Our work builds on the systematic review and meta-analysis by Raisi-Estabragh et al.1, which demonstrated major limitations of existing CMR reference ranges. The analysis revealed significant variation between key CMR normal reference range publications and illustrated the importance of multiple population-related factors and technical factors in driving these differences. Furthermore, limited sample sizes and technical heterogeneity precluded derivation of reference ranges from summary level data of existing reports. It was clear that wider representation of different populations and standardisation of image analysis were urgently needed to establish such reference distributions.
Following on from this work, we sought to establish a CMR data bank of healthy adults, which may be analysed de novo using uniform segmentation methods and analysis software. We hypothesised that in this way, major technical sources of variation could be eliminated and permit pooling of CMR metrics from different source studies and presentation of population stratified reference ranges. We invited authors of all papers included in our systematic review paper to contribute. The Healthy Hearts Consortium comprises contributions from researchers from these published works who agreed to share their data and who had appropriate ethics in place.
The characteristics of these cohorts and the source papers is summarised in Table 1.
Special note on UK Biobank
The UK Biobank reference range paper (2017) by Petersen et al.2 included in the original meta-analysis has been reconstructed with several improvements. First, at the time of the original study, the health record linkage in UK Biobank was not fully established and thus many of the disease exclusions in this study were based on self-report at UK Biobank assessments. Second, the source paper limited to participants of European ancestry, this is because at that time the number of participants of other ethnic background who had completed the UK Biobank Imaging protocol was too small. Third, healthy participant bias has been highlighted as an important consideration in the UK Biobank, with the cohort having a generally healthier exposure profile than the general population. Constructing reference ranges based on a “hyper-healthy” cohort could be potentially problematic when applied to the general population, as many individuals may be incorrectly classified as having “abnormal” metrics. To address these issues, we used the Health Survey for England to create a healthy subset of the UK Biobank with similar characteristics to the general population. We additionally, referred to national linked health records for disease exclusions. Overall, this resulted in a sample of over 7,000 CMR scans from the UK Biobank.
Table 1. Summary of characteristics of the Healthy Hearts Consortium cohorts
Cohort | Author, year of publication | n | Age | Sex | Ethnicity | Scanner Vendor | Scanner model | Magnet Strength |
UK Biobank | Petersen et al.2 2017 | 7,672 | Mean: 64.2 SD: 10.0 Range: 44 to 83 | Men: 3,751 (49%) Women: 3,921 (51%) | White = 6,189 (81%) Black = 340 (4%) South Asian = 510 (7%) Chinese = 155 (2%) Mixed/Other = 478 (6%) | Siemens | Magnetom Aera | 1.5 T |
SHIP | Bulow et al. 3 2018 | 627 | Mean: 45.1 SD: 11.9 Range: 21 to 81 | Men: 297 (47%) Women: 330 (53%) | White | Siemens | Magnetom Avanto | 1.5 T |
Utrecht | Eikendal et al.4 2016 | 128 | Mean: 28 SD: 3.1 Range: 22 to 34 | Men: 64 (50%) Women: 64 (50%) | White | Philips | Achieva | 3 T |
Heidelberg | Riffel et al.5 2019 | 190 | Mean: 49.3 SD: 12.4 Range: 21 to 74 | Men: 105 (55%) Women: 85 (45%) | White | Philips | Achieva | 1.5 T |
Pisa | Aquaro et al.6 2017 | 291 | Mean: 44.3 SD: 15.5 Range: 18 to 77 | Men = 144 (49%) Women: 147 (51%) | White = 280 Black = 1 | General Electric | Excite HDX 1 (< 1%) Signa Artist 99 (34%) Signa CVi 191 (66%) | 1.5 T = 290 (> 99%) 3 T = 1 (< 1%) |
Singapore | Le et al.7 2016 | 180 | Mean: 44.7 SD: 13.5 Range: 21 to 70 | Men: 91 (51%) Women: 89 (49%) | Chinese | Philips | Ingenia | 3 T |
Table 1 footnote. SHIP: Study of health in Pomerania; T: Tesla; SD: standard deviation. The sample size indicates number of cases included in the consortium after application of exclusion criteria.
References
- Raisi-Estabragh Z, Kenawy AAM, Aung N, et al. Variation in left ventricular cardiac magnetic resonance normal reference ranges: systematic review and meta-analysis. Eur Heart J Cardiovasc Imaging. 2021;22(5):494-504. doi: 10.1093/ehjci/jeaa089.
- Petersen SE, Aung N, Sanghvi MM, et al. Reference ranges for cardiac structure and function using cardiovascular magnetic resonance (CMR) in Caucasians from the UK Biobank population cohort. J Cardiovasc Magn Reson 2017;19:1. doi: 10.1186/s12968-017-0327-9
- Bülow R, Ittermann T, Dörr M, et al. Reference ranges of left ventricular structure and function assessed by contrast-enhanced cardiac MR and changes related to ageing and hypertension in a population-based study. Eur Radiol R. 2018;28:3996–4005. Doi: 10.1007/s00330-018-5345-y
- Eikendal ALM, Bots ML, Haaring C, et al. Reference values for cardiac and aortic magnetic resonance imaging in healthy, young caucasian adults. PLoS One 2016;11:1–17. doi: 10.1371/journal.pone.0164480
- Riffel JH, Schmucker K, Andre F, et al. Cardiovascular magnetic resonance of cardiac morphology and function: impact of different strategies of contour drawing and indexing. Clin Res Cardiol 2019;108:411–429. Doi: 10.1007/s00392-018-1371-7
- Aquaro GD, Camastra G, Monti L, et al. Reference values of cardiac volumes, dimensions, and new functional parameters by MR: A multicenter, multivendor study. J Magn Reson Imaging 2017;45:1055–1067. doi: 10.1002/jmri.25450
- Le TT, Tan RS, De Deyn M, et al. Cardiovascular magnetic resonance reference ranges for the heart and aorta in Chinese at 3T. J Cardiovasc Magn Reson 2016;18:21. doi: 10.1186/s12968-016-0236-3