Table 2, Summary of studies included in the evidence review - Evidence review for CT and MRI indications for intervention

Table 2Summary of studies included in the evidence review

Study	Population	Analysis	Prognostic variables	Confounders	Outcomes	Limitations
Aortic stenosis – LVEF <50% on cardiac MRI
Everett 2020⁸⁸ N=440 UK, Germany, USA, Canada, South Korea Prospective cohort study	Severe AS scheduled for AVR: 36% in NYHA class III/IV Mean age 69.67 years	Multivariate Cox regression model	LVEF <50% on cardiac MRI	Extracellular volume percentage, age, gender, LGE on cardiac MRI and peak aortic jet velocity	All-cause mortality following aortic valve intervention – median follow-up 3.8 years	Risk of bias: very high Indirectness: Population - all already had an indication for intervention as scheduled for aortic valve intervention
Hwang 2020¹²³ N=43 South Korea Prospective cohort study	Severe AS scheduled for AVR: mean NYHA class 2.1 Mean age 65.9 years	Multivariate Cox proportional hazard regression analysis	LVEF <50% on cardiac MRI	Univariate results only	Cardiovascular death, hospitalisation for cardiac causes, non-fatal stroke and symptomatic aggravation (worsening NYHA class) following AVR– median follow-up 38.8 months	Risk of bias: very high Indirectness: Population - all already scheduled for AVR so no uncertainty as to whether there is an indication for intervention prior to cardiac MRI Outcome - composite of multiple outcomes in the protocol
Lindsay 2016¹⁵⁸ N=187 UK Retrospective cohort study	Those undergoing TAVI for AS: >70% with symptoms at rest or marked limitation of physical activity and median aortic valve area on echocardiography 0.60 cm² in whole cohort, though unclear for those included in this analysis Median age for whole cohort was 81 years, not clear for those included in this analysis	Cox regression analysis	LVEF 30–49% on cardiac MRI LVEF <30% on cardiac MRI	Univariate results only	All-cause mortality following TAVI – median follow-up 850 days for whole cohort, though unclear for those analysed here	Risk of bias: very high Indirectness: Population - all already had an indication for intervention as scheduled for TAVI Prognostic factor - splits LVEF into two separate thresholds compared with the same referent rather than using a single threshold. Also some uncertainty as to whether measured on cardiac MRI or echocardiography, though overall details suggest this is cardiac MRI measurements
Aortic stenosis – myocardial fibrosis on cardiac MRI
Agoston-Coldea 2019⁶ N=52 Romania Prospective cohort study	Severe AS undergoing AVR: 28.8% with NYHA class ≥III Mean age 66 years	Multivariable Cox regression model	Late gadolinium enhancement (LGE) on cardiac MRI	Age, 6 minute walking distance, E/E’ ratio, LVEF and LAS	Major adverse cardiac events (sudden cardiac death, non-fatal myocardial infarction, sustained ventricular arrhythmias, third-degree AV block and hospitalisation for heart failure) – median follow-up 386 days	Risk of bias: very high Indirectness: Population - indication for intervention already present: severe AS patients undergoing AVR Outcome - composite of multiple outcomes including some in the protocol as well as additional ones
Barone-Rochette 2014²² N=154 Belgium Prospective cohort study	Severe AS undergoing surgical AVR: 27% in NYHA class III/IV Mean age 74 years	Multivariate Cox proportional hazards model	LGE (myocardial fibrosis) on cardiac MRI	NYHA class III/IV and left bundle branch block	All-cause mortality following surgical AVR – median follow-up 2.9 years	Risk of bias: very high Indirectness: Population - all already scheduled to have AVR so population is not those where there is uncertainty about whether or not intervention is indicated
Christensen 2017⁵⁷ N=78 Denmark Prospective cohort study	Asymptomatic severe AS Mean age 74 years for whole cohort, including some not included in fibrosis analysis	Multivariate Cox proportional hazards analysis	Fibrosis on cardiac MRI	Age, gender and aortic mean gradient	Unplanned hospital admission (for atrial fibrillation, heart failure or acute coronary syndrome), aortic valve replacement or death – median follow-up 358 days	Risk of bias: very high Indirectness: Outcome - composite of three separate outcomes listed in the protocol
Dweck 2011⁸⁴ N=143 UK Prospective cohort study	Moderate or severe AS: symptomatic status unclear Mean age 67.2 years	Multivariate Cox proportional hazards regression	Midwall fibrosis LGE pattern on cardiac MRI Infarct fibrosis LGE pattern on cardiac MRI	LVEF, indexed LV end-diastolic volume and subsequent AVR – full list unclear but these variables are suggested based on those reported in the table	All-cause mortality (mixed medical/surgical treatment) – mean follow-up 2 years	Risk of bias: very high Indirectness: Population - unclear whether indication for intervention was uncertain in all patients, as includes some that underwent AVR which may have been scheduled prior to cardiac MRI Outcome - includes those with and without surgery during follow-up, whereas ideally aimed to look at results for operative and non-operative mortality separately
Everett 2020⁸⁸ N=440 UK, Germany, USA, Canada, South Korea Prospective cohort study	Severe AS scheduled for AVR: NYHA class III/IV in 36% Mean age 69.67 years	Multivariate Cox regression model	LGE on cardiac MRI	Extracellular volume percentage, age, gender, LV ejection fraction <50% and peak aortic jet velocity	All-cause mortality following AVR – median follow-up 3.8 years	Risk of bias: very high Indirectness: Population – all already scheduled for aortic valve intervention so no uncertainty about whether there is indication for intervention.
Herrmann 2018¹¹⁸ N=46 Germany Prospective cohort study	Symptomatic severe AS referred for AVR Mean age 68.3 years	Multivariate Cox proportional hazards regression	Mild fibrosis on cardiac MRI Severe fibrosis on cardiac MRI	Varied depending on model Model 1: age and sex Model 2: EuroSCORE	All-cause mortality – follow-up was 10 years 9 months in 57/58 enrolled patients (46 had data for fibrosis and unclear whether the one patient that was lost to follow-up was part of this analysis)	Risk of bias: very high Indirectness: Population - all were symptomatic severe AS undergoing AVR, so already have an indication for intervention prior to cardiac MRI Prognostic factor - specific severity of fibrosis on cardiac MRI compared with no fibrosis rather than comparing any fibrosis with no fibrosis
Hwang 2020¹²³ N=43 South Korea Prospective cohort study	Severe AS scheduled for AVR: mean NYHA class 2.1 Mean age 65.9 years	Multivariate Cox proportional hazard regression analysis	Diffuse myocardial fibrosis on cardiac MRI	Atrial fibrillation, anaemia and mild renal dysfunction	Cardiovascular death, hospitalisation for cardiac causes, non-fatal stroke and symptomatic aggravation (worsening NYHA class) following AVR– median follow-up 38.8 months	Risk of bias: very high Indirectness: Population - all already scheduled for AVR so no uncertainty as to whether there is an indication for intervention prior to cardiac MRI Outcome - composite of multiple outcomes in the protocol
Lee 2018¹⁵⁵ N=127 South Korea Prospective cohort study	Moderate or severe AS: proportion with severe AS was 62.2% and with any typical AS symptoms was 54.5% Mean age 68.8 years 69% had AVR during follow-up	Multivariate Cox regression analysis	LGE on cardiac MRI	EuroSCORE II, prior use of diuretics and being within highest native T1 value tertile	All-cause mortality and unexpected hospitalisation for heart failure during follow-up (mixed medical and surgical treatment)	Risk of bias: very high Indirectness: Population - includes a large proportion that were already deemed to have an indication for intervention regardless of cardiac MRI Outcome - composite outcome of multiple outcomes in protocol. Also includes those with and without operation in the analysis, whereas ideally aimed to analyse operative and non-operative outcomes separately.
Musa 2018¹⁸⁷ N=613 UK Prospective cohort study	Severe AS undergoing AVR: proportion with NYHA class ≥III was 40.1% Median age 74.6 years	Multivariate Cox proportional hazards model	LGE on cardiac MRI (LV myocardial scar)	Varied depending on the outcome All-cause mortality post-intervention: RV ejection fraction on cardiac MRI, LVEF on cardiac MRI, indexed atrial volume on cardiac MRI, atrial fibrillation, LV maximal wall thickness, STS score, LV stroke volume score on cardiac MRI, coronary artery disease, aortic valve area on echocardiography and age Cardiovascular mortality post-intervention: gender, previous coronary artery disease, LVEF on cardiac MRI, atrial fibrillation and age	All-cause mortality post-intervention Cardiovascular mortality post-intervention Median follow-up was 3.6 years	Risk of bias: very high Indirectness: Population - all already scheduled for AVR so does no uncertainty about whether intervention is indicated
Rajesh 2017²²⁵ N=109 India Prospective cohort study	Severe AS with/without symptom: 16.5% were in NYHA class III/IV Mean age 57.3 years 34.9% had AVR	Multivariate logistic regression analysis	LGE on cardiac MRI	Age >62 years, NYHA class III/IV, current smoker, modified Simpsons LVEF, LV mass on cardiac MRI, peak velocity and valvuloarterial impedance	Mortality, LVEF drop ≥20%, new-onset heart failure or hospitalisation for cardiovascular causes and new-onset arrythmia (mixed medical/surgical treatment – mean follow-up 13 months	Risk of bias: very high Indirectness: Population - 35% already deemed to have indications for intervention regardless of cardiac MRI results Outcome - composite of multiple factors listed in protocol, as well as some not listed in protocol. Also includes medically managed and surgically managed patients in the same analysis, whereas ideally aimed to analyse postoperative and non-operative outcomes separately.
Aortic stenosis – coronary artery disease on CT
Carstensen 2016⁴⁰ N=104 Denmark Prospective cohort study	Asymptomatic moderate-severe AS Mean age 72 years	Cox regression analysis	Significant stenosis (>50% luminal diameter) of 1, 2 or 3 vessels on CT OR Significant stenosis (>50% luminal diameter) of 1, 2 or 3 vessels or atheromatosis on CT	No multivariable analysis, unadjusted RR calculated from number of events reported in each group	Indication for AVR during follow-up – median follow-up 2.3 years	Risk of bias: very high Indirectness: None identified Note: Cohort overlaps with Larsen 2016¹⁵²
Larsen 2016¹⁵² N=116 Denmark Prospective cohort study	Asymptomatic mild-severe AS: mean aortic valve area on echocardiography was 1.01 cm² Mean age 72 years	Cox proportional hazards regression model	Coronary artery disease >70% stenosis on CT	Univariate results only	Indication for AVR during follow-up – median follow-up 27 months	Risk of bias: very high Indirectness: None identified Cohort overlaps with Carstensen 2016⁴⁰
Utsunomiya 2013²⁷⁵ N=64 Japan Prospective cohort study	Asymptomatic mild-severe AS: 45% being severe cases Mean age 74 years	Cox regression analysis	Multivessel obstructive coronary artery disease on CT	Age, gender, baseline systolic and diastolic blood pressure, peak transaortic velocity ≥4 m/s, aortic valve area on CCTA, LVEF on CCTA, LV mass index on CCTA and aortic valve calcium score	Cardiac events (cardiac death, AVR, non-fatal myocardial infarction and heart failure requiring urgent hospitalisation) – median follow-up 29 months	Risk of bias: very high Indirectness: Population - unclear whether there is uncertainty regarding indication for intervention in all patents, as includes mild-severe asymptomatic AS patients, with only 45% being asymptomatic severe Outcome - composite of multiple outcomes specified in the protocol.
Aortic stenosis – aortic valve area on CT
Clavel 2015⁶² N=269 France Prospective cohort study	AS patients undergoing CT and echocardiography in same episode of care: 45% with NYHA class III/IV and mean aortic valve area 0.94 cm2 Mean age 76 years	Multivariable Cox proportional hazards regression model	Aortic valve area ≤1.2 cm² on CT Aortic valve area ≤1.0 cm² on CT	Age-adjusted Charlson score index, sex, symptoms, mean gradient and LVEF	Mortality under medical management – mean follow-up 3.2 years	Risk of bias: high Indirectness: None identified
Aortic stenosis – aortic valve calcium score on CT
Akodad 2018⁸ N=118 France Prospective cohort study	Those undergoing TAVI for AS: >50% NYHA class ≥3 and mean gradient consistent with severe AS. Mean age 83.2 years	Multivariate logistic regression	Calcium score >6,000 HU on CT	Adjusted but list of variables included unclear	All-cause mortality, stroke, myocardial infarction, heart failure or rehospitalisation for cardiac causes - 1 month post-TAVI Rehospitalisation (unclear if all or only cardiac causes) - 1 month post-TAVI	Risk of bias: very high Indirectness: Population - all had TAVI so already an indication for intervention Prognostic factor - threshold of 6,000 HU used different to suggested thresholds in protocol and same one used for men and women Outcome - composite of multiple outcomes in protocol as well as some not listed in protocol
Aksoy 2014⁹ N=21 included in analysis that underwent AVR USA Retrospective cohort study	Low-flow low-gradient severe AS undergoing surgical aortic valve replacement (AVR) Mean age and further details for the subgroup undergoing surgical AVR unclear	Cox proportional hazards analysis	Calcium score >2027 on CT	No multivariable analysis within this subgroup, unadjusted estimate of HR calculated using data provided in the paper	Mortality postAVR – 30 days post-AVR	Risk of bias: very high Indirectness: Prognostic factor - same threshold used for men and women rather than a separate one as in protocol
Clavel 2014⁶³ N=794 USA, France, Canada Prospective cohort study	At least mild AS under conservative management: 27% with heart failure symptoms and mean gradient 35 mmHg Mean age 73 years	Multivariate Cox proportional hazards model	Severe aortic valve calcification (≥2065 in AU in men and ≥1274 AU in women) on CT	Age, sex, NYHA class ≥III, diabetes, history of coronary artery disease, indexed aortic valve area, mean gradient and LVEF	Mortality under conservative management – mean follow-up 1.7 years	Risk of bias: very high Indirectness: Population - unclear if there was uncertainty about whether to intervene as includes mild-severe AS under conservative management
Fischer-Rasokat, 2020⁹⁴ N=650 Germany Retrospective cohort study	Severe AS in as TAVI registry. Categorised as low-flow, low-gradient (LFLG), paradoxical LFLG, normal-flow, low-gradient	Multivariate Cox proportional hazards model	Aortic valve calcium score (low/high) on CT Threshold ≥1200 AU in women and ≥2000 AU in men.	BMI, GFR, dyslipidaemia, LV hypertrophy, mean pressure gradient, aortic valve area index, balloon expandable valve, rapid pacing, residual AR.	All-cause mortality at 1 year after TAVI	Risk of bias: very high Indirectness: Population - all had TAVI so already an indication for intervention
Larsen 2016¹⁵² N=115 Denmark Prospective cohort study	Asymptomatic mild-severe AS: mean aortic valve area on echocardiography was 1.01 cm² Mean age 72 years	Cox proportional hazards regression	Severe aortic valve calcium density (>300 AU/cm² for women and >475 AU/cm² for men) on CT	Only univariate results available	Indication for AVR during follow-up – median follow-up 27 months	Risk of bias: very high Indirectness: Prognostic factor – calcium density relative to area rather than calcium score of the valve.
Ludwig 2020¹⁶² N=526 Germany Retrospective cohort study	Severe low LVEF low-flow, low-gradient (LFLG) and paradoxical LFLG AS undergoing TAVI Median age 79.9 years in LFLG and 82.2 in pLFLG subgroups	Multivariate Cox proportional hazards model	Aortic valve calcium density on CT (based on total calcium in the annular plane and the LVOT: high, medium, low	Age, BMI, diabetes, COPD, atrial fibrillation, prior myocardial infarction (for pLFLG only), non-TF access.	Mortality up to 3 years after TAVI	Risk of bias: high Indirectness: Population - all had TAVI so already an indication for intervention Prognostic factor – calcium density relative to area rather than calcium score of the valve.
Pawade 2018²¹² N=215 UK, Canada, France, Spain, USA Multicentre registry with primarily prospective data	Various AS presentations, including mild-severe with symptom status varying between patients (only includes those where decision on whether to perform an intervention had not been made prior to CT in outcome analysis) Mean age 77 years	Cox proportional hazards regression	Severe aortic valve calcium (≥1274 AU for women and ≥2065 AU for men) on CT	Age, sex, Vmax ≥4 m/s and aortic valve area <1.0 cm²	Death or AVR during follow-up – median follow-up 1029 days	Risk of bias: very high Indirectness: Outcome - composite of two separate outcomes listed in the protocol. Also unclear whether AVR captures only unplanned intervention as in our protocol, or whether some were planned procedures following CT results.
Utsunomiya 2013²⁷⁵ N=64 whole cohort (n=29 in asymptomatic severe subgroup) Japan Prospective cohort study	Whole cohort: Asymptomatic mild-severe AS (45% being severe cases) Mean age 74 years Asymptomatic severe subgroup: Mean age and other details for this subgroup not reported	Cox regression analysis	Aortic valve calcium score ≥723 on CT – whole cohort Aortic valve calcium score ≥1266 – asymptomatic severe subgroup	No multivariable analysis, unadjusted estimates of HR calculated using KM curves and number at risk or other details reported in the paper	Cardiac events (cardiac death, AVR, non-fatal myocardial infarction and heart failure requiring urgent hospitalisation) Non-AVR cardiac events (cardiac death, non-fatal myocardial infarction and heart failure requiring urgent hospitalisation) Median follow-up for whole cohort was 29 months, but was not clear for the asymptomatic severe subgroup	Risk of bias: very high Indirectness: Whole cohort Population - unclear if there is uncertainty about whether to intervene, as includes mixture of mild-severe asymptomatic AS with only 45% severe Prognostic factor - threshold is quite different to that specified in the protocol and the same one has been used for men and women, rather than using a separate threshold Outcome - composite of multiple outcomes listed in the protocol Asymptomatic severe subgroup: Prognostic factor - threshold is the same one has been used for men and women, rather than using a separate threshold Outcome - composite of multiple outcomes listed in the protocol
Yoon 2020²⁹¹ N=1034 Denmark, France, Germany, Israel, Italy, the Netherlands, Switzerland, and USA Mixed prospective/ret rospective cohort study	Bicuspid aortic valve undergoing TAVI for symptomatic severe AS Mean age 74.7 (9.3)	Multivariate Cox proportional hazards model	Excess leaflet calcification on CT (more than the median value for the cohort, >382 mm³)	Age, STS score, peripheral vascular disease, prior AF, calcified raphe, aortopathy, non-TF access.	All-cause mortality after TAVI Median follow-up 360 days Cardiovascular mortality	Risk of bias: high for all-cause mortality, very high for cardiovascular mortality Indirectness: Population - all had TAVI so already an indication for intervention Prognostic factor – calcium density relative to area rather than calcium score of the valve.
Aortic regurgitation – regurgitant fraction and regurgitant volume on cardiac MRI
Kockova 2019¹⁴⁰ N=104 Czech Republic Prospective cohort study	Asymptomatic moderate-severe or severe aortic regurgitation Mean age 44 years	Multivariable Cox proportional hazards regression model	Aortic regurgitant fraction <34% on cardiac MRI Aortic regurgitant volume <45 ml on cardiac MRI	MRI-derived LV volumes or their indices	Aortic valve surgery during follow-up – median follow-up 587 days	Risk of bias: very high Indirectness: None identified
Myerson 2012¹⁹¹ N=113 UK Retrospective cohort study	Asymptomatic moderate or severe chronic aortic regurgitation Mean age 49 years	Multivariable Cox proportional hazards regression model	AR fraction ≤33% on cardiac MRI AR volume ≤42 ml on cardiac MRI	Appears to be adjusted for regurgitant volume and LV end-diastolic volume, though this is unclear	Development of an indication for surgery during follow-up – mean follow-up 2.6 years	Risk of bias: very high Indirectness: None identified
Mitral regurgitation – regurgitant volume on cardiac MRI
Myerson 2016¹⁹⁰ N=109 UK Prospective cohort study	Asymptomatic moderate or severe chronic organic mitral regurgitation Mean age 64.8 years	Cox proportional hazards regression model	Mitral regurgitant volume ≤55 ml on cardiac MRI	Univariate results only	Indication for surgery during follow-up – mean follow-up 2.5 years	Risk of bias: very high Indirectness: None identified
Penicka 2018²¹³ N=258 Belgium and Czech Republic Prospective cohort study	Asymptomatic, chronic moderate and severe organic MR attributable to flail or prolapse	Cox proportional hazards regression model	Mitral regurgitant volume per 10 mL on cardiac MRI	Age, sex, and LVESVI on MRI.	All-cause mortality Indication for mitral valve surgery – median follow-up 5.0 (IQR 3.5–6.0) years	Risk of bias: very high Indirectness: None identified
Tricuspid regurgitation – right ventricular function on cardiac MRI
Park 2016²¹¹ N=75 South Korea Prospective cohort study	Severe isolated functional tricuspid regurgitation (TR) undergoing TR surgery: 54.7% in NYHA class III/IV Mean age 59.3 years	Multivariate/un ivariate Cox proportional hazards model (depending on prognostic factor)	Right ventricular ejection fraction (RVEF) per 5% higher (continuous) RVEF <46% on cardiac MRI Right ventricular end systolic volume index (RV-ESVI) per 10 ml/m² increase (continuous) RV-ESVI ≥76 ml/m² All on cardiac MRI	Continuous variable analyses for RVEF and RV-ESVI are adjusted for age, sex, NYHA class, haemoglobin level and glomerular filtration rate Results for other prognostic factors are unadjusted	Cardiac death following TR surgery – median follow-up 57 months	Risk of bias: very high Indirectness: Population - all underwent intervention for severe functional TR so does not represent population where there is uncertainty about whether there is an indication for intervention Outcome - only includes cardiac deaths and not all deaths.

From: Evidence review for CT and MRI indications for intervention

Cover of Evidence review for CT and MRI indications for intervention

Evidence review for CT and MRI indications for intervention: Heart valve disease presenting in adults: investigation and management: Evidence review F.

NICE Guideline, No. 208.

London: National Institute for Health and Care Excellence (NICE); 2021 Nov.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.