Editorial NexoScience
A replacement heart valve slides into place, the balloon comes down, and the team watches the screen. Almost every one of these procedures leaves a hairline gap somewhere around the rim of the device, and a thin jet of blood slips backward through it with every beat. For years that small backflow was filed under harmless. A large pooled analysis says it is not.
The gap has a name, paravalvular leak, a stream of blood that escapes around the outside of an implanted valve instead of passing through it. It is common after transcatheter aortic valve implantation (TAVI), the procedure that threads a replacement aortic valve up through an artery instead of opening the chest. Researchers reported November 14, 2022 in the journal Structural Heart that patients left with any leak at all, including the mildest grade, went on to die and to land back in hospital more often than patients whose valves sealed cleanly.
Mild leaks are not the harmless category they were assumed to be. Set against patients with no leak or only a trace of one, those with a mild leak carried roughly a third higher risk of dying during follow-up (hazard ratio 1.35). They were also readmitted to hospital more often, and they died of cardiovascular causes more often. Until now the evidence on mild leaks had pointed in both directions, which is why the grade was widely treated as a cosmetic finding on the discharge echo.
The risk grows the longer patients are followed. Analyses of this kind normally assume a risk gap that stays steady over time. Here it did not. The statistical check for that assumption failed, and a flexible model showed the distance between leaking and sealed valves widening as the years passed instead of closing. The direction matters, because it means the headline figure probably understates the harm rather than inflating it. It also fits an earlier trial in intermediate-risk patients that found no association at first, yet whose survival curves for mild leaks began to drift apart from the rest after about two years.
The damage tracks the size of the leak. When patients were sorted into three groups rather than two, the pattern lined up in order: a mild leak came with about a third more deaths than no leak at all, a moderate or severe leak with more than twice as many (hazard ratios 1.36 and 2.30). The same graded pattern held for readmission and for cardiovascular death. The danger from a moderate or severe leak also showed up with both balloon-expandable and self-expandable devices, so it is not a quirk of one valve design.
The team rebuilt the raw data from published survival curves. Instead of pooling summary results, the researchers searched PubMed/MEDLINE, EMBASE and Google Scholar for TAVI studies that published at least one Kaplan-Meier curve, the stepped graph that tracks how many patients in a group are still alive as time passes. Software then worked those curves backward into individual patient timelines, which were merged into a single dataset. Thirty-eight studies cleared the bar, together covering more than 25,000 patients, most of them in their eighties.
A new leak lands on a heart that has had no time to adapt. A leaky native aortic valve usually goes bad over years, and the left ventricle thickens and stretches to absorb the extra volume, which is why some people stay free of symptoms for a long time. A leak created during a procedure arrives all at once, on a ventricle that has had no warning. That may explain why readmissions cluster in the months right after the implant, before the heart has built any compensation, while the risk of dying keeps climbing long afterwards.
The authors are explicit about what this analysis cannot do. They describe the whole approach as hypothesis generating, able to show an association with biological plausibility but not to prove that the leak causes the deaths. Most of the pooled studies were unadjusted cohorts, and the authors acknowledged that patients who ended up with a leak may simply have been sicker to begin with, which could account for worse survival on its own. Because the analysis works from published graphs rather than original records, the researchers could not correct for age, sex, other illnesses or heart function. They also warn that few patients were still being followed into the second and third year, so the long-term stretch of the result is the least solid part of it.
The next question is what to do about a leak that used to be left alone. The authors argue that real effort should go into avoiding any degree of leak during the procedure, that open surgery may be the better option for patients whose anatomy makes a leak likely, and that re-intervention deserves consideration even for a mild leak when a patient keeps coming back with heart failure. That calculation gets sharper every year, because TAVI is being offered to younger patients at lower surgical risk, people who will live with the same valve, and the same small jet of blood, for decades.
The findings come from the study ‘Impact of Paravalvular Leak on Outcomes After Transcatheter Aortic Valve Implantation: Meta-Analysis of Kaplan-Meier-derived Individual Patient Data’, published by the researchers who carried out the analysis on November 14, 2022 in the journal Structural Heart, the official journal of the Society for Cardiovascular Angiography and Interventions. DOI: 10.1016/j.shj.2022.100118.
