By Michael Crawford, MD, Editor

SYNOPSIS: A study of all patients who underwent transcatheter aortic valve replacement over eight years in France was used to develop a futility score that would help predict who would not live one year after the procedure. This simple clinical score based on comorbidities predicted who would live one year with 95% specificity.

SOURCE: Lantelme P, Lacour T, Bisson A, et al. Futility risk model for predicting outcome after transcatheter aortic valve implantation. Am J Cardiol 2020;130:100-107.

Transcatheter aortic valve replacement (TAVR) is a mature technology that carries solid success rates in high-risk surgical patients. However, some patients survive less than a year despite a technically successful procedure.

Lantelme et al identified such patients using simple clinical variables to establish a futility score (FS). Using a national database, they studied all patients in France who underwent TAVR between 2010 and 2018. The resulting population of 20,443 patients was divided into a derivation cohort (10,221) and a validation cohort (10,222). The primary outcome was all-cause death. Procedural futility was defined as death occurring less than one year after TAVR. Those who died during the first year were compared to those who did not.

From 33 clinical variables, the authors constructed a FS using multivariate logistic regression. The mean age of the population was 83 years, and the mean follow-up interval was two years. The annual death rate was 15% (53% of deaths occurred during the first year, which was 18% of all TAVR patients). The final FS consisted of 14 clinical characteristics: age, heart failure, pulmonary edema, stroke, atrial fibrillation, vascular disease, chronic kidney disease, liver disease, lung disease, anemia, cancer in the last five years, metastatic cancer, depression, and poor nutrition. The maximum point score was 17, with all but pulmonary edema (3) and metastatic cancer (2) representing one point each. The scores were divided into three segments: low risk (0-4), moderate risk (5-8), and high risk (> 8).

The low-risk group was 47% of the population, the moderate-risk group was 46% of the population, and the high-risk group was 7% of the population. The one-year death rate among the three groups was 11% (low risk), 22% (moderate risk), and 43% (high risk). The FS (AUC, 0.67) outperformed the EuroSCORE II (AUC, 0.63), the Charlson Comorbidity Index (AUC, 0.56), and the Frailty Index (AUC, 0.49) for identifying futility. Similar results were obtained in the derivation and validation cohorts. The authors concluded the FS could be a relevant tool for helping identify patients unlikely to benefit from TAVR.


A mentor of mine once said that just because you can do something does not mean you have to. Another way of expressing this is asking whether it is acceptable to die without a TAVR if you suffer from severe aortic stenosis. Accordingly, the European Society of Cardiology guidelines do not recommend aortic valve replacement if the patient’s life expectancy is less than one year. This sounds reasonable, but predicting death is not easy. Other investigators have proposed various approaches to identifying those with a high risk of futility.

The FS derived from this study is one approach that performed well, but none of the proposed approaches carries an AUC higher than about 0.7 for identifying those who will not survive a year. Lantelme et al found 18% of their TAVR patients died within one year. An FS score of > 8 translated to a one-year death rate of 43%, which is similar to that reported for medical therapy of severe aortic stenosis.

The FS includes important comorbidities, but clinicians are going to struggle to remember the 14 included in the score. Most of the comorbidities are only worth one point, except for pulmonary edema and metastatic cancer. If those conditions are present, it would not take much more to cross 8 points. Without those two conditions, it would take at least eight comorbidities to hit the high-risk point. Thus, it is not surprising that only 7% of patients were high risk. Outside the high-risk group, the specificity of surviving one year was 95%. However, 22% of the one-year deaths occurred in moderate-risk patients who cardiologists probably would not exclude from TAVR. Therefore, no score is going to be an absolute cutpoint for not offering TAVR. There remains plenty of room for the collective clinical judgment of the heart team.

There were weaknesses to this study. It was retrospective and observational, and the authors used administrative data they did not check against the clinical record. There are no data available about left ventricular function, valve characteristics, or medical therapy. Also, it may be reasonable in a high-risk FS patient to consider TAVR to reduce or eliminate symptoms that are interfering with the patient’s enjoyment of his or her last year. This is a complex issue, but this simple comorbidity score concept may help with these difficult decisions.