The voyage of discovery consists, not in seeking new landscapes, but in having new eyes.

--Marcel Proust

Introduction

Cardiovascular disease (CVD) due to atherosclerosis remains the leading cause of death worldwide, accounting for 15.2 million fatalities in 2016 [1]. It is well known now that atherosclerosis is a chronic inflammatory disease that involves multiple interconnected biological processes including endothelial dysfunction, monocyte infiltration, LDL oxidation, inflammatory protease activity, smooth muscle migration, neovascularization, intraplaque hemorrhage, and calcification [2]. Atheroma may be further classified as stable and unstable based on their underlying structural and/or molecular phenotype. Unstable plaque phenotypes, such as plaque rupture or erosion, are commonly present in plaques underlying ACS, as demonstrated by high-resolution intravascular imaging methods such as intravascular ultrasound (IVUS) or optical coherence tomography (OCT) [3]. In addition, certain IVUS-derived structural measures (high plaque burden > 70%, minimal luminal area < 4 square mm, or IVUS-virtual histology demarcated thin-capped fibroatheroma (TCFA)) may identify vulnerable plaques that will progress to cause ACS [4]. However, even when incorporating advanced plaque measures such as endothelial shear stress [5, 6], accurate prediction of events arising from vulnerable plaques is insufficient. As plaque pathobiology such as inflammation is a major driver of coronary events [7,8,9], an imaging method to specifically report on coronary arterial inflammation or other pathobiology could improve the detection of high-risk plaques.

Basis for Molecular Imaging

Cardiovascular molecular imaging has emerged as an important research and evolving clinical tool that allows in vivo visualization of key molecules and cells underlying plaque destabilization and progression [10,11,12]. Molecular imaging requires two components: an imaging hardware detection system (e.g., catheter and console) and an injectable molecular imaging agent (e.g., peptide, antibody, small molecule conjugated to a signal generating moiety) that targets a specific molecule or cell and provides a target-specific signal detectable by the desired hardware system. Although non-invasive molecular imaging via positron emission tomography (PET) has witnessed exciting advances in terms of coronary artery visualization using PET tracers, such as 18-fluorodeoxyglucose or 18-sodium fluoride [13, 14], PET imaging is still of unclear application to CAD patients, as the coronary arteries reside deep within the chest, are of small caliber (3–4 mm in diameter, approximately the resolution of PET), overlie the metabolically active heart, and are subject to cardiorespiratory motion and pulsatile blood flow [15, 16].

To enable high-resolution molecular imaging of CAD, where plaques of interest are of 1–5 mm thickness, intravascular near-infrared fluorescence (NIRF) had emerged as a translational intracoronary optical imaging modality that allows high-resolution detection of plaque molecules and cells [17]. More recently, hybrid imaging systems that fully integrate NIRF optical fibers with either intravascular ultrasound (IVUS) or optical coherence tomography (OCT) have been developed [18•, 19, 20•, 21•], enabling co-registered and quantitative molecular and structural imaging. In addition, a first human study using NIR autofluorescence (NIRAF)-OCT has already been performed [22••], demonstrating that NIR fluorescence imaging is feasible in CAD patients. These developments offer the potential to open a new horizon for personalized medical therapy of patients harboring high-risk coronary plaques [10, 23]. The focus of this review is to showcase recent advances in intravascular molecular imaging, with a focus on intravascular NIRF imaging.

Intravascular Imaging of High-Risk, Vulnerable Plaques

While our initial understanding of vulnerable coronary artery plaques derived from autopsy studies, over the past decades, multiple intracoronary imaging techniques have emerged, including IVUS, OCT, and near-infrared spectroscopy (NIRS). These methods have enabled in vivo clinical assessment of plaque structure and lipid-rich necrotic cores. It is important to note NIRS measures plaque lipid solely and is distinct from NIRF (fluorescence) molecular imaging, the focus of this review (Table 1). For more in-depth reviews on standalone IVUS, OCT, and NIRS, please refer to a recent intravascular imaging topical collection [24, 25].

Table 1 Comparison of different intravascular imaging modalities in detecting vulnerable plaques
Full size table

Foundations of Near-Infrared Fluorescence High-Resolution Molecular Imaging

Near-infrared fluorescence (NIRF) molecular imaging is a clinically emerging optical imaging modality due to its high sensitivity, ability to deliver light down flexible fibers, relatively low cost, and the availability of chemical engineered NIR fluorophores that can report on a wide range of in vivo molecular processes [26]. NIRF imaging involves administering targeted or activatable fluorophores, injectable imaging agents that localize and bind to or interact with molecular or cellular target of interest in atheroma [23]. Excitingly, a number of NIRF molecular imaging agents are undergoing testing in patients with cancer (e.g., bevacizumab-IRDye800CW) and should provide an opportunity for further translation to the cardiovascular domain [27••, 28]. Illumination light from the near-infrared spectrum (650–900 nm) is focused on to the arterial wall via optical fibers and is used to excite fluorophores from their ground state (S0) to an excited state (S1, S2). The excited fluorophores then may re-emit light at a longer wavelength (fluorescence emission) and return to the ground state, available for another round of excitation. The emitted NIR fluorescence is collected by intravascular optical fibers housing the appropriate emission filters that exclude illumination light. Emission light (fluorescence) is then captured using a high sensitivity charge-coupled device (CCD) camera [17, 23]. Compared with visible light, near-infrared light-based imaging possesses several features that render it favorable for in vivo imaging: (A) lower absorption by hemoglobin allowing NIR light penetration through blood; (B) reduced scattering allowing deeper tissue penetration of light; and (C) reduced tissue autofluorescence, allowing more sensitive detection of injectable NIRF molecular imaging agents [29, 30].

The first intravascular, real-time in vivo NIRF catheter-based molecular imaging approach was described by Jaffer et al. in 2008 [31]. After injecting a cysteine protease-activatable NIRF molecular imaging agent (ProSense750/VM110), the NIRF fiber was percutaneously inserted into rabbit iliac arteries (1.5–2.5 mm in diameter) and was able to detect plaque inflammation through blood [31]. This non-rotational prototype, however, operated in a one-dimensional fashion, sensing the NIRF signal from a limited section of the circumferential arterial wall in a spectroscopic-type mode. To overcome these limitations, our group next developed a two-dimensional rotational NIRF intravascular catheter with automated pullback, to enhance the axial/longitudinal anatomical accuracy of the generated images. This allowed the construction of real-time, in vivo images of stented rabbit aortas [32••]. The group found increased stent-induced inflammatory activity around the edges of implanted stents, enabling a new in vivo to approach to assess stent-induced vascular injury and healing. However, this 2D NIRF approach employed a standalone imaging catheter, and required a second intravascular imaging catheter (i.e. IVUS) to provide anatomical information. This limitation provided the motivation to develop multimodal NIRF imaging systems.

Hybrid Intravascular Near-Infrared Fluorescence Molecular Imaging

To enable simultaneous molecular and structural imaging via a single catheter, intravascular imaging systems combining NIRF and OCT have been developed [18•, 33]. This has enabled exact co-registration of molecular and structural imaging signals and furthermore allows for distance correction of NIRF signal based on the catheter distance to the lumen-wall border (Fig. 1) [11].

Fig. 1
figure 1

Serial in vivo dual-modality optical coherence tomographic-near-infrared fluorescence (OCT-NIRF) and intravascular ultrasound (IVUS) imaging was completed at weeks 8 and 12 after balloon injury to rabbit aorta. ProSenseVM110, a fluorophore that detects cathepsin protease activity, was injected 24 h prior to imaging. (A) Longitudinal co-registered NIRF-IVUS imaging at 8 and 12 weeks. (B) Axial NIRF-OCT fusion image (yellow/white: high near-infrared fluorescence; blue/black: low near-infrared fluorescence) at the location of the white dotted line in (A), second row (near-infrared fluorescence, 12 weeks). Matched cross-sectional fluorescence microscopy (red: ProSense VM110; green: autofluorescence) and histopathology demonstrates increased ProSense VM110 NIRF signal within a moderate fibrofatty atheroma (H&E) associated with cathepsin B immunostain. Scale bars, 1 mm (reprinted from Bourantas et al. [11], with permission from the European Society of Cardiology)

Full size image

Hybrid NIRF-OCT Molecular-Structural Imaging

OCT provides exquisite spatial resolution, and two hybrid-dual modal OCT-NIRF systems (Fig. 1) have been developed and validated in vivo. The first one utilized a 2.4 Fr dual-clad fiber with single-mode OCT core and an inner NIRF cladding. This system enabled co-illumination of 750 nm NIRF and 1320 nm OCT light, and exact co-registration of the returned NIR fluorescence and OCT backscatter signals [18•]. This NIRF-OCT catheter system enabled molecular imaging of fibrin of a stent implanted in a cadaveric human coronary artery, and imaging of plaque inflammation in vivo in the rabbit iliac artery [18•]. Similarly, Lee et al. [33] developed a high-speed intravascular OCT-NIRF platform with a probe diameter of 2.5 Fr and image acquisition rate of 100 frames/s. Using this hybrid system, it was possible to detect ICG deposition in macrophage-rich plaques using an atheroma rabbit model [33], and later specific detection of macrophages was demonstrated [34]. This system was then further validated in pig coronary arteries, demonstrating the ability to image ICG leakage into atheroma and behind implanted DES [35•].

Hybrid NIRF-IVUS Imaging

For the past three decades, IVUS has been the worldwide dominant intravascular imaging approach. Compared with OCT, IVUS has greater depth penetration (5–6 mm) but lower spatial resolution (100–250 μm), lower contrast for discriminating plaque components, and is able to detect moderate backscatter from blood with 40–60 MHz transducers. While backscatter may complicate detection of the luminal border, IVUS does not require blood clearance. [36]. This feature makes an NIRF-IVUS combination appealing as both modalities can be used through blood. In a first proof-of-principle, Abran et al. [19•] engineered and validated a bimodal NIRF-IVUS catheter system in ex vivo rabbit atherosclerosis, demonstrating the ability to image ICG uptake in plaques. Next, Dixon and Hossack [37] designed a hybrid platform using side by side NIRF and IVUS imaging design with a larger probe size of 4.2 Fr, and performed ex vivo coronary artery imaging and NIRF signal correction. In the first in vivo demonstrations of NIRF-IVUS, Bozhko et al. demonstrated the ability to image ICG endothelial leakage and stent fibrin in vivo. More recently, Bertrand et al. demonstrated that NIRF-IVUS could detect locally administered NIRF inflammation agents in atheroma [38]. Future NIRF-IVUS designs are under development aiming to reduce the NIRF-IVUS catheter diameter to < 3 Fr, suitable for intracoronary use.

Hybrid Tri-Modality NIRF-OCT-IVUS Imaging

In 2017, Li et al. [39] designed a tri-modal catheter using IVUS, OCT, and NIRF technologies. The tri-modal catheter diameter of the probe is 1.0 mm (3.9 Fr), small enough to fit in 5 Fr guide, but likely too large for routine clinical use in its present form. The system is able to obtain all three IVUS, OCT, and NIRF imaging simultaneously. The group conducted phantom and ex vivo experiments using pig and rabbit arteries. This integrated tri-modal system has a great potential to identify vulnerable plaques through NIRF molecular imaging and provide the joint advantages of OCT and IVUS in a single system.

Applications of NIRF Molecular Imaging to Detect High-Risk Plaques

NIRF Molecular Imaging of Inflammatory Protease Activity

Inflammatory proteases, secreted by macrophages or activated smooth muscle cells, promote atheroma progression and plaque rupture. Two widely studied classes of proteases are cathepsins and matrix metalloproteinases (MMP). Cysteine cathepsins are a group of proteases that are mainly found in lysosomes. There are 11 human cysteine cathepsins (B, C, F, H, K, L, O, S, V, X, and W) that vary in their distribution and specific function [40]. In atherosclerosis, the elastolytic and collagenolytic activity of cysteine cathepsins especially S, K, B, L, and F have been linked to extracellular matrix degradation and hence, both positive and negative remodeling. It was shown that early human atherosclerotic lesions expressed cathepsins K and S [41], and deficiency in cathepsins K and S in a mouse model reduced atherosclerosis [42, 43]. ProSense/VM110 is one of the most widely investigated NIRF protease-activatable imaging agents in animal models. At baseline, ProSense/VM110 is optically silent. Once ProSense/VM110 circulates and localizes to the tissue of interest (in this case, plaques with cathepsin proteases), it is cleaved by cathepsins yielding highly fluorescent fragments [44]. ProSense activatable fluorophores are available in two forms: ProSense 680, activated by Cathepsin B, L, and S with emitted light of 700 nm, and ProSense 750/VM110, activated by Cathepsins B, L, S, K, V, and D with emitted light of 780 nm [44]. Another potential protease target for NIRF imaging is the class of metalloproteases (MMPs), which are linked to unstable/vulnerable atherosclerotic plaques [45].

Intravascular NIRF imaging has been able to detect plaque protease activity in a number of scenarios. An initial series of studies demonstrated the ability to image plaque inflammation at high resolution in vivo, using new NIRF catheter systems [18•, 31, 32••]. Thereafter, Calfon Press et al. [46•] demonstrated that everolimus-eluting stents, the most prevalent type of DES, can suppress plaque inflammation in vivo and reduce plaque macrophages, strengthening the potential of DES to serve as local therapy for vulnerable plaque. In a more recent application using serial NIRF-OCT and IVUS imaging, Osborn et al. [47] have demonstrated preliminarily that in rabbit atheroma, the presence of plaque inflammation independently predicts plaque progression. These preclinical studies on plaque inflammation using NIRF imaging will provide a framework for clinical studies once a NIRF catheter is approved by the FDA.

Oxidized Low-Density Lipoprotein

The presence of oxidized LDL (oxLDL) indicates oxidative stress, a high-risk plaque feature that localize within fibroatheroma lipid-rich core. LO1 is an auto-antibody that binds to copper-oxidized LDL in mice, rabbits, and humans [48]. By labeling the LO1 antibody with a NIRF dye AF750, Khamis et al. [49••] synthesized a NIRF agent called LO1-750 (excitation light of 750 nm). This enabled in vivo molecular imaging in mice using fluorescence molecular tomography (FMT), which is a non-intravascular NIRF imaging apparatus, combined with micro-computed tomography (CT). The group found increased accumulation of LO1-750 in the aorta and subclavian arteries of LDLr−/− mice when compared with the wild type (WT). When compared with MMPSense signal, LO1-750 showed higher NIRF signal ratio. Using ex vivo intravascular NIRF imaging, oxLDL was imaged in rabbit plaques [49••]. This is a step forward to be able image oxidative plaque stress in living subjects.

Cellular Inflammation in Areas of Impaired Endothelial Permeability

Impaired endothelial permeability has been linked to superficial plaque erosion and atherothrombosis. A macrophage-targeted NIRF iron oxide nanoparticle CLIO-CyAm7 was developed to investigate in vivo mechanisms of impaired endothelial function in an atheroma rabbit model [50•]. Using NIRF imaging, Stein-Merlob et al. showed that CLIO-CyAm7 nanoparticles deposited in plaque cells in areas of impaired endothelial barrier function [50•]. It was also shown that iron oxide nanoparticles had high concentration in thrombosed plaques when compared with non-thrombosed plaques [50•].

Molecular Imaging of Fibrin Deposition

Fibrin is the end product of coagulation cascade and is involved in both venous and arterial thrombosis [51]. FTP11-Cy7 is a NIRF fluorophore developed to specifically bind to fibrin in living subjects [52]. In 2012, Hara et al. [52] demonstrated that FTP11 binds to human in vitro thrombi and in vivo murine thrombi using non-invasive NIRF imaging. Later, the same group used FTP11 to detect fibrin deposition in unhealed stents and demonstrated increased fibrin deposition on DES compared with BMS [53•]. The group concluded that a substantial percentage of stent struts that appear to be covered by OCT imaging are actually covered by fibrin (Fig. 2). This may help better identify stents more prone to stent thrombosis. Additional applications of fibrin imaging are anticipated in elucidating culprit plaques in ACS patients, and in subjects with myocardial infarction and non-obstructive arteries (MINOCA), and identify high-risk plaques with subclinical fibrin-based thrombosis.

Fig. 2
figure 2

In rabbit aortas and at day 7, drug-eluting stents (DES) displayed greater fibrin deposition than bare metal stents (BMS), using the FTP11-CyAm7 fibrin NIRF molecular imaging agent. (A) In vivo 2D near-infrared fluorescence map with BMS (left) and DES (right), (B) corresponding fluorescence microscopy after longitudinally opening stents, (C) fluorescence reflectance imaging (FRI), and (D) fluorescence microscopy from the outside. In vivo axial NIRF-OCT images of day 7 (E) bare metal stent and (F) day 7 drug-eluting stent at the distal edge (arrowhead in A). Carstairs’ stain shows that some areas of BMS surface areas are fibrin negative (G), but almost all areas of DES tissue coverage are fibrin positive (H). (I) In vivo NIRF-fibrin signal from the proximal to distal stent edge of BMS and DES (n = 7 BMS, n = 7 DES). Scale bars = 1 mm (reprinted from Hara et al. [53], with permission from European Society of Cardiology)

Full size image

NIRF Imaging of Atherosclerosis and Human Studies

NIRF Imaging of Impaired Plaque Endothelial Function Using Indocyanine Green in Patients

At present, intravascular NIRF molecular imaging has not yet been evaluated in patients, pending the availability of a clinically approved catheter. However, clinical NIRF molecular imaging will be facilitated by the availability of the targeted NIRF imaging agent, indocyanine green (ICG). ICG was first introduced in the 1950s as a blood flow dye and currently serves as a routine clinical retinal imaging agent. Unexpectedly, in 2011, ICG was demonstrated to function as a targeted molecular imaging agent, where it accumulated in lipid-rich macrophages in vitro, and was detectable rapidly by intravascular NIRF sensing in vivo in rabbit atheroma [54••] and subsequently in pig coronary atheroma [35•, 55••]. This paved the way for the first human trial testing whether ICG could be injected into living subjects and target human atheroma. The BRIGHT-CEA trial (Indocyanine Green Fluorescence Uptake in Human Carotid Artery Plaque) in 2016 injected ICG into patients undergoing carotid endarterectomy, and then, plaques were resected 99 min after ICG injection [55••]. Resected carotid plaques were then imaging using ex vivo NIRF macroscopic imaging and intravascular NIRF-OCT. The group demonstrated that ICG accumulated in plaque areas with impaired endothelial integrity, disrupted fibrous cap, and areas of neovascularization and intraplaque hemorrhage. The availability of ICG as an FDA-approved agent will accelerate first-in-human targeted NIRF imaging studies.

NIRAF-OCT Imaging of Intraplaque Hemorrhage

Concurrently with the development of targeted NIRF molecular imaging approaches, it was recognized that plaque NIR autofluorescence (NIRAF) might represent a new measure of high-risk plaques [56]. NIRAF is distinct from NIRF molecular imaging, in that NIRAF detects endogenous autofluorescence and does not require the administration of imaging agents. In addition, the excitation wavelength of light (633 nm) is significantly lower than that used for targeted NIRF molecular imaging (750–800 nm). NIRAF is detected after NIR light (633 -nm excitation in clinical subjects) illuminates plaques that possess certain endogenous NIR fluorophores, making its adoption into clinical use faster in the near future. Recognizing the possibility to advance the clinical potential of NIRAF imaging, a recent clinical trial explored the use of intracoronary NIRAF-OCT in CAD patients [22••]. Ughi et al. [22••] conducted the first human study on 12 patients with coronary artery disease undergoing PCI using a hybrid NIRAF-OCT catheter, designed to detect NIRAF in non-culprit lesions. This catheter received an investigational device exemption (IDE) from the FDA prior to clinical study. NIRAF was detected in 1–2 plaques per coronary artery, a relatively specific signal pattern. Elevated NIRAF signal was associated with high-risk morphologic phenotype like fibroatheroma, plaque rupture, and in-stent restenosis (Fig. 3) [22••]. While detection of NIRAF indicates components of intraplaque hemorrhage (e.g., bilirubin, protoporphyrin IX) [57], the complete set of molecules and biological processes underlying 633 nm NIRAF remains to be elucidated. Nonetheless, this landmark NIRAF-OCT study demonstrated the safety and utility of NIRAF to detect advanced plaques in CAD patients and provides a foundation for future intracoronary NIRAF studies, as well as targeted NIRF molecular imaging at longer wavelengths (750–800 nm).

Fig. 3
figure 3

OCT-NIRAF imaging of thin-cap fibroatheroma (TCFA) rupture. (A) Coronary angiogram of left anterior descending (LAD) artery. (B) 2D NIRAF map presenting a focal region of high NIRAF signal in the ostial LAD. (C, D, E) OCT-NIRAF cross-sections from sites in (B) with increased NIRAF, revealing subclinical fibrous cap rupture on OCT imaging. (F) Magnification of a cholesterol crystal below the cap, co-localized with high NIRAF, and (G, H) magnified views of the rupture site. In (G), the rupture site (arrowhead) is covered by a small white luminal thrombus (arrow) and the arrow in (H) points to the site of the thin-cap rupture, representing co-localized and very high focal NIRAF signal. (I) 3D cutaway rendering showing that the highest NIRAF spot appears focally within a large lipid pool (arrow), and the remaining portion of the vessel shows diffuse disease that was negative for NIRAF. Scale bars on OCT images and magnifications are equal to 1 mm and 0.5 mm, respectively; scale bar in (B) is equal to 5 mm (reprinted from Ughi et al. [22], with permission from the American College of Cardiology)

Full size image

Practical Use of Intracoronary NIRF Imaging in the Cardiac Catheterization Laboratory

Intravascular NIRF molecular imaging is likely to be applied to assess high-risk non-culprit plaques in patients already undergoing culprit-lesion PCI. As PCI patients are increasingly undergoing IVUS or OCT for stent optimization [24], we envision that patients would undergo injection of a NIRF molecular imaging agent at the start of PCI, with subsequent NIRF-IVUS or NIRF-OCT at the end of the PCI. This would enable at least single-vessel coronary arterial phenotyping of plaques proximal and distal to the culprit zones. The longer term goal of additional NIRF information [27••] would help to better predict patient-specific risk based on the degree of abnormal plaque pathobiology, and then to prescribe specific medications based on the risk and the specific abnormal biology, enabling personalized medical therapy of CAD patients.

Conclusion

NIRF intravascular molecular imaging is a clinically translatable powerful approach to visualize high-risk plaque molecular processes, including macrophages, inflammatory protease activity, oxidized LDL molecules, and impaired endothelial barrier function. Hybrid intravascular catheters including NIRF-OCT or NIRF-IVUS technologies can depict microstructural and molecular changes of the arterial wall with high resolution and high sensitivity. At present, ICG is the only FDA-approved fluorophore for use in humans, but additional clinical NIR fluorophores to be approved for human use are emerging from the field of cancer NIRF molecular imaging. Following validation human studies, future clinical trials will determine the value of intravascular NIRF molecular imaging in patients, to enable personalized assessments of risk and targeted prescription of CAD pharmacotherapy.