Abstract & Commentary
Risk Model for Transcatheter Aortic Valve Replacement
By Jeffrey Zimmet, MD, PhD
Associate Professor of Medicine, University of California, San Francisco, Director, Cardiac Catheterization Laboratory, San Francisco VA Medical Center
Arnold SV, et al. Predictors of poor outcomes after transcatheter aortic valve replacement: Results from the PARTNER trial. Circulation 2014; May 23. [Epub ahead of print.]
The introduction of transcatheter aortic valve replacement (TAVR) has been a transformative step in the treatment of severe aortic stenosis. Patients who are ineligible or very high-risk for open surgery are now routinely treated with this less-invasive option. The patients eligible for this procedure are by nature medically complex, most often are of advanced age, and have comorbidities that in and of themselves can present a continued threat to life and well-being, even after successful TAVR. Recognition of this problem has not been lacking. In the PARTNER trial using the Edwards Sapien device, patients in the inoperable cohort who received the device still had a more than 30% risk of death at 1 year. In the "high-risk" cohort where the transcatheter procedure was compared with surgery, more than 24% of TAVR patients were dead at 1 year. It appears obvious that procedures might be avoided in patients who will not derive a mortality benefit. But mortality is not the whole story here. In any such cohort of elderly patients with extensive comorbidities, simply prolonging life without also improving quality of life (QoL) cannot be considered a victory. Attempts to predict which patients will benefit from TAVR should include both survival and QoL.
In their recently published work, Dr. Arnold and colleagues present the results of their prediction model for poor outcomes after TAVR, applied to the PARTNER trial database. For the purposes of this study, poor outcome was defined as the combined endpoint of death, very poor QoL, or a significant decrease in QoL compared with baseline. Patients in the trial and in the continued-access database were assessed with the Kansas City Cardiomyopathy Questionnaire and with the 6-minute walk test at predetermined time points (baseline, 1, 6, and 12 months) following randomization. For the initial model, outcomes were examined at the 6-month time point. Poor outcome was defined specifically as death or poor health status (Kansas City Cardiomyopathy Questionnaire overall summary [KCCQ-OS] score < 45) or a decline in health status (decrease of ≥ 10 in the KCCQ-OS score from baseline to 6 months).
Out of 2137 patients who received TAVR with the Edwards Sapien device through the PARTNER trial or the continued access registry, 704 (33%) were found to have a poor 6-month outcome by this definition. Of these, 19% died, while 12% showed poor QoL on the KCCQ scale and 2% demonstrated decreased QoL. A second expanded definition model looked at outcomes at 1 year, and defined poor outcome as death, KCCQ score < 60, or a decrease of at least 10 points in the KCCQ score compared with baseline. In both models, the most striking predictors of poor outcomes were lower mean aortic valve gradients and poor performance on the 6-minute walk test. Other significant predictors included O2-dependent lung disease, kidney disease, and worse baseline cognitive dysfunction. Interestingly, the Society of Thoracic Surgery predicted risk of mortality (STS-PROM) score was not a predictor in these models.
The 6-month model was validated by stratifying patients into low- (< 25%), intermediate- (25% to < 50%), and high-risk (≥ 50%) groups based on their predicted risk of poor outcomes. At the 6-month time point, 31% of high-risk patients died, while an additional 24% had very poor QoL or a decline in QoL. In the intermediate-risk group, 17% had died, and an additional 20% had a very poor QoL or a decline in QoL, while among low-risk patients, only 8% had died and 10% had a very poor QoL or a decline in QoL. Stratification into low-risk, intermediate-risk, high-risk, and very high-risk groups was performed with the 1-year model, with similar results. The authors concluded that validated predictive models can identify patients at high risk for a poor outcome after TAVR, but their discrimination ability was only moderate.
Commentary
One of the great strengths of the PARTNER database is the inclusion of validated measures of patient health status (such as the KCCQ and the 6-minute walk test) along with hard clinical endpoints. In turn, the novelty of this prediction model is the inclusion of patient-reported outcomes alongside mortality as an essential piece of the puzzle. The inclusion of QoL measures focuses on issues that are often lost in the clinical decision-making process, but that are clearly important to patients. The fact that one-third of all TAVR patients experienced a poor outcome at 6 months post-procedure should serve as a caution for clinicians who are involved in patient education and selection for these procedures. As for many prediction models, however, the question of actual clinical utility looms large. An accompanying editorial in Circulation is appropriately titled "Can we predict who will be alive and well after TAVR? Is that useful to individual patients?" This model is certainly a step in the right direction, but as is often the case, direct application to patient care is some distance away.
