Worst view or average several views?
Our colleague and cath lab expert, Dr. Barry Uretsky from the University of Arkansas, Little Rock, started this conversation with one of the oldest and probably most important questions about coronary angiography, and that is, how should we report the degree of narrowing of the coronary angiogram, more specifically, a lumenogram? Should we use the worst view, i.e., the most narrowed diameter, or should we consider all the views and report the averaged value? Our contributing experts, myself included, question the value of reporting a percent diameter at all, given the well-known large variance of different angiographers’ readings of the angiogram. Is it time to give up the percent stenosis in favor of a more qualitative but truer reading?
Dr. Uretsky asks, “Mort, I had an interesting conversation with my fellows today on how to best estimate visually percent angiographic diameter stenosis. I have taught fellows for a long time that one should estimate the severity based on multiple views and draw an average. I emphasized that it is particularly important for eccentric lesions where the percent stenosis may be quite different in different views, say 90% in one view and 30% in another. I used an example of a 90% concentric lesion with a diameter of 2mm and lumen area of 3.14mm2 vs an eccentric lesion with a 90% stenosis in one view (diameter 2mm) and 0% stenosis in the second view, and diameter of 4mm, which, using a prolate ellipsoid model, produces a lumen area of 6.28mm2. Obviously, the coronary flow would be much more reduced in the former than the latter lesion.
“The fellows then challenged me, stating that ‘Kern’ (The Cardiac Catheterization Handbook, 6th edition, 2015, page 131) recommended one should use the view with the worst stenosis for this evaluation. I was surprised, since ‘Kern’ has emphasized the importance of coronary flow and the use of fractional flow reserve (FFR) to help in determining the physiological significance of a lesion. It seems to me that estimating the stenosis by integrating all views that clearly show the lesion is a better angiographic method of estimation. My questions are the following:
- Are you aware of any formal recommendations on the proper method of angiographic visual estimation?
- Do the considerations noted above recommend an estimate from multiple angiographic views, or do you recommend the worst view method, and if so, what is the reason? What do you think other labs do?”
Mort Kern, Long Beach, California: Barry, this is a great question and very worthy of discussion. The visual angiographic interpretation of a stenosis varies +/-20% among expert observers due exactly to the poorly defined notion of what the “anatomic lumen reduction” represents. A concentric narrowing seen in multiple views often shows a similar degree of narrowing and is usually reported as the % narrowing estimated as the worst from several projections. But an eccentric lesion, say 80% in one view and 20% in an orthogonal view, is a dilemma (Figure 1). My recommendation is to use the worst view and report that, not estimate an average of two or more views. Lesion eccentricity often leads to a visual (%) vs functional (FFR) mismatch, which is frequent in eccentric lesions and much less so in concentric narrowings.
To your questions:
- I am not aware of formal recommendations other than the “one-man consensus” as noted in The Cardiac Catheterization Handbook: “The degree of an angiographic narrowing is estimated as the percentage lumen reduction of the most severely narrowed segment compared with the adjacent angiographically normal vessel segment, seen in the worst x-ray projection.”1 The evaluation and averaging of multiple degrees of visual narrowings, with reporting some qualitative visual “averaged” value contributes to the disparity (+/-20% range) of readings among expert observers. I believe that the fellows and most attendings currently report the most severe narrowing in the worst view for the purposes of the written cath report.
- My reason for adopting the worst view is because that is what I was taught. So old school, I know. As an angiographic nihilist (…nihilism is defined as a belief that traditional values have no worth…), I don’t have to tell you I am a fan of FFR. FFR provides the answer for any such visual dilemma or indecision regarding true degree of narrowing. With that declaration, I look forward to hearing from our colleagues.
George Vetrovec, Virginia Commonwealth University, Richmond, Virginia: I agree that I am not at all aware of a “gold” standard of visual assessment. I also agree that a “visual” average is the best estimate. I would add that lack of density in the session area of a segment with no apparent diameter stenosis often is the result of a very eccentric lesion seen in a 90-degree alternate view. And FFR is the best to know flow!
Aaron V. Kaplan, Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire: My two thoughts: 1) The cornerstones of our knowledge base are derived from studies utilizing angiographic core labs. Each core lab has its own specific protocols, which I believe typically utilize the “worst” view. I would defer to other group members who manage core labs to comment (especially if they corroborate my statement).
2) Caution should be used when reporting visual estimates using % diameter stenosis, as this implies an accuracy that can only be obtained utilizing sophisticated quantitative coronary analysis (QCA) algorithms. I advocate using descriptive terms, e.g., “high grade” >70%, intermediate (30-70%), and low grade (<30%), which are more reflective of the precision (or lack thereof) of our assessment.
Mort Kern, Long Beach, California: Aaron, I concur wholeheartedly. I have suggested more than once we drop the % narrowing in exchange for mild, moderate, severe designation, which is about as precise as one can be visually.
Jim Blankenship, Hershey, Pennsylvania: The idea of rating lesions as mild, moderate, or severe makes sense. One would probably add a grade for “critical” or “flow-limiting”. The traditional teaching has been to call the lesion by its worst view. I do that for concentric lesions. For obviously eccentric lesions, I usually mentally average several different views. That is, for a lesion that is 70% in one view and 30% in the orthogonal view, I will call it “50%”, or report it as “30-50% in different views”.
Lloyd Klein, Chicago, Illinois: Mort, wait…you mean angiographic interpretation is an art?
I also was taught that eccentric lesions should be interpreted in their worst view. This never made sense to me, since the area of an ellipse is calculated by both the two diameters. Perhaps it makes more sense if we consider the critical importance of minimal diameter in hemodynamics.
Still, there is an art to it. Is the discrepancy really eccentricity or is it poor visualization, calcification, overlap, thrombosis, bifurcation geometry, recanalization, etc.? Does the lesion match what we know about the patient, the stress test, the regional function, and so on? Can an anatomic measure ever be conclusive for this question? I doubt it.
Bonnie Weiner, Worchester, Massachusetts: I disagree. First, as far as core labs are concerned, some of the quantitative systems incorporate a videodensitometric parameter to take into account the eccentricity of lesions, so “worst view” isn’t always what is reported. The increasingly available 3-D angiographic evaluations add both accuracy and precision in these measurements. There are good studies that have correlated visual estimates of diameter stenoses with core lab estimates. A general rule of thumb is that the core lab determinations are around 15% lower (there is a range to this, but rarely is it more than 20%). It is not hard to recalibrate the eye to this, but most do not. That having been said, I don’t like the more generic descriptions. Many of the outcome studies do use visual estimates, and there are clearly differences between 50% stenoses and 70+ % stenoses. Again, a good rule of thumb is that if you can see a lumen, the stenosis cannot be more than 80% and is likely closer to 70%. If you include TIMI flow, then if flow is “normal” again, the stenosis cannot be more than 90-95%. We have talked in terms of diameter stenosis for so long, that to abandon it would I think be unwise.
John Hodgson, MetroHealth Medical Center, Cleveland, Ohio: Having done some core lab work, it is typically reported in the worst view, although on rare occasions, the protocol calls for analyzing two views and averaging. Of course, none of these methods integrate the density of contrast in the lesion, which I generally use to “modify” my severity reading. We did some work in the 80’s using densitometry, which worked well, but is impractical for day-to-day use.
I teach our fellows to use general ranges (i.e., 50-75) when describing lesions. I also use an overall modifier, such as “diffuse mild disease up to 20% narrowing”. So, I basically agree with Aaron and Mort. About 8 years ago, I helped ACC with a data standards paper where we specified how to report lesions (Table 1).2,3
Mitchell W. Krucoff, Duke University, Raleigh, North Carolina: As Bonnie and others point out, there is a well-defined frame shift between core lab and visual estimates, as the heuristic cues that live doctors use and the digital definitions that computer algorithms use lead to numerically different places. That being said, these differences have never been related to superior correlation with clinical outcomes, physiologic measures, etc. Core labs have become essential to device evaluations, because their results are more consistent/less variable, with lower inter- and intra-observer variability, than what you get with interventional cardiologists.
I was also taught the “take the worst view” as the gospel of angiography. We knew before, but now FFR clearly proves that [the worst view approach] is way oversimplified. There are two additional points that belong in this e-dialogue:
- Averaging views sounds nice, but angiographic views may create pseudo-asymmetry through subtle degrees of foreshortening, either of the lesion or of the adjacent reference vessel.
- Having spent months in the 70’s with Dr. Bill Roberts slicing coronaries into 1-mm segments (no IVUS [intravascular ultrasound] back then!), we also need to remember the lesion “severity” is also highly affected by amount of disease in the adjacent “normal” reference vessel segment. Today, for example, IVUS will tell you that an angiographic 50% lesion is much worse than that in India, where the whole reference vessel is packed with 40% diffuse disease.
So indeed there is a lot to the “art” of reading/interpreting angiographic lesions related to clinical syndromes. What should really scare us all are how many studies have “validated” non-invasive diagnostic tests using angiographically documented lesions as the “gold” standard (certainly more like tin…).
Sam Butman, Verde Valley Medical Center, Cottonwood, Arizona: This is truly a most excellent, albeit fundamental, question of what we do, what we have been taught, and oy, what we have taught so many others. I am in the Kern et al camp, only because I am as old as Mort and whoever else uses the worst view stenosis. I also favor the generic tight, severe, moderate, and mild description because as we ALL know, with the high inter- and intra-observer variation among expert angiographers, to say more is just silly and implies “we know.”
Having said that, good luck putting those words on your daily ACC [cath] form. So, we are back to numbers. We need a truly scientific base here, and while interesting and even pleasing, surely a consensus [reading] is not the best method. Mort, as another in this erudite group...fix it!
Mort Kern, Long Beach, California: Sam, I did, when I said use FFR.
Carl Tommaso, Chicago, Illinois: This discussion brings to mind a story that I need to share. In 1970, I was a third-year fellow. At the AHA Scientific Session was an evening session entitled “How to grade a coronary stenosis”. The first speaker was either Dick Conti or Carl Pepine from the University of Florida, who described their method, which was to use a caliper and compare the stenosis to a normal diameter of the coronary in two orthogonal views, plug this into a formula they had derived, and report the stenosis to two decimal places. The next speaker was from the Cleveland Clinic, who noted they were a very busy lab and did a visual estimate of the stenosis, and reported it as 0, 30-50%, 50-75%, 75-90%, 99%, or 100%. The final speaker was Dr. Goffredo Gensini in private practice from Syracuse, who said, “Gentlemen, there are only three classes of coronary stenosis: 1. That’s a beautiful coronary artery; 2. That’s a terrible coronary artery, and 3. [As he slapped his forehead]Oh, my God.”
I share this story, because 45 years later we are still asking the same question and still have the same answers!
Jeff Popma, Boston, Massachusetts: Our core lab averages two views by protocol. But, as a clinician, I typically report the “worst” view. Clinicians’ eyeballs have been retrained. The QCA algorithms have become better at tracking the true stenosis with less rigorous smoothing algorithms. The discrepancies are not as bad as they once were. The core lab QCA algorithms are much superior to the “online” algorithms, which are still 1st-2nd generation algorithms. At the end of the day, I trust my eye in the lab, and I grew up with QCA.
Recently, we learned from ABSORB III that the clinicians do overestimate reference vessel diameter (RVD) by about 0.25mm versus QCA; but again, the QCA algorithms were developed for reproducibility and not clinical relevance. When in doubt about RVD, IVUS or OCT [optical coherence tomography] is best. I am sympathetic to the idea of ordinal rankings. I learned this from Lowell Satler and Gus Pichard 20 years ago at the Washington Heart Center when I was a kid in the lab. FFR has taken away almost all the need for quantitative measurements in the lab; any lesion is game for FFR to know clinical importance for those lesions between 40-70%. The good news is that by our review of cath reports from all you guys in clinical trials, the days of 90% pre and 0% post are gone. Everyone is trying harder.
Ajay Kirtane, Columbia University, New York City, New York: Very well said, Jeff. I couldn’t agree more, and “growing up” in Mike Gibson’s core lab right next to yours, I have always questioned the overall rationale of visually assessed “percent diameter stenoses”, both from an accuracy standpoint as well as from a clinical one. It always struck me when reviewing cath reports and films that we used to see lesion severity “called” by what made clinical therapeutic sense for the patient. But this doesn’t necessarily imply that operators were always gaming the angiogram.
The reason I was taught during fellowship by many mentors to assess lesions in multiple views including the “worst” view (but without foreshortening or some artifact that could make a less significant lesion look worse), was so that I didn’t miss an eccentric stenoses that could be clinically significant. Notably, this was not to treat unnecessary or clinically irrelevant disease. Rather, this was to give clinically appropriate patients (patients who came to the lab for symptoms, significant ischemia, etc.) the benefit of the doubt when they were in the lab. As always, the decision to treat (or not) was based upon the clinical picture in the context of pre-cath ischemic testing or FFR if the angiogram was equivocal. And this was before FAME and other trials substantiated this practice. That’s why decisions to treat solely based upon visually assessed lesion severity of “70%” or greater generally never really made sense to me.
Thankfully, as a field, we have gone away from this practice. I concur that collectively we are doing much better now. As a result, with the use of a more realistic/objective assessment of the purpose of the angiogram, combined with the appropriate use of FFR and (let’s not forget) sound clinical judgement based upon the patient’s presenting symptoms/syndrome, we have somewhat quieted the undercurrents we started to hear 10 years ago for online QCA or QCA reviews for every case. QCA is a great tool for standardization across studies and can reduce the variability in what stenosis severity is called visually in the cath lab, but as we well know, QCA %DS [diameter stenosis] cannot and should not be the only arbiter of the clinical significance of a lesion.
Mitchell W. Krucoff, Duke University, Raleigh, North Carolina: For all the attractive simplicity of the tripartite “min/moderate/severe” approach, in the same way, it is silly to report stenoses to 3 decimal points. We should be cautious about oversimplification as well. I was trained with numeric ranges (0-25, 25-50, 50-75, 75-99, 100) in Dr. Kenny Kent’s lab when he was at Georgetown University. From a laminar flow dynamics (the old conversion of edge-edge diameter reduction 50% = 75% cross-sectional area with perfusion pressure loss across the lesion) theoretical basis, 0-25 and 25-50% implied clinically insignificant plaque mass. When I moved to Duke University in 1987, lesions were characterized numerically for the cardiac database, and 50% represented a clinically significant lesion. But, in both of these systems, quantification supported statistical analyses of lesion severity as a continuous variable, and from the BARI study to the era of stents, that notion was extended to late lumen loss and other important surrogate measures (in core labs with reproducible methods, etc.).
Historically, I think it is important we don’t throw out the baby with the bathwater. This is important not only for clinical guidelines, but also for developing technology. In labeling, is FFR for moderate lesions OK, or is >40% and <90% more likely to create predictable/generalizable instructions for use? And, of course, none of this discussion has touched on the length of a lesion in conjunction with flow reduction.
Allen Jeremias, Columbia University, New York City, New York: I think Ajay’s points are well taken and, while we base most of our clinical decision making on the angiogram, I am less and less convinced that this is the right thing to do. I’m not sure how many of you noticed this recent Circulation4 publication of an international survey study called ISIS (maybe the worst name for a study in the history of clinical trials), which surveyed almost 500 experienced interventional cardiologists to review 5 angiograms and assess lesion severity. All lesions were in the intermediate range and had a QCA done that was not revealed to the participants. It turns out that the inter-observer variability for EVERY single lesion ranged from 30-90%, independent of the actual lesion severity by QCA. Obviously there is always some degree of variability, but it is difficult to explain how one interventional cardiologist can call a stenosis 30% and the next interventional cardiologist calls it 90% (Figure 2).
These operators also had to make a decision whether to treat each lesion medically or by PCI, or if they needed additional studies (i.e., FFR or intravascular imaging). As it turns out, in 70% of the cases, the operators were confident making a treatment decision based on the angiogram alone and they were right about HALF the time (based on an FFR cut-off point of 0.80, Figure 3). It seems a little concerning to me that our ability to make the right decision by looking at an angiogram is no better than a coin toss.
Ajay Kirtane, Columbia University, New York City, New York: It is very important to note, though, that the study in question only included focal intermediate stenoses, NOT severe stenoses or non-obstructive disease, and did not include a clinical context or antecedent ischemic testing. What you suggest is far less of a concern when looking at truly severe stenoses, prognostically important disease, or patients with prior ischemic testing. I’d even bet that if an intermediate stenosis study like this one were extended beyond angiography alone to include clinical presentations, etc., there would likely be more concordance with respect to in-lab decision making, further testing, and treatments. The “right decision” cannot be determined by looking at the angiogram out of clinical context.
Allen Jeremias, Columbia University, New York City, New York: True, these were all intermediate lesions based on QCA, BUT this is obviously not what the participants thought when they graded these stenoses either 30% or 90%. At the end of the day, the ad hoc lesion severity at the time of the angiogram is what matters for treatment decisions, not a QCA done later in the core lab. Clinical history will certainly help, but how often is that clouded by what we see on the angiogram?
Jeff Moses, Columbia University, New York City, New York: Allen’s points are well taken, but we are rediscovering the wheel. At Lenox Hill in the late 80’s/early 90’s, we invested a lot of time and money in online QCA and reported [the stenosis diameter] to the single digits. Then CFR [coronary flow reserve] and FFR came along, making all this pseudo angiographic precision a waste of time for clinical decision making. The paper you quote is precisely why all lesions eyeballed at <90% have high rates of FFR >0.8 (see FAME). We need a “number” for audits, as the guidelines prescribe, with 70% as the threshold. If we drop it to a semi quantitative scale as suggested, it will just add another opportunity for confusion. Our clinical decisions are made with a combination of lesion severity, clinical syndrome, and functional data. That’s as far as angiography can take us. As far as the “worst” view, that’s the best standard, as getting an orthogonal view with minimal foreshortening and overlap is rarely achieved. Given all the limitations of our “number”, why add to a situation that’s already not so easy to explain to the laymen who ask, “How much is it blocked, doc?”
Gregg W. Stone, Columbia University, New York City, New York: I would agree that the angiogram is a terrible discriminator of the true cross-sectional narrowing of a lesion, either compared to the adjacent reference segment or cross-sectional measurement (which of course can’t be appreciated at all by the angiographic lumenogram). There is a great deal of inter-observer variability with angiography. QCA reduces some of the inter-observer variability, but issues of foreshortening and eccentricity remain. In this regard, IVUS is much better, and in general, the absolute lumen cross-sectional area has been more useful than the minimal lumen diameter (MLD) in any one direction in correlating with FFR, although even IVUS has, at best, only a modest correlation with FFR.
In terms of dimensions, it appears OCT is closest to “the truth”, with QCA underestimating vessel dimensions, while IVUS overestimates. But at the site of greatest plaque burden, because of limited penetration, OCT doesn’t visualize the entire vessel in a large proportion of cases. Finally, while FFR is the gold standard for ischemia, the reason ischemia is important (as relates to its relationship with cardiac death and MI [myocardial infarction]) is that it is likely a surrogate for vulnerable plaque; more severe lesions have greater plaque burden, which are more likely to be fibroatheromas, and have abnormal shear stress and endothelial dysfunction. But that’s a whole other discussion. Bottom line — unless the angiogram is truly normal or shows severe disease, the expert (read: competent?) interventionalist should use physiologic lesion assessment and/or intravascular imaging to make fully informed treatment decisions.
Jeff Moses, Columbia University, New York City, New York: Exactly. Concordant non-invasive stress testing obviates the need for most FFR when they are part of the picture.
The bottom line
Mort Kern, Long Beach, California: The angiogram is the oldest coronary imaging tool and the initial basis for nearly all important decisions in invasive and interventional cardiology. Yet after more than 50 years, today we still have differences of opinion and lack of guidelines on how to interpret a coronary angiographic narrowing. After hearing from our colleagues, I believe we can agree on a couple of major points.
- When reporting stenosis severity, use the worst view, i.e., the most severe narrowing of any view. Add descriptive comment to enhance concentric vs eccentric narrowing and other features which play into the clinical meaning of the angiogram.
- Use a reporting nomenclature that provides a range of stenosis (Table 1) for quantitation of an angiogram that cannot be more precisely quantified visually.
- QCA may have its special rules on defining stenosis with modes for combining views, added algorithms for stenosis edge detection, and area computation, but its use is mostly for research consistency.
- Finally, if the decision to treat a patient from an angiogram needs to be confirmed (due to uncertainty in the clinical presentation, stress test, etc.), use FFR to provide objective data to justify your intervention in the absence of other persuasive information.
- Kern M, Sorajja P, Lim MJ. The Cardiac Catheterization Handbook. 6th ed. Philadelphia, PA: Elsevier; 2016: 131.
- Douglas PS, Hendel RC, Cummings JE, Dent JM, Hodgson JM, Hoffmann U, et al; American College of Cardiology Foundation (ACCF). ACCF/ACR/AHA/ASE/ASNC/HRS/NASCI/RSNA/SAIP/SCAI/SCCT/SCMR 2008 Health Policy Statement on Structured Reporting in Cardiovascular Imaging. J Am Coll Cardiol. 2009 Jan 6; 53(1): 76-90. doi: 10.1016/j.jacc.2008.09.005.
- Hendel RC, Budoff MJ, Cardella JF, Chambers CE, Dent JM, Fitzgerald DM, et al; American College of Cardiology (ACC); American Heart Association (AHA). ACC/AHA/ACR/ASE/ASNC/HRS/NASCI/RSNA/SAIP/SCAI/SCCT/SCMR/SIR 2008 Key Data Elements and Definitions for Cardiac Imaging: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Data Standards (Writing Committee to Develop Clinical Data Standards for Cardiac Imaging). J Am Coll Cardiol. 2009 Jan 6; 53(1): 91-124. doi: 10.1016/j.jacc.2008.09.006.
- Toth GG, Toth B, Johnson NP, De Vroey F, Di Serafino L, Pyxaras S, et al. Revascularization decisions in patients with stable angina and intermediate lesions: results of the international survey on interventional strategy. Circ Cardiovasc Interv. 2014 Dec; 7(6): 751-759. doi: 10.1161/CIRCINTERVENTIONS.114.001608.