Fetal Cardiac Growth: New Z-Score Ranges From 3,000 Normal Pregnancies

Saturday, October 17, 2009: 9:45 AM
201 (Washington Convention Center)
Venugopal Amula, MD1, Thomas Riggs, MD, PhD1, Wesley Lee, MD2, Nancy Cutler, MD1, Richard Bronsteen, MD2 and Christine Comstock, MD2, (1)Pediatric Cardiology, William Beaumont Children's Hospital, Royal Oak, MI, (2)Department of Obstetrics and Gynecology, Division of Fetal Imaging, William Beaumont Hospital

Purpose: Z-scores provide a method for interpreting fetal cardiac biometry by expressing how many standard deviations a specific measurement lies above or below the expected mean. For a normal distribution (“bell-shaped curve”), - 2 z to + 2 z range comprises 95 % of the data. A prior study of 130 fetuses has established preliminary standards using fetal echocardiography (Schneider, 2005).  Their study suggested that non-linear regression models, using natural logarithms of cardiac dimensions and fetal biophysical measurements were superior to simple linear models.  Our investigation evaluated linear and non-linear models and described new Z-score standards using a large population from a single institution.

Methods: Five fetal cardiac dimensions: right ventricle (RV), left ventricle (LV), aorta (AO), pulmonary artery (PA) and cardiac circumference (Circ) were measured in mm from 3,000 pregnant women with normal singleton fetuses at gestational age 16-41 weeks. Regression equations were derived predicting cardiac dimensions from gestational age (GA), femoral diaphysis length (FDL) or bi-parietal diameter (BPD). Testing for significant differences between correlation coefficients from different models used Fisher’s Z-test. Residuals from the regression equations were evaluated using a Kolmogorov-Smirnov Goodness-of-fit test to determine whether they deviated from a normal distribution.

Results: The best regression equations were based on FDL, although BPD and GA were only slightly worse .Non-linear models were not significantly better than linear models and femoral diaphysis length was a better predictor than estimated gestational age.

AO = -0.597 + 0.119*FDL, r = 0.945

PA = -0.583 + 0.130*FDL, r = 0.931

RV = -0.677 + 0.227*FDL, r = 0.953

LV = -0.359 + 0.216*FDL, r = 0.934

Circ = -0.0147 + 2.049*FDL, r = 0.964 

Conclusion: From our larger sample size, we have greater confidence and precision in identifying normal ranges and our simple linear model allows easier calculation and visual representation of Z-scores.  Easy and accurate computation of Z-Scores will help to identify problems in fetal cardiac development with ongoing implications for management.  Future use of these nomograms may aid in detection, serial evaluation and management of critical morphologic heart lesions.