Risk Prediction Model Personalizes Revascularization Options for FREEDOM-Like Patients

(UPDATED) A risk prediction tool that incorporates multiple clinical variables, including age, kidney function, and left ventricular ejection fraction among others, can help personalize the best revascularization option for patients with multivessel coronary artery disease and type 2 diabetes, according to the results of a new study.

The model drew on data from FREEDOM, a large-scale randomized trial that showed surgery to be superior to PCI in patients with multivessel CAD and diabetes. For their new analysis, researchers—including the two FREEDOM principal investigators—report that their model can estimate the long-term risk of both major adverse cardiovascular events and angina and, for the cohort as a whole, supports the trial’s original conclusion that CABG surgery is the preferred strategy, particularly in those with a history of smoking.

But in an important new finding, the risk prediction model was able to identify 45% of patients in the trial who would be expected to have similar risks of MACE at 5 years if treated with either PCI or surgery. 

Lead investigator Mohammed Qintar, MD (Saint Luke’s Mid America Heart Institute, Kansas City, MO), said the researchers imagine the risk-prediction tool, once it has been prospectively validated and tested at the point of care, could be used as part of the heart team discussion with patients deciding between PCI and CABG. “[I see it] definitely not as a stand-alone tool but rather as a complimentary validated tool,” he said. “The next step forward is to obviously prospectively validate the use of this tool in real life and study outcomes associated with its use.”

Michael Farkouh, MD (Peter Munk Cardiac Centre, Toronto, Canada), who led FREEDOM along with Valentin Fuster, MD (Icahn School of Medicine at Mount Sinai, New York), said the risk score is an attempt to select the best revascularization strategy for patients with multivessel disease and diabetes. The risk model can help identify patients who would clearly benefit from bypass surgery as well as those who might fare equally well with PCI, which is important when factoring in the availability of healthcare resources and patient preference.

In no way, said Farkouh, is the risk model meant to diminish CABG surgery in FREEDOM-like patients. “No one did better with PCI,” said Farkouh. “The question is can we identify those that really need bypass and make sure we don’t miss those patients.”

Not everyone is persuaded that this tool adds anything beyond the kind of individualized decision-making heart teams should already be doing. Arnold Seto, MD (Long Beach VA Medical Center, CA), who was not involved in the study, noted to TCTMD that patients, as well as the referring interventionalists, have a strong preference for avoiding CABG surgery given higher morbidity associated with the procedure but said that decision is best made by a multidisciplinary team.

“At extreme ends, people who have a very high SYNTAX score and a couple of chronic total occlusions, even an interventionalist will say, ‘I can’t revascularize this—you should go with the surgery,’” said Seto. “For the low-risk patient, such as those with three simple focal lesions, the interventionalist will often go ahead and fix it. This is why the societies all recommend a heart team approach to help decide on revascularization with PCI or CABG. That’s probably the best way to go about it.”

Depending on the patient profile, the risk of MACE with PCI and CABG surgery can vary, he added. “It’s not always clear-cut,” said Seto. “We make our best guess but that’s based on the discussion between what the surgeon and interventionalist think.”

“In this day and age, a real heart team approach to patients with multivessel CAD and diabetes mellitus does not exist in many places,” countered Qintar. “The TAVR and [transcatheter mitral valve replacement] heart team is more developed. In an ideal world, a heart team approach with the help of personalized risk models will help patients make more informed decisions and will help them get the therapy that will likely benefit them the most.” 

No Patient Is Truly Average

The FREEDOM trial, published in 2012, randomized 1,900 patients with multivessel CAD and diabetes to CABG surgery or PCI. At 5 years, the primary composite endpoint of death, stroke, and MI was significantly lower among patients treated with surgery (18.7% vs 26.6% with PCI; P = 0.005). Longer-term follow-up out to nearly 8 years revealed that surgery retained a significant survival advantage.

As a result, US and European guidelines state CABG is the preferred revascularization approach for FREEDOM-like patients who are good candidates for surgery.

“In every trial, there are patients who get benefit, patients who are neutral, and patients who are harmed,” said Qintar. The trial reports the mean effects, or overall results, and while FREEDOM showed superiority of CABG over PCI, it is not known if the individual patient benefited, was harmed, or had no treatment effect, he said. “Medicine is moving into precision and personalized medicine, and this project is trying to achieve just that.”   

The researchers devised two models with variables available at the time of treatment. The model for predicting MACE risk included age, body mass index, history of smoking, history of MI, history of stroke, insulin status, estimated glomerular filtration rate, and left ventricular ejection fraction. The risk model for predicting angina included age, sex, ACS presentation, hemoglobin levels, angina frequency at baseline, and SYNTAX score. The risk model for MACE had a C-statistic of 0.67, indicating a modest discrimination between observed and predicted events, while the angina model had a C-statistic of 0.64.

The MACE risk-prediction model included a strong interaction with smoking status, such that all patients who had a history of smoking had a lower risk of events with CABG compared with PCI. Overall, 54.5% of patients would be predicted to have a lower risk of MACE with CABG surgery, while 45.5% of the FREEDOM patients would be predicted to have an equivalent risk of MACE with either PCI or CABG. No patient had a lower risk of MACE with PCI.

The angina risk-prediction model had a treatment interaction with the SYNTAX score where all patients with scores > 22 would be predicted to have better angina relief with surgery. Those with lower SYNTAX scores (≤ 22) would be expected to achieve a similar degree of angina relief with CABG or PCI. The model estimated that 65% of patients would have a greater reduction of angina symptoms with CABG while 35% would have a similar extent of angina relief with either type of revascularization procedure.

Cardiothoracic surgeon Robbin Cohen, MD (University of Southern California, Los Angeles), who was not involved in the study, told TCTMD that although FREEDOM showed a clear win for CABG surgery over PCI, no study applies to every patient.

“Are there patients in this group—even though surgery is never inferior—who could be stented instead of having a big operation and still have equivalent, or close to equivalent results? That’s a well-meaning hypothesis,” Cohen said. “In the real world it’s true we see some patients who are diabetics and who have three-vessel coronary disease but for one reason or another you ask yourself if it’s really the right thing to operate. They might have other comorbidities, such as cancer. We often get referred patients who according to FREEDOM should do better with surgery but just aren’t great surgical candidates.”

Seto said the “general sense” among interventional cardiologists is that patients with a SYNTAX score < 22 can undergo PCI given equivalent outcomes with surgery and less morbidity. For those with a SYNTAX score 23-32, it might be “worth discussing with the patients their preferences for worse outcomes with PCI but less morbidity,” he said. Noninterventional cardiologists and certainly surgeons might counsel these same patients differently, arguing that those with an intermediate SYNTAX score might derive greater benefit with CABG surgery, said Seto.

Does This Help Inform Heart Team Discussions?

At this stage, Seto said he’s uncertain whether the risk-prediction tool devised by Qintar and colleagues helps inform the heart team discussion.

“I don’t think the data are strong enough to overcome the clinician’s better sense of confidence in their PCI ability, their confidence in treating [chronic total occlusions], for example, or the surgeon’s confidence in taking on a larger number of comorbidities,” he told TCTMD. The C-statistic is also relatively weak, said Seto. “I would argue that this model, built on only the results of the FREEDOM trial, is more limited than a model that somehow incorporated the results of the many other prior multivessel PCI versus CABG trials performed in the past, especially SYNTAX.”

To TCTMD, Qintar said there are data from Canada suggesting that 40% of patients with diabetes and multivessel CAD are undergoing PCI, but the reasons for this are uncertain. Providing a risk-prediction model—they are calling this one the FREEDOM score—can help frame the decision better for patients and physicians.

“Physicians are often experienced in estimating risks but they are far from perfect,” said Qintar. “Our patients are getting more complicated as they survive longer, and it is getting harder to estimate risks. We think that validated risk models help tremendously in deciding revascularization strategies. In left main disease and multivessel disease without diabetes, the SYNTAX score has been fundamental in these decisions and they should be calculated. We believe that the FREEDOM score will help patients and heart team alike in making the best treatment decisions.

For Cohen, the risk prediction tool is a means to ask if physicians can do better for the individual given the results of a large randomized controlled trial. As a surgeon who believes in the FREEDOM trial, and who believes that surgery is the best approach for patients with diabetics with multivessel disease, Cohen said he is quite comfortable with research seeking to identify patients who could avoid surgery if comparable outcomes could be achieved.

“They didn’t say the [risk prediction] tool invalidates FREEDOM,” he said. “They were careful not to do that. All they said is that in the era of the heart team approach, here is a tool that might even improve our ability to come up with the best treatment for every individual patient. That’s a well-meaning, thoughtful statement.”

In an editorial, James Blankenship, MD (Geisinger Medical Center, Danville, PA), and H. Lester Kirchner, PhD (Geisinger Health System), state that at “first blush” the risk-prediction models might be a “transparent attempt by interventionalists to justify recommending PCI to diabetic patients with multivessel disease, thumbing their noses at randomized controlled trials, cohort studies, meta-analyses, and editorials that uniformly support CABG in this patient population.”

They suggest that “perhaps” a personalized risk model might be a useful tool, noting that personalized risk models, not unlike the CHA₂DS₂-VASc score, can identify patients with the most to gain or lose from treatment. While they point out that diabetic patients with multivessel CAD as a whole do better with CABG than PCI, a fact consistently observed over the past 25 years, it is “plausible“ FREEDOM-like patients might do equally well with either revascularization strategy, the editorialists agree.

“It is the clinician’s responsibility to present to the patient objective data regarding the merits of different approaches as well as clinical judgment,” write Blankenship and Kirchner. “Use of the personalized risk model by Qintar et al may help clinicians provide more objective recommendations and may help patients make more rational decisions regarding revascularization strategies.”

They point out, however, that CABG has been shown to be superior regardless of insulin-requirement status, the presence of chronic kidney disease, lung disease, or peripheral vascular disease, presentation (ACS vs stable ischemic heart disease), and age, sex, and ejection fraction.