Statistical Analysis Plan: CMR Longitudinal Strain Analysis in Aortic Stenosis

CMR Longitudinal Strain Analysis in Aortic Stenosis

STATISTICAL ANALYSIS PLAN

November 2017

Spath NB¹, Gomez M², Everett R¹, Shah ASV¹, Semple S³,  Chin CLW⁴,  Newby DE^1,5, Dweck MR^1,5

1. BHF/University Centre for Cardiovascular Science, University of Edinburgh, UK

2. Hospital del Mar Medical Research Institute, Universitat Autònoma de Barcelona, Spain

3. Clinical Research Imaging Centre, University of Edinburgh, UK

4. Department of Cardiovascular Science, National Heart Center, Singapore

5. Department of Cardiology, Royal Infirmary of Edinburgh, UK

Correspondence:

Dr Nicholas B Spath

BHF/University Centre for Cardiovascular Science

Chancellor’s Building

University of Edinburgh

Edinburgh EH16 4SB

United Kingdom

Tel: +44 7841 044 129

Fax:   +44 131 242 6379

E-mail: [email protected]

Contents

1Introduction

1.1Background

1.2Study objectives

1.2.1Hypothesis

2Statistical analysis plan

2.1.1Objectives

2.1.2General Statistical Principles

2.1.3Analysis population

2.1.4Software

3List of analyses

3.1Baseline Characteristics

3.2Primary outcomes

3.2.1Outcome summary

3.2.2Primary analyses

4References

1Introduction

1.1Background

There is interest in developing non-invasive measures of myocardial health for patients with aortic stenosis in order to identify the early stages of left ventricular (LV) decompensation and optimise the timing of aortic valve replacement (AVR). Cardiac magnetic resonance (CMR) imaging techniques are well established in detecting and quantifying replacement fibrosis and LV remodelling in the context of aortic stenosis using extracellular volume (ECV) and late gadolinium-enhancement (LGE) []. CMR strain analysis is emerging as a potentially valuable tool in this field [] using steady-state free precession acquisition obviating the need for contrast imaging. Across a prospective cohort of patients with aortic stenosis, ranging from mild to severe, attending the Edinburgh Heart Centre, we propose to evaluate the relationship between CMR strain analysis and established markers of AS severity, LV remodelling and clinical outcomes.

This analysis will include anonymised cardiac MRI data from all stable patients with at least mild aortic stenosis (aortic jet velocity ≥2 m/s) who attended the Edinburgh Heart Centre between March 2012 and August 2014 who were invited to participate in our recently published prospective observational cohort study []. All participants with fully analysable CMR datasets will be included (complete 2 chamber, 3 chamber, 4 chamber cines, full short-axis cine stack). Partially complete datasets will be included for global longitudinal measurements only. Images will be analysed by two independent operators, both experienced in CMR image analysis, using commercially available software (CVI4.2^®, Tissue Tracking Module, Circle Cardiovascular Imaging, Canada) at the Edinburgh Imagine Facility at the Queen’s Medical Research institute.

1.2Study objectives

To evaluate the determinants and clinical association of CMR strain analysis in patients with aortic stenosis.

1.2.1Hypothesis

CMR global longitudinal strain analysis will positively correlate with established imaging markers of LV remodelling in the context of aortic stenosis (extracellular volume, late-enhancement, LV volume, ejection fraction, LV mass, diastolic dysfunction) as well as clinical endpoints (mortality, heart failure admission).

2Statistical analysis plan

2.1.1Objectives

The objective of this SAP is to describe the statistical analyses contributing to the final report and primary publication(s) of the analyses described below.  

2.1.2General Statistical Principles

Baseline data will be summarised as proportions for categorical data. Parametric and non-parametric continuous data will be presented as mean and standard deviations or median and interquartile range respectively.

2.1.3Analysis population

The data analysed in this study is derived from the dataset from our previously published data in aortic stenosis, above. Participants in this study consented to use of their anonymised data in future research studies at this centre.

2.1.3.1Control cohort

The control cohort were age and sex-matched to study participants with aortic stenosis at the time of recruitment. Control patients with imaging evidence of myocardial infarction will be excluded. Control data will be used as a comparator against aortic stenosis data, but not included in the extended multivariate analysis.

2.1.3.2Aortic stenosis cohort

The aortic stenosis cohort includes patients with mild, moderate and severe aortic stenosis, as well as patients with severe aortic stenosis who had been referred for AVR at the time of recruitment. Clinical endpoints will be collected between March 2012 and September 2015 with mortality data collected from the National Records of Scotland (NRS) database.

2.1.4Software

Image analysis will be performed in CVI4.2^®, Tissue Tracking Module, Circle Cardiovascular Imaging, Canada. Statistical analyses will be performed in R (Vienna, Austria). Images will be analysed by one of two independent operators.

3List of analyses

3.1Baseline Characteristics

Summary statistics will be provided for the baseline characteristics for both control patients and patients with varying severity of aortic stenosis. The following variables where available will be reported:

•Age (years) – mean(standard deviation)•Sex (male/female) n(%)•Cardiovascular risk factors (hypertension, hyperlipidaemia, diabetes mellitus) n(%)•Past medical history (ischaemic heart disease, previous coronary revascularisation) n(%)•Medical therapy on admission (dichotomous) n(%)oAspirinoBeta blockeroAngiotensin converting enzyme (ACE) inhibitor / angiotensin receptor blocker (ARB)•Statin•CMR (EF, mass, volumes, iECV, LGE, ECV) and Echo markers of LV function and aortic stenosis severity

Aortic stenosis patients will be divided into three groups a priori; by tertiles according to % global longitudinal strain. Continuous variables will be compared using parametric and non-parametric tests as appropriate. Categorical variables will be compared using the chi-square or fisher’s exact test where appropriate.

3.2Primary outcomes

3.2.1Outcome summary

The following primary analyses will be presented:

•Multivariate linear regression analysis to assess relationship between global longitudinal strain (GLS) and clinical parameters, established imaging markers of LV remodelling in aortic stenosis.•Multivariate cox regression analysis to evaluate the relationship between GLS and clinical outcomes of mortality and heart failure admission adjusted for both clinical and imaging confounders.

3.2.2Primary analyses

Clinical and MRI Parameters

Model 1 – GLS & Clinical Parameters

a) Age + Sex

b) Age + Sex + Co-morbidities

c) Age + Sex + Co-morbidities + AS severity

d) Age + Sex + Co-morbidities + AS severity + HS Troponin I (baseline)

  

Model 2 – GLS & CMR Parameters

a) LV ejection fraction (CMR)

b) LV ejection fraction (CMR) + Indexed ECV

c) LV ejection fraction (CMR) + Indexed ECV + presence of LGE (midwall or infarct)

Model 3 – GLS & Clinical + CMR Parameters

a) [Model 1d + Model 2c] 

Clinical Outcomes

Model 4 – GLS and All-cause mortality (ACM)

a) GLS (unadjusted)

b) GLS + Age + Sex

c) GLS + Age + Sex  + LVEF

d) GLS + Age + Sex + LVEF (CMR) + LGE (midwall or infarct)

Model 5 – GLS and Composite Endpoint 1 [All-cause mortality (ACM)/Heart Failure Admission]

a) GLS (unadjusted)

b) GLS + Age + Sex

c) GLS + Age + Sex  + LVEF

d) GLS + Age + Sex + LVEF (CMR) + LGE (midwall or infarct)

4References

1. Everett RJ, Stirrat CG, Semple SIR, Newby DE, Dweck MR, &Mirsadraee S. Assessment of myocardial fibrosis with T1 mapping MRI. Clinical Radiology. 2016;71(8),768–778.

2. Hwang JW, Kim SM, Park SJ, Cho EJ, Kim EK,Chang SA, et al. Assessment of reverse remodeling predicted by myocardialdeformation on tissue tracking in patients with severe aortic stenosis: acardiovascular magnetic resonance imaging study. Journal of Cardiovascular Magnetic Resonance. 2017;19(1), 984.

3. ChinCWL, Everett RJ, Kwiecinski J, Vesey AT, Yeung E, Esson G, et al. MyocardialFibrosis and Cardiac Decompensation in Aortic Stenosis. J Am Coll Cardiol: Cardiovascular Imaging
.2017;10(11), 1320–1333.