The Final Meta-Analysis?

John G. Webb; Uri Landes


Eur Heart J. 2019;40(38):3154-3155. 

Just a decade ago, transcatheter aortic valve implantation (TAVI) was thought so dubious a procedure that it could not be evaluated in patients who might otherwise undergo surgical aortic valve replacement (SAVR). Consequently, the first major TAVI trials compared TAVI with medical management in patients refused surgery. In patients with a 50% mortality at 1 year, TAVI surprised even its advocates, with one of the largest absolute reductions in mortality (20%) reported in any field of medicine. With this initial success it was thought defensible to challenge open surgical valve replacement in patients at extreme, high, and subsequently intermediate surgical risk. At each risk stratum, these trials were replicated for each of the two major transcatheter balloon-expandable annular and self-expanding supra-annular valve platforms.

There have been many meta-analyses,[1–6] punctuating each step at each surgical risk level, with each of the two main valve types. Each meta-analysis rapidly fades in relevance as the next larger meta-analysis comes along. However, finally we have the major randomized trials comparing TAVI and SAVR in the lower and 'last' surgical risk strata: PARTNER 3, EVOLUT Low Risk, and NOTION.[7–9]

In this issue of the European Heart Journal, Siontis et al. examined 14 publications from seven landmark trials with 8020 patients randomized between TAVI and SAVR and with outcomes reported for a period of at least 1 year.[10] Compared with SAVR, across all risk categories and up to 2 years, TAVI was associated with a relative risk reduction in mortality of 12%, and of 17% when transfemoral access was utilized (which was the case in >90% of the patients). In addition, the risks of stroke, acute kidney injury, major bleeding, and new-onset atrial fibrillation were 19, 44, 54, and 66% lower, respectively. These advantages were demonstrable irrespective of which of the two transcatheter heart valve platforms were utilized. Most importantly, these benefits were consistent across the entire surgical risk spectrum.

So is this the final meta-analysis? Clearly there are other questions still to be answered. Up until now, major TAVR trials tell us very little about patients at 'increased risk for TAVI' such as patients with bicuspid valves, unfavourable root anatomy, or peripheral vascular disease. These trials tell us relatively little about younger patients with the potential for longevity where differences in atrioventricular conduction block, pacemakers, coronary access, durability, and repeatability increase substantially in relative importance.

The new question that will begin to redefine heart team discussions in coming years may well be 'why does this patient require open heart surgery'? Yet there are still many reasons to argue superiority for open surgery, in selected patients. SAVR may be indicated in the presence of TAVI-specific risks such as bulky or subannular calcium, bicuspid anatomy, advanced atrioventricular block, aorto-ilio-femoral disease, endocarditis, etc. There may be other conditions that are best managed surgically, such as complex coronary disease, mitral or tricuspid valve disease, or aortopathy. In the absence of practically any experience with TAVI in young patients it may be reasonable to favour surgery. Figure 1 suggests one possible 'TAVI first' scenario

Figure 1.

One possible scenario for heart team decision-making in a 'TAVI first' scenario.

So is this the final meta-analysis comparing TAVI and SAVR? In a sense this does conclude and punctuate an era where surgical risk has been the major determinant of who and who should not be eligible for TAVI. TAVI is associated with significantly better survival and less stroke across the spectrum of surgical risk. The bulk of secondary outcomes also favour TAVI, regardless of surgical risk. In the final analysis, surgical risk, per se, should no longer determine TAVI candidacy. The old paradigm of 'SAVR if possible, TAVI if necessary' no longer holds. Equally valid might be a new paradigm; 'TAVI if possible, SAVR if necessary'. The truth is probably somewhere in between.