Introduction

Fetal growth restriction (FGR) is a condition in which a fetus has not reached its growth potential¹. Management of FGR includes measurement of fetal and maternal uterine artery circulatory resistance from Doppler ultrasound². However, it is not possible to directly assess fetal oxygenation. Magnetic resonance imaging (MRI) is increasingly providing information on placental function in-vivo to support clinical decision-making. Although no animal model truly recapitulates human pregnancy, pregnant sheep have been used to study maternal-fetal interactions and validate MRI measurements^3,4. Non-invasive fetal oxygen saturation (FO₂) measurement would be clinically valuable in pregnancies with FGR.

The purpose of this study was to: 1) use a placentome-specific multi-compartment MRI model to estimate FO₂ using MRI; and 2) validate MRI FO₂ against invasively collected blood gas samples.

Methods

The study was approved by the Animal Ethics Committee of the South Australian Health and Medical Research Institute.

Carunclectomy surgery
Non-pregnant ewes (n=10) were assigned to have most of their endometrial caruncles removed via carunclectomy⁵ (CX) under general anesthesia (induction, diazepam (0.3mg/kg) & ketamine (7mg/kg); maintenance, 2.5% isoflurane).

MRI
At 109-111 (Control, n=10; CX, n=10) and 139-141 (Control, n=4; CX, n=5) days gestation, ewes were anesthetised as per surgery and underwent MRI scans on a 3T Siemens Skyra Scanner (Erlangen,Germany). DW-MRI was performed at 7 b‐values (b = 0, 10, 20, 30, 50, 70, 100, 200, 300, 500, 600 s.mm^-2) and T2-relaxometry at 10 echo times (TE) = (81,90,96,120,150,180,210,240,270,300 ms). Data was also acquired at b‐value 50 s.mm^-2 and 200 s.mm^-2 for TE = (81, 90, 120, 150, 180, 210, 240 ms).

Signal-Model
The sheep-specific signal model is of the form:

$$S({\bf b},{\bf{T_E}})= S_0 \left[ e^{{-\bf b}d^*} \Big( f e^{-{\bf {T_E}}R_2^{f_b}}+ v e^{-{\bf {T_E}}R_2^{m_b}}\Big) + (1-f-v)e^{{-\bf b}d-{\bf {T_E}} R_2^{ts}} \right], \quad (1)$$

where $$$S$$$ is the measured MR signal and $$$S_0$$$ is the signal with $$$b=0$$$. The five model parameters are the feto-placenta blood volume fraction $$$f$$$, trophoblast diffusivity $$$d$$$, pseudo-diffusivity $$$d^*$$$, feto-placental blood relaxation $$$R_2^{f_b}=1/T_2^{f_b}$$$ and maternal blood volume fraction $$$v$$$. We used literature-based values for maternal blood relaxation $$$R_2^{m_b}=(150ms)^{-1}$$$ and tissue relaxation $$$R_2^{ts}=(46ms)^{-1}$$$.

Feto-placental Oxygenation
The dependence of FO₂ on the effective $$$T_2^{f_b}$$$ is characterised by the Luz-Meiboom model⁶:
$$\frac{1}{T_2^{f_b}}= \frac{1}{T_{2,0}} + K_0 \Big( 1- \frac{FO_2}{100\%} \Big) + K_1 \Big( 1-\frac{FO_2}{100\%} \Big)^2 \quad (2),$$
where $$$T_2^{f_b}$$$ is the measured $$$T_2^{f_b}$$$ value of partially oxygenated blood calculated in Eq.(1). $$$T_{2,0}$$$, $$$K_0$$$ and $$$K_1$$$ are held fixed at literature⁷ values of 148.4ms, 1.4 s^-1 and 104.4 s^-1, respectively.

Image-Analysis
The placentomes where manually segmented from the baseline non-diffusion weighted image. To reduce motion, we applied a non-rigid registration⁸. Two-way ANOVA was performed to examine the effect of group and gestational age on MRI parameters. Linear regression was performed to compare the relationship between FO₂ against reference fetal descending aorta oxygenation measured by catheterisation.

Results

The linear relationship between fetal arterial blood gas and MRI placental based solutions to Luz-Meiboom model was significant for Controls (y=1.64x-91.3, R² = 0.88, P = 0.0002), for CXs (y = 1.17x - 54.3, R² = 0.59, P = 0.017) and when all animals were considered together (y = 1.37x - 69.4, R² = 0.73, P < 0.0001). A strong correlation was observed between fetal DAo measurements and fetal MRI oxygen saturation measurements in Controls (r=0.94, P < 0.0001) and when all animals considered together (r=0.85, P <0.0001); and a medium correlation in CXs (r=0.77, P = 0.0003). Bland Altman analysis showed a mean difference of 28.7% where the limits of agreement ranged from 21.7% to 35.7% for Controls (Figure 3). The mean difference for the CX data was 40.8% and the limits of agreement were wider from 24.1% to 57.5%.

We observed a significant decrease in $$$f$$$ between Controls and CX in late gestation (Table 1). $$$T_2^{f_b}$$$ and FO₂ were lower in CXs at both gestational ages. $$$d$$$ was lower in CXs, but only in early gestation. $$$d^*$$$ increased with gestation in Controls.

Discussion

We propose a model of separating fetal and maternal circulations to measure oxygen saturation in sheep placentomes. The proposed model measures the oxygenation of fetal blood in the placenta and we show the value is physiologically higher than the sample from the fetal DAo, which is mixed with blood circulating from the fetal body. These differences in feto-placental and DAo saturation result in the consistent bias seen here. We assumed that maternal and fetal blood are intra-capillary but that the maternal blood stays nearly 100% saturated with oxygen. This assumption is not valid, especially in CXs where there is an increased oxygen gradient in the maternal placentome capillaries as a result of the appositional villous structure and is likely also related to the increased bias seen at the lowest oxygen saturations of CXs (Figure 3). Our MRI FO₂ results are consistent with what is known about oxygen distribution in fetus⁹ and show that placental MRI represents a valid means to assess fetal oxygenation status.

Conclusion

This study supports the use of multi-compartment placental MRI in pregnancy using sheep as a model of human pregnancy. Results demonstrated a link between MRI FO₂ and gold-standard fetal DAo oxygen saturation, supporting the possibility of using MRI for diagnosis of FGR associated with fetal hypoxia.

Acknowledgements

This research was supported by the Wellcome Trust (210182/Z/18/Z, 101957/Z/13/Z, 203148/Z/16/Z), the EPSRC (NS/A000027/1) and an ARC Future Fellowship (Level 3; FT170100431) to JLM.