Coronary Angiography: Is it Time to Reassess?

Historical Perspective

The introduction of selective coronary angiography by Mason Sones in 1958 stands as a milestone in clinical cardiology. The procedure continues as a cornerstone in the evaluation of the coronary arteries and is indicated not only for the diagnosis of coronary artery disease (CAD), but also to assess its severity. These data are then integrated with other clinical information to help guide treatment decisions. Traditionally, assessment of coronary lesion severity has been accomplished by a visual estimate obtained from simple inspection of the angiogram. While this method of assigning CAD diagnosis, and more specifically attempting to quantify its severity, has long been recognized to have important limitations,¹ simple visual estimation remains the most commonly used form of lesion evaluation and is taken by many operators to be their reference standard.

It is important to understand that among the many factors contributing to limit coronary blood flow (e.g., diastolic pressure time, microvascular resistance, etc.), the minimal luminal cross-sectional area available for flow is critically important. At rest blood flow remains unchanged until the cross-sectional area reduction is very severe (i.e. >80% stenosis) and increases resistance (Figure 1). Maximal blood flow, however, becomes impaired when the reduction in cross-sectional area approximates 50%, and this became the definition of “significant CAD”. This reduction in area also operates over the length of the lesion, so lumen area reduction actually represents only one factor in a complex geometry within a lesion that is difficult to measure in the clinical setting. For this reason, and because experimental models centered on the “idealized” focal stenosis, clinicians substituted an approximation of the most severe appearing obstruction.

Figure 1.

Comparison of the hypothetical relationship between % reduction in cross-sectional area (vertical axis) and % reduction in diameter stenosis (horizontal axis) estimates. At the point of significance suggested by the arrow, Poiseuille’s formula states that the change in resistance is inversely proportional to the square of the change in diameter. Beyond this point, resistance increases based upon R=8ŋL/πr4 (R=resistance, ŋ=viscosity, L=length). Adjunctive methods of lesion assessment and their lower thresholds for significant obstruction are shown at a hypothetical % diameter reduction. These include fractional flow reserve (FFR), minimal luminal diameter (MLD), minimal luminal area (MLA), and coronary flow reserve (CFR).

But CAD patients frequently have long and/or multiple stenoses in the same coronary artery. Our lab and others have shown that compared with a single focal stenosis, multiple consecutive stenoses, as well as an increase in stenosis length, result in a greater reduction in maximal flow.^2-6 Because the area reduction over the length of the complex lesion and multiple lesions was difficult to measure with precision even with multiple views, clinicians defaulted to estimating the maximal percent diameter narrowing in the worst view. Additionally considerable variability occurs in the relation between myocardial flow and percent diameter stenosis⁷ suggesting that other factors (e.g. microvascular dysfunction) also operate. Nevertheless “percent diameter stenosis” is most commonly used to define the presence of obstructive CAD.

Limitations of Coronary Angiography

Limitations of visually assessing coronary artery stenoses were realized long ago and attempts at improving this assessment followed. Using quantitative coronary angiography (QCA) improved our ability to more accurately estimate the percent stenosis of a lesion and its length which contribute to resistance to blood flow.⁸ While this technique is a well validated tool for accurately and reproducibly defining coronary lesion severity, these validations were mostly done in the cine film era using high dose radiography and high speed filming (60 fps) rates. Its use, as originally validated, remains mostly in the realm of specialized research using experimental models and in clinical trials. With the transition to so-called “lossless” compression digital angiography, use of lower dose radiography, and lower cine capture rates (15 fps), the information captured has been compromised.^9,10 The information available with these changes (particularly the lower film rates) markedly limits the ability to select multiple, optimally filled contrast segments, in several planes as required for high quality coronary angiography to quantitatively assess the coronary lesion (QCA). Thus attempts at classical QCA are not used in daily clinical practice. Other methods used in an attempt to improve the estimation of lumen narrowing such as calipers were met with mixed results and also have not achieved wide penetration into routine practice.^11,12

Contemporary Assessment of Coronary Angiography

It has been nearly two decades since the clinical assessment of CAD severity has been compared with a more quantitative approach. It is unclear if the technological advancements and adaptation of digital imaging have impacted our ability to accurately discern the degree of coronary disease by visual assessment. In this issue, Nallamothu and colleagues¹³ compared routine clinical assessment of CAD to QCA of over 200 coronary lesions from randomly selected patients prior to percutaneous coronary intervention (PCI). They extracted the clinical assessment of each epicardial lesion from catheterization and clinical reports. Quantitative assessments of each coronary lesion were performed from the single best view by an angiographic core lab. The investigators then compared the findings obtained by the two techniques as both continuous and categorical variables.

Overall their QCA revealed a median percent stenosis of 80% (IQR 80-90%).¹³ The mean percent diameter stenosis assessed clinically was about 8% higher than that obtained by QCA. Interestingly there was a large difference between the two analysis strategies for the visually assessed severe stenoses (70-90%). Use of QCA in this group of stenoses resulted in a severity of 50-70% in more than a quarter of the patients. A similar trend was noted in those patients who were clinically categorized as having 90 to <100% stenosis. There were no differences according to lesion location, angiogram quality, or whether stress testing or fractional flow reserve (FFR) was performed. When viewed from an institutional perspective, the mean difference between the two techniques varied by as much as two-fold (5.6 vs. 11.2%). The number of patients per site is small, however, and any inferences drawn from this analysis are hypothesis generating only.

It is important to emphasize that since all of the patients studied were undergoing PCI, this analysis does not include patients with a stenosis <50%, patients treated only medically, and none who were referred for coronary bypass. Furthermore, despite the fact that these patients were all from high volume centers voluntarily participating in a National Registry, stress testing was done in only about half of the patients, and FFR was used in only 7%.

Perspective

To put these findings in perspective, a comparison with prior studies suggests that the difference between the clinically and QCA assessed percent diameter stenosis may have actually improved. Earlier work suggested the difference between the two techniques was actually higher, although the reverse was true when evaluating lesions of <50% stenosis.¹⁴ As mentioned, there were no lesions <50% in the current study. In another study, also in patients undergoing PCI, the mean percent diameter stenosis was 87.9 ± 9.9% when assessed visually and 64.6 ± 9.2% when measured by QCA.⁸ Thus the mean difference of 8.2% from the current study seems small in comparison, and perhaps a goal of complete concordance is not realistic. It is possible however, that the difference between visually and QCA assessed percent diameter stenosis seen in the current study differs from that seen in an earlier era because of differences in angiographic acquisition. The use of lossy compression in the digital age may in part explain the smaller difference seen in the current study.^9,10

The authors should be commended for approaching this topic and with such scientific rigor. While the current work is important in providing a contemporary picture of the most commonly used method for coronary lesion assessment, it should be reassuring to note that the difference in these two techniques appears relatively small. It suggests that the focus from these results might be best placed on quality control and indeed it sets a high standard for others to achieve. It is also an important message to send to the non-medical community. At a time when it seems that almost daily there are accusations of medical impropriety within the cardiology community, this work sends a strong message that we as a profession are policing ourselves.¹⁵ If these findings can be replicated in a different cohort and one without the PCI referral bias, this work could set the stage for improvements in cardiac catheterization quality control. The variation in average percent stenosis by institution in the current study provides an example of data that might be amenable to use as a quality metric. A center whose average visual assessment of lesion severity is too far from the QCA mean might then be eligible for closer oversight.

Once a visual estimation of CAD severity is made, it remains for the individual operator to decide on the need for further testing prior to choosing a treatment strategy. Clearly the lower the category of stenosis severity, the greater the difference between clinical and QCA lesion assessment (Figure 2). It is in these lesions that adjunctive diagnostic strategies are likely to play the most important role. There are other tools that, when used in conjunction with the clinical assessment of CAD, can help guide treatment decisions, especially in patients with angiographically borderline (or intermediate) stenoses (e.g. 40-70%). These include intravascular ultrasound, optical coherence tomography, angioscopy, coronary flow reserve, and FFR. The former methods all improve anatomical assessment of the lesion, and the latter provide physiological evaluation of the coronary vascular disease. They all, however, involve additional time, expense, and patient risk. It is these factors, at least in part, that have allowed the simple visual assessment of lesion severity to remain the most commonly used technique to evaluate CAD.

Figure 2.

Difference between visual and QCA estimates of mean % diameter stenosis by severity category. The horizontal axis shows stenosis category and the vertical axis summarizes the mean differences between visual and QCA estimated % diameter stensosis.

We are moving to an era where not only has the additional evaluation of visually assessed disease already been validated, but it may at some point even be mandated. We have learned the value of using FFR to assess coronary lesions from the FAME studies in patients undergoing PCI. Indeed when significant lesions as assessed by visual estimation undergo evaluation by FFR, their functional significance is often far less. When percent stenosis was visually estimated to be between 70-90% in FAME, one in five lesions was found not to be significant by FFR (>0.80). Even in severe disease felt to be >90% occlusive, fully 4% were found by FFR not to be hemodynamically significant.^16,17 FAME-2 has extended the validation of FFR to the interventional treatment of patients with stable coronary disease.¹⁸ The goal of having complete agreement between visual and QCA estimations of CAD severity is likely not reachable, but the results of the current study suggest that the gap may not be great. With ongoing evaluation such as that provided by Nallamothu et al, and use of adjunctive strategies when appropriate, our patients will continue to receive the high quality care that they deserve.