PEER REVIEW HISTORY ...€¦ · femoral cutdown eligible? Carotid access is becoming popular as an...

PEER REVIEW HISTORY

BMJ Open publishes all reviews undertaken for accepted manuscripts. Reviewers are asked to

complete a checklist review form (http://bmjopen.bmj.com/site/about/resources/checklist.pdf) and

are provided with free text boxes to elaborate on their assessment. These free text comments are

reproduced below.

ARTICLE DETAILS

TITLE (PROVISIONAL) Outcomes comparison of different surgical strategies for the

management of severe aortic valve stenosis: study protocol of a

prospective multicentre European registry (E-AVR Registry)

AUTHORS Onorati, Francesco; Gherli, Riccardo; Mariscalco, Giovanni; Girdauskas, Evaldas; Quintana, Eduardo; Santini, Francesco; De Feo, Marisa; Sponga, Sandro; Tozzi, Piergiorgio; Bashir, Mohamad; Perrotti, Andrea; Pappalardo, Aniello; Ruggieri, Vito; Santarpino, Giuseppe; Rinaldi, Mauro; Silva, Ronaldo; Nicolini, Francesco

VERSION 1 – REVIEW

REVIEWER Stephen Fremes Sunnybrook Health Sciences Centre University of Toronto Toronto Ontario Canada

REVIEW RETURNED 04-Jul-2017

GENERAL COMMENTS Major Comments In terms of organization, the submission seems like 2 protocol submissions, which are similar but not identical. It is preferable if there is a single merged protocol. The study will be very important, but would be much more valuable if TF TAVR was also included. And as the landscape is changing so quickly, will be of greater importance by the time this project is completed. The Registry is described, and in the second part of the protocol, the investigators describe aims. The investigators have not described any hypotheses, general or testable. The authors have identified 30-day cardiovascular mortality as the primary outcome. I expected that 1 year, 2 year, or 5 year mortality would be the primary outcome. Please comment. The authors have described the overall sample size, but not the sample size for different subgroups. The authors have not provided sample size calculations for any of the study aims. Is there a central core echo lab? Will the central core lab review preoperative and postoperative echos?

on June 29, 2020 by guest. Protected by copyright.

http://bmjopen.bm

j.com/

BM

J Open: first published as 10.1136/bm

jopen-2017-018036 on 10 February 2018. D

ownloaded from

http://bmjopen.bmj.com/site/about/resources/checklist.pdf

http://bmjopen.bmj.com/

Are events centrally adjudicated? And can the CEC be blinded to some of the key procedural factors? What will happen if there are new RCTs or new implantable devices in terms of this Registry? Will such patients be excluded from this Registry? Will they be included? Will the participating centres be disallowed from participating in other investigational protocols? This question also applies to new protocols for medical treatment of valvular heart disease patients. Additional Comments The Registry is an “all-comers” study. As I initially read through the protocol, I wondered whether that was necessary. As the field is changing quickly, it is probably worthwhile to recruit all patients. Page 5, Line 18: The authors say that 5 years will provide definite answers – I disagree. 5 years is OK for inoperable, intermediate risk, and maybe for elderly low risk, but not for an “all-comers” study. Information about valve durability will not apparent after only 5 years of F/U – which is a very important question for low risk SAVR (or TAVR) patients. Much longer F/U is recommended. Page 5, Line 34: Is it not possible to maintain the blinding of some of the analysis team? Page 6, Line 25: The authors mention a “number” of sutureless aortic valves. Is that true? The authors should list them. I know of Intuity (Edwards) and Percevel (Sorin). Are there multidisciplinary Heart Teams in every centre who help decide on the mode of treatment of all AS patients? Page 9, Line 10: Patients will be followed for 5 years – is that a minimum, mean or median of 5 years? Page 9, Lines 12-9: Are TF TAVR cases performed through a femoral cutdown eligible? Carotid access is becoming popular as an alternative access site – are these cases eligible? Are autografts or homografts eligible, or only mechanical prostheses and xenografts? Are patients with annular or aortic enlargement eligible? Are patients with composite replacement eligible? Are patients with ascending aortic replacement not requiring circ arrest eligible? Are patients having AFib ablations eligible? Preoperative Diagnosis: Is the Registry only for patients with aortic stenosis? For AS/AI? For pure AI? Do patients need to have a specific gradient, dimensionless index, or AVA? Are low flow, low gradient AS patients eligible? Are patients with acute endocarditis, or undergoing emergency operation for valve thrombosis eligible? Follow-up: Is F/U local in person, local by telephone, or central? Consent: It is frequently difficult to obtain consent for a research study prior to an emergency operation. Will patients who underwent emergency surgery without consent for participation in the Registry, be approached for consent after the surgery? Page 10 Data Management and Monitoring: I am unsure of the meaning of the last sentence in the paragraph.


http://bmjopen.bm

j.com/

BM



ownloaded from


Statistical Methods: What and where is the Core laboratory? Who is the study statistician? The statistics section mentions a number of statistical tests, but does not provide text such as which test will be used for which comparison. Page 11 Interim Analysis: Are there statistical penalties for the interim analyses? Page 11 Early Secondary Outcomes: The authors mention, “device success, early safety, clinical efficacy, time-related valve safety”. How are these outcomes defined? Page 12 Late Secondary Outcomes: It isn’t clear whether perioperative MACCE outcomes are part of the late MACCE outcome. Page 12 Laboratory parameters: Will hs-troponin be collected?? Will BNP be collected? Page 20 Length of stay in the intensive care unit: It is better to initially collect ICU LOS in hours rather than days. Steering Committee: What is the frequency of the Steering Committee meetings? DSMB: Is there a DSMB? Page 49: “Evaluation of limits and benefits of combined SAVR and CABG vs SAVR and staged PCI or TAVR and staged/reversed stage/combined PCI”. The meaning of this sentence is not clear. Page 49: “Identification of indication criteria for SAVR based on baseline characteristics”. The meaning of this sentence is not clear. Page 49: “Analysis of costs related to hospitalization of AS-patients treated with different surgical modalities (e.g. sutureless valves vs stented/stentless, sternotomy vs minimally-invasive, standard SAVR vs surgical TAVR, etc) and evaluations of cost/effectiveness”. The authors do not describe much in terms of the acquisition of cost data. Page 50 Line 25: The authors describe audit of 10% of the data. Earlier (page 10 line 27) 40% is mentioned.

REVIEWER Marechaux Sylvestre Groupement des hôpitaux de l'Institut Catholique de Lille

REVIEW RETURNED 18-Jul-2017

GENERAL COMMENTS This paper is the design of a study aiming at comparing in the real life outcomes of different surgical strategies for patients being referred in the participating centers for aortic valve replacement. This study comes in a timely manner. The question is highly relevant. The paper is well written, outcomes are clearly defined. The sample size in the context of a multicenter study is large and appropriate. My only question is how the authors are sure that all surgical strategies will be sufficiently represent, especially for propensity matching comparisons of outcome data between groups.


http://bmjopen.bm

j.com/

BM



ownloaded from


VERSION 1 – AUTHOR RESPONSE

Reviewer: 1

Reviewer: Stephen Fremes; Institution and Country Sunnybrook Health Sciences Centre - University

of Toronto; Toronto; Ontario, Canada.

1.In terms of organization, the submission seems like 2 protocol submissions, which are similar but

not identical. It is preferable if there is a single merged protocol.

A: We thank the reviewer for his suggestion. A single merged protocol is now reported, and a SPIRIT

checklist added to the submission.

C: Second part was completely deleted. A SPIRIT Checklist is now added.

2.The study will be very important, but would be much more valuable if TF TAVR was also included.

And as the landscape is changing so quickly, will be of greater importance by the time this project is

completed.

A: We thank for the appreciation and agree with the reviewer that the enrollment of TF TAVR would

have been of greater interest. Unfortunately, some of the interventional cardiologists of the involved

centers denied to participate in the registry and to provide data about their percutaneous TAVR

practice (also in view of their enrollment in other studies, some RCTs). Therefore, the enrollment of

only few percutaneous TAVR would have certainly biased - at least in our opinion - all the analyses

about future comparisons between different techniques. Furthermore, surgical TAVR will be all

included in the registry, therefore some conclusions about the long-term safety, efficacy, durability,

and functional results (e.g. NYHA, QoF) achieved with transcatheter valves could certainly be

reached (in light of the fact that surgical TAVR and percutaneous TAVR share the same “platforms” in

terms of prosthetic materials, storage, deployment, etc).

Furthermore, it has to be considered that “surgical” TAVR are still performed in a routine fashion in

the enrolled Institutions. For all these reasons EAVR-Investigators decided – at least for the purpose

of this registry – to exclude TF TAVR (and other percutaneous TAVR) to reduce bias, given also the

ambitious aim to provide a clear picture of short-, mid-, and long-term results of the different surgical

options currently available for the treatment of SAVS.

C: None.

3.The Registry is described, and in the second part of the protocol, the investigators describe aims.

The investigators have not described any hypothesis, general or testable.

A: According to previous comment, the second part has been completely deleted, and a single

merged protocol with declared aim and hypothesis is now reported. Thus, a sample size calculation

based on the main hypothesis has been reported in the text.

C: In the Methods section, “Rationale of the study and aim”, it has been specified: “…[OMISSIS]…The

primary aim of the study is a 5-year comparison between SAVR and surgical TAVR: we hypothesize

to report a 10% superiority in terms of all-cause mortality in favor of SAVR vs TAVR.”.

Accordingly, in the Methods section, “Statistical Analysis” paragraph, we have now added the

sentence: “Given the PARTNER TRIAL 5-year all-cause mortality risk (67.8% after TAVR and 62.4%

after SAVR) – accepting a type 1 error of 5% - the overall number of patients needed to achieve 80%

power (1-beta) for a mortality odds of 1.1 (10% mortality difference) is 2866 patients (i.e. 1433

patient/group) (7). Therefore, the expected number of 8000 patients is far beyond the requested

sample size of the primary objective of the registry, and further accounts for risk-adjustment

methodologies and the expected 1.5% (historical data) of lost to follow-up.”

Finally, in the Ethics and Dissemination section:…[OMISSIS]….Several studies are planned at the

moment:


http://bmjopen.bm

j.com/

BM



ownloaded from


Primary study:

1) A 5-year study comparing all-cause mortality between SAVR and surgical TAVR. We expect to

report a 10% superiority of SAVR vs. TAVR according to sample size calculation and literature data

(7). This study will also report echocardiographic data, functional status, quality of life, incidence of

cardiovascular mortality, reinterventions on the aortic valve, and incidence of structural valve

deterioration between “all-comers” surgical TAVR and SAVR. The study is expected 6-years after the

start of data collection and it is aimed at being presented in a major European cardiology

journal…[OMISSIS]…

4.The authors have identified 30-day cardiovascular mortality as the primary outcome. I expected that

1 year, 2 year, or 5 year mortality would be the primary outcome. Please comment.

A: We thank the reviewer for his valuable suggestion, that we take. Therefore 5-year all-cause

mortality is now the primary endpoint. That also in light of the fact that the sample size calculation and

power analysis – based on another comment from this reviewer – were made on literature data on 5-

year all-cause mortality of SAVR and TAVR.

C: Abstract (Methods and Analysis): “Primary outcome is 5-year all-cause mortality. Secondary

outcomes aim at establishing “early” 30-day all-cause and cardiovascular mortality, as well as major

morbidity, and “late” cardio-vascular mortality, major morbidity, structural and non-structural valve

complications, quality of life and echocardiographic data.”

Early and late endpoints paragraph: “…[OMISSIS]…Primary outcome of the E-AVR registry: 5-year

all-cause mortality …[OMISSIS]… Early secondary outcomes: all-cause mortality, cardiovascular

mortality, stroke, …[OMISSIS]… Late secondary outcomes (collected starting from discharge to the

end of the 10th year after the index procedure): cardiovascular mortality, all-cause mortality (from 1 to

4 years after surgery, then from 6 to 10 years), stroke, …[OMISSIS]…”

5.The authors have described the overall sample size, but not the sample size for different subgroups.

The authors have not provided sample size calculations for any of the study aims.

A: We have now reported the main hypothesis and aim of the registry, as well as the sample size

calculation for that. Sample size calculations for subgroup analyses are waived given the exploratory

nature of these sub-studies (clearly stated in SPIRIT checklist).

C: In the Methods section, “Statistical Analysis” paragraph, we have now added the sentence: “Given

the PARTNER TRIAL 5-year all-cause mortality risk (67.8% after TAVR and 62.4% after SAVR) –

accepting a type 1 error of 5% - the overall number of patients needed to achieve 80% power (1-beta)

for a mortality odds of 1.1 (10% mortality difference) is 2866 patients (i.e. 1433 patient/group) (7).

Therefore, the expected number of 8000 patients is far beyond the requested sample size of the

primary objective of the registry, and further accounts for risk-adjustment methodologies and the

expected 1.5% (historical data) of lost to follow-up.”

6.Is there a central core echo lab? Will the central core lab review preoperative and postoperative

echos?

A: Although a Central core lab will enhance the quality of the study, this achievement was impossible

to reach during study planning. Therefore, we decided to maintain Institutional echo data. This

concept was added as a Limitation to the text (although it should be considered that Echo-Lab of the

involved Institutions are all ISO9001 and 3rd level certified by national – and sometimes European –

societies)

C: “Strengths and Limitations of this study” paragraph: “…[OMISSIS]… Limitations include absence of

a Central Core Laboratory for echocardiographic assessment….[OMISSIS]…”


http://bmjopen.bm

j.com/

BM



ownloaded from


Early and Late endpoints paragraph: “Echocardiographic data of prosthesis performance are defined

according to the Valve Academic Research Consortium-2 definitions (36). Although absence of a

Central Core Echocardiographic Lab is a limitation, all the echocardiographic data will come from 3rd

level national and/or international certified Institutional Echo Labs. ”

7.Are events centrally adjudicated? And can be blinded to some of the key procedural factors?

A: Based on this comment, we have better specified now that data will be collected at single

Institutions. As already reported however, data will undergo auditing from the Central Core Lab on a

regular basis. Unfortunately, the study plan is “open”, thus no blinding towards data was considered.

Indeed, we have specified also that Central Core Lab will be blinded towards the Centre.

We think that this might not be a major issue, given that data will be audited by an external Statistical

Core Lab, who will perform the entire set of statistical analyses.

C: Data Management and Monitoring paragraph “…Events and outcome variables will be adjudicated

after agreement of two local E-AVR Investigators, and collected at local Institutions. In the event of

controversy on outcome adjudication between the two local E-AVR Investigators, the outcome will be

discussed and adjudicated after a final consult inside the E-AVR Steering Committee….”

Early and Late endpoints paragraph: “Collection of data is under the responsibility of the Representing

Member inside the Steering Committee from each of the participating centres. Data will be audited

from the Central Core Laboratory on a regular basis, as reported previously.”

Data Collection paragraph: “Participating Centre: Each participating centre will be anonymized by

identification with a capital letter. The correspondence between centres and capital letters will only be

known by the PI of the study. The Central Core Laboratory analyzing the data will be blinded towards

the surgical teams.”

8.What will happen if there are new RCTs or new implantable devices in terms of this Registry? Will

such patients be excluded from this Registry? Will they be included? Will the participating centres be

disallowed from participating in other investigational protocols? This question also applies to new

protocols for medical treatment of valvular heart disease patients.

A: In case of new devices or new protocols of medical treatment, there is no preclusion to consider

these “new devices/protocols” in the Registry, given that all the analyses will be propensity-scored or

risk-adjusted. Therefore, there is already the plan to consider all the potential biases in future

analyses, and to manage them accordingly with several statistical techniques.

Obviously, patients eventually enrolled in future RCTs might be excluded from the Registry according

to the potential strict RCT’s rules. In this case, the number (and causes) of patients excluded from the

registry will be highlighted in future publications. However, no Centre has planned at the moment to

participate in any forthcoming RCT, and there is the agreement to try to avoid future participation in

RCTs dealing with SAVS.

C: In the interest of space, no specification on hypothetical future problems has been added to the

text.

9.The Registry is an “all-comers” study. As I initially read through the protocol, I wondered whether

that was necessary. As the field is changing quickly, it is probably worthwhile to recruit all patients.

A: Based on other comments from this reviewer we have better detailed that the Registry is an “all-

comers” study on the treatment of SAVS. However, given that the primary objective of the study – as

for other comments – has been clearly identified and based on a SAVR vs surgical TAVR

comparison, some restrictions seems to be necessary to avoid “uncorrectable bias”: for these reasons

we decided to collect ALL patients coming with a SAVS±coronary disease, but without need for

concomitant surgical procedures (other than a surgical aortic valve replacement ± myocardial

revascularization). For the same reasons aortic root diseases are not considered (they are a different

disease, at least in our opinion), as well as pure aortic regurgitations.


http://bmjopen.bm

j.com/

BM



ownloaded from


C: Based on several reviewer’s comments we have added a “better-detailed” list of inclusion and

exclusion criteria. In “Criteria for registry-enrolment” paragraph: “The following inclusion and exclusion

criteria will be considered:

Inclusion criteria

Age >18 yy

Isolated SAVS with or without concomitant aortic valve regurgitation

Isolated prosthetic aortic dysfunction

SAVS + coronary artery disease (CAD)

Prosthetic aortic dysfunction + CAD

Elective, urgent and emergent procedures

Endocarditic aetiology

Exclusion criteria

Patients undergoing concomitant mitral valve surgery, or tricuspid valve surgery, or aortic surgery (i.e.

composite aortic valve and ascending aorta replacement with o without circulatory arrest), or atrial

fibrillation surgery, or any other associated cardiac surgical procedure (with the exception of CABG)

Concomitant aortic root procedure (i.e. Bentall operation, David operation, homografts, autografts)

SAVR with techniques of aortic annular enlargement

Porcelain aorta

Pure aortic valve regurgitation

Percutaneous TAVR requiring surgical cut-down (i.e. failure to comply with a full percutaneous

approach, thus configuring a “hybrid procedure”)

Patient refusal”

The same specifications were added to the flowchart (Fig.1).

10. Page 5, Line 18: The authors say that 5 years will provide definite answers – I disagree. 5 years is

OK for inoperable, intermediate risk, and maybe for elderly low risk, but not for an “all-comers” study.

Information about valve durability will not apparent after only 5 years of F/U – which is a very

important question for low risk SAVR (or TAVR) patients. Much longer F/U is recommended.

A: We thank the reviewer for his valuable comment. We agree with this suggestion. Accordingly, what

we have previously considered a “minimum follow-up of 5 years” has been clarified that it will be

extended to a “maximum of 10-year follow-up” for each patient.

C: Rationale of the study paragraph: “...[OMISSIS]… For the purpose of this study, patients will be

consecutively enrolled for a 2-year period, and will be followed-up for a minimum of 5 years after the

index surgical treatment. Maximum follow-up length will be 10-years after surgery.”

Criteria for registry enrolment paragraph: “Every patient will be followed up at 30 days, 6 months, 1

year, and yearly thereafter up to a minimum of 5 years after the index surgical procedure (Figure 2).

Afterwards yearly follow-up will be closed at the completion of the 10th year from surgery for each

patient.”

11. Page 5, Line 34: Is it not possible to maintain the blinding of some of the analysis team?

A: We appreciated and take the reviewer’s suggestion. A Central Core Lab with no commitment will

manage the entire set of data (by auditing and by performing all the future statistical analyses).

However, there will be blinding against the surgical team because each institution will be identified in

the Database with a Capital Letter.

C: Data collection paragraph: “…Participating Centre: Each participating Centre will be anonymized

by identification with a capital letter. The correspondence between Centres and capital letters will only

be known by the PI of the study. The Central Core Laboratory analyzing the data will be therefore

blinded towards the surgical teams.”

12. Page 6, Line 25: The authors mention a “number” of sutureless aortic valves. Is that true? The

authors should list them. I know of Intuity (Edwards) and Percevel (Sorin).


http://bmjopen.bm

j.com/

BM



ownloaded from


A: We agree with the reviewer’s statement. We better explained this concept and rephrase the

sentence

C: “There are on the market two types of “sutureless” valves (i.e. Sorin Perceval and Edwards Intuity)

at the moment - aimed at reducing some surgical drawbacks such as cross-clamp time and

myocardial ischemia-reperfusion injury (13,15-20) – and it is possible that new “sutureless” valves will

enter the market in the next future.”

13.Are there multidisciplinary Heart Teams in every centre who help decide on the mode of treatment

of all AS patients?

A: In all Institutions the local Heart Team already works in allocating the patient to the best treatment

option. Therefore, this concept has been better clarified in the text.

C: In the Rationale of the study paragraph of the Methods and Analysis section is now reported: ”

Patient allocation to a specific surgical procedure will be based on the local Heart Team decision at

each Institution, according to standard clinical practice and current guidelines [2].”

14.Page 9, Line 10: Patients will be followed for 5 years – is that a minimum, mean or median of 5

years?

A: according to a previous comment, we have now better specified that a minimum of 5-year follow-up

will be considered.

C: Rationale of the study paragraph: “...[OMISSIS]… For the purpose of this study, patients will be

consecutively enrolled for a 2-year period, and will be followed-up for a minimum of 5 years after the

index surgical treatment. Maximum follow-up length will be 10-years after surgery.”

Criteria for registry enrolment paragraph: “Every patient will be followed up at 30 days, 6 months, 1

year, and yearly thereafter up to a minimum of 5 years after the index surgical procedure (Figure 2).

Afterwards yearly follow-up will be closed at the completion of the 10th year from surgery for each

patient.”

15. Page 9, Lines 12-9: Are TF TAVR cases performed through a femoral cutdown eligible? Carotid

access is becoming popular as an alternative access site – are these cases eligible? Are autografts or

homografts eligible, or only mechanical prostheses and xenografts? Are patients with annular or aortic

enlargement eligible? Are patients with composite replacement eligible? Are patients with ascending

aortic replacement not requiring circ arrest eligible? Are patients having AFib ablations eligible?

Preoperative Diagnosis: Is the Registry only for patients with aortic stenosis? For AS/AI? For pure AI?

Do patients need to have a specific gradient, dimensionless index, or AVA? Are low flow, low gradient

AS patients eligible? Are patients with acute endocarditis, or undergoing emergency operation for

valve thrombosis eligible?

A: We thank the reviewer for this important comment. Therefore we have better specified the inclusion

and exclusion criteria. In particular, only emergent surgery, endocarditis and coexistence of aortic

regurgitation are inclusion criteria. Similarly, any SAVS (i.e including low flow low gradient;

paradoxical low flow low gradient; regardless of gradients, AVA or AVAi…) will be included. All the

other preoperative conditions and surgical options reported by the reviewer are exclusion criteria from

the registry.

C: Rationale of the study and aim: “Therefore, the rationale of this European multicenter observational

open registry is to prospectively collect data on baseline characteristics, treatment options,

perioperative management and postoperative outcome of all patients consecutively undergoing

surgical treatment of SAVS (regardless of gradients, AVA or AVAi)±CAD or aortic prosthetic

dysfunction±CAD at 17 European university or non-university tertiary hospitals located in six

European countries (France, Germany, Italy, Spain, Switzerland, and United Kingdom).”

Paragraph “Criteria for registry enrolment” now reports:

Inclusion criteria

Age >18 yy


http://bmjopen.bm

j.com/

BM



ownloaded from


Isolated SAVS with or without concomitant aortic valve regurgitation

Isolated prosthetic aortic dysfunction

SAVS + coronary artery disease (CAD)

Prosthetic aortic dysfunction + CAD

Elective, urgent and emergent procedures

Endocarditic aetiology

Exclusion criteria

Patients undergoing concomitant mitral valve surgery, or tricuspid valve surgery, or aortic surgery (i.e.

composite aortic valve and ascending aorta replacement with o without circulatory arrest), or atrial

fibrillation surgery, or any other associated cardiac surgical procedure (with the exception of CABG)

Concomitant aortic root procedure (i.e. Bentall operation, David operation, homografts, autografts)

SAVR with techniques of aortic annular enlargement

Porcelain aorta

Pure aortic valve regurgitation

Percutaneous TAVR requiring surgical cut-down (i.e. failure to comply with a full percutaneous

approach, thus configuring a “hybrid procedure”)

Patient refusal

Same specifications are now reported also in the flow-chart (Figure 1)

16. Follow-up: Is F/U local in person, local by telephone, or central?

A: We thank for this comment. Follow up will be local in person, and follow-up outcomes further

checked by linking with regional Social Security Death and Events Master files (in those nations

where these files are available). Only if data are missed or absent, variables will be collected by direct

phone contact with general practitioners, and in case of persistently missing-data by phone contact

with patients and families. These specifications are added to the new version of the paper.

C: Data Management and Monitoring paragraph: “…[OMISSIS]… Baseline characteristics, operative

details and outcome data pertaining hospitalization will be prospectively collected from hospital

registries. Variables and events occurring after the index hospital discharge will be collected from

outpatient clinics at the individual Institutions, and linking with regional Social Security Death and

Events Master files where available. In case of absent/missing data, variables and events will be

collected by direct phone contact with general practitioners, and only if persistently missed by phone

contact with patients and families.”

17. Consent: It is frequently difficult to obtain consent for a research study prior to an emergency

operation. Will patients who underwent emergency surgery without consent for participation in the

Registry, be approached for consent after the surgery?

A: We thank the reviewer for the question. Informed consent for emergent cases will be collected from

the family before surgery, and from the patient after surgery, in light of the long-term follow-up

foreseen for these patients. However, “postoperative” consent (in emergencies) will be waived in case

of death or severe neurological damage precluding adequate informed consent.

C: Informed consent paragraph: “…[OMISSIS]…In case of emergent surgery, informed consent will

be collected from the patient’s family (or legal representative) before surgery, as well as from patient

after surgery (if unable to give it before intervention). This consent will be waived in case of death or

severe neurological damage precluding adequate postoperative patient informed

consent...[OMISSIS]….”


http://bmjopen.bm

j.com/

BM



ownloaded from


18. Data Management and Monitoring: I am unsure of the meaning of the last sentence in the

paragraph.

A: We thank the reviewer for pointing out the “unclear” sentence. The sentence only (and “unclearly”)

stigmatizes that the entire set of future statistical analyses performed by the central statistical core-lab

will be sent to all co-authors for “transparency” and clinical “interpretation” of data during the process

of writing a scientific paper.

C: The sentence is now: “The entire set of statistical analyses will be available to all E-AVR

researchers for the interpretation of data.”

19.Page 10 Statistical Methods: What and where is the Core laboratory? Who is the study

statistician? The statistics section mentions a number of statistical tests, but does not provide text

such as which test will be used for which comparison.

A: Based on this comment we have detailed the Core Laboratory in the “Data Management and

Monitoring” paragraph. The Core Laboratory is the “Unit for Clinical Research and Biostatistics,

Verona University Hospital, Verona, Italy”, whose Director is Ms G. Bisoffi, Biostatistician. However,

her name is not added to the text as co-Author at the moment because the single biostatistician

responsible for each analysis has not been already identified by Ms Bisoffi (however, a statement on

her statistical support is added to the Acknowledgements). Furthermore, the previously unexplained

statistical tests have been now clarified in the text of Statistical Methods paragraph

C: In “Data Management and monitoring” the following sentence is now reported “Data will be

collected into a dedicated datasheet with predefined variables..[OMISSIS]... Storage, analysis and

auditing of data will be accomplished by an independent Central Core Laboratory (Unit for Clinical

Research and Biostatistics, Verona University Hospital, Verona, Italy).”

In “Statistical Methods” paragraph it is now reported “Univariate analysis will be performed using the

Mann-Whitney U test and Student’s t-test for continuous variables (pending the not-normal or normal

distribution respectively), the Kruskall-Wallis test (independent multilevel ordinal variables), Wilcoxon

test (for paired variables), Fisher exact test and Chi-square test (for dichotomous/nominal variables)

and Kaplan-Meier test (for time-dependent dichotomous variables). Multivariable analyses will be

performed using logistic regression method (for categorical dependent variable), classification tree

analysis (for target variables with a discrete set of value), linear regression (for continuous dependent

variable) and ordinal regression methods (for ordinal dependent variable), as well as Cox-proportional

hazards method (to test the effects of covariates on time-dependent dichotomous variables).”

20.Page 11 Interim Analysis: Are there statistical penalties for the interim analyses?

A: We thank the reviewer for the precious comment. The statistical method to correct for interim

analysis p-values has been specified in the text and referenced

C: Statistical analysis paragraph: “Interim analyses are planned at different time-points (see Ethics

and Dissemination). Critical p-values of accomplished interim analyses will be corrected according to

the Armitage-McPherson adjustment (36).”

21. Page 11 Early Secondary Outcomes: The authors mention, “device success, early safety, clinical

efficacy, time-related valve safety”. How are these outcomes defined?

A: We have better specified the significance of these outcome, which are concordant with last VARC-

2 definitions.

C: Composite outcome subparagraph: “..[OMISSIS]….according to VARC-2 definitions (36), this

includes: 1) device success (absence of procedural mortality with correct positioning of a single

prosthesis and with intended performance of the prosthesis); 2) early safety at 30 days (composite

endpoint of all-cause mortality, all strokes, life-threatening bleeding, acute kidney injury stage 2 or 3,

coronary obstruction requiring intervention, major vascular complication or valve-related dysfunction

requiring repeat procedure); 3) clinical efficacy after 30 days (composite endpoint of all-cause

mortality, all strokes, hospitalization for valve-related symptoms or worsening congestive heart failure,

NYHA class III or IV, valve related dysfunction);


http://bmjopen.bm

j.com/

BM



ownloaded from


4) time-related valve safety (composite endpoint of structural valve deterioration requiring repeat

procedure, prosthetic valve endocarditis, thrombosis, thrombo-embolic events or valve-related VARC

bleeding ).

22. Page 12 Late Secondary Outcomes: It isn’t clear whether perioperative MACCE outcomes are

part of the late MACCE outcome.

A: Late MACCE did not include perioperative MACCE.

C: In Follow-up MACCE subparagraph: “…[OMISSIS]…defined as a composite end-point occurring

after the 30-day time-point (considered as hospitalization, 30th day if discharged home, or during

“rehab-hospitalization” at any time point if never discharged home), and including any of the following

adverse events: death from cardiovascular cause, stroke, myocardial infarction, follow-up repeated

revascularization”

23.Page 12 Laboratory parameters: Will hs-troponin be collected?? Will BNP be collected?

A: Although of great interest for the scientific community, hs-troponin and BNP assays are not

planned because of the extreme variability in the generations of laboratory test of each institution. As

example, some centers still have 1st or 2nd generation troponin assay, therefore data are not

comparable between the Institutions.

C: None

24.Page 20 Length of stay in the intensive care unit: It is better to initially collect ICU LOS in hours

rather than days.

A: We thank the reviewer for the comment. The unit of measurement has changed into “hours”.

C: Data Collection paragraph: “..[OMISSIS]..Length of stay in the intensive care unit: number of hours

spent in the intensive care unit from surgery.”

25. Steering Committee: What is the frequency of the Steering Committee meetings?

A: We thank the reviewer for the question. The established plan of E-AVR Steering Committee

meetings has been reported in the text.

C: Ethics and Dissemination paragraph: “…[OMISSIS]… Data collection, analysis and writing process

will be monitored by the Steering Committee of the E-AVR, which is made up of the Principal

Investigator and a Representing Member from each of the participating centres. It is expected that

periodical E-AVR Steering Committee meetings will occur, every 6 months for the first 2 years, yearly

thereafter up to the end of follow-up.”

26.DSMB: Is there a DSMB?

A: No Data Safety Monitoring Board has been considered for the purpose of this registry-trial,

because E-AVR Steering Committee by means of their individuals will audit continuously data at a

local level, and another six-month data auditing by the external Central Core Statistical Laboratory is

already planned.

C: None

27.Page 49: “Evaluation of limits and benefits of combined SAVR and CABG vs SAVR and staged

PCI or TAVR and staged/reversed stage/combined PCI”. The meaning of this sentence is not clear.

A: Based on a previous comment, this second part of the protocol has been completely deleted, and a

single merged protocol with a SPIRIT checklist is now submitted.

C: The sentence (and the entire second part of the protocol) were deleted.

28. Page 49: “Identification of indication criteria for SAVR based on baseline characteristics”. The

meaning of this sentence is not clear.


http://bmjopen.bm

j.com/

BM



ownloaded from



single merged protocol with a SPIRIT checklist is now submitted.


29. Page 49: “Analysis of costs related to hospitalization of AS-patients treated with different surgical

modalities (e.g. sutureless valves vs stented/stentless, sternotomy vs minimally-invasive, standard

SAVR vs surgical TAVR, etc) and evaluations of cost/effectiveness”. The authors do not describe

much in terms of the acquisition of cost data.


single merged protocol with a SPIRIT checklist is now submitted. Cost-effective analysis is no more

considered as a potential sub-study of the registry.


30. Page 50 Line 25: The authors describe audit of 10% of the data. Earlier (page 10 line 27) 40% is

mentioned.

A: We apologize for the typing error.

C: The new version of the protocol the correct value of 40% is reported. “…[OMISSIS]… Auditing of

the dataset will be performed every six months by checking the data of 40 % of patients.”

Reviewer: Marechaux Sylvestre

Institution and Country

Groupement des hôpitaux de l'Institut Catholique de Lille

This paper is the design of a study aiming at comparing in the real life outcomes of different surgical

strategies for patients being referred in the participating centers for aortic valve replacement. This

study comes in a timely manner. The question is highly relevant. The paper is well written, outcomes

are clearly defined. The sample size in the context of a multicenter study is large and appropriate. My

only question is how the authors are sure that all surgical strategies will be sufficiently represent,

especially for propensity matching comparisons of outcome data between groups.

A: We thank the reviewer for his overall appreciation of the study/registry. Based also on Editor’s

request and another reviewer’s request a sample size calculation is added to the text. Furthermore, a

primary objective (therefore a main outcome study) has been clearly identified, and all the sub-studies

planned to date should be considered in their “exploratory” nature. Therefore sample size calculation

was waived for these “sensitivity” sub-studies. However, as easily deductable from the sample size

calculation, the expected number of enrolled patients (>8000) is far beyond the requested sample size

of 2866 patients (to reach 80% power to detect a 10% difference in all-cause mortality at 5 years),

and it accounts for risk-adjustment methodologies and the expected 1.5% of lost to follow-up (based

on historical data).

Coming back to the reviewer comment, we are sure to have all the surgical techniques sufficiently

represented in the Registry, because most of the enrolled centres are high-volume centres and still

have different options of surgical treatment of SAVS (transcatheter and surgical, via minimally

invasive or full-sternotomy approaches, stented or stentless valves, etc.), which are differently used in

their daily practice. We think therefore that merging data from these high-volume Centres will help to

overcome the “local bias”, and help to have a large number of patients treated with each technique,

amenable to different statistical tools (e.g. risk-adjustments and propensity-scores) able to reach

conclusive data.

C: C: In the Methods section, “Rationale of the study and aim”, it has been specified:

“…[OMISSIS]…The primary objective of the study is a 5-year comparison between SAVR and

surgical TAVR: we hypothesize to report a 10% superiority in terms of all-cause mortality in favor of

SAVR vs TAVR.”.


http://bmjopen.bm

j.com/

BM



ownloaded from


Accordingly, in the Methods section, “Statistical Analysis” paragraph we have now added the

sentence: “Given the PARTNER TRIAL 5-year all-cause mortality risk (67.8% after TAVR and 62.4%

after SAVR) – accepting a type 1 error of 5% - the overall number of patients needed to achieve 80%

power (1-beta) for a mortality odds of 1.1 (10% mortality difference) is 2866 patients (i.e. 1433

patient/group) (7). Therefore, the expected number of 8000 patients is far beyond the requested

sample size of the primary objective of the registry, and further accounts for risk-adjustment

methodologies and the expected 1.5% (historical data) of lost to follow-up.”

We hope You and the Reviewers will now react favorably to the study. Please let us know if You need

additional information or have any questions. We greatly appreciate your consideration of the

manuscript. Looking forward to hearing from you at Your earliest convenience, I send my best

Kindest Regards.

Francesco Onorati, MD PhD

Division of Cardiac Surgery University of Verona Medical School

On behalf of E-AVR Investigators


REVIEWER Stephen Fremes Sunnybrook Health Sciences Centre University of Toronto Ontario Canada

REVIEW RETURNED 06-Oct-2017

GENERAL COMMENTS The authors have significantly revised the study design of the E-AVR Registry. I did review the earlier submission. The use of 5 year all-cause mortality for the primary outcome, makes this a much better proposal. Also, the planned F/U for a minimum of 5 years and a maximum of 10 years is also important. The authors have included SS calculations. The control group that the authors chose does not make sense to this reviewer - the 5 year results of the PARTNER 1 trial. (Not sure if the authors used the TAVR or surgical AVR results for this but less important). The likely patients in E-AVR will likely be of much lower risk than this patient cohort (the high risk PARTNER study). Also, looking forward, as there is a trend to use TAVR in patients of lower risk; it is likely that the surgical patients in E_AVR will generally be of lower risk than in the past, except for patients who are young and very complicated for various reasons. The authors will be using Cox PHM. The authors need to test for proportional hazards. The authors should consider evaluating clinical outcomes centrally.


http://bmjopen.bm

j.com/

BM



ownloaded from


The authors should consider evaluating some proportion of the qualifying echoes and postop echoes centrally. I understand that reviewing all echoes centrally may not be of great value. The authors will use a study ID that contains a lot of personal information (city code, initials, and DOB). I suggest that the authors use a study ID that includes a code for country, centre or city, and then consecutive numbers. The proposed study is valuable but would of much greater value if the investigators were also collecting data for TF TAVR (perc or cutdown). The authors talk about collaborations - maybe they are thinking about using collaborations to get TF TAVR results.


Reviewer: 1



1. The use of 5 year all-cause mortality for the primary outcome, makes this a much better proposal.

Also, the planned F/U for a minimum of 5 years and a maximum of 10 years is also important.

A: We thank the reviewer for his overall appreciation to the new version of the protocol.

C: None

2. The authors have included SS calculations. The control group that the authors chose does not

make sense to this reviewer - the 5 year results of the PARTNER 1 trial. (Not sure if the authors used

the TAVR or surgical AVR results for this but less important). The likely patients in E-AVR will likely be

of much lower risk than this patient cohort (the high risk PARTNER study). Also, looking forward, as

there is a trend to use TAVR in patients of lower risk; it is likely that the surgical patients in E_AVR will

generally be of lower risk than in the past, except for patients who are young and very complicated for

various reasons.

A: We agree with the reviewer opinion and thank for the precious comment. Therefore, our

Biostatistician (now added also in the co-Author’s list) performed a new power analysis, based on a

very recent literature paper (2017) reporting outcome of SAVR in intermediate risk patients (those

who will also be treated by TAVR, according to 2017 international Guidelines): this patient population

will account for the majority of patients enrolled in our prospective surgical registry. Accordingly, with

our hypothesis of 10% superiority of SAVR over surgical TAVR in terms of 5-year mortality,

considering a 1.5% censoring based on historical data, a 2-year enrollment period, and a 5 year

follow-up length, the resulting sample size of 6493 patients will result in a 80% power with 5%

significance level. The referenced literature study was also added in the reference list (n.36).

C: Statistical Methods paragraph:… “The sample size calculation is based on a recent published

study (36), specifically on the 5-year all-cause mortality rate (24.6%) of the 620 intermediate risk

patients, our target population, who underwent SAVR. We hypothesize that the 5-year mortality rate

following SAVR is 10% inferior compared to the surgical TAVR, and calculate the sample size using a

conservative one-sided log-rank test. Assuming that patients will enter the study uniformly over the 2

years accrual time, 5 years total follow-up time, 80% statistical power, 5% significance level and 1.5%

(historical data) probability of patient’s right censoring occurs. The resulting sample size is 6493

subjects (3246 in the control group and 3247 in the treatment group). Therefore the expected number

of 8000 patients over a 2-year enrolment period is far beyond the requested sample size of the

primary endpoint of the trial.”


http://bmjopen.bm

j.com/

BM



ownloaded from


3. The authors will be using Cox PHM. The authors need to test for proportional hazards.

A: We completely agree with this comment. Again our biostatistician replied to this query.

C: Statistical Methods paragraph:…“Cox-proportional hazards method will test the effects of

covariates on time-dependent dichotomous variables; the model’s proportional hazard assumption will

be checked using the Schoenfeld residuals test.”

4. The authors should consider evaluating clinical outcomes centrally.

A: The E-AVR Steering committee completely took this reviewer’s suggestion. The final decision was

that all clinical outcomes will be adjudicated centrally.

C: Data management and monitoring paragraph: “Events and outcome variables will be adjudicated

centrally by a Central Core Laboratory (Unit for Clinical Research and Biostatistics, Verona University

Hospital, Verona, Italy ). In the event of controversy on outcome adjudication, this will be discussed

and adjudicated after a final consult between the Central Core Laboratory and the E-AVR Steering

Committee.”

5. The authors should consider evaluating some proportion of the qualifying echoes and postop

echoes centrally. I understand that reviewing all echoes centrally may not be of great value.

A: We thank the reviewer for this suggestion. Indeed, based on our resource allocation and the

expected very high number of follow-up echoes, the Steering Committee decided that 5% of

echocardiographic exams will be reviewed centrally.

C: Early and Late Endpoints paragraph: “Echocardiographic data of prosthesis performance are

defined according to the Valve Academic Research Consortium-2 definitions (38). All

echocardiographic data will be collected from 3rd level nationally and/or internationally certified

Institutional Echo Laboratories: 5% of these echocardiographic exams will be reviewed centrally (Unit

for Clinical Research and Biostatistics, Verona University Hospital, Verona, Italy) by third level

certified echocardiographers.”

6.The authors will use a study ID that contains a lot of personal information (city code, initials, and

DOB). I suggest that the authors use a study ID that includes a code for country, centre or city, and

then consecutive numbers.

A: We thank the reviewer for his suggestion, which was totally accepted.

C: Data management and monitoring paragraph: “Each patient enrolled in the Registry will be

anonymized by the generation of a code consisting of the initials of the enrolling Country (2 letters),

enrolling Centre (2 letters), and then consecutive number (considered at thousands)(e.g. Mr. XY, third

patient enrolled in London = UKLO0003)”

7. The proposed study is valuable but would of much greater value if the investigators were also

collecting data for TF TAVR (perc or cutdown). The authors talk about collaborations - maybe they are

thinking about using collaborations to get TF TAVR results.

A: We completely agree with the reviewer’s opinion. For this reason, we have tried since the

beginning to build up a comprehensive registry encompassing also percutaneous or cutdown TAVR.

However, as already stated during the first round of review, some interventional cardiologists of the

enrolled centers denied participation (mostly because already involved in different registries and/or

RCTs). Therefore, we decided at that time to have a “pure” surgical registry. There’s no doubt that

enrollment of “cardiological” TAVR will add more strength and value to our registry. However, there is

only a preliminary project at the moment aimed at merging our registry with a concurrent multicenter

Italian registry enrolling only percutaneous TAVR, with a similarly planned follow-up of 5 years.

Honestly, whether this merging will occur in the future or not cannot be ratified at this stage

C: None


http://bmjopen.bm

j.com/

BM



ownloaded from




manuscript. Looking forward to hearing from you at Your earliest convenience, I send my best

Kindest Regards.





REVIEWER Stephen Fremes Schulich Heart Centre Sunnybrook Health Sciences Centre University of Toronto Canada

REVIEW RETURNED 05-Nov-2017

GENERAL COMMENTS Overall, the authors have been very responsive revising the manuscript according to earlier comments. Page 9 line 37: "Percutaneous TAVR requiring surgical cut-down (i.e. failure to comply with a full percutaneous approach, thus configuring a “hybrid procedure"". I am not sure what exactly is meant here - I think that the authors are excluding all TF TAVR, either by a completely percutaneous approach or by surgical cutdown. I suggest the authors reword this exclusion criteria for clarity. Page 11 line 45: "We hypothesize that the 5-year mortality rate following SAVR is 10% inferior compared to the surgical TAVR, and calculate the sample size using a conservative one-sided log rank test." This contradicts page 8 line 12: " we hypothesize to report a 10% superiority in terms of all-cause mortality in favor of SAVR vs TAVR". I think that the authors expect that SAVR survival will exceed that of TAVR. Please ensure that these 2 statements are consistent. The sample size calculations still require some clarity (starting on page 11, statistical methods). Firstly, I doubt that there will be equal numbers of SAVR and TAVR patients, which I assume is what the authors mean when they say control and treatment groups. I anticipate that there will be many more SAVR patients than TAVR (as the TAVR patients are all TAVR using a non-TF approach). The authors should reference the program they used for SS calculations. Did they test for a difference in the HR of 10% or an absolute difference of 10%? In RCTs, SS are often based on the total number of events needed to be observed, based on anticipated control event rate, HR and alpha and beta. Did the authors determine a certain number of events to be observed? Another important adjustment to the SS is LTFU.


http://bmjopen.bm

j.com/

BM



ownloaded from


In any case, the authors intend to enroll approximately 8000 patients, of which a certain proportion will be SAVR and the remainder TAVR. They should include a statement similar to the following, "Considering the estimated event rate of 25% in the TAVR patients at 5 years, and an anticipated LTFU of __%, and an improvement with SAVR of 10% (either a relative or absolute improvement with SAVR according to what the authors consider important), alpha 0.05 (1 or 2 tailed?) the registry will have approximately ____% power." The authors could then perform similar calculations for other key comparisons. The authors have included a very detailed statistical analysis plan. They mention PS - should also include the variables to be included in the derivation of the PS, and that analyses with PSM cohorts will be performed accounting for the matched nature of the data. I think there authors also need to decide whether the primary analysis comparing mortality between SAVR and TAVR should be an unadjusted or an adjusted analysis. Clearly both should be performed, although the TAVR group will likely be of higher overall risk except for some emergency SAVR patients such as extensive endocarditis, thrombosed valves, etc. - the adjusted analyses are likely more meaningful. I consider the registry will be very valuable but mainly for comparative outcomes between the different surgical cohorts. The total SS is substantial - we don't know the expected numbers of the different surgical cohorts. That is important to determine the power for some of the surgical comparisons. The study will be important. The editors should decide whether they would like a statistical reviewer to be assigned at this juncture.


Reviewer: 1



1. Query (Q). 1: Page 9 line 37: "Percutaneous TAVR requiring surgical cut-down (i.e. failure to

comply with a full percutaneous approach, thus configuring a “hybrid procedure"". I am not sure what

exactly is meant here - I think that the authors are excluding all TF TAVR, either by a completely

percutaneous approach or by surgical cutdown. I suggest the authors reword this exclusion criteria for

clarity.

Answer (A): We completely agree with the reviewer suggestion and changed the sentence based on

his comment

Changes (C ): page 9 line 37 :”…. Percutaneous TAVR (regardless of the technique, i.e. either by a

completely percutaneous approach or by surgical cut-down)

2. Q 2: Page 11 line 45: "We hypothesize that the 5-year mortality rate following SAVR is 10% inferior

compared to the surgical TAVR, and calculate the sample size using a conservative one-sided log

rank test." This contradicts page 8 line 12: " we hypothesize to report a 10% superiority in terms of all-

cause mortality in favor of SAVR vs TAVR". I think that the authors expect that SAVR survival will

exceed that of TAVR. Please ensure that these 2 statements are consistent.


http://bmjopen.bm

j.com/

BM



ownloaded from


A:We have better explained that the hypothesis (and trial design) is to detect an expected 10%

inferior 5-year all-cause mortality event rate after SAVR compared to TAVR.

C: Page 8: “ The primary aim of the study is a 5-year comparison between SAVR and surgical TAVR:

we hypothesize to report a 10% inferior 5-year all-cause mortality event rate in SAVR, i.e., we expect

that SAVR survival will exceed by 10% (absolute value) that of surgical TAVR.”.

Similarly, and based on the suggested sentence of the reviewer (see later Query n.5), on page 11 it is

reported now: “It is intended to enrol 8000 patient, of which 60 to 70% will be SAVR and the

remainder TAVR (historical data based on Institutional practices). Considering the estimated event

rate of 25% in the TAVR patients at 5 years (36), a 2-year accrual time, an anticipated loss-to-follow-

up rate of 1.5% (historical data), and the target power of 80% at a 0.05 one-sided log-rank test

significance level to detect the hypothesized 10% inferior (absolute improvement) 5-year all-cause

mortality rate in favour of SAVR, the calculations showed the overall sample size of the registry to

meet the targeted power for all expected scenarios of estimated proportions of SAVR and TAVR

patients.”

3. Q 3: The sample size calculations still require some clarity (starting on page 11, statistical

methods). Firstly, I doubt that there will be equal numbers of SAVR and TAVR patients, which I

assume is what the authors mean when they say control and treatment groups. I anticipate that there

will be many more SAVR patients than TAVR (as the TAVR patients are all TAVR using a non-TF

approach).

A:We thank for the suggestion. It is now specified in the paper the expected proportion of SAVR and

TAVR, and the study showed to be powered for all scenarios considering the various estimated

proportions of SAVR and TAVR patients. Based on historical data from our Institutions, the reviewer is

right when assessing that the proportion of SAVR and TAVR will not be equal. We expect that 60-

70% of patients will be SAVR, and 30-40% TAVR. Indeed, it is possible that the later percentage will

improve during the 2-year accrual time, based on the recent extension of TAVR indication to

intermediate-risk patients. Therefore, the expected minimum number of >8000 patients will suffice to

test our hypothesis. Thus sample size calculation has been better explained according to the

reviewer’s suggested sentence (see later), and the expected proportion of SAVR and TAVR has been

clearly reported in the text

C: Page 11 Statistical Methods paragraph: “It is intended to enrol 8000 patient, of which 60 to 70%

will be SAVR and the remainder TAVR (historical data based on Institutional practices).”

4. Q 4: The authors should reference the program they used for SS calculations. Did they test for a

difference in the HR of 10% or an absolute difference of 10%?

A: The employed program for SS calculation has been added to the text, and we have also explained

that a 10% absolute difference was considered for this registry.

C: Page 12: “Statistical calculations were accomplished with PASS 14.0 Power Analysis and Sample

Size Software (2015) statistical package (NCSS, LLC; Kaysville, Utah, USA -

ncss.com/software/pass).”; Page 8: “…..we hypothesize to report a 10% inferior 5-year all-cause

mortality event rate in SAVR, i.e., we expect that SAVR survival will exceed by 10% (absolute value)

that of surgical TAVR”; furthermore, on pages 11-12: “…Considering the estimated event rate of 25%

in the TAVR patients at 5 years (36), a 2-year accrual time, an anticipated loss-to-follow-up rate of

1.5% (historical data), and the target power of 80% at a 0.05 one-sided log-rank test significance level

to detect the hypothesized 10% inferior (absolute improvement) 5-year all-cause mortality rate in

favour of SAVR, the calculations showed the overall sample size of the registry to meet the targeted

power for all expected scenarios of estimated proportions of SAVR and TAVR patients.”


http://bmjopen.bm

j.com/

BM



ownloaded from


5. Q 5: In RCTs, SS are often based on the total number of events needed to be observed, based on

anticipated control event rate, HR and alpha and beta. Did the authors determine a certain number of

events to be observed? Another important adjustment to the SS is LTFU. In any case, the authors

intend to enroll approximately 8000 patients, of which a certain proportion will be SAVR and the

remainder TAVR. They should include a statement similar to the following, "Considering the estimated

event rate of 25% in the TAVR patients at 5 years, and an anticipated LTFU of __%, and an

improvement with SAVR of 10% (either a relative or absolute improvement with SAVR according to

what the authors consider important), alpha 0.05 (1 or 2 tailed?) the registry will have approximately

____% power." The authors could then perform similar calculations for other key comparisons.

A: We thank the reviewer for asking to be clearer when reporting the employed SS calculation for the

primary endpoint, and to be clearer also about secondary endpoints. The authors are aware of other

types of study design: the interest here is a study design where a fixed number of patients are

enrolled and followed for a specified period of time. Specifically, the study is powered to detect a

minimal hypothesized mortality rate difference between the SAVR and TAVR groups at 5-year time-

period. Therefore, we have reported the SS calculation of the primary endpoint, resembling the

suggested sentence of the reviewer. Moreover, we also rephrased the sentence “1.5% (historical

data) probability of patient’s right censoring occurs” to make it clearer it is referred to lost-to-follow-up.

When secondary endpoints are considered, we have better expressed the achieved power with

current SS only for 30-day all cause mortality, given its “rough” clinical significance and the “well-

reported” event rate in recent literature studies comparing SAVR and TAVR (reference n. 37), given a

meaningful hypothesis to test for. However, in the interest of space (due to the allowed limited length

of a paper), and considering that a lot of early and late secondary endpoints will be collected

(requested by VARC-2 criteria), and due to the fact that event rates reported in the literature for most

of these secondary endpoints are too different from study to study, we consider that it is difficult to

establish an hypothesis for these secondary endpoints. For the above mentioned reasons, we

decided (since the beginning of the Registry design) to not report any achieved power for all the other

secondary endpoints: these secondary endpoints should be considered for their nature, i.e. as

exploratory analyses, potentially useful for SS calculations of future clinical trials.

Indeed, a sentence clarifying the nature of these secondary endpoints is also added to the text.

C: Page 11-12 Statistical methods paragraph: “….Considering the estimated event rate of 25% in the

TAVR patients at 5 years (36), a 2-year accrual time, an anticipated loss-to-follow-up rate of 1.5%

(historical data), and the target power of 80% at a 0.05 one-sided log-rank test significance level to

detect the hypothesized 10% inferior (absolute improvement) 5-year all-cause mortality rate in favour

of SAVR, the calculations showed the overall sample size of the registry to meet the targeted power

for all expected scenarios of estimated proportions of SAVR and TAVR patients.

Similar calculation shows that, considering the estimated 30-days all-cause mortality event rate of

5.0% after SAVR and of 5.5% after TAVR (0.5% absolute difference) in intermediate-risk patients

(37), the overall sample size of the registry also meets the targeted 80% power to detect a difference

as small as 0.5% in 30-days all-cause mortality event rate in favour of SAVR. Other secondary

endpoints will serve as exploratory analyses, possibly useful for sample size estimation of future

clinical trials.”

6. Q 6: The authors have included a very detailed statistical analysis plan. They mention PS - should

also include the variables to be included in the derivation of the PS, and that analyses with PSM

cohorts will be performed accounting for the matched nature of the data.

A: we totally agree with this statement, and thank the reviewer for the request of clarification. In the

interest of space – given that several sub-studies deserve propensity-score method for statistical

analysis – a single sentence detailing factors entering PSM is reported at the end of the list of planned

studies.


http://bmjopen.bm

j.com/

BM



ownloaded from


C: Page 29, Ethics and Dissemination paragraph: “…For all the above-mentioned studies considering

propensity-score method, demographic data, gender, EuroSCORE-2. STS Score, NYHA functional

class, frailty scale, left ventricular function, comorbidities, and all those baseline characteristics

(anthropometric data, key laboratory tests, echocardiographic parameters and surgical factors

included) having a different distribution at univariate analyses between the 2 patient-populations

and/or potentially acting as bias will be included in the derivation of the propensity score.

Furthermore, the variable “participating centre” will always enter the propensity-score model, in order

to account for “undetectable” differences between participating centres (e.g. perioperative care

protocols, institutional protocols, ethnicity, environmental factors, etc. not collected in the Registry).

Subsequent analyses with matched cohorts will be performed accounting for the matched nature of

the data.”

7. Q 7: I think there authors also need to decide whether the primary analysis comparing mortality

between SAVR and TAVR should be an unadjusted or an adjusted analysis. Clearly both should be

performed, although the TAVR group will likely be of higher overall risk except for some emergency

SAVR patients such as extensive endocarditis, thrombosed valves, etc. - the adjusted analyses are

likely more meaningful.

A:We agree with the suggestion. We have better explained in the text that both adjusted and

unadjusted analyses will be performed

C:Page 26: “…Either adjusted and unadjusted analyses will be performed, although the adjusted

analysis will be of clearer value, given the expected different risk-profiles of SAVR and TAVR patients.

Adjustment will be made by entering as covariates all demographic and anthropometric data, risk

scores, comorbidities, and those baseline characteristics (key echocardiographic and

surgical/technical factors included) having a different distribution at univariate analyses between the 2

patient-populations.”



manuscript. Looking forward to hearing from you at Your earliest convenience, I send my Best

Regards and wish You and the Editorial Staff a Merry Christmas and a Happy New Year.





http://bmjopen.bm

j.com/

BM



ownloaded from


PEER REVIEW HISTORY ...€¦ · femoral cutdown eligible? Carotid access is becoming popular as an...

Documents

Transcript of PEER REVIEW HISTORY ...€¦ · femoral cutdown eligible? Carotid access is becoming popular as an...