Acute Coronary Syndromes (ACSs) are the leading cause of death in the industrialized world, representing over 25% of all deaths in the US alone [1
]. The importance of early aggressive intervention can’t be over emphasized. Specific plaques called Thin-Capped Fibroatheromas (TCFAs) lead to most ACSs [4
]. When these plaques rupture, they release thrombogenic necrotic material into the blood, a clot forms, and the vessel occludes in about 20% of the cases. Detecting TCFAs prior to rupture, particularly risk stratifying those 20% predisposed to ACS progression, is a central focus of cardiovascular research [7
]. TFCAs are characterized by a necrotic (lipid-laden) core, a thin intimal cap, and minimal protrusion into the lumen [4
]. In addition, they have a loss of cap collagen/smooth muscle, immature angiogenesis, and intermittent inflammation. Plaque strain analysis and autopsy data has demonstrated these plaques have caps less than 75 µm in width and a necrotic core which is soft (high strain) making them susceptible to rupture [14
]. It is a currently held belief that Optical Coherence Tomography (OCT), in its current state, can identify TCFAs reliably [11
]. Studies do strongly support the ability to identify the thin intimal caps [21
]. However, contrary to currently held beliefs, the evidence is relatively weak it can reliably identify necrotic cores (or even pure lipid cores).
Part of the confusion lies in the fact, as will be discussed, that the ‘diffuse boundary’ criteria currently used to identify ‘lipid’ plaque is insufficient. Further leading to the confusion is that what need to be identified based on histopathological studies of ACS is necrotic cores, but OCT investigators have been looking predominantly for lipid cores (which is made worse by the literature incorrectly using the terms necrotic and lipid interchangeably). A necrotic core (as opposed to a lipid collection) contains a large amount of cellular debris, cells, and microvessels, in addition to lipids, particularly free cholesterol
(example in Figure 1). From Figure 1, it can be seen that unlike pure lipid, there are a large number of scattering elements within the core. This paper attempts to clarify misconceptions that current OCT techniques can reliably identify TCFAs and provides alternative OCT approaches to achieve this objective.
Before discussing the basis of the belief that OCT can identify necrotic plaques, a brief overview of OCT is given. OCT is analogous to ultrasound, measuring the backreflection of infrared light rather than sound [20
]. OCT identifies small thin-capped plaques better than any currently available imaging modalities. Cardiovascular OCT has a resolution of approximately 20 µm, catheters less than 1 mm in diameter, and an acquisition rate faster than 30 frames/second. In the early 90‘s we published the first work on OCT cardiovascular imaging, which focused on vulnerable plaque assessment [21
]. Though we have published a large number of studies since, aspects of this paper are particularly relevant to the issues of OCT plaque characterization discussed in the current paper.
The belief current OCT approach can define core composition accurately is primarily based on one study that claimed lipid plaque (the term lipid plaque again would later incorrectly be used interchangeably with necrotic core) had a diffuse intimal lipid border (1992, Yabushita et al.) that could be used as a marker for lipid cores [25
]. We have previously expressed concerns in short editorials but this paper performs a more detailed analysis [23
]. As lipid does not backscatter, some criteria needed to be introduced to differentiate lipid areas from areas where signal was lost from rapid surface attenuation [23
]. In other words if there was no backreflection below the intima it could be either because it was lipid (which gives no backreflection) or because light did not penetrate. The current paper argues that in OCT’s current form and how it is applied, it can’t reliably identify TCFAs because it can’t accurately identify necrotic cores. The Yabushita et al. paper claimed that lipid cores (again we are actually looking for necrotic cores) could be differentiated from signal loss by having diffuse rather than sharp intimal-core borders. There are several lines of evidence that do not support this assertion. First, these criteria only had a 70% predictive power for lipid plaque [25
]. The problems should have been obvious from the author’s own observations “False-positive OCT diagnoses of lipid-rich plaques often contained histological evidence of small amounts of lipid present within a predominantly fibrous plaque. These lesions, perceived as lipid-rich by OCT, were interpreted as fibrous plaque by histopathology
, resulting in a relatively low sensitivity of the OCT criteria for diagnosing fibrous plaques (71% to 79%).” Second, as will also be discussed below, there is no theoretical reason why an intimal-core interface should be diffuse. This is a point we have both argued and demonstrated before and will be discussed here. We have demonstrated (and will show in this paper) both fat cells and many lipid plaques have borders that are sharp, consistent with the fact that large mismatch in refractive index lead to sharp borders. On the other hand, when many lipid crescents or calcium deposits are present in the intima this will result in phenomena known as multiple scattering, the physics of which is described below, that will make the intima-core border diffuse. This is independent of the core composition. Third, over 2/3 of the specimens for the study were from elastic and not coronary arteries, which distinct mechanisms of rupture. The only image of a coronary plaque in that paper had no histopathology for unknown reasons. It is unclear how many lipid plaques, if any, came from coronary arteries. Fourth, the OCT readers had minimal experience, an engineer and student. Fifth, they did not look at necrotic plaque but instead looked at lipid plaque. According to the paper, the authors “classified all of the plaques in the validation set as fibrous, fibrocalcific, or lipid rich”. This is not a clinically useful categorization. Sixth, several other papers have confirmed the only limited utility of the ‘diffuse border’ criteria but it still continues to be used as the standard for identifying a plaque as having a lipid core. Similar results were obtained by a second group in 2006, but 2 additional studies performed the same year had even less promising results; one of these studies reported only a 45% sensitivity and 83% specificity for identifying lipid-filled plaques [26
]. From this, along with data below, it is likely that the results of the Yabushita et al. reflect a higher incidence of intimal lipid or calcium in necrotic plaques, which lead to the intima-lipid border appearing diffuse on imaging. This makes it an indirect relationship rather than a direct measure, contributing to the relatively low sensitivity [33
It should be noted that our conclusions are also consistent with the 2012 Consensus Standards for Acquisition, Measurement, and Reporting of Intravascular Optical Coherence Tomography Studies. The problem is these important conclusions were not stated prominently in the paper [11
]. First the report defined “A necrotic core by OCT is a signal-poor region within an atherosclerotic plaque, with poorly delineated borders, a fast IVOCT signal drop-off, and little or no OCT signal backscattering, within a lesion that is covered by a fibrous cap.” It was then noted “At present, there are no definitive published studies directly comparing OCT lipid pool–containing plaques with necrotic core by histology, and as a result, the Evidence Level was determined to be Low for OCT delineation of necrotic core.” In other words, calling a lesion a necrotic core by OCT imaging is not based on any data in the literature. OCT’s ability to identify necrotic cores is untested to date. This is consistent with the current paper and previously expressed concerns [23
]. Unfortunately, these important conclusions were not emphasized in the conclusions of the paper and most still believe that diffuse borders identify necrotic cores, an unsubstantiated conclusion we have noted numerous times [11
From our initial paper that established OCT for cardiovascular imaging (and imaging in non-transparent tissue in general), several points relevant to the current manuscript will be raised. First, it demonstrated OCT’s ability to identify intimal caps at micron scale resolutions, which few now would dispute. Second, the reason we were able to image in non-transparent tissue was that we identified using the 1300 nm wavelength range for imaging in arteries (830 nm had been used in the eye) to more than double the penetration. At this wavelength, lipid scattering and absorption
is negligible. Third, it looked at plaques with high lipid content and thin caps. We did not differentiate lipid collections from necrotic regions because the importance of this would not be clear till several years later. But of significance to this paper and illustrated below, all high lipid plaques in that paper, except for one, had well-defined and not diffuse borders (the one plaque provides insight and will be discussed below). This we will see is inconsistent with the Yabushita et al. Fourth, it demonstrated that pure lipid was transparent to the infrared light while water based tissue such as intima and smooth muscle backscattered signal. But a point which seems to be missed from that paper, but relevant to the current discussion, is that fat cells (but not pure lipid) led to rapid deterioration of the signal. This seems counterintuitive (as lipid is transparent) but it is due to the mismatches in the refractive index between the inside and outside of the cells (1.35 versus 1.55) as well as their size (this is similar to the mechanism where we demonstrated why blood has poor penetration) [21
]. These principles would support that intima-lipid borders would generally be sharp unless the intima contained significant lipid or calcium collections. This type of intima would lead to diffuse intimal-core borders by a process known as multiple scattering [23
]. We have discussed this most recently in a 2011 Circulation paper as well as other prior work but it will be readdressed in far more detail here [27
It is essential to identify high risk TCFA accurately for selective intervention to prevent ACS. To do so, thin capped necrotic versus non-necrotic plaque need to be differentiated. Current OCT approaches do not accurately identify necrotic cores for the reasons described. We will demonstrate this both by discussing the underlying physics involved and as well as presenting relevant OCT images. In the discussion, we will conclude by describing potentially superior OCT approaches.