Facebook Twitter YouTube

Another Dimension to Survival: Predicting Mortality and Ventilator-Dependency with Fetal MRI Versus Prenatal Ultrasound In Patients with Congenital Diaphragmatic Hernia

Sunday, October 21, 2012: 7:46 AM
Versailles Ballroom (Hilton Riverside)
Arin L. Madenci1, Anna R. Sjogren1, Marjorie C. Treadwell2, Maria F. Ladino-Torres3, Robert A. Drongowski4, Jeannie Kreutzman4, Steven W. Bruch4 and George B. Mychaliska4, (1)University of Michigan Medical School, Ann Arbor, MI, (2)Division of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, University of Michigan, Ann Arbor, MI, (3)Section of Pediatric Radiology, Department of Radiology, University of Michigan, Ann Arbor, MI, (4)Section of Pediatric Surgery, Department of Surgery, University of Michigan, Ann Arbor, MI

Purpose: A major determinant of survival in patients with CDH is the degree of pulmonary hypoplasia. Prenatal methods of lung assessment include lung-to-head ratio (LHR) by ultrasound, observed-to-expected LHR (o:eLHR) by ultrasound, and percent-predicted lung volume (PPLV) by fetal MRI (fMRI). This study addresses the comparative effectiveness of LHR, o:eLHR, and PPLV in predicting mortality and ventilator-dependency.

Methods: We retrospectively reviewed 67 patients born with CDH between 2004-2008. Prenatal ultrasounds were assessed to determine LHR and o:eLHR. PPLV was obtained from imaging reports of patients who also underwent fMRI. Postnatal data included discharge mortality and ventilator-dependency at day-of-life (DOL) 30 for survivors. Univariate generalized linear mixed modeling (GLMM), a longitudinal analysis random-effects technique that accounts for patient-specific repeated measures, was utilized to assess variation of LHR and o:eLHR with estimated gestational age (EGA). Simple linear regression modeling was used to predict PPLV with EGA. Next, repeated measures of LHR and o:eLHR for a given patient were summarized using the minimum value and simple logistic regression was used to model minimum-LHR (min-LHR), minimum-o:eLHR (min-o:eLHR), and PPLV as predictors of mortality and ventilator-dependency. Log-transformation was employed to stabilize the variance of the logistic regression models. Models were compared using Akaike's information criterion (AIC), with lower AIC indicating better fit. To further quantify the effect on mortality and ventilator-dependency of LHR and o:eLHR without requiring a summary statistic (i.e. minimum value), simple GLMM was again utilized. All analysis was conducted using SAS, with alpha-level=0.05.

Results: Thirty-seven patients underwent 82 prenatal ultrasounds, while 26 of this sub-cohort underwent an fMRI study. Survival-to-discharge was 78.4% (29/37) and 76.9% (20/26) in the ultrasound and fMRI cohorts, respectively. Median EGA at ultrasound and fMRI were 30.2 (range=18.1-38.5) and 23.0 weeks (range=20.0-38.0), respectively. Using GLMM to account for repeated measures, EGA predicted LHR (p=0.02), but not o:eLHR (p=0.12). Using linear regression, EGA did not predict PPLV (p=0.72). Univariate logistic regression revealed min-LHR, min-o:eLHR, and PPLV as significant predictors of mortality and ventilator-dependency. The PPLV models had the best fits, based on AIC.[Figure 1] Receiver-operating-characteristic (ROC) analysis illustrated highest area-under-the-curve for the PPLV mortality-prediction model.[Figure 2] Univariate GLMM for repeated ultrasound measures showed o:eLHR (p=0.04) but not LHR (p=0.06) predicted mortality. Both LHR (p=0.02) and o:eLHR (p=0.02) predicted ventilator-dependency at DOL30.

Conclusion: PPLV and o:eLHR were independent of EGA, in contrast to LHR. Min-LHR, min-o:eLHR, and PPLV each independently predicted mortality and ventilator-dependency. PPLV was a slightly more predictive and discriminative measure for these outcomes. Given that PPLV and o:eLHR were independent of EGA and predictive of both mortality and ventilator dependency, these measures were found to be superior to LHR.  When assessing fetuses with CDH, o:eLHR using ultrasound or PPLV utilizing fetal MRI may be most helpful for counseling regarding postnatal outcomes.