Depo-larization signatures map gold nanorods withinbiological tissue. Nat Photonics 2017;11:583–8.

20. Yahagi K, Kolodgie FD, Otsuka F, et al. Patho-physiology of native coronary, vein graft, and in-stent atherosclerosis. Nat RevCardiol 2016;13:79–98.

21. Niccoli G, Montone RA, Di Vito L, et al. Plaquerupture and intact fibrous cap assessed by optical

mailto:[email protected]

http://refhub.elsevier.com/S1936-878X(19)30599-6/sref1































































































12

coherence tomography portend different out-comes in patients with acute coronary syndrome.Eur Heart J 2015;36:1377–84.

22. Galis ZS, Sukhova GK, Lark MW, Libby P.Increased expression of matrix metalloproteinasesand matrix degrading activity in vulnerable regionsof human atherosclerotic plaques. J Clin Invest1994;94:2493–503.

23. Ridker PM, Everett BM, Thuren T, et al. Anti-inflammatory therapy with canakinumab foratherosclerotic disease. N Engl J Med 2017;377:1119–31.

24. Duewell P, Kono H, Rayner KJ, et al. NLRP3inflammasomes are required for atherogenesis andactivated by cholesterol crystals. Nature 2010;464:1357–61.

KEY WORDS atherosclerosis, collagen,inflammation, macrophage, opticalcoherence tomography, polarized light

APPENDIX For supplemental methodsand figures, please see the online version ofthis paper.

















intravascular polarimetry in patients with coronary artery...

Documents