Rapid Fire Abstracts
Andres F. Moreno Rojas, MD
Post Doctoral Clinical Fellow
Stanford University
Andres F. Moreno Rojas, MD
Post Doctoral Clinical Fellow
Stanford University
Alexandria Jensen, PhD
Senior Biostatistician, Quantitative Sciences Unit
Stanford University
Aya Akino, MD
3D lab
Stanford University
Frandics Chan, MD, PhD
Professor
Stanford University
Ali Syed
Clinical Assistant Professor
Stanford University
Shreyas S. Vasanawala, MD, PhD
Professor
Stanford University
Shiraz A. Maskatia, MD
Clinical Professor
Stanford University School of Medicine
Anna Seehofnerova, MD
Clinical Assistant Professor
Stanford University
Sujatha Buddhe, MD
Professor
Stanford University
Different methods have resulted in varying normative models for pediatric cardiac MRI (CMR) measurements. However, most of these methods have utilized datasets that encompass patients across the pediatric spectrum, assuming a singular model can provide accurate reference values for all pediatric patients. Using CMR measurements from healthy children, we compared our proposed model with previously published z-score models in children under 2 years of age.
Methods: Children under 2 years of age with normal intra-cardiac anatomy who underwent cardiac MRI with 4D flow imaging for non-shunt indications were included in the study. Body surface area (BSA) was calculated using the Haycock method. Z-scores were derived for cardiovascular measurements of interest (LV end-diastolic volume, RV end-diastolic volume, LV end-diastolic mass) using three models (Buddhe, Buechel, Olivieri). Model comparisons were performed through evaluation of (a) overlaid graphs of measurement versus BSA with Z=0 curves for all models and Z=+/-2 for the Buddhe model and (b) pairwise scatterplots of the Z-score models with corresponding Spearman correlation coefficients.
Results: Table 1 provides characteristics of the three models and underlying datasets used for comparison in this study. Our model (Buddhe) is derived from a total of 46 patients (46% female; median age 49.5 days). Unlike the Buechel and Olivieri studies, which included patients across the pediatric age spectrum, the Buddhe study focused on pediatric patients in the first 2 years of life. Overall, the Buddhe and Buechel z-score curves were similar for all three CMR measurements of interest across all BSAs (Figure 1). The Olivieri z-score curves were similar for LV diastolic mass but diverged for LV and RV end-systolic volumes, especially at higher BSAs. Corresponding correlation coefficients were highest when comparing the Buddhe and Buechel models (mean r=0.94) and similarly lower when comparing either the Buddhe or Buechel models to the Olivieri models; correlation was generally found to be higher for female patients as compared to male patients across all CMR measurements. The accompanying scatterplots suggest systematic differences, with the Olivieri model showing smaller spread of z-scores across all BSAs (Figure 2).
Conclusion:
Normative models built using datasets with pediatric patients of all ages showed marked differences from the proposed model built using only patients 2 years of age or under, suggesting a separate model should be used to provide accurate reference values for young patients.
Figure 1: Comparison of z-score curves for the Buddhe model (Z=0 solid grey line; Z=-2, +2 dashed grey lines) versus similar curves for the Buechel model (Z=0 light pink [female patients] and light blue [male patients]) and Olivieri model (Z=0 dark pink [female patients] and dark blue [male patients]). BSA = body surface area; LV = left ventricular.
Fig1.pdf
Figure 2: Scatterplots of pairwise z-score comparisons for the Buddhe, Buechel, and Olivieri models, stratified by female (dark pink) and male (dark blue) patients. The black line denotes the y=x line. Spearman correlations were calculated for male and female patients, respectively.
Fig2.pdf
