Purpose or Case Report: Amniotic fluid volume (AFV) measured by maximum vertical pocket (MVP) or amniotic fluid index (AFI), is a key indicator of fetoplacental well-being and is assessed by ultrasound (US) after 20 weeks’ gestational age (GA). Fetal magnetic resonance imaging (MRI) is sometimes performed in isolation, and there is no standard assessment of AFV by MRI. Our purpose is to assess the robustness of US AFV indices and interrater variability by MRI in patients with known normal indices at same day US.
Methods & Materials: Following IRB approval, consecutive singleton fetuses 20 weeks’ GA or greater undergoing same day MRI and US from July 2018 to September 2019 were identified, and the electronic medical record was reviewed. Exclusion criteria were abnormal MVP (<2cm or >8cm) or non-diagnostic axial T2-weighted MRI of the maternal uterus. Following brief training, four pediatric radiologists with expertise in fetal imaging blinded to US independently reviewed MR images, measuring MVP and AFI. Interobserver variability for MVP and AFI was measured calculating the intraclass correlation coefficients (ICCs), using general linear mixed models with a random effect. Agreement between US and MR for each index was measured by ICC and Bland-Altman analysis.
Results: 200 fetuses divided equally by 2nd and 3rd trimester were included (49.5% male, mean GA 27.3 ± 4.7 weeks). Most MR exams were performed supine (85%, 170/200), with the remainder left lateral decubitus. Erroneous indices upon review of US images excluded some cases for comparison, leaving 198 and 167 pairs for MVP and AFI analysis, respectively. Overall interrater correlation of MVP and AFI was strong (0.81 and 0.89, respectively), with better correlation in the 2nd trimester (0.89 and 0.97 for MVP and AFI, respectively) than 3rd trimester (0.79 and 0.87). There was less strong ICC between US and MR MVP (0.46) and AFI (0.60). Modest Bland-Altman agreement between US and MRI for MVP (Bias:-1.61, Precision:1.44, LOA:-4.49, 1.27) was better than for AFI (Bias:-3.31, Precision:4.09, LOA:-11.49, 4.81). 38 patients (19%) with normal MVP and 33 patients (20%) with normal AFI by US had polyhydramnios by MRI, with no oligohydramnios cases.
Conclusions: Interrater ICC of MR MVP and MR AFI is high, though intermethod ICC between same day MR and US MVP and AFI is modest. This suggests that MRI MVP and AFI can be reliably measured, though MR normative ranges are different from US. If normal US metrics are applied to MR measures of AFV, polyhydramnios may be overdiagnosed.