Background: Aortic stenosis is the most common valvular heart disease in Western countries. A Norwegian study estimates that the prevalence of aortic stenosis is 0.2% in adults aged 50 to 59, 1.3% in adults aged 60 to 69, and up to 9.8% in patients 80 to 89.

Aortic stenosis is generally caused by calcification of the aortic valve that ultimately can lead to heart failure. The three most important symptoms are chest pains, shortness of breath on exertion, and fainting. The disease may be asymptomatic for long periods of time, but once symptoms appear (severe aortic stenosis), an untreated individual has an average life expectancy of 2 to 3 years. The only effective treatment is aortic valve replacement (AVR) surgery.

Objective: The National System for Managed Introduction of New Health Technologies within the Specialist Health Care Service in Norway commissioned us to perform single technology assessment(s) on the use of sutureless aortic valve replacement in treatment of aortic stenosis (Nye metoder ID2015 042). One company, Livanova, Sorin group, provided a submission file. Based on the commission and the submission file, our assessment has been restricted to effectiveness, safety and cost effectiveness of Perceval sutureless aortic valve replacement (Perceval) compared to traditional aortic valve replacement (traditional AVR) for treating adult operable patients with severe aortic stenosis. We have evaluated the submitted documentation in relation to the best available published evidence.

Clinical effectiveness and safety: The clinical documentation submitted by the firm consisted of 25 studies included after a systematic search. We excluded six studies with transcathether aortic valve implantion (TAVI) as comparator, but included comparisons to other types of sutureless valves. We identified one additional study based on an independent systematic search, leaving 20 studies to be assessed. We aimed to identify the best available evidence for the outcomes long- and short-term mortality, morbidity, quality of life, resource use and adverse events. We evaluated internal validity of the studies based on the submitted information and a simplified risk of bias analysis. Data from all assessed studies, are presented in an appendix to this report, but only studies considered to represent best available evidence, were assessed in depth based on full text inspection. RevMan 5 was used to pool data from comparative studies. We assessed the quality (confidence in estimates) of the best available evidence based on the guidelines provided by The Grading of Recommendations Assessment, Development and Evaluation (GRADE).

The firm submitted an economic model consisting of three elements: (1) a hierarchical, random-effects Bayesian meta-analysis of clinical data from studies used to estimate pooled clinical parameters; (2) a probabilistic, patient-level simulation model that used clinical outcomes from the meta-analysis to determine the life-time effectiveness (30-day mortality, life-years gained, QALYs) and costs of Perceval compared to traditional valves based on 10,000 simulated patients; and (3) a five-year budget impact model to translate the cost-effectiveness results into a budget impact statement. The model examined six treatment groups consisting of four isolated AVR groups (full-sternotomy with traditional vs. Perceval valve; minimally invasive surgery with traditional vs. Perceval valve) and two groups undergoing concomitant surgical procedures via full-sternotomy (traditional vs. Perceval valve).

The model relied heavily on a published study (not included in the clinical effect documentation) relating outcomes of aortic valve replacement surgery to cross-clamp time (CCT). To capture the independent effects of surgical procedure and valve type on CCT, the submission included data from seven published studies, only one of which was part of the submitted clinical evidence on effect. The data were pooled using Bayesian meta-analyses in order to estimate relative effects of valve type and surgical procedure on cross-clamp time (CCT), and the baseline mean values and associated distributions for adverse events for the model reference group (CCT < 60 min). Cost data were retrieved, when possible, from Norwegian sources. A healthcare-payer perspective was used for the analysis.

Clinical effectiveness and safety: Among the 20 assessed studies there were no randomized controlled trials (RCTs). Except for two studies from Canada, the assessed studies were based on European case series. The only study for which we identified an entry in a trial registry was the single-arm CAVALIER study (NCT01368666). As some studies are overlapping, we cannot give an estimate on the total number of patients in the assessed studies. Ten studies were non-comparative and ten were comparative. Eight studies compared Perceval with traditional AVR and two studies compared Perceval with other types of sutureless valves.

Based on a simplified risk of bias assessment we considered four propensity score matched cohort studies comparing Perceval with traditional AVR (1033 patients in total) and the single arm CAVALIER study (658 patients receiving Perceval) to represent the best available evidence for the predefined indication and outcomes. We considered the remaining studies to represent very low quality evidence and did not assess them further.

Based on pooled data from the four propensity score matched studies, it is uncertain whether Perceval reduces, increases or has a similar 30-day mortality compared with traditional AVR. There were 19 deaths in the Perceval group (N=484) and 22 in the traditional AVR group (N= 549). A random effects meta-analysis provided a risk ratio of 1.09; 95% CI 0.58 to 2.06 (GRADE quality of evidence: low).

There may be small or no differences in hemodynamic measures at 30 days between Perceval and traditional AVR (mean gradient (mm Hg) -0.73; 95% CI -1.75, 0.30; GRADE quality of evidence: low). We also found that Perceval may provide lower cross-clamp time and cardiopulmonary bypass time (mean difference, respectively = -22.53; 95% CI -34.28 to -10.78 and -26.83; 95% CI -32.10 to -21.55; GRADE quality of evidence: low). Postoperative differences in functional status (NYHA class) was not reported in any of the comparative studies. None of the studies reported quality of life data. No conclusions could be made with regard to the influence of Perceval on intensive care unit or hospital length of stay due to very low quality of evidence.

No published comparative studies were available to allow for subgroup analyses based on surgical procedure (minimally invasive versus full sternotomy). The same type of adverse events, if reported, were present in both groups of the comparative studies. No adverse events occurred at a rate higher than 10% in the Perceval arm of the propensity matched comparative studies. The need for pacemaker implantation was higher in the Perceval group compared to traditional AVR (risk ratio = 1.62; 95% CI 0.98 to 2.67; GRADE quality of evidence: low). No conclusions could be made regarding Perceval versus other sutureless valves because there were no propensity score matched studies. No conclusions could be made with regard to differences in long-term outcomes.

Short-term (30-day) adverse events, including death (3.7%), stroke (2.2%), major bleeding (4.5%), and the need for permanent pacemaker implantation (11.8%) were common in the single-arm CAVALIER study (N=614). Freedom from valve or procedure related death among patients available to follow up decreased from 97.2% (95% CI 95.9 to 98.5) after one year (N=554) to 89.5% (95% CI 85.1 to 93.8) at four years (N=83).

Health economic results: The results of the submitted cost-effectiveness simulations are mainly based on data not included in the submitter’s effect evidence. Based on the model, Perceval can be cost-effective (less costly and slightly more effective) relative to traditional sutured valves for the three types of surgical procedures considered: isolated full sternotomy (FS), isolated minimally invasive surgery (MiS), and concomitant surgery with full sternotomy (CONC).

For isolated FS procedures the estimated effect gains for Perceval relative to traditional valves are a 2.1% reduction in mortality, a 0.13 increase in life-years gained, and a 0.11 increase in QALYs gained. The estimated gains associated with Perceval are slightly lower for isolated MiS and slightly higher for CONC procedures. The largest estimated gains come with a switch from FS with a traditional valve to MiS with Perceval, with a 2.9% reduction in 30-day mortality, a 0.19 increase in life-years gained and a 0.15 increase in QALYs gained. This supports the idea that there are independent gains from a MiS rather than FS procedure and from using Perceval rather than traditional sutured valves.

Estimated costs are lower using Perceval valves compared to sutured valves across all surgical procedures. Estimated savings for Perceval compared to trational valves are approximately NOK 133,300 with a full sternotomy and NOK 114,350 for minimally invasive surgery. The estimated savings for concomitant procedures using Perceval is NOK 206,900. As with effects, the largest estimated cost savings occur with a switch from FS with a traditional valve to MiS with Perceval, a saving of approximately NOK 181,600.

The five-year budget impact compares total costs for annual aortic replacement surgery for 698 patients in two scenarios: (1) no use of sutureless valves and (2) a gradual, linear market penetration by Perceval of 15% over five years. The budget impact analysis also shows cost savings with Perceval of 1.33%, 2.01%, 2.72%, 3.43% and 4.15% in years 1 to 5, respectively. The total five-year saving with the specified gradual introduction of Perceval is approximately NOK 44,660,000. The analysis is based on the assumption that 50% AVR procedures are minimally invasive.

Sensitivity analysis of both the cost-effectiveness results and the budget impact analysis showed that the base case results were robust for analyses reflecting uncertainty in the simulated outcomes in the model and for a variation in the assumed base case Perceval price (NOK 32,500) within a price range from NOK 25,000 to 40,000.

Effectiveness and safety: Our major objection to the submitted material is the low quality level of currently available evidence. There is one ongoing highly relevant RCT (PERSIST‐AVR trial, NCT02673697) with planned enrollment of 1234 patients. Primary data from this trial is anticipated to be available in 2019. We considered the most appropriate argument for including non-randomized studies, at this time, is that it is early in the life cycle of the technology, and that there may be a need for a temporary decision on whether to offer this technology based on best available evidence and/or evaluate the need for additional trials. Thus, we have focused on identifying the best available evidence. More definitive conclusions can be made when results of the ongoing RCT are available.

There are several new methods available for treatment of operable patients with severe aortic stenosis, including other types of sutureless procedures. In addition, both sutureless and transcathether based procedures (TAVI) have been suggested for patients with severe aortic stenosis and an intermediate to high operative risk as well as patients with anatomical characteristics not suited for traditional AVR. This may provide new options for patients with unmet needs, but it also increases the need for additional clinical trials, i.e. trials comparing sutureless AVR to TAVI.

Health economics: The economic analysis relies on a model that relates clinical outcomes for aortic valve replacement surgery to cross-clamp time (a surrogate endpoint) and surgical technique. The model itself is well-constructed, relevant for the Norwegian context, and exhibits, as far as we can tell, internal validity. The data used in the model, however, were mostly from studies that were not part of the clinical evidence submitted by the firm and were therefore not graded for quality. The one study that was included in the submitted evidence was considered to be of very low quality.

Two competing effects could influence model-estimated savings when using Perceval instead of traditional valves. Savings across all procedure types may be lower than suggested in the cost-effectiveness analysis if the reduction in time needed for surgery cannot be fully translated into additional operations. On the other hand, savings estimated in the five-year budget impact analysis are based on the assumption that approximately half of valve replacement procedures are minimally invasive surgeries. Because aortic valve replacement in Norway is usually performed as a full-sternotomy, and savings using Perceval are higher compared to traditional valves for FS than MiS, the actual savings may tend to be higher than reported.

Effectiveness and safety: The quality of the available evidence comparing Perceval sutureless AVR to traditional AVR is low to very low. More robust conclusions will be available upon publication of primary data from an ongoing RCT expected in 2019. Based on best available evidence, it is uncertain whether Perceval AVR reduces, increases or has a similar 30-day mortality compared with traditional AVR. Perceval AVR may reduce perioperative cardiac bypass time and cross-clamp time, and may provide little or no difference in hemodynamic function at 30 days compared to traditional AVR. However, no firm conclusions can be made with regard to superiority of either method.

Health economics: Based on the cost-effectiveness and budget impact analyses performed by the firm Perceval can be cost-saving compared to traditional sutured valves for isolated full sternotomy or minimally invasive valve replacement surgery, and for concomitant surgeries with full sternotomy. Model estimates of clinical effect indicate that there may be small gains connected with Perceval. Estimates from the five-year budget impact analysis show cost savings with expanded use of Perceval. Because the data used in the model were not based on the assessed comparative studies, there remains uncertainty about the likelihood and validity of the results. More robust conclusions will be possible on publication of the ongoing RCT.