Poster Session 2 - C9
1Jiawei Xu, 2An Qi Duan, 3,4Johannes Keunen, 5,6Sharon Portnoy, 5,6John Sled, 3,4John Kingdom, 5,6Christopher K. Macgowan, 1,7.Mike Seed
1 Dept. of Physiology, University of Toronto; 2 Institute of Medical Science, University of Toronto; 3 Dept.of Obstetrics and Gynecology, University of Toronto; 4 Division of Maternal-Fetal Medicine, Sinai Health System; 5 Dept.of Medical Biophysics, University of Toronto; 6 Translational Medicine, Hospital for Sick Children; 7 Division of Cardiology, Hospital for Sick Children, Toronto, ON, Canada
Background: Fetal anemia is a serious pregnancy complication characterized by a low hematocrit value. The modern treatment for severe fetal anemia is intrauterine transfusion (IUT). The current standard-of-care technique for monitoring anemic fetuses - Doppler ultrasonography of MCA-PSV - becomes less predictive for the timing of subsequent IUTs, potentially exposing patients to unnecessary and risky fetal sampling. Previous studies have shown that fetal hematocrit can be accurately assessed by measuring the MRI relaxation properties of fetal blood in vivo. This study aims to study the feasibility and accuracy of using fetal MRI for measuring fetal hematocrit in fetal anemia. We hypothesized that 1. MRI-predicted hematocrit is accurate within 5% in anemic fetuses before and after IUT and 2. MRI measurement is more accurate than MCA-PSV for detecting fetal anemia.
Methods: Patients with suspected fetal anemia in their second or third trimester requiring intrauterine transfusion (IUT) at Mount Sinai Hospital were scanned using a 1.5T MRI system (Siemens, Erlangen) before or after IUT. T1 and T2 measurements of umbilical vein blood were obtained using Modified Look-Locker inversion recovery and T2 preparation sequences, respectively. Hematocrit was calculated from T1 and T2 using a mathematical model developed by our group. The calculated hematocrit was compared against laboratory hematocrit obtained via routine fetal blood sampling from the umbilical cord performed before and after IUT. Data were analyzed using generalized estimating equation (GEE) and Bland-Altman analyses.
Results: A total of 17 fetuses (19-36 weeks’ gestation) were scanned before or after their 1st to 4th IUT: 15 with suspected fetal anemia and 2 with suspected polycythemia. 3 of the 15 fetuses with suspected fetal anemia were scanned multiple times: 2 were scanned twice, and 1 was scanned 3 times for a total of 21 MRI scans. Based on the laboratory hematocrit, 14 out of 17 cases of suspected fetal anemia had moderate or severe anemia while the remaining 3 were normal or had mild anemia. The T1 value range of all measurements ranged from 1213 – 2180 ms. GEE analysis showed a correlation of -0.9 (95%CI: -0.96 - -0.77) between T1 and the laboratory hematocrit. In addition, quantitative analysis showed that the mean difference between the MRI-predicted and laboratory hematocrit was 0.06 (SD = 0.05). Bland-Altman analysis confirmed an excellent agreement between MRI and laboratory hematocrit with a bias of -0.018. MRI-predicted hematocrit measured prior to IUT correctly identified all anemic fetuses. MCA Doppler correctly identified 9 out of 16 cases of suspected fetal anemia using MCA-PSV >1.50 MoM as the cut-off for moderate/severe anemia (Doppler was not available in one case). The positive predictive value (PPV) was 90% (95%CI: 56-100%) while negative predictive value 33% (95%CI: 4-78%) for MCA-PSV Doppler.
Conclusion: Based on the current findings, MRI measurement of fetal hematocrit is feasible and accurate in the anemic fetuses over a wide range of hematocrit. A larger sample size is needed for further confirmation.