Perceval Sutureless Implant Versus Standard-Aortic Valve Replacement: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Perceval Sutureless Implant Versus Standard-Aortic Valve Replacement (NCT02673697), a randomized na study. The primary outcome is Proportion of Participants With Freedom From Major Adverse Cardiac and Cerebrovascular Events (MACCE).

2. Clinical Question

3. Trial Design

NCT02673697 was a randomized, parallel-group, non-blinded trial comparing the Perceval valve with other stented biological valves. The registry describes the study as prospective, stratified, multicenter, international, and post-market. The sponsor was Corcym S.r.l. The registry lists the phase as NA.

The study began on 2016-03-22 and its primary completion date was 2020-01-17. Its status is TERMINATED, and enrollment is listed as 914. The reason for termination and whether the study ended according to its planned stopping process are not reported in the registry record.

The registry’s brief description says the trial had a flexible sample size to be determined adaptively. It describes planned enrollment of up to 1234 subjects, with accrual potentially ending earlier at approximately 900 or 1050 subjects, and recruitment at approximately 60 worldwide investigational sites where the device was commercially available. These are descriptions of the planned design; the record does not explain which sample-size pathway was used or why enrollment ended at 914.

Random allocation was reported, but the allocation procedure, concealment method, and stratification factors are not reported in the registry record. Masking was listed as none. The investigator-selected comparator tissue valve means the control arm could include different approved bovine or porcine valves; the distribution of those valve types is not reported. The record does not provide the full protocol, sample-size justification, or detailed operational account of the adaptive design.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: A total of 914 patients were enrolled in the PERSIST-AVR trial. Among the 914 patients enrolled, 910 underwent randomization. 4 patients were not randomized due to consent withdrawal (2 cases) and physician decision (2 cases)

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Per Protocol Population

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Per Protocol Population

6. Primary Endpoint

The registry lists the primary outcome as Proportion of Participants With Freedom From Major Adverse Cardiac and Cerebrovascular Events (MACCE), measured over 1 year post-operatively.

The primary endpoint was the proportion of participants with freedom from MACCE at 1 year post-operatively, based on adjudication by a Clinical Event Committee. The composite included all cause death, myocardial infarction, stroke, and valve re-intervention. A participant counted as free from the endpoint only if none of the specified events occurred during the endpoint period, according to the study’s endpoint definition.

Composite endpoints are used in clinical trials to combine several clinically important events into a single outcome. This can provide a broader summary of event-free status than examining only one component. However, the components may differ in frequency and clinical meaning, and a composite result does not show how much each component contributed. Component-specific primary outcome results are not reported in the registry record.

The endpoint is reported as a proportion with a 95% confidence interval for each arm. The Perceval estimate was 0.917 (95% CI 0.888 to 0.942), and the estimate for other stented biological valves was 0.92 (95% CI 0.892 to 0.945). These arm-specific intervals describe uncertainty around each arm’s estimated proportion; they do not, by themselves, give a confidence interval for the difference between arms. The primary analysis population was the Per Protocol Population.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

General statistical explanation: a trial estimate is a summary of the observed data in a defined population. For this endpoint, the estimate is a proportion of participants classified as free from the specified events during the follow-up period. The registry reports an estimate and 95% confidence interval for each arm, rather than a posted numerical treatment contrast with an interval.

What do the arm-specific confidence intervals describe?

In general, a confidence interval expresses the statistical uncertainty around an estimated quantity under a specified method. A narrower interval indicates greater precision for that quantity, while a wider interval indicates less precision. The intervals apply separately to the estimated proportion in each arm: 0.917 with 95% CI 0.888 to 0.942 for Perceval, and 0.92 with 95% CI 0.892 to 0.945 for other stented biological valves. They are not an interval for the difference between those proportions. A direct treatment comparison requires an estimate and uncertainty calculation for a contrast, using the specified analysis method.

What does a non-inferiority analysis ask?

General statistical explanation: non-inferiority analyses assess whether the result for an intervention is not worse than a comparator by more than a prespecified margin. The margin defines the largest difference the study will regard as compatible with non-inferiority. The interpretation depends on the direction of the contrast, the margin, the analysis population, and the uncertainty around the estimated contrast. A result cannot be fully evaluated from two arm-specific estimates alone.

The registry labels its reported analysis Bayesian and non-inferiority and gives p=0.991. It does not report the margin, the posterior quantity or other definition represented by p, the estimated contrast, or an interval for that contrast. Because the meaning of the reported value is not explained, it should be quoted as recorded rather than assigned a conventional frequentist or Bayesian interpretation. Details of the model, prior distributions, and decision threshold are not reported in the registry record.

How should the analysis population be considered?

The primary endpoint is designated for the Per Protocol Population. In general, a per-protocol analysis focuses on participants meeting specified requirements for inclusion in that analysis; the exact requirements depend on the protocol. The registry record does not describe those requirements, identify exclusions from this population, or post a corresponding intention-to-treat primary analysis. It is therefore not possible to assess from the record how population definitions or exclusions affected the estimates.

Randomization supports a fair comparison when participants are analyzed in a way that preserves the randomized groups, but the specific analysis populations and handling of departures from treatment assignment matter. The record states that allocation was randomized and masking was none. It does not report allocation concealment or the detailed approach to protocol deviations. These omissions limit assessment of the analysis in relation to the design.

What does the adaptive sample-size description establish?

The registry describes a flexible sample size determined adaptively, enrollment up to 1234, and the possibility of earlier stopping at approximately 900 or 1050 subjects. This records a planned design feature, but does not give the adaptive decision rules, timing of assessments, criteria for stopping, or the statistical adjustments used to account for those decisions. Interim analyses and their methods are not reported in the registry record. The observed enrollment of 914 does not, on its own, identify which planned pathway was followed.

In general, sample-size planning is distinct from analysis of the observed trial results. A planning illustration based on an observed effect is not observed power and is not a substitute for a prespecified design calculation. The registry description does not report the trial’s sample-size justification or the assumptions used in its adaptive plan. No sample-size rationale can be inferred from enrollment alone.

What is missing for a complete statistical appraisal?

The registry record does not report the non-inferiority margin, a numerical treatment contrast, a contrast interval, the detailed Bayesian model, missing-data methods, or a complete statistical analysis plan. It also does not provide primary endpoint component results or a randomized-population estimate. These details would clarify the analysis assumptions, the precision of the between-arm comparison, and the robustness of the reported result. General statistical principles can explain why those details matter, but cannot fill in the trial-specific information that is absent.

10. Analyses Reported by the Sponsor

The registry reports a Bayesian non-inferiority analysis with p=0.991. It does not provide a numerical between-arm estimate, an interval for that estimate, the non-inferiority margin, or a definition of the reported p value. Those omissions prevent a complete interpretation of the registry-reported analysis based on the posted record alone.

A two-arm recalculation is not available for the primary endpoint: the record indicates that values are missing or not estimable for a two-arm comparison for this continuous-endpoint data format. The available arm-specific proportions and their confidence intervals are not a substitute for that recalculation. No recalculated result should therefore be read as confirming or contradicting the Bayesian analysis.

The registry record does not establish that the reported analysis is a sponsor-authored analysis, nor does it give enough methodological detail to reproduce it. The appropriate reading is to distinguish the reported Bayesian result from the arm-specific descriptive estimates and from any unavailable recalculation. The record lists no discrepancy between a sponsor result and a recalculation, but no recalculated comparison is posted for this endpoint.

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists two outcome measures and one analysis. The primary endpoint is freedom from MACCE at 1 year post-operatively, and the listed secondary outcome is Surgical Times. The record does not describe a multiplicity adjustment, a hierarchy among outcomes, or a strategy for controlling the chance of false-positive findings across multiple comparisons. It is not reported whether any multiplicity procedure was prespecified.

In general, when multiple outcomes or comparisons are assessed, the chance of at least one apparently notable result can differ from the chance associated with a single planned test. A prespecified approach can explain how results across outcomes are to be interpreted. The absence of such information in the registry record means the handling of multiple outcomes cannot be evaluated; it does not establish that no plan existed outside the record.

Missing-data methods for the primary outcome are not reported in the registry record. The record does not say how participants without an assessable one-year endpoint were handled, whether follow-up was complete, or whether any sensitivity analyses were conducted. Those details matter because an endpoint estimate can depend on how unavailable outcome information is treated. No assumptions about missing outcomes should be made from the arm estimates alone.

The primary analysis population is identified as the Per Protocol Population, but eligibility rules for that population and the number excluded from it are not reported. A separate intention-to-treat result is not posted. The record also does not explain the relationship between the primary analysis denominators and the safety at-risk counts. These population details limit evaluation of whether the reported estimate captures all randomized participants and how departures from assigned treatment were handled.

The trial description refers to adaptive sample size, but interim analyses, stopping rules, and any adjustments associated with adaptive decisions are not reported in the registry record. This leaves uncertainty about how the planned flexibility was implemented and how it relates to the final analysis. The registry does not report whether multiplicity, missing-data, or analysis-population decisions were addressed in a protocol or statistical analysis plan.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: other non-cardiovascular sae (29.3%), arrhythmia or conduction disorder (22.2%), bleeding (8.7%), other cardiovascular sae (8.7%), stroke (4.7%).

The registry lists serious adverse event figures for participants at risk in each group. For Perceval, 450 were at risk and 390 were listed as affected; for other stented biological valves, 446 were at risk and 342 were listed as affected. The record also lists deaths of 32 and 24, respectively. These are safety summaries and should not be confused with the per-protocol primary endpoint estimates or treated as a comparison adjusted for differences in follow-up or population.

The listed serious adverse event terms include Other non-cardiovascular SAE at 29.3%, Arrhythmia or conduction disorder at 22.2%, Bleeding at 8.7%, Other cardiovascular SAE at 8.7%, Stroke at 4.7%, and Myocardial Infarction at 3.4%. The record does not state the denominator for these term percentages or whether participants could be represented in more than one category. They should not be added or interpreted as mutually exclusive categories.

Further information about event timing, severity grading, attribution, and statistical comparison is not reported in the registry record. The counts alone do not establish a causal relationship to a valve or provide a complete account of comparative safety.

13. What the Result Does, and Does Not, Mean

The registry reports one-year freedom-from-MACCE estimates of 0.917 for Perceval and 0.92 for other stented biological valves in the Per Protocol Population. It also reports a Bayesian non-inferiority analysis with p=0.991. These are the central posted results, but their interpretation is constrained by the absence of a reported numerical treatment contrast, a contrast interval, and a stated non-inferiority margin.

The result does not establish that the two valve strategies are identical. Similar-looking arm estimates do not prove equivalence, and non-inferiority is a distinct statistical claim that depends on a prespecified margin and an analysis of the treatment contrast. The record does not provide the information needed to independently determine how the Bayesian result maps to that claim. The reported p value should not be interpreted as a conventional p value or posterior probability because its definition is not reported.

The estimates describe the specified endpoint and population, not every possible outcome after surgery. MACCE combines all cause death, myocardial infarction, stroke, and valve re-intervention, while separate primary results for those components are not reported. The estimates also apply to the one-year postoperative timeframe and do not characterize outcomes outside that period.

The safety figures offer counts of participants at risk and affected, deaths, and percentages for selected serious adverse event terms. They do not by themselves establish which group had a safer outcome, because the record does not provide comparative safety analyses, event timing, or full detail on the categories. They also do not replace the primary endpoint analysis.

Finally, the registry lists the study as TERMINATED with enrollment of 914, while its description gives a flexible planned sample size. The reason for termination and the relationship between actual enrollment and the planned adaptive stopping approach are not reported. The result should therefore be understood as the posted registry record, with these statistical and reporting limits, rather than as a complete account of the trial protocol or its full analysis.

14. Why This Trial Matters Statistically

This trial raises a useful statistical question: how should a non-inferiority result be assessed when the registry provides arm-specific endpoint estimates and labels an analysis Bayesian, but does not report the margin or the treatment contrast? The posted estimates give readers a view of the observed one-year proportions, while the missing inferential details prevent a full independent appraisal of the non-inferiority conclusion.

The outcome is a composite of death, myocardial infarction, stroke, and valve re-intervention. Composite endpoints can make a trial’s primary question more inclusive, but their interpretation depends on understanding the components and their contributions. The registry reports the composite estimate but does not post component-specific primary outcome results. That limits the ability to see whether the overall endpoint reflects a consistent pattern across its parts.

The design description also highlights the importance of transparent reporting for adaptive sample-size plans. The record describes flexible enrollment and possible earlier stopping, yet does not state the decision rules or how they were implemented. Enrollment of 914 is reported, but without stopping details it is not possible to connect that number to a particular planned decision pathway.

For clinical researchers and biostatisticians, the record illustrates why an analysis label and a single reported statistic are not enough to reconstruct a trial’s evidence. The analysis population, non-inferiority margin, contrast estimate, interval, missing-data handling, and adaptive design methods all help determine what a result means. When those details are absent, careful interpretation means separating the descriptive estimates from the reported inferential claim and being explicit about what remains unresolved.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

Several trial-specific limitations of interpretation arise from the registry record. The primary endpoint result is reported for the Per Protocol Population, but the record does not describe that population’s inclusion rules, exclusions, or size beyond the arm-specific denominators of 407 and 412. It does not post an intention-to-treat primary result, so the relationship between the per-protocol findings and all randomized participants cannot be assessed from the posted results.

The non-inferiority analysis is not fully documented. Although the registry identifies a Bayesian method and gives p=0.991, it does not report the margin, an effect estimate for the comparison, a confidence or credible interval, or the definition and decision threshold for the reported p value. The model and prior assumptions are also not reported. This prevents reconstruction of the analysis or an independent check of its conclusion.

The trial is described as having a flexible, adaptively determined sample size, with enrollment up to 1234 and potential earlier stopping at approximately 900 or 1050 subjects. The final enrollment is 914, but the reason for termination and the actual stopping criteria used are not reported. Interim analyses and any statistical adjustments for adaptive decisions are not reported either.

The primary endpoint is a composite, but component-specific primary outcome results are not posted. Missing-data methods and outcome completeness are not described. The registry also does not report the comparator valve distribution, detailed randomization procedures, or the stratification factors. Masking is listed as none, and the record does not describe additional steps to address possible bias in outcome assessment beyond stating that the primary endpoint was based on Clinical Event Committee adjudication.

These limitations concern what can be evaluated from the registry record; they do not establish that the unreported procedures were absent from the full trial documentation.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Perceval Sutureless Implant Versus Standard-Aortic Valve Replacement — NCT02673697.
• Registry record last updated 2024-03-06; status last verified 2024-03.
• Results first posted 2022-07-27.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT02673697 is a randomized, parallel-group, non-blinded trial comparing the Perceval sutureless valve with other stented biological valves in surgical treatment of aortic valve disease. The record lists enrollment of 914, sponsor Corcym S.r.l., and status TERMINATED. The primary endpoint was freedom from MACCE at 1 year post-operatively in the Per Protocol Population.

The arm-specific estimates were 0.917 (95% CI 0.888 to 0.942) for Perceval and 0.92 (95% CI 0.892 to 0.945) for other stented biological valves. A Bayesian non-inferiority analysis is reported as p=0.991. The record does not include the non-inferiority margin or a numerical treatment contrast. It also describes an adaptive sample-size plan, but does not explain the stopping decision or the reason for termination. The posted data therefore support descriptive reading of the arm estimates, while leaving important aspects of the non-inferiority analysis and trial implementation unresolved.