Few debates in aortic valve surgery have persisted for so long while remaining so resistant to a clear-cut answer as the choice between mechanical and biological prostheses in patients aged 50 to 70 years. European and American guidelines differ in their recommendations for this age group, and real-world clinical practice shows substantial inter-centre variability which, in simple terms, seems to reflect institutional habits and inertia more than robust evidence.
In this context, the systematic review and meta-analysis by Trevis et al., recently published in Heart, deserves careful consideration.
The literature search, conducted in PubMed, Embase and the Cochrane Library up to August 2025, identified 30 studies —29 retrospective studies and a single randomised controlled trial (RCT)— including a total of 120,844 patients. The primary endpoint was overall survival; secondary endpoints included stroke, major bleeding and reoperation rates. Data were pooled using an inverse-variance random-effects model, and results were reported as hazard ratios (HRs) with 95% confidence intervals.
Regarding overall survival, the pooled analysis of 26 studies involving 107,842 patients showed a statistically significant survival advantage in favour of mechanical prostheses (HR = 0.88; 95% CI = 0.81-0.94; p = .001), corresponding to a 12% relative reduction in mortality. This finding, however, should be interpreted in context: statistical heterogeneity was high (I² = 76%), largely driven by the study by Lu et al.; after excluding this study in the sensitivity analysis, I² decreased to 52% while the effect remained in favour of mechanical valves (HR = 0.90; 95% CI = 0.84-0.96; p = .002).
The subanalysis restricted to the 16 matched studies (using PSM or IPW) and the only available randomised trial, together including 95,363 patients, is the most robust finding of the study: the mechanical valve advantage persisted (HR = 0.90; 95% CI = 0.87-0.94; p < .0001), with virtually no heterogeneity (I² = 6%, Tau² < .001). When the analysis was limited to crude, unmatched data, the difference lost statistical significance (HR = 0.87; 95% CI = 0.69-1.11; p = .22; I² = 89%), highlighting the importance of indication bias in the available literature.
The subanalysis according to procedure type is also relevant: in isolated AVR (13 studies, n = 93,223), the mechanical valve advantage reached statistical significance (HR = 0.89; 95% CI = 0.83-0.95; p = .005; I² = 41%), whereas in studies including concomitant procedures the difference was attenuated and no longer statistically significant (HR = 0.89; 95% CI = 0.78-1.02; p = .09; I² = 78%). No statistically significant differences were observed between subgroups (p = .78), suggesting that concomitant procedures do not substantially modify the effect, although they do introduce heterogeneity.
With regard to stroke, no significant differences were observed between valve types, either in the overall analysis (HR = 1.07; 95% CI = 0.91-1.27; p = .37; 13 studies, 21,317 patients) or in the subanalysis of matched studies and the randomised trial (HR = 1.10; 95% CI = 0.92-1.32; p = .27). Heterogeneity was low (I² = 0%), with the methodological caveat noted by the authors themselves: the absence of statistical heterogeneity may reflect imprecision in individual studies with wide confidence intervals rather than true homogeneity.
Mechanical prostheses were associated with a significantly higher risk of major bleeding (HR = 1.60; 95% CI = 1.43-1.78; p < .001; 15 studies, 18,897 patients). This finding was confirmed in the subanalysis of matched studies and the RCT (HR = 1.58; 95% CI = 1.40-1.78; p < .001; I² = 0%), showing very high consistency across studies. This remains the most relevant trade-off of mechanical prostheses and the factor that most often influences clinical decision-making and patient preferences. The authors note, however, that many of these studies reflect earlier-generation anticoagulation regimens with higher INR targets (up to 3.5) which may overestimate bleeding risk in contemporary practice.
Bioprostheses showed significantly higher reoperation rates (HR = 0.44 for the mechanical option; 95% CI = 0.33-0.57; p< .001; 24 studies, 32183 patients), a finding that persisted in the matched-study subanalysis (HR = 0.45; 95% CI = 0.32-0.62; p < .001). Heterogeneity was high (I² = 86%), partly influenced by the study by Chan et al.; its exclusion reduced heterogeneity to 75% without changing the direction of the effect (HR = 0.42; 95% CI = 0.32-0.54; p < .001). Although residual heterogeneity limits the precision of the quantitative estimate, the direction of the result is unequivocal and consistent with the biology of structural valve deterioration (SVD), which in younger patients often appears before 15 years after implantation.
COMMENTARY:
The main findings are not surprising: mechanical prostheses are associated with an approximately 12% reduction in overall mortality (HR = 0.88; 95% CI = 0.81-0.94), with no significant differences in stroke rates, but with a significantly higher risk of major bleeding (HR = 1.60). Bioprostheses, in turn, accumulate substantially higher reoperation rates (HR = 0.44 for mechanical valves, meaning that biological valves are associated with almost twice as many reoperations).
The analysis of matched studies, which markedly reduces statistical heterogeneity, preserves the survival benefit of mechanical prostheses, with an I² of 6%, lending greater robustness to this estimate. This is worth emphasising, because the overall analysis is affected by the expected heterogeneity that arises when studies spanning four decades of surgical history are combined, including very different generations of prostheses and evolving professional attitudes and clinical practice.
The underlying problem remains unchanged: almost all the evidence is observational and largely reflects outcomes from prostheses implanted between the 1980s and 2010. During that period, both biological prostheses and anticoagulation strategies for mechanical valves have evolved substantially. The PROACT trial showed that an INR range of 1.5-2.0 was safe in high-risk patients with an aortic On-X mechanical prosthesis, with significantly lower bleeding rates than in the control group. The ongoing LIMIT trial aims to assess the safety of reduced INR targets (1.5-2.5) versus conventional targets (2.0-3.5) in bileaflet mechanical prostheses, and its results will be decisive in updating the risk-benefit balance. If the bleeding risk of mechanical prostheses is substantially reduced with contemporary anticoagulation, the net survival advantage of this option becomes even more clinically relevant.
On the biological side, the narrative of transcatheter valve-in-valve (ViV) as a rescue strategy for structural degeneration remains attractive. It offers a way to defer or avoid conventional redo surgery in failed bioprostheses, making it increasingly prominent in discussions with patients. However, the available data are entirely observational, subject to substantial selection bias towards higher-risk surgical patients, and lack sufficient follow-up to assess the long-term durability of haemodynamic results. The recently funded REPEAT trial should clarify whether this strategy holds up against conventional redo surgical valve replacement. Until then, using it as the central argument to tilt the decision towards a bioprosthesis in younger patients may be premature, although its development will be essential before ViV feasibility can be used as a formal argument in primary prosthesis selection at younger ages.
The UK NASCA registry data cited by the authors are striking: in patients younger than 60 years, bioprosthesis use ranges from 7% to 82% depending on the centre. This variation does not reflect sophisticated personalised medicine; for the most part, it reflects surgeon preference and institutional culture. In Spain, the situation is not very different: the increasing trend towards bioprostheses in this age group is driven more by patient expectations (particularly the wish to avoid lifelong anticoagulation) than by evidence demonstrating better long-term outcomes.
From the surgeon’s perspective, this obliges us to be particularly careful during the informed consent process. Explaining to a 58-year-old patient that choosing a bioprosthesis means accepting a real probability of reoperation within the following decade (with morbidity and mortality that, although improved, are not negligible) requires time and honesty. The possibility of ViV does not eliminate that risk: it transforms it and postpones it, but does not make it disappear. And if, once properly informed, that patient prefers a bioprosthesis while accepting that risk, the decision is legitimate and respectable. What is not acceptable is for that conversation not to take place.
Ultimately, the study by Trevis et al. provides a rigorous update of the evidence in an area where uncertainty remains the rule. Its results should not be interpreted as a universal prescription in favour of mechanical valves, but rather as a warning against the growing tendency to use bioprostheses in younger patients. What is clear is that it once again confirms the glaring absence of contemporary randomised trials in this field. While we await the results of ongoing trials, the best tools available remain well-informed clinical judgement, transparent communication with the patient and surgical planning that preserves future options.
REFERENCE:
Tasoudis PT, Varvoglis DN, Vitkos E, et al. Mechanical versus bioprosthetic valve for aortic valve replacement: systematic review and meta-analysis of reconstructed individual participant data. Eur J Cardiothorac Surg. 2022;62:ezac268.
