0006
Fetal MRI-based body and adiposity quantification for small for gestational age perinatal risk stratification1Sagol Brain Institute, Tel-Aviv Sourasky Medical Center, Tel-Aviv, Israel, 2Department of Radiology, Tel-Aviv Sourasky Medical Center, Tel-Aviv, Israel, 3Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel, 4School of Computer Science and Engineering, The Hebrew University of Jerusalem, Jerusalem, Israel, 5Sagol School of Neuroscience, Tel-Aviv University, Tel-Aviv, Israel, 6Department of Obstetrics and Gynecology, Lis Hospital for Women, Tel-Aviv Sourasky Medical Center, Tel-Aviv, Israel, 7School of Neurobiology, Biochemistry and Biophysics, The George S. Wise Faculty of Life Sciences, Tel-Aviv University, Tel-Aviv, Israel, 8Neonatal Intensive Care Unit, Dana Dwek Children's Hospital, Tel-Aviv Sourasky Medical Center, Tel-Aviv, Israel, 9Edmond and Lily Safra Center for Brain Sciences, The Hebrew University of Jerusalem, Jerusalem, Israel
Synopsis
Keywords: YIA, Fetus
Motivation: Small for gestational age (SGA) fetuses are undernourished and at higher risk for adverse outcomes; however, conventional assessment methods exhibit limited sensitivity.
Goal(s): To stratify perinatal risk using MRI-based body composition metrics.
Approach: TruFISP and 2-points Dixon images were used to compute the total fetal volume (TFV), fat-to-body volume ratio (FBVR) and adipose tissue fat signal fraction (FSF) using deep-learning segmentation.
Results: SGA fetuses (N=40) with lower FBVR were more likely to require obstetric interventions because of non-reassuring status, while those with reduced TFV were prone to adverse neonatal outcomes. The model’s sensitivity/specificity rates are 85.7%/87.5% and 82.35%/86.4%, respectively.
Impact: Quantifying fetal body composition through MRI can offer additional insights into the severity of small for gestational age complicated pregnancies and may help in stratifying perinatal risk.
Introduction
SGA fetuses face worrisome perinatal adverse outcomes.1,2 However, distinguishing between fetuses with pathological smallness (FGR) and those with constitutional smallness is challenging. Ultrasound is typically used for prognostication using biometric measurements and Doppler velocimetry.3,4 Nevertheless, the conventional assessment exhibits limited concordance with perinatal outcomes.5,6 While various etiologies exists for FGR, placental insufficiency is the most common, resulting in inadequate fetal nutrition.7Throughout the third trimester, fetuses gain fat rapidly, which is crucial for postnatal life adaptation. Fetal undernutrition may manifest as reduced adiposity, which is potentially indicative of severe pathology and elevated risk for postnatal morbidity.8–10 In this context, the fetal body composition emerges as a promising prognostic marker. MRI fat-water separation properties have been recently used for the evaluation of fetuses.9,11–13 We hypothesized that MRI quantification of fetal body composition and specifically lower adiposity would correlate with malnutrition and adverse outcomes. Moreover, we evaluated the diagnostic utility of MRI-based body composition metrics compared with conventional sonographic criteria.
Methods
In this prospective IRB-approved study, participants with SGA-complicated pregnancies were enrolled following consent. Exclusions included genetic or structural abnormalities, poor image quality, or pregnancy termination. Monitoring continued until infant discharge, recording obstetric and neonatal outcomes. The primary outcomes were obstetric interventions for non-reassuring fetal status (NRFS) and composite adverse neonatal outcomes (CANO), including respiratory support need, intraventricular hemorrhage, necrotizing enterocolitis, or extended NICU stay post 35 gestation weeks. FGR sonographic criteria included an estimated fetal weight or abdominal circumference below the 3rd centile or 3rd to 10th centiles with abnormal Doppler results.14MRI was conducted using 3T MRI scanners (Prisma, Vida, or Skyra; Siemens Healthineers). Following dedicated neuroimaging, True Fast Imaging with Steady State Free Precession (TruFISP) (TR/TE=445.7/0.64ms, flip angle=43°, FOV=400x400mm, resolution=0.78x0.78x2mm, acquisition time=57 seconds, without breath holding) and T1-weighted two-point Dixon (TR/TE1/TE2=4-4.24/1.34/2.57ms, flip angle=9°, FOV=400x400mm, resolution=1.25-1.4x1.25-1.4x2mm, acquisition time=18-22 seconds, with a single breath hold) were acquired. The entire fetal anatomy was captured in both sequences. Fat and water-only images were automatically generated using the scanner's built-in software. The entire fetal body and subcutaneous adipose tissue were segmented using in-house developed neural networks15,16 and were manually refined as needed. TFV, FBVR, and FSF were computed as outlined in Fig. 1.
TFV, FSF, and FBVR were adjusted by dividing each parameter by the gestational age. Univariate and multivariate logistic regressions with backward elimination were used to study the association between TFV, FBVR, FSF and outcomes. The association between the sonographic criteria for FGR and perinatal outcomes was evaluated using Fisher's exact test. Sensitivity, specificity, positive and negative predictive values, and accuracy were calculated. Furthermore, the Mann-Whitney test, followed by Bonferroni multiple comparisons correction, was used to assess differences between favorable and unfavorable outcomes. A p-value <0.05 was determined as statistically significant.
Results
A total of 45 participants were recruited; five were excluded due to abnormalities on MRI (N=2), poor image quality (N=2), and pregnancy termination (N=1). The final cohort included 40 participants (26 [61.9%] female and 14 [35%] male fetuses), 10/40 (25%) had obstetric interventions due to NRFS, and 17/40 (42.5%) experienced CANO.FBVR (odds ratio [OR] 0.39, 95% confidence interval [CI] 0.2-0.76) and FSF (OR 0.95, CI 0.91-0.99) were significantly linked with NRFS interventions. TFV was not significantly associated with NRFS (P=0.057). Conversely, TFV (OR 0.69, CI 0.56-0.86) and FSF (OR 0.96, CI 0.93-0.99) were significantly associated with CANO, whereas FBVR was not (P=0.08). Fig. 2 shows individual body composition data overlaid on 142 appropriately developing fetuses for reference.
The predictive models derived from this study showed that FBVR alone may predict the need for NRFS interventions with a sensitivity/specificity of 85.7%/87.5%. Similarly, TFV alone predicted CANO with a sensitivity/specificity of 82.35%/86.4%. In contrast, the sensitivity/specificity of sonographic criteria for predicting obstetric interventions were 100%/33.3% and did not significantly predict CANO (P=0.145).
In the study of interventions for NRFS, FBVR showed a statistically significant difference between the two groups. However, TFV and FSF did not differ significantly, (P≥0.099). TFV was significantly different between fetuses with and without CANO, whereas FBVR and FSF did not, (P≥0.132). These findings are detailed in Fig. 3, and Fig. 4 depicts six illustrative cases of fetuses with and without perinatal adverse events.
Discussion
Reduced FBVR and TFV are linked to an increased likelihood of obstetric interventions for NRFS and CANO. MRI shows promise in enhancing the prognostication of SGA-complicated pregnancies. The MRI body composition measurements utilized in this study are rapid and rely on widely accessible acquisition methods, facilitating their adoption in other fetal MRI centers.Acknowledgements
No acknowledgement found.References
1. Nohuz, E., Rivière, O., Coste, K. & Vendittelli, F. Prenatal identification of small-for-gestational age and risk of neonatal morbidity and stillbirth. Ultrasound Obstet. Gynecol. Off. J. Int. Soc. Ultrasound Obstet. Gynecol. 55, 621–628 (2020).
2. Madden, J. V, Flatley, C. J. & Kumar, S. Term small-for-gestational-age infants from low-risk women are at significantly greater risk of adverse neonatal outcomes. Am. J. Obstet. Gynecol. 218, 525.e1-525.e9 (2018).
3. Lees, C. C. et al. ISUOG Practice Guidelines: diagnosis and management of small-for-gestational-age fetus and fetal growth restriction. Ultrasound Obstet. Gynecol. Off. J. Int. Soc. Ultrasound Obstet. Gynecol. 56, 298–312 (2020).
4. Melamed, N. et al. FIGO (international Federation of Gynecology and obstetrics) initiative on fetal growth: best practice advice for screening, diagnosis, and management of fetal growth restriction. Int. J. Gynaecol. Obstet. Off. organ Int. Fed. Gynaecol. Obstet. 152 Suppl, 3–57 (2021).
5. Meler, E. et al. Severe smallness as predictor of adverse neonatal outcome in suspected late SGA fetuses: systematic review and meta-analysis. Ultrasound Obstet. Gynecol. Off. J. Int. Soc. Ultrasound Obstet. Gynecol. (2022) doi:10.1002/uog.24977.
6. Schreiber, V. et al. Definitions matter: detection rates and perinatal outcome for infants classified prenatally as having late fetal growth restriction using SMFM biometric vs ISUOG/Delphi consensus criteria. Ultrasound Obstet. Gynecol. Off. J. Int. Soc. Ultrasound Obstet. Gynecol. 61, 377–385 (2023).
7. Salafia, C. M. et al. Intrauterine growth restriction in infants of less than thirty-two weeks’ gestation: associated placental pathologic features. Am. J. Obstet. Gynecol. 173, 1049–1057 (1995).
8. Larciprete, G. et al. Intrauterine growth restriction and fetal body composition. Ultrasound Obstet. Gynecol. Off. J. Int. Soc. Ultrasound Obstet. Gynecol. 26, 258–262 (2005).
9. Rabinowich, A. et al. Reduced adipose tissue in growth-restricted fetuses using quantitative analysis of magnetic resonance images. Eur. Radiol. (2023) doi:10.1007/s00330-023-09855-y.
10. Padoan, A. et al. Differences in fat and lean mass proportions in normal and growth-restricted fetuses. Am. J. Obstet. Gynecol. 191, 1459–1464 (2004).
11. Giza, S. A., Miller, M. R., Parthasarathy, P., de Vrijer, B. & McKenzie, C. A. Comparison of modified two-point dixon and chemical shift encoded MRI water-fat separation methods for fetal fat quantification. J. Magn. Reson. Imaging 48, 274–282 (2018).
12. Giza, S. A. et al. Measuring fetal adipose tissue using 3D water-fat magnetic resonance imaging: a feasibility study. J. Matern. neonatal Med. 33, 831–837 (2020).
13. Giza, S. A. et al. Water-fat magnetic resonance imaging of adipose tissue compartments in the normal third trimester fetus. Pediatr. Radiol. 51, 1214–1222 (2021).
14. Gordijn, S. J. et al. Consensus definition of fetal growth restriction: a Delphi procedure. Ultrasound Obstet. Gynecol. Off. J. Int. Soc. Ultrasound Obstet. Gynecol. 48, 333–339 (2016).
15. Specktor-Fadida, B. et al. Deep learning-based segmentation of whole-body fetal MRI and fetal weight estimation: assessing performance, repeatability, and reproducibility. Eur. Radiol. (2023) doi:10.1007/s00330-023-10038-y.
16. Avisdris, N. et al. Automatic fetal fat quantification from MRI. in International Workshop on Preterm, Perinatal and Paediatric Image Analysis 25–37 (Springer, Cham, 2022)
Figures
