Association of Fetal Left Ventricular Size with Congenital Diaphragmatic Hernia Outcomes

Background: Fetal lung measurements are widely utilized to predict outcomes in neonates with congenital diaphragmatic hernia (CDH), but may underrepresent the cardiovascular impairments that exacerbate CDH morbidity. Preliminary studies have demonstrated an association between CDH and abnormalities of fetal left-sided heart structures in the context of mediastinal compression. However, it remains unclear whether fetal left ventricle (LV) cavity size correlates with postnatal outcomes and whether specific cut-offs of LV size warrant concern. We further explored the association between fetal LV size and CDH with a novel use of internal controls, and queried whether adverse outcomes correlated with specific LV size thresholds.

Methods: We reviewed available fetal echocardiography images and postnatal electronic health records for all children with prenatally diagnosed CDH across two hospital systems from 2012–2020. Normal control fetuses who underwent fetal echocardiography from January–August 2020 were identified based on normal echocardiographic findings and absence of anatomic anomalies. LV end-diastolic length (LVED-L) and area were measured by three independent reviewers using the same methodology, with LV end-diastolic volume (LVED-V) calculated via the Bullet method (5/6 area x length). Gestational age (GA), biparietal diameter (BPD), and femur length were recorded alongside LV parameters. The linear association of CDH with echocardiographic findings controlling for GA and BPD was estimated with multivariable regression. Normalization against controls was achieved using z-scores derived from individual 95% confidence intervals along the regression models. The primary clinical outcome measure was a composite metric of ECMO or death. Optimal diagnostic cutoffs for LVED-L and LVED-V for select clinical outcomes were obtained using ROC curves.

Results: We identified prenatal echocardiograms in 21 neonates with CDH, in addition to 113 normal control fetal echocardiograms. On multivariable regression controlling for GA and BPD, fetuses with CDH demonstrated significantly reduced LVED-L (P <0.001, R2=0.76) and LVED-V (P <0.001, R2=0.86) compared to control fetuses. LVED-L was on average 0.30cm shorter and LVED-V was on average 1.2mL smaller in fetuses with CDH. Based on standardized regression coefficients, the presence of CDH had a stronger influence on LVED-V versus LVED-L, suggesting CDH may exert greater compression against the cardiac short axis (Figure 1). Among neonates with CDH, a fetal LVED-V z-score < -3.5 as derived from the GA-normalized control regression model was predictive of the composite outcome of ECMO or death with 92% sensitivity and 75% specificity (Figure 2; AUC=0.92, 95%CI=0.78-1.00, P=0.01).

Conclusion: CDH is associated with reduced fetal LVED-L and LVED-V, independent of GA and BPD. Fetal LVED-V measurements greater than 3.5 standard-deviations below GA-normalized controls may predict postnatal ECMO or death. Additional prospective analysis that incorporates fetal lung measurements is warranted to further validate the relative performance of fetal LV size as an accurate and independent predictor of CDH outcomes.