In the 14 years since the first transcatheter aortic valve replacement in 2002, devices and delivery methods have undergone several generations of improvement, but one facet of the procedure remained largely unchanged: When prospective patients underwent preprocedural assessment to gauge their risk level, the long-standing approach to quantify their disease severity and 30-day mortality risk was to run their clinical and demographic numbers through the risk calculator developed by the Society of Thoracic Surgeons.
Although determining a patient’s Society of Thoracic Surgeons (STS) risk score using a formula based on extensive experience performing surgical aortic valve replacement (SAVR) by open surgery made a lot of sense during an era when the preeminent question was how transcatheter aortic valve replacement (TAVR) compared with SAVR, it also carried the inherent limitation of estimating a patient’s risk when undergoing TAVR based on SAVR’s track record.
That limitation is now gone.
In December 2015, a team of cardiothoracic surgeons and interventional cardiologists assembled by the STS and the American College of Cardiology placed a new risk calculator online to estimate a prospective TAVR patient’s risk for dying in hospital following a TAVR procedure. The panel developed this risk calculator with data from nearly 14,000 U.S. TAVR patients enrolled in the STS/ACC Transcatheter Valve Replacement Registry during November 2011–February 2014 and then validated it with data from another nearly 7,000 U.S. TAVR patients who underwent their procedure during March-October 2014. This created the first mortality-risk calculator for TAVR patients based entirely on experience with such patients ( JAMA Cardiology. 2016 Apr;1:46-52 ).
In May 2016, a second, independent risk calculator will go live, also based exclusively on experience in TAVR patients, that estimates a patient’s risk for either dying or having a worsened or unimproved and poor quality of life during the 6 months following TAVR. This risk calculator , developed by a team led by researchers based at Saint Luke’s Mid America Heart Institute in Kansas City, Mo., used data collected from more than 2,000 TAVR patients enrolled in the PARTNER trial or in the continued-access PARTNER registry ( Circulation. 2014 June 24;129:2682-90 ), and is offered to users by Health Outcomes Sciences , a Kansas City–based company that’s affiliated with Saint Luke’s.
Although use of these two TAVR-specific risk calculators during their early days of availability remains relatively light, TAVR experts see them as marking a new era in the workup of TAVR candidates.
“There is universal agreement that risk models must be developed based on the history of patients who actually received the treatment,” said Dr. Fred H. Edwards , the cardiothoracic surgeon at the University of Florida, Jacksonville, who led the team that developed the transcatheter valve therapy (TVT)-derived in-hospital mortality risk calculator. “Everyone realized that the STS score and the EuroScore [another operative-risk calculator historically used for prospective TAVR patients] were inadequate to extrapolate to TAVR patients.” But development had to wait until an adequately-sized experience with TAVR had accumulated. “We started the process [to develop the TVT risk calculator] when we reached about 10,000 patients” enrolled in the registry, after which it took “close to 2 years” to produce the finished product, Dr. Edwards said in an interview.
The TVT in-hospital mortality predictor gradually goes mainstream
“There is consensus that the new TVT calculator will be more reliable than the STS operative-risk model” for assessing patients being considered for TAVR, but it has not yet gained widespread use “because it is so new,” noted Dr. Edwards.
“It will take a while to incorporate it into routine practice, but I think it will be used quite a bit,” especially for “trickier and harder cases,” commented Dr. George Dangas , a professor of medicine at the Mount Sinai School of Medicine in New York and an interventional cardiologist who performs TAVR.
He also predicted that prospective patients and their families will become frequent users of the TVT in-hospital risk calculator. He sees the new risk tool as a complement to the STS risk score rather than something to replace it.
“Patients find it useful to receive an estimate of their surgical risk, and they’ll want to compare that” with their TAVR risk. “It helps to know both,” Dr. Dangas said in an interview. “Patients will likely compute it themselves.”
He foresees fairly quick integration of the new TVT score into heart team discussions as well. “The STS score will always be part of the discussion, but over time as people grow accustomed to the TVT score they will incorporate it as well. The [TAVR] community has to figure out how to use the two scores in combination.”
“There has been need for a risk calculator specific to TAVR since the patient population has only some overlap with the patients who underwent SAVR in the past,” said Dr. John D. Carroll , professor and director of interventional cardiology at the University of Colorado at Denver, Aurora, and a member of the TVT Registry panel that developed the calculator. “The STS score does not include variables that are frequently found in TAVR patients, such as liver disease, pulmonary hypertension, very-advance age, and frailty.”
A big plus for the new TVT risk calculator is its derivation from thousands of real-world, U.S. patients, Dr. Carroll noted. “It provides a patient-specific risk assessment rather than an assessment based on average patients from TAVR clinical trials.” Dr. Carroll added that since its introduction he has used the TVT risk calculator for some of his own patients being considered for TAVR.
The Saint Luke’s TAVR risk calculator
Although experts foresee an important role for profiling a patient’s predicted gain or loss in quality of life following TAVR as another element in the discussions between heart teams and patients, the risk calculator becoming available in May from Health Outcomes Sciences faces some significant issues, they said.
The Saint Luke’s Mid America team that developed the calculator used as their endpoint for a poor patient outcome 6 months after TAVR the combined rate of either death or two different measures of impaired quality of life: a Kansas City Cardiomyopathy Questionnaire–Overall Summary Scale ( KCCQ-OS ) score of less than 45 (measured on a scale of 0-100, where higher scores reflect better quality of life and function) or at least a 10-point reduction from baseline in a patient’s KCCQ-OS score.
By predicting a patient’s likelihood of emerging from TAVR with either a low or significantly worsened quality of life and function, this score is a “complement” to the TVT in-hospital mortality score, said Dr. Suzanne V. Arnold , a cardiologist at Saint Luke’s Mid America and one of the lead developers of this risk calculator.
“In-hospital mortality is an assessment of whether a patient will make it through the TAVR procedure. Our model is more about failure to recover after the procedure,” she explained in an interview. She and others also say that the Saint Luke’s calculator is a measure of probable futility when performing TAVR – that is, the likelihood that a patient will either not live long after the procedure or gain enough benefit from it to make performing the TAVR procedure an attractive option.
Dr. Arnold and her associates derived and validated the risk model using data collected from the “high risk” patients enrolled in the initial PARTNER trial and registry, so for the time being it remains primarily applicable to high-risk patients, who have classically been defined as patients with a STS risk score of 8% or greater (patients who have at least an 8% predicted risk of death during the 30 days following SAVR). She and her associates plan to see if they can validate the Saint Luke’s calculator in intermediate risk patients with aortic stenosis (usually defined as patients with a STS risk score of 4%-8%) with data collected in trials that enrolled these patients, such as PARTNER 2 . At the ACC’s annual meeting in April in Chicago, Dr. Arnold presented a report in which she and her associates further validated their model using data collected from the high-risk patients enrolled in the CoreValve TAVR trial. This validation confirmed that the model worked “beautifully,” Dr. Arnold said.
The Saint Luke’s risk calculator uses six data entries: whether or not the patient has diabetes, atrial fibrillation, or requires oxygen support at home; whether the patient has no, mild, or moderate-to-severe dementia; the patient’s mean aortic gradient; and the patient’s score on the Kansas City Cardiomyopathy Questionnaire-12 (an abbreviated version of the 23-question KCCQ-OS scale). The result it produces – the likelihood of death or poor quality of life at 6 months – is “another piece of data to help the physician, surgeon, and patient make a decision” on whether to proceed with TAVR, she explained. “I would probably advise that if the predicted risk is high, the patient consider undergoing balloon aortic valvuloplasty” instead of TAVR “to see whether the patient has some recovery before committing to TAVR, but there are no hard and fast rules,” she acknowledged.
Dr. Arnold gave this recent case she had as an example of how running the Saint Luke’s risk calculator helped decision making:
A potential TAVR patient was a man in his 60s with end-stage renal disease, oxygen-dependent lung disease, and poor functional status. Applying the risk calculator to this patient “changed the conversation a surgeon and I had with the patient,” she said. “We let him know that TAVR might not make him feel better and it wouldn’t fix his kidneys or his lungs. Given his [relatively young] age, the patient still wanted to proceed, and we performed balloon aortic valvuloplasty, after which he seemed to do much better. He eventually underwent TAVR and did okay.”
This was a good application of the Saint Luke’s prediction model, Dr. Arnold said. As a consequence of the risk quantification, it provided that “the patient more fully understood the risk of the TAVR procedure” and had a better understanding of his prospects for recovery.
The heart teams at Saint Luke’s Mid America that perform TAVR began running this risk calculator on every high-risk patient evaluated in their valve clinic, about five patients each week, starting in mid-March 2016. The results are displayed in the patient’s chart next to the STS risk score.
By late April 2016, the staff at Health Outcomes Services, the commercial partner to Saint Luke’s that will offer the calculator online to the public, planned to have this risk calculator ready for routine use online by sometime in May. They said the company will make access to it available to any physician, surgeon, or other member of a heart team at no charge.
Pros and cons of the Saint Luke’s risk calculator
While others applauded the creation of the Saint Luke’s risk calculator and the broader range of outcomes it predicts, they also questioned the generalizability of a risk assessment that is modeled against the selected patients enrolled in the PARTNER trial and registry, and some uncertainty on how to best use this information to inform clinical decision making.
“We certainly need to assess benefit as well as risk,” commented Dr. Edwards. The Kansas City–led group “is solidly on the right track,” and “I hope people will use their model,” he said.
In fact, the TVT Registry panel he heads is currently moving toward producing a similar expansion of their in-hospital mortality risk calculator that takes into account midterm changes in KCCQ status. Dr. Edwards said he hopes this expanded version of the TVT calculator might be ready in another year or two.
“KCCQ data are hard and time-consuming to collect, and clinicians who participate in the TVT Registry do not always do it,” he explained. “It’s an administrative burden” to make a KCCQ assessment both at baseline and at 6 or 12 month follow-up after TAVR.
“We’ve done a lot of education with registry participants to improve KCCQ data collection.” Dr. Edwards said he hopes that within another year this aspect of patient assessment will occur for about 90% of patients enrolled in the TVT Registry.
Dr. Edwards also highlighted the inherent limitations of applying the Saint Luke’s risk-assessment model, developed and validated in patients enrolled in the PARTNER and CoreValve trials and PARTNER registry, an aggregate of roughly 5,000 selected patients, to the more diverse patients seen in routine practice today and entered into the TVT Registry.
“The advantages of the TVT Registry are the huge numbers and the all-comers population,” he said. “You need to ask whether the randomized trial populations are truly representative.”
That’s a concern shared by Dr. Dangas, who also highlighted the challenges of assessing the risk a patient faces from noncardiovascular comorbidities and how that might affect a decision of whether or not to perform TAVR.
“These are elderly patients with many comorbidities. I’m not sure how well the noncardiac comorbidities were captured in the databases” for the PARTNER and CoreValve trials and registry, he said. “I’m not sure the noncardiac comorbidities were as well figured out” in those trials run several years ago as they might be today. “It’s not exactly what TAVR is about in 2016.” Despite his skepticism, Dr. Dangas acknowledged that a scoring formula that reliably captures and follows data collected by the KCCQ “would be useful.”
Another challenge is using a risk assessment tool that takes into account patient “frailty” as a way to judge whether TAVR might be “futile” for a specific patient, Dr. Dangas said.
“It’s a tough discussion to have prospectively with a patient. The question is valid, but how confident can we be prospectively, at the time of TAVR, that the procedure will be futile within a year?” He envisions that with more time and data, researchers will create a reliable risk assessment formula to gauge a patient’s midterm benefit from TAVR, but for now the Saint Luke’s scoring formula probably needs “more refinement,” he said.
“Patients need guidance by an individualized assessment that takes into account not only the risk they face from TAVR but also the benefit – the likelihood that they will not only be alive following TAVR but that they will gain improvement in functionality and in quality of life,” said Dr. Carroll. “This is especially important for patients who have other conditions that might prevent them from fully benefiting from TAVR, such as being on dialysis, having significant lung disease, or other conditions that could limit their ability to become more functional and feel better after TAVR.”
Although Health Outcomes Services is a for-profit company, it plans to make access to the Saint Luke’s risk calculator, which they call the “TAVI [transcatheter aortic valve implantation] Risk Calculator,” free to any health professional who applies for access, with a promised turnaround for access after application of about a day, said Joy Efron , the company’s vice president for commercialization. They do not plan to provide calculator access to patients. According to Ms. Efron, Health Outcomes Services decided to make their calculator available at no charge in the hope of potentially interesting some users in an upgraded, more extensive version of the calculator bundled with related products and features that are available as a package for a monthly subscription fee.
Dr. Edwards, Dr. Dangas, Dr. Carroll, and Dr. Arnold had no relevant disclosures. Saint Luke’s Mid America Heart Institute, where Dr. Arnold works, was an early investor in Health Outcomes Services, the company that is marketing the Saint Luke’s risk calculator.
On Twitter @mitchelzoler