Dissecting the contributions of the peripheral chemoreflex and myocardial hypoxia to fetal heart rate decelerations in near‐term fetal sheep

Brief repeated fetal hypoxaemia during labour can trigger intrapartum decelerations of the fetal heart rate (FHR) via the peripheral chemoreflex or the direct effects of myocardial hypoxia, but the relative contribution of these two mechanisms and how this balance changes with evolving fetal compromise remain unknown. In the present study, chronically instrumented near‐term fetal sheep received surgical vagotomy (n = 8) or sham vagotomy (control, n = 11) to disable the peripheral chemoreflex and unmask myocardial hypoxia. One‐minute complete umbilical cord occlusions (UCOs) were performed every 2.5 min for 4 h or until arterial pressure fell below 20 mmHg. Hypotension and severe acidaemia developed progressively after 65.7 ± 7.2 UCOs in control fetuses and 49.5 ± 7.8 UCOs after vagotomy. Vagotomy was associated with faster development of metabolic acidaemia and faster impairment of arterial pressure during UCOs without impairing centralization of blood flow or neurophysiological adaptation to UCOs. During the first half of the UCO series, before severe hypotension developed, vagotomy was associated with a marked increase in FHR during UCOs. After the onset of evolving severe hypotension, FHR fell faster in control fetuses during the first 20 s of UCOs, but FHR during the final 40 s of UCOs became progressively more similar between groups, with no difference in the nadir of decelerations. In conclusion, FHR decelerations were initiated and sustained by the peripheral chemoreflex at a time when fetuses were able to maintain arterial pressure. After the onset of evolving hypotension and acidaemia, the peripheral chemoreflex continued to initiate decelerations, but myocardial hypoxia became progressively more important in sustaining and deepening decelerations.

r Reflex control of fetal heart rate was disabled by vagotomy to unmask the effects of myocardial hypoxia in chronically instrumented fetal sheep. Fetuses were then subjected to repeated brief hypoxaemia consistent with the rates of uterine contractions during labour.
r We show that the peripheral chemoreflex controls brief decelerations in their entirety at a time when fetuses were able to maintain normal or increased arterial pressure.
r The peripheral chemoreflex still initiated decelerations even after the onset of evolving hypotension and acidaemia, but myocardial hypoxia made an increasing contribution to sustain and deepen decelerations.

Introduction
Fetal heart rate (FHR) monitoring is the only method available to monitor fetal wellbeing continuously during labour but, unfortunately, remains poorly able to identify reliably those fetuses at risk of hypoxia-ischaemia and subsequent lifelong neurodevelopmental disability (Nelson et al., 1996;Westgate et al., 2007). The interpretation of intrapartum decelerations, brief falls in FHR associated with uterine contractions, is central to the overall interpretation of FHR patterns, yet their physiological origins remain controversial and incompletely understood . It has been known intuitively for decades that severe fetal hypoxaemia can trigger intrapartum decelerations by a vagal reflex mechanism followed by the direct effects of myocardial hypoxia (Ball & Parer, 1992;Barcroft, 1946;Hon et al., 1961;Méndez-Bauer et al., 1963). However, the relative contributions of these two mechanisms during intrapartum decelerations and, crucially, how this balance changes with evolving fetal compromise over time, have never been established. The reflex response to fetal hypoxaemia is mediated via the peripheral chemoreflex, triggering increased parasympathetic activity that reduces FHR, and increased sympathetic activity that causes peripheral vasoconstriction to centralise blood flow and support arterial pressure (Giussani, 2016;Giussani et al., 1993;Lear, Kasai, Booth, et al., 2020;. We, therefore, sought to dissect the roles of the peripheral chemoreflex and myocardial hypoxia during FHR decelerations in near-term fetal sheep at 0.8 of gestation, when fetal maturation of the CNS and the cardiovascular system are broadly comparable to the term human fetus (Morrison et al., 2018), as detailed further in the Methods. We have recently shown that the FHR and cardiovascular responses to brief (1 min) complete umbilical cord occlusions (UCOs) are almost entirely mediated by the peripheral chemoreflex, with a negligible contribution from the baroreflex (limited to the first 3−4 s and a minor fall in FHR by 9.7-16.9 beats/min) that is related to the physical occlusion of the umbilical arteries (Lear, Kasai, Booth, et al., 2020). Furthermore, complete 1 min UCOs do not trigger the Bezold-Jarisch reflex (which had been hypothesised to be triggered by impaired ventricular preload) (Lear et al., 2021a(Lear et al., , 2021b. Brief, complete UCO can therefore be considered a robust methodology to stimulate the peripheral chemoreflex specifically (Lear et al., 2021a;Lear, Kasai, Booth, et al., 2020).
We, therefore, studied chronically instrumented fetuses that received 1 min UCOs repeated every 2.5 min (i.e. after 1.5 min reperfusion) until fetal cardiovascular compromise developed, defined as severe arterial hypotension (<20 mmHg). Hypotension is key to the development of hypoxic-ischaemic injury (de Haan et al., 1997;Georgieva et al., 2021;Lear et al., 2023;. Peripheral chemoreflex control of FHR was prevented by performing bilateral cervical vagotomy, removing all myocardial parasympathetic innervation to unmask the contribution from myocardial hypoxia. We instrumented fetal sheep to monitor cardiovascular parameters comprehensively, in order to understand the impact of vagotomy on the cardiovascular adaptation to labour-like hypoxaemia. We also measured EEG activity and cortical impedance (a measure of cortical swelling; Williams et al., 1991) to investigate any subsequent effect on the neurophysiological adaptation.

Ethical approval
All procedures, including termination protocols, were approved by the Animal Ethics Committee of the University of Auckland (number 22069), following the New Zealand Animal Welfare Act 1999, and the Code of Ethical Conduct for animals in research established by the Ministry of Primary Industries, Government of New Zealand, and comply with the ARRIVE guidelines (Percie du Sert et al., 2020).

Rationale for the use of near-term fetal sheep
The near-term fetal sheep at 0.85 of gestation (∼125 days; term is 147 days) is broadly equivalent to the neural maturation of the term human neonate (Back et al., 2012;Morrison et al., 2018;van den Heuij et al., 2016), as observed in the pattern of peak brain growth, oligodendrocyte maturation and both cortical and white matter myelination (Back et al., 2012;Barlow, 1969;Dobbing & Sands, 1970;McIntosh et al., 1979). The compartmentalization of fetal behaviour and cycling between fetal sleep and/or behavioural state is likewise seen in both species near term Nijhuis et al., 1982;Pillai & James, 1990). The maturational trajectory of fetal cardiomyocytes and their shift from proliferation to terminal differentiation during late gestation is also similar in humans and sheep during the final trimester (Burrell et al., 2003;Dimasi et al., 2023;Jonker et al., 2015). Furthermore, the autonomic system follows a broadly similar development in sheep and humans towards the end of gestation, with a progressive decline in baseline FHR related to increasing parasympathetic tone in both species, along with increasing FHR variability and maturation of renal sympathetic nerve activity (Booth et al., 2011;Koome et al., 2014;Méndez-Bauer et al., 1963;Walker et al., 1978). A similar trajectory of axonal and myelination development is observed in the vagus nerve during the third trimester (Hasan et al., 1993;Pereyra et al., 1992;Sachis et al., 1982), while growth of sympathetic efferent J Physiol 601.10 fibres into the myocardium begins at mid-gestation and is nearing maturation by term in fetal sheep (Lebowitz et al., 1972). The development of the thoracic sympathetic trunk is similar in fetal humans and sheep (Nourinezhad et al., 2013). These comparative studies collectively support that the maturation of multiple organ systems (excluding the respiratory system) in near-term fetal sheep at 0.85 of gestation is broadly similar to the term human fetus, making it an appropriate model for studying intrapartum fetal cardiovascular physiology. Multiple parallels between the FHR patterns in near-term fetal sheep experiments and human labour have been observed (Georgieva et al., 2021;Tarvonen et al., 2022).

Surgical procedures
Nineteen Romney/Suffolk fetal sheep were randomly assigned to receive vagotomy (n = 8) or sham vagotomy (control, n = 11) and surgically instrumented at 122.9 ± 3.0 days of gestation (Kasai et al., 2019;Lear, Kasai, Booth, et al., 2020). In twin pregnancies, only one fetus was instrumented. Surgical procedures were performed as previously described (Lear et al., 2021a;Lear, Kasai, Booth, et al., 2020). Ewes were given oxytetracycline (20 mg/kg; Phoenix Pharm Distributors, Auckland, New Zealand) intramuscularly 30 min before surgery for infection prophylaxis. Anaesthesia was induced by intravenous propofol (5 mg/kg; AstraZeneca, Auckland, New Zealand) and maintained using 2−3% isoflurane (Medsource, Ashburton, New Zealand) in oxygen. Fetuses were partly exteriorised via a mid-line abdominal incision and uterotomy. Catheters were placed in the left femoral artery and vein [to sit in the abdominal aorta and inferior vena cava (IVC), respectively], right brachial artery and amniotic sac. These were used to measure mean arterial pressure (MAP), ICV pressure (IVCP), sample preductal arterial blood and measure amniotic pressure to correct for maternal position. Ultrasonic flow probes (size 3S; Transonic Systems, Ithaca, NY, USA) were placed around the right femoral artery and left carotid artery to measure femoral and carotid blood flow (FBF and CaBF, respectively). Subcutaneous electrodes (Cooner Wire, Chatsworth, CA, USA) over the right shoulder and left fifth intercostal space measured ECG activity. Bilateral cervical vagotomy or sham vagotomy was performed via two separate incisions made lateral to the trachea (Lear et al., 2021a). An inflatable silicone occluder was placed around the umbilical cord to facilitate UCOs (18HD; In Vivo Metric, Healdsburg, CA, USA). Two pairs of electrodes were placed on the dura over the parasagittal parietal cortex (10 and 20 mm anterior and 10 mm lateral to bregma) to measure EEG activity. An additional pair of electrodes were placed bilaterally on the dura (15 mm anterior and 12.5 mm lateral to bregma) to measure cortical impedance. Cortical impedance provides a measure of cortical oedema and swelling. The impedance of a tissue rises concomitantly as cells depolarise and fluid shifts from the extracellular space to the intracellular space in association with the development of cytotoxic oedema (Williams et al., 1991). A reference electrode was sewn over the occiput.
Gentamicin was administered into the amniotic sac (80 mg; Pfizer, Auckland, New Zealand) and incisions closed. The maternal skin incision was infiltrated with a long-acting analgesic, bupivacaine plus adrenaline (AstraZeneca). Fetal leads were exteriorised through the maternal flank, and a maternal long saphenous vein was catheterised for postoperative care.
Postoperative care and signal acquisition. Ewes were housed together in separate metabolic cages with ad libitum access to food and water, in temperature-controlled rooms (16 ± 1°C, humidity 50 ± 10%) with a 12 h-12 h light-dark cycle. Daily intravenous antibiotics were administered to the ewe for 4 days (600 mg benzylpenicillin sodium; Novartis, Auckland, New Zealand; and 80 mg gentamicin). Fetal catheters were maintained patent by continuous infusion of heparinised saline (20 U/ml at 0.2 ml/h).
All signals were digitised at a sampling rate of 4096 Hz before being decimated to lower sampling rates and stored using customised LabVIEW-based acquisition software (National Instruments, Austin, TX, USA) (Lear, Kasai, Booth, et al., 2020). Fetal blood pressures, corrected for maternal position by subtraction of amniotic pressure, were recorded using Deltran blood pressure transducers (DPT-100; Utah Medical Products, Athlone, Ireland), low-pass filtered with a fifth-order Butterworth filter set at 20 Hz and saved at 64 Hz. The CaBF and FBF were measured using a Perivascular Flowmeter (TS420 module on a T402 console; Transonic Systems), low-pass filtered with a second-order Butterworth filter at 0.1 Hz and saved at 64 Hz. The ECG was analog filtered with a first-order high-pass filter at 1 Hz and an eighth-order low-pass Bessel filter at 100 Hz and saved at 1024 Hz. The EEG signal was low-pass filtered through an inverse Chebyshev filter set at 512 Hz and saved at 1024 Hz. Cortical impedance was measured by injecting a sinusoidal current with a frequency of 200 Hz and amplitude of 0.2 μA through the two cortical impedance electrodes. Cortical impedance was calculated by measuring the voltage drop across the front two EEG electrodes (Williams et al., 1991). and can include a degree of umbilical cord compression . Rapid (<1 s) complete UCO as used in the present study is unlikely to be very common during human labour, except during umbilical cord prolapse. However, deep decelerations occur only during complete or near-complete UCO (Itskovitz et al., 1983). Thus, complete UCO achieves a highly consistent and reproducible pattern of severe hypoxaemia with progressive fetal compromise that represents one of the key targets for monitoring in labour.
Occlusion of upstream vessels, such as the common internal iliac artery, is possible in fetal sheep models but typically does not achieve severe hypoxaemia owing to significant collateral arterial supply (Ball et al., 1994), whereas occlusion of the descending aorta results in severe hypoxaemia but leads to hindlimb ischaemia in the ewes (Jensen et al., 1987a(Jensen et al., , 1987b. Repeated brief complete UCOs are therefore an appropriate methodology to model fetal severe hypoxaemia characteristic of labour, whether secondary to uteroplacental compression or umbilical cord compression. Protocol. Experiments began 4−5 days after surgery, at 126.9 ± 2.7 days (0.85) of gestation. No animals were in labour during these experiments. Fetuses were exposed to brief repeated UCOs induced by rapid inflation of the umbilical cord occluder with a volume of saline known to occlude the umbilical cord completely. Each UCO lasted 1 min before the occluder was deflated rapidly and completely to allow a 1.5 min period of reperfusion. The UCOs were repeated at this rate for 4 h (total of 96 UCOs) or until MAP fell below 20 mmHg on two successive UCOs.

Data analysis
Data were analysed in quarters consisting of six UCOs to account for the unequal duration of individual experiments owing to individual fetuses reaching the defined endpoints at different times. Data were processed into continuous 1 s means. We calculated absolute FHR, from which we calculated the deceleration nadir (the lowest absolute FHR recorded either during or within 30 s after the end of UCO) and the timing of that nadir. We also calculated the change in FHR during each UCO relative to baseline FHR (i.e. FHR 1 s before the start of each UCO) and from this we calculated deceleration amplitude (the maximal depth of decelerations relative to the baseline FHR during or within 30 s after the end of UCO). For display purposes, change in FHR, deceleration nadir, deceleration amplitude and nadir timing were averaged across the six UCOs per quarter in each fetus. Ventricular contractility was estimated by calculating the maximal rate of rise of arterial blood pressure (dP/dt max ) from arterial pressure waveforms (De Hert et al., 2006;Morimont et al., 2012). Arterial dP/dt max and IVCP were also processed with a 5 s median filter. Carotid and femoral vascular conductance (CaVC and FVC, respectively) were calculated as (blood flow)/(MAP − IVCP). The EEG power was calculated on the power spectrum between 0.5 and 20 Hz and logarithmically transformed for presentation [in decibels; 20 × log(power)]. The EEG spectral edge frequency was calculated as the frequency below which 90% of EEG power was present (Szeto, 1990;Williams & Gluckman, 1990). Blood flow, vascular conductance, arterial dP/dt max , EEG power, spectral edge frequency and cortical impedance are displayed relative to their baseline values.
The final group sizes were reduced in the following cases owing to failed recordings: control group, FBF and FVC in three fetuses (n = 8) and cortical impedance in two fetuses (n = 9); vagotomy group, IVCP in one fetus (n = 7), arterial dP/dt max in one fetus (n = 7), CaBF and CaVC in one fetus (n = 7) and FBF and FVC in two fetuses (n = 6). Arterial blood samples could not be collected in one fetus in the vagotomy group (n = 7).

Statistical analysis
Statistical analysis was performed using SPSS (IBM, Armonk, NY, USA). Changes during UCOs were evaluated by repeated-measures ANOVA, with vagotomy, the quarter of the experiment and the six UCOs of each quarter as independent factors and time as a repeated factor. Six non-overlapping time epochs were investigated: 0−20, 20−40 and 40−60 s during UCOs and 0−30, 30−60 and 60−90 s between UCOs during the reperfusion periods. If interactions between the effects of vagotomy and quarter were found, a separate repeated-measures ANOVA on each quarter was performed. Exploratory analysis was performed during the first UCO alone. The within-individual correlation between the minimum MAP and deceleration nadir, deceleration amplitude and the timing of the deceleration nadir across the four quarters of the study was calculated as described by Bland and Altman (1995a). Three time points during the UCO series were selected for analysis and presentation of arterial biochemistry outcomes, taken 126.9 ± 36.7 118.7 ± 37.4 0.8970 Spleen weight (g) 6.6 ± 1.8 5.9 ± 1.5 0.3733 Kidney weight (g) 11.9 ± 1.5 12.0 ± 1.6 0.9449 Left and right kidneys and adrenals were averaged. The statistical analysis performed was one-way ANOVA. Data are mean ± SD. Abbreviations: F, female; M, male; S, singleton; T, twin.
at 30 min after the start of UCOs, at the approximate midpoint of the UCO series and after the final UCO (Start, Middle and Final, respectively; Table 2) and were evaluated by repeated-measures ANOVA. We also assessed the between-individual correlation of pH, base excess (BE) and lactate across both groups as described by Bland and Altman (1995b). For this analysis, we did not limit our analysis to the time points shown in Table 2 but used all samples available across the UCO series (data not shown) to test the hypothesis that vagotomy would be associated with more rapid development of metabolic acidaemia. Post-mortem outcomes were assessed by one-way ANOVA. Statistical significance was accepted when P < 0.05. Exact P-values are provided unless tests returned P < 0.0001. Data are presented as means ± SD; n in all cases refers to the number of fetal sheep in each group.

Results
All fetuses were healthy based on our laboratory standards before experiments. Fetal characteristics and post-mortem findings are shown in Table 1 Overall, there were no differences between groups in the number of UCOs (P = 0.1498), the final MAP (P = 0.1381) or the severity of metabolic acidaemia between groups (Table 2). In analysis of the correlation between individuals during UCOs using the method of Bland and Altman (1995b) for repeated measurements, there was an effect of time on pH (P < 0.0001, R 2 = 0.586, n = 18), BE (P = 0.0021, R 2 = 0.306, n = 18) and lactate (P < 0.0001, R 2 = 0.760, n = 18), with no overall effect of vagotomy (P = 0.1211, P = 0.2942 and P = 0.1639, respectively). However, there was an interaction effect between time and vagotomy for pH (P = 0.0009, R 2 = 0.329, n = 18), BE (P < 0.0001, R 2 = 0.657, n = 18) and lactate (P = 0.0001, R 2 = 0.375, n = 18), suggesting that the decrease in pH and BE and increase in lactate occurred significantly faster in the vagotomy group than in control fetuses.

Fetal heart rate
All UCOs in the control group were associated with rapid decelerations in FHR (Figs 1 and 2). Across both groups, decelerations became deeper with successive quarters (0-20 s, P < 0.0001; 20−40 s, P = 0.0004; 40−60 s P < 0.0001; Figs 1-3). Vagotomy impaired the FHR response to UCOs, resulting in a higher FHR at 0−20 s compared with control fetuses during all UCOs throughout the experiment (P < 0.0001). Interactions between the effects of vagotomy and quarter were observed at 20−40 s (P = 0.0001) and 40−60 s during UCOs (P = 0.0001), reflecting higher FHR in the vagotomy group at 20−40 s during UCOs in the first, second and third quarters (P < 0.0001, P = 0.0003 and P = 0.0043, respectively) and a higher FHR in the vagotomy group at 40−60 s of UCOs in the first and second quarters (P < 0.0001 and P = 0.0162, respectively).
During reperfusion between UCOs, FHR was higher at 30−60 s in the vagotomy group throughout the experiment (P < 0.0001). Interactions between the 10.2 ± 0.7 12.6 ± 0.9 0.1310 13.8 ± 0.9 11.9 ± 1.1 11.9 ± 0.9 Blood gases presented during occlusions were selected from early in the experiment, in the middle of the experiment and after the final occlusion. Control (C), effects of vagotomy and quarter were observed at 0−30 s (P < 0.0001) and 60−90 s during reperfusion (P < 0.0001), leading to higher FHR after vagotomy at 0−30 s during reperfusion in the first and second quarters (P < 0.0001 and P = 0.0008, respectively) and at 60−90 s during reperfusion in the first, second and third quarters (P < 0.0001, P = 0.0001 and P = 0.0020, respectively).
We then assessed the change in FHR relative to the immediate baseline before the start of each UCO (Fig. 2). During UCOs, the relative fall in FHR at 0−20 s of Figure 1. Fetal heart rate and mean arterial pressure during repeated umbilical cord occlusions The control group is shown in black (n = 11) and the vagotomy group in red (n = 8), with the periods of umbilical cord occlusion shown in grey shading. * P < 0.05 control versus vagotomy (ANOVA, with vagotomy, the quarter of the experiment and the six umbilical cord occlusions of each quarter as independent factors and time as a repeated factor). Please refer to the main text and Statistical Summary Document for exact P-values. Data are 1 s means ± SD; dotted lines represent the SD.
UCOs was less in the vagotomy group compared with the control group throughout the experiment (i.e. FHR fell more slowly after vagotomy, P = 0.0003). An interaction between the effects of vagotomy and quarter was observed at 40−60 s of UCOs (P = 0.0409), such that there was a greater relative fall in FHR after vagotomy during the first and second quarters (P = 0.0009 and P = 0.0302, respectively) and a borderline effect in the third quarter (P = 0.0501).
During reperfusion, FHR relative to pre-UCO baseline was greater at 30−60 s of reperfusion in the vagotomy group throughout the experiment (P < 0.0001). Interactions between the effects of vagotomy and quarter were observed at 0−30 s (P = 0.0007) and 60−90 s of reperfusion (P < 0.0001), leading to a higher relative FHR after vagotomy at 0−30 s of reperfusion in first quarter (P = 0.0403) and at 60−90 s of reperfusion in the first and second quarters (P < 0.0001 and P = 0.0102, respectively).

Figure 2. Evolution of relative fetal heart rate changes during umbilical cord occlusions
A, Fetal heart rate from the four quarters of the experiment superimposed for each group to allow assessment of the changes over time. The data displayed are 1 s means from each quarter; the data are the same as displayed in Fig. 1, without SD. B, Morphology of decelerations assessed relative to the immediate baseline before the start of each umbilical cord occlusion in the control group (n = 11, black) and the vagotomy group (n = 8, red). Data displayed are the average of the six occlusions in each quarter; statistical analysis was performed on the unaveraged data. Statistics performed were ANOVA, with vagotomy, the quarter of the experiment and the six umbilical cord occlusions of each quarter as independent factors and time as a repeated factor. Data are 1 s means ± SD; dotted lines represent the SD. C, The data displayed in B have been superimposed without the SD for each group to allow assessment of the evolution of deceleration morphology throughout the experiment. Finally, we assessed features associated with the nadir of FHR decelerations (Fig. 3). The nadir became progressively deeper with each quarter in both groups (P < 0.0001) and was higher in the vagotomy group compared with the control group (P = 0.0435). An interaction between vagotomy and quarter was observed (P = 0.0059), mediated by a higher nadir during the first (P < 0.0001) and second quarters (P = 0.0142) in the vagotomy group. The amplitude of decelerations (i.e. relative to FHR immediately before UCOs) increased with each quarter (P < 0.0001) and was increased further by vagotomy throughout the experiment compared with the control group (P = 0.0080). An interaction between vagotomy and quarter was observed (P = 0.0475), leading to a greater amplitude during the first (P = 0.0001), second (P = 0.0136) and third quarters (P = 0.0159) in the vagotomy group. The timing of the nadir after the onset of UCOs became later with each quarter across both groups (P < 0.0001) and was not significantly altered by vagotomy (P = 0.0712).
Throughout the experiment, analysis of within-individual correlations showed that the minimum MAP in each quarter was associated with the nadir of decelerations (control, P < 0.0001, R 2 = 0.736, n = 11; vagotomy, P < 0.0001, R 2 = 0.897, n = 8) and the timing of deceleration nadir (control, P = 0.0004, R 2 = 0.570, n = 11; vagotomy, P < 0.0001, R 2 = 0.750, n = 8) in both the control and vagotomy groups. The amplitude of decelerations was also associated with minimum MAP in the control group (P < 0.0001, R 2 = 0.723, n = 11), but not in the vagotomy group (P = 0.1216).

Blood pressures
Both groups initially showed hypertension during UCOs, followed by a transition to a biphasic pattern of an The deceleration nadir is the lowest absolute fetal heart rate (FHR) reached during UCOs; deceleration amplitude is the lowest fall in FHR relative to the baseline FHR immediately before each UCO; and nadir timing is the time at which the nadir was reached after the start of each UCO. The control group is shown with white circles (n = 11) and the vagotomy group with red circles (n = 8). The statistical analysis performed was ANOVA, with vagotomy as the independent factor and the quarter of the experiment as a repeated factor. Individual data points are shown overlaid with means ± SD. B, relationship between minimum mean arterial pressure and the deceleration nadir, deceleration amplitude and nadir timing in each quarter. Statistical analysis was performed by within-individual regression using the method of Bland and Altman (1995a). The regression line included in the figure is for illustrative purposes only. initial increase then fall in MAP during UCOs (Fig. 1). The initial increase in MAP at 0−20 s of UCO in both groups became blunted with successive quarters (P < 0.0001), while the fall in MAP between 20 and 60 s of UCOs became progressively faster and more severe with successive quarters (20-40 s, P < 0.0001; 40−60 s, P < 0.0001). Vagotomy was associated with an impaired MAP response during UCOs, with a lower MAP at 20−40 s of UCO (P = 0.0154) and at 40−60 s of UCOs (P = 0.0046) in the vagotomy group compared with the control group throughout the experiment. However, vagotomy did not affect the initial increase in MAP observed at 0−20 s of UCOs at any point in the experiment (P = 0.9688).
The MAP at 0−90 s between UCOs during reperfusion fell progressively with successive quarters across both groups (0-30 s, P < 0.0001; 30−60 s, P < 0.0001; 60−90 s, P < 0.0001). Interactions between the effects of vagotomy and quarter were observed at 0−30 s of reperfusion (P = 0.0175), leading to higher MAP in first quarter (P = 0.0193) and lower MAP in the third quarter after vagotomy (P = 0.0455). The MAP was higher at 30−60 s of reperfusion in the vagotomy group throughout the experiment (P = 0.0108).
The IVCP increased progressively during UCOs, and no effect of vagotomy during UCOs was observed (Fig. 4). During reperfusion, an interaction between the effects of vagotomy and quarter was observed at 0−30 s of reperfusion (P = 0.0414), reflecting increased IVCP after vagotomy in the fourth quarter (P = 0.0192).

Ventricular contractility
Arterial dP/dt max , an index of ventricular contractility, increased during the first UCO. Thereafter, it remained elevated between UCOs and fell progressively during each UCO. An effect of quarter on arterial dP/dt max was observed throughout all UCO and reperfusion epochs (all P < 0.0001), indicating a progressive decline in arterial dP/dt max both during and between UCOs throughout the experiment in the control and vagotomy groups (Fig. 4). During UCOs, arterial dP/dt max was lower in the vagotomy group at 20−40 s during UCOs throughout the experiment (P = 0.0285).

Femoral blood flow and vascular conductance
The FBF and FVC fell rapidly during each UCO, with partial resolution during reperfusion in both groups (Fig. 5). During UCOs in the main analysis, no effect of vagotomy was observed for FBF, but an interaction between the effects of vagotomy and quarter was observed at 20−40 s (P = 0.0272), leading to a slower fall in FVC during the first quarter (P = 0.0272). Exploratory analysis of the first UCO suggested that FBF was higher in the vagotomy group at 0−20 s (P = 0.0068), 20−40 s (P < 0.0001) and 40−60 s during the UCO (P = 0.0435) and that FVC was higher at 0−20 s (P = 0.0103) and 20−40 s during the UCO (P < 0.0001), with a borderline effect at 40−60 s (P = 0.0514).

Carotid blood flow and vascular conductance
The control group initially maintained CaBF during UCOs, before showing a delayed fall that became greater with repeated UCOs (Fig. 6). During UCOs, an interaction between the effects of vagotomy and quarter was observed at 20−40 s during UCOs (P = 0.0384), reflecting higher CaBF after vagotomy during the first quarter (P = 0.0233). Exploratory analysis of the first UCO suggested that CaBF was higher in the vagotomy group at 0−20 s (P = 0.0010), 20−40 s (P < 0.0001) and 40−60 s during the UCO (P = 0.0011). During reperfusion, an interaction between the effects of vagotomy and quarter was observed at 0−30 s between UCOs (P = 0.0089), reflecting greater CaBF after vagotomy during the first quarter (P = 0.0154). In the control group, a delayed fall in CaVC was observed during each UCO. This was prevented by vagotomy throughout the experiment, with greater CaVC at 0−20 s (P = 0.0493), 20−40 s (P = 0.0086) and 40−60 s (P = 0.0142) during UCOs.

Neurophysiological adaptation
The EEG power reduced in a delayed fashion during each UCO in both groups, followed by a recovery towards baseline values during the reperfusion periods (Fig. 7). An effect of quarter was observed on EEG power throughout all UCO and reperfusion epochs (all P < 0.0001, except P = 0.0009 at 40−60 s during UCOs), indicating progressively more severe suppression of EEG power throughout the experiment.
Spectral edge frequency reduced in a delayed fashion during the first UCO in both groups and remained reduced until the end of the experiment (Fig. 7). A relative rise in spectral edge frequency was observed during UCOs in both groups towards the end of the experiment. An effect of quarter was observed in both groups at 0−20, 20−40 and 40−60 s during UCO epochs (P = 0.0001, P = 0.0398 and P = 0.0161, respectively), probably indicating a progressive earlier relative increase in spectral edge frequency. An effect of quarter was also observed at 30−60 s between UCOs (P = 0.0086).
Cortical impedance increased progressively during each UCO in both groups (Fig. 8). An effect of quarter was observed on cortical impedance throughout all UCO and reperfusion epochs (all P < 0.0001), indicating J Physiol 601.10 a progressive rise throughout the experiment. There were no effects of vagotomy on EEG power, spectral edge frequency or cortical impedance during either UCOs or the reperfusion periods (see Statistics Summary Document for full details).

Discussion
In the present study, we sought to dissect the relative roles of the peripheral chemoreflex and the direct negative chronotropic effects of myocardial hypoxia during FHR decelerations induced by brief hypoxaemia. Previous

Figure 4. Arterial dP/dt max and inferior vena cava pressure during repeated umbilical cord occlusions
The maximal rate of rise of arterial blood pressure (dP/dt max ) provides an index of ventricular contractility and is shown as change relative to baseline. The control group is shown in black (n = 11) and the vagotomy group in red (n = 7); the periods of umbilical cord occlusion are shown in grey shading. * P < 0.05 control versus vagotomy (ANOVA, with vagotomy, the quarter of the experiment and the six umbilical cord occlusions of each quarter as independent factors and time as a repeated factor). Please refer to the main text and Statistical Summary Document for exact P-values. Data are 1 s means ± SD; dotted lines represent SD.

Figure 5. Femoral blood flow and femoral vascular conductance during repeated umbilical cord occlusions
Femoral vascular conductance is the inverse of vascular resistance, hence a fall in vascular conductance reflects femoral vasoconstriction. Both femoral blood flow and vascular conductance are shown relative to baseline. The control group is shown in black (n = 8) and the vagotomy group in red (n = 6); the periods of umbilical cord occlusion are shown in grey shading. * P < 0.05 control versus vagotomy (ANOVA, with vagotomy, the quarter of the experiment and the six umbilical cord occlusions of each quarter as independent factors and time as a repeated factor). Please refer to the main text and Statistical Summary Document for exact P-values. Data are 1 s means ± SD; dotted lines represent SD. J Physiol 601.10

Figure 6. Carotid blood flow and carotid vascular conductance during repeated umbilical cord occlusions
Carotid vascular conductance is the inverse of vascular resistance, hence a fall in vascular conductance reflects carotid vasoconstriction. Both carotid blood flow and vascular conductance are shown relative to baseline. The control group is shown in black (n = 11) and the vagotomy group in red (n = 7); the periods of umbilical cord occlusion are shown in grey shading. * P < 0.05 control versus vagotomy (ANOVA, with vagotomy, the quarter of the experiment and the six umbilical cord occlusions of each quarter as independent factors and time as a repeated factor). Please refer to the main text and Statistical Summary Document for exact P-values. Data are 1 s means ± SD; dotted lines represent SD. exploratory experiments in fetal goats, fetal sheep and neonatal/postnatal rabbits all show that vagotomy delays the onset of bradycardia during severe hypoxaemia (Barcroft, 1946;Westgate et al., 2007). Reports in human fetuses also show that anti-cholinergics abolish some, but not all, intrapartum decelerations (Bradfield, 1961(Bradfield, , 1962Hon et al., 1961;Méndez-Bauer et al., 1963). These studies imply that myocardial hypoxia has a role in some intrapartum decelerations. The present study is the first systematic evaluation of the relative contributions Figure 7. EEG power and spectral edge frequency during repeated umbilical cord occlusions Spectral edge frequency is the frequency below which 90% of EEG power occurs. Both EEG power and spectral edge frequency are shown relative to baseline. The control group is shown in black (n = 11) and the vagotomy group in red (n = 8); the periods of umbilical cord occlusion are shown in grey shading. Please refer to the main text and Statistical Summary Document for outcomes of statistical analysis. Data are 1 s means ± SD; dotted lines represent SD. J Physiol 601.10 of the peripheral chemoreflex and myocardial hypoxia to FHR decelerations caused by labour-like hypoxaemia.

Figure 8. Cortical impedance during repeated umbilical cord occlusions
Cortical impedance is a measure of cortical cell swelling and is shown relative to baseline. The control group is shown in black (n = 9) and the vagotomy group in red (n = 8); the periods of umbilical cord occlusion are shown in grey shading. Please refer to the main text and Statistical Summary Document for outcomes of statistical analysis. Data are 1 s means ± SD; dotted lines represent SD.

Control of fetal heart rate during and between decelerations
In intact control fetuses, UCOs were associated with rapid FHR decelerations. During the first two quarters of UCOs in intact control fetuses, the initial relative fall in FHR was faster (Fig. 2) and absolute FHR was lower throughout UCOs (Fig. 1) than in vagotomised fetuses. This is likely to indicate that reflex parasympathetic mechanisms (i.e. the peripheral chemoreflex) were the dominant mechanism initiating and sustaining FHR decelerations during the first and second quarters, when fetuses were able to maintain arterial pressure. This, in turn, suggests that myocardial hypoxia is unlikely to be an important contributor to decelerations in fetuses showing effective cardiovascular adaptation to labour-like hypoxaemia, and thus is not a feature of the majority of uncomplicated human labours.
As fetuses developed progressive cardiovascular compromise, hallmarked by worsening arterial hypotension, FHR continued to fall faster in intact control fetuses at the start of decelerations in the third and fourth quarters (Fig. 2). This finding illustrates that the peripheral chemoreflex remained able to be activated at the start of labour-like hypoxaemia even after the onset of evolving hypotension and severe acidaemia. In contrast, the present study suggests that myocardial hypoxia made an increasing contribution in sustaining the later portion of decelerations after the onset of progressive fetal compromise, contributing to the last 20 s of decelerations in the third quarter and the last 40 s of decelerations in the fourth quarter (Fig. 1). This is illustrated by the lack of difference between the control and vagotomy groups at these time points in Fig. 1. At the end of the experiment in the setting of severe hypotension (fourth quarter), despite the initial faster fall in FHR in control fetuses, the overall deceleration morphology (Fig. 1) was very similar between the two groups. This suggests that myocardial hypoxia progressively became the dominant regulator of decelerations during fetal compromise. This is supported by the similar deceleration nadir and timing of the nadir in both groups at this time (Fig. 3).
In the control group, the nadir and amplitude of decelerations increased in association with worsening hypotension, while the timing of the nadir also became increasingly delayed (Fig. 3). Similar patterns were observed for the deceleration nadir and the timing of the nadir in the vagotomy group, which supports the hypothesis that deepening decelerations with progressively later nadirs are likely to be mediated by myocardial hypoxia and reflect increasing risk of hypotension. This might well contribute to the increased risk of fetal compromise associated with late decelerations clinically. Interestingly, the deceleration amplitude was greater in the vagotomy group in the first to third quarters and did not increase with hypotension (Fig. 3). This greater amplitude was related to the marked tachycardia between UCOs, which subsided by the fourth quarter.
These findings emphasise that the peripheral chemoreflex is a robust adaptation, capable of being reactivated continually during successive decelerations even after the onset of progressive cardiovascular compromise. After the initial fall in FHR, myocardial hypoxia progressively becomes the major factor sustaining the later portions of 1 min decelerations, as severe fetal hypotension and acidaemia develop. Myocardial hypoxia is not well studied in the fetus (Lumbers et al., 1986;Stowe et al., 1985), but realistically, it must include the combined effects of myocardial hypoxia, acidosis and hypoglycaemia. Studies of adult cardiac ischaemia indicate that bradycardia in this setting is likely to reflect modulation of multiple ionic currents (Carmeliet, 1999;Du & Nathan, 2007), with ATP-sensitive K + channels and adenosine accumulation playing important roles (Clemo & Belardinelli, 1986a, 1986bFukuzaki et al., 2008;Saito et al., 2005;Wesley et al., 1986) leading to a reduced rate of diastolic depolarisation and prolonged nodal conduction (Senges et al., 1979).
During the first and second quarters, FHR during the reperfusion period was considerably higher in the vagotomy group than in the control group, probably reflecting the unopposed effects of high circulating catecholamines in the absence of parasympathetic tone (Galinsky et al., 2014Recher et al., 2021). In turn, this indicates that there is high parasympathetic activity between UCOs that markedly restrains FHR in control fetuses Recher et al., 2021;Tournier et al., 2022). This tachycardia in the vagotomy group progressively reduced in magnitude, with less difference compared with the control group in the third and fourth quarters. We have previously shown that blood catecholamine levels increase markedly and early during repeated brief UCOs (Galinsky et al., 2014; and remain high after the onset of arterial hypotension (Galinsky et al., 2014). This indicates that the attenuation of tachycardia is likely to reflect impaired cardiac responsiveness to β-adrenergic stimulation, potentially attributable to intracellular acidosis and failing high-energy phosphate levels.

Cardiovascular adaptation and evolving hypotension
The control group initially showed sustained hypertension during UCOs, predominantly mediated by intense peripheral vasoconstriction. Vascular conductance is the inverse of vascular resistance, hence the rapid fall in FVC with each UCO (Fig. 5) denotes rapid peripheral vasoconstriction that was sufficiently intense to reduce FBF practically to zero. This is known to be mediated initially by the rapid activation of the sympathetic efferent arm of the peripheral chemoreflex, then augmented by release of humoral factors including adrenal catecholamines, vasopressin and angiotensin II (Broughton-Pipkin et al., 1974;Fletcher et al., 2000;Galinsky et al., 2014;Giussani et al., 1993;Jones & Robinson, 1975;Perez et al., 1989). A moderate fall in CaVC was also observed during UCOs in intact control fetuses. The carotid artery supplies both cerebral and non-cerebral vascular beds, and thus it is not clear from the present study whether this represents active cerebral vasoconstriction to maintain cerebral blood flow at near-baseline values or preferential vasoconstriction of non-cerebral beds.
The control group then progressively transitioned to a biphasic pattern of an initial increase then fall in MAP during UCOs, which worsened with repeated UCOs. The MAP fell below inter-occlusion baseline values during the second quarter and below pre-experiment baseline during the third and fourth quarters (de Haan et al., 1997;Westgate et al., 1999). Progressive hypotension appeared most closely related to falling arterial dP/dt max (Fig. 4), which provides an estimate of ventricular contractility (De Hert et al., 2006;Morimont et al., 2012). This falling ventricular contractility is likely to be related to intracellular myocardial acidosis, depletion of myocardial glycogen and evolving myocardial injury (Dawes et al., 1959;Gunn et al., 2000;Shelley, 1961). In contrast, intense peripheral vasoconstriction was maintained throughout all four quarters (Fig. 5), indicating that impaired peripheral vascular tone does not contribute to fetal hypotension in this setting . Thus, the present findings strongly support the concept that the onset of fetal hypotension during labour is mediated by impaired myocardial contractility.
Importantly, vagotomy was associated with impaired maintenance of MAP and more rapid onset of hypotension. Although the final minimum MAP was not different between groups (essentially by experimental design), overall, the vagotomy group developed greater hypotension over a longer portion of each UCO than the control group. This appeared to be related to faster impairment in ventricular contractility after vagotomy (arterial dP/dt max ; Fig. 4), while peripheral vascular tone was not impaired (Fig. 5). Moreover, although the final severity of metabolic acidaemia was not worse after vagotomy by the end of the experiment, there was a significant interaction between vagotomy and time for pH, BE and lactate concentrations, such that metabolic acidaemia developed more rapidly in vagotomised fetuses. We also observed a slower fall in IVCP in the first 30 s after UCOs during the fourth quarter in the vagotomy group J Physiol 601.10 ( Fig. 4), which might represent impairment of combined ventricular output, leading to greater venous pooling.
Collectively, these findings highlight that the failure to reduce FHR rapidly at the start of decelerations after vagotomy hastened oxygen and high-energy phosphate depletion and, thereby, impaired ventricular contractility and combined ventricular output while also accelerating the development of metabolic acidaemia. These data support the concept that intrapartum decelerations do indeed represent a physiological adaptation to reduce myocardial work, conserving limited oxygen and substrate supplies, particularly myocardial glycogen reserves, and thus helping to protect the fetus from cardiovascular compromise and hypoxia-ischaemia. This concept has been assumed for many years, but surprisingly hard to prove objectively (Parer, 1984). Our finding that the peripheral chemoreflex was still able to reduce FHR rapidly at the end of our UCO series, therefore, suggests that it continues to act to prolong fetal survival, albeit in a more limited capacity even after severe cardiovascular compromise is established.
The FHR decelerations during hypoxaemia probably have additional beneficial effects. There is evidence that parallel activation of sympathetic and parasympathetic systems has synergistic effects on myocardial function, enhancing adrenergic inotropic effects and, thereby, helping to sustain cardiac output despite bradycardia (Joyce & Wang, 2022;Koizumi et al., 1982;Paton et al., 2005). Consistent with this concept, in near-term fetal sheep exposed to moderate hypoxaemia, combined ventricular output does not fall in parallel with the fall in FHR (Cohn et al., 1980(Cohn et al., , 1982Itskovitz et al., 1991). Although, during severe hypoxaemia, combined ventricular output does ultimately fall after 2 min, this was not prevented by pretreatment with atropine sulphate to prevent bradycardia (Cohn et al., 1980). In the early phases of the present study, maintenance of ventricular output in the control group despite bradycardia might, in turn, explain why there was no further increase in MAP during UCOs after vagotomy and only a modest increase between UCOs despite marked tachycardia.
These studies support the overall concept that, consistent with our previous discussion, ventricular contractility rather than FHR per se appears to be the key determinant of combined ventricular output during cardiovascular adaptation and compromise attributable to severe hypoxaemia. Sympathetic stimulation also enhances the actions of acetylcholine via accentuated antagonism, helping to explain why bradycardia occurs consistently during fetal hypoxaemia despite the very large increase in adrenergic activity (Joyce & Wang, 2022;Levy, 1971). Additionally, parasympathetic activation might well have direct cardioprotective properties mediated via muscarinic activation and ATP-sensitive K + channels (Gourine & Gourine, 2014).
The onset of hypotension in control fetuses was associated with a parallel fall in CaBF (Fig. 6), indicating progressive cerebral hypoperfusion. During the first UCO, vagotomy increased CaBF and FBF. This is similar to our recent report during three 1 min UCOs repeated every 5 min (Lear et al., 2021a). This was, however, short lived in the present study, and for most UCOs the CaBF and FBF were not different from the control group. Thus, despite vagotomy impairing maintenance of MAP, CaBF was not impaired to a similar extent. The vagotomy group did show attenuation of the falls in CaVC during UCOs observed in control fetuses (Fig. 6), which might represent preferential vasodilatation of cerebral vasculature in the vagotomy group. This might have mitigated cerebral hypoperfusion during impaired arterial pressure after vagotomy.
Throughout the UCO series there was delayed suppression of EEG power during each UCO, which became greater with more severe hypotension (Fig. 7). This was associated with a switch towards lower spectral edge frequency (which is less metabolically demanding) from the first UCO onwards. These represent endogenous neuroprotective responses to reduce cerebral metabolism and EEG activity, and thus limit the severity of neural injury, mediated by local release of neuroinhibitory mediators, including adenosine (Hunter et al., 2003), GABA (Tan et al., 1996) and the neurosteroid allopregnanolone (Nguyen et al., 2004;Yawno et al., 2007). Furthermore, cortical impedance increased progressively during the UCO series, indicating progressive cortical cell swelling (Fig. 8). These neurophysiological adaptations to UCOs were not impaired or altered by vagotomy compared with intact control fetuses. This is consistent with the maintenance of CaBF in the vagotomy group despite impaired cardiovascular adaptation, highlighting that cerebral perfusion is strongly protected during hypoxaemia.

Alternative reflexes mediating fetal heart rate decelerations
Although our study was not designed specifically to investigate other reflex mechanisms that might contribute to decelerations, the present results help to shed light on their contributions. The baroreflex (which would trigger bradycardia with peripheral vasodilatation) is widely hypothesised to explain rapid variable decelerations during human labour and believed to be stimulated by repeated UCOs, resulting in a passive increase in total peripheral resistance from removal of the low-resistance placental bed (Ball & Parer, 1992). We have previously shown that the first 3−4 s of rapid UCOs was consistent with a brief but minor baroreflex-mediated fall in FHR, but thereafter the majority of the fall in FHR was consistent with peripheral chemoreflex activation and correlated with falling cerebral oxygenation as measured by near-infrared spectroscopy (Lear, Kasai, Booth, et al., 2020). Although we did not perform a similar analysis here, the pattern of rapid, intense femoral vasoconstriction throughout UCOs in the present study is not consistent with any meaningful baroreflex activation (Lear, Kasai, Booth, et al., 2020).
The Bezold-Jarisch reflex (triggering bradycardia, hypotension and peripheral vasodilatation) has also been hypothesised to be activated either by impaired ventricular preload secondary to UCO or as a result of severe fetal acidaemia and cardiovascular compromise (Frasch, 2021;Rosén & Mårtendal, 2015). As we and others have previously reported, ventricular preload appears well protected even during complete loss of umbilical venous return (Lear et al., 2021a(Lear et al., , 2021c. The pattern of increased IVCP during UCOs in the present study throughout the present UCO series is further evidence for protected ventricular preload. This is probably supported by intense peripheral vasoconstriction to reduce the size of the arterial compartment, which thus transfers blood volume to the venous compartment. Here, we also showed that peripheral vasodilatation was never observed during UCOs in the setting of fetal adaptation or compromise. Thus, the present findings provide compelling evidence that the Bezold-Jarisch reflex does not contribute to FHR decelerations or the pathophysiology of fetal cardiovascular compromise (Lear et al., 2021a).
Overall, the cardiovascular patterns observed throughout UCOs in the present study are highly consistent with the peripheral chemoreflex and also consistent with our previous reports (Lear et al., 2021a;Lear, Kasai, Booth, et al., 2020). These findings emphasise that the peripheral chemoreflex is the dominant initial mediator of decelerations during labour-like hypoxaemia Lear, Westgate et al., 2021;Ugwumadu & Arulkumaran, 2022), with a secondary contribution of myocardial hypoxia in the setting of fetal compromise.

Strengths and limitations
The major strength of the present study is the ability to monitor cardiovascular function in utero comprehensively, without confounding from anaesthesia (which impairs the FHR response to hypoxaemia) (Varcoe et al., 2020), allowing us to assess fetal responses that cannot be tested in humans. Contractions from the late first stage of human labour onwards occur at a rate of four to five per 10 min (broadly equivalent to one per 2.5 min) (Bakker et al., 2007;Caldeyro-Barcia & Poseiro, 1960). A systematic review showed that the median durations of the late first stage (from 5 cm cervical dilatation onwards) and second stage in nulliparous women are between 3.8 and 4.3 h and between 0.2 and 1.1 h, respectively (Abalos et al., 2018). Our experimental design of repeated UCOs at a frequency of four per 10 min for a mean of 2.7 h (controls) and 2.1 h (vagotomy) and maximum of 4 h is therefore consistent with human labour. The reader should consider that the highly structured protocol of repeated complete UCOs examined here does not reflect the heterogeneous nature of intrapartum contractions during human labour. Nevertheless, it provides a highly consistent degree and timing of hypoxaemia in both groups. The severity of hypotension targeted in this study is associated with cortical and subcortical neuronal death, focal cerebral infarctions (de Haan et al., 1997) and subendocardial injury (Gunn et al., 2000) and is therefore highly relevant to understanding the pathogenesis of intrapartum hypoxia-ischaemia.
An important limitation is that we do not have a direct, tissue-level measure of myocardial hypoxia. Furthermore, coronary artery blood flow increases markedly during fetal hypoxaemia (Court et al., 1984;Itskovitz et al., 1991;Jensen et al., 1987a;Thornburg & Reller, 1999), which is important in maintaining myocardial oxygen consumption and substrate supply during moderate hypoxaemia (Fisher et al., 1982). Increased coronary perfusion is attributable, in part, to peripheral chemoreflex-mediated parasympathetic stimulation, at least in adults (Feigl, 1969;Pen et al., 2022). This might be another mechanism through which vagal activation is cardioprotective. Nonetheless, sinoaortic denervation, which abolished bradycardia, did not impair the increase in coronary artery blood flow during moderate hypoxaemia in near-term fetal sheep (Itskovitz et al., 1991), emphasizing that other mechanisms, including β 2 -adrenergic stimulation and local adenosine and nitric oxide release, are also important (Thornburg & Reller, 1999). Conversely, vagotomy is likely to have increased the rate of myocardial oxygen consumption during UCOs by preventing bradycardia. The present study might therefore somewhat overestimate the magnitude of the contribution of myocardial hypoxia to FHR decelerations. Given these limitations, it is not possible to establish the exact times when myocardial hypoxia becomes important during decelerations.
In principle, vagotomy would also impair other reflexes, such as the baroreflex and Bezold-Jarisch reflex. Nevertheless, as discussed above, the present and previous studies strongly suggest that they are not important for adaptation to UCOs (Lear et al., , 2021a(Lear et al., , 2021bLear, Kasai, Booth, et al., 2020). Vagotomy also removes afferents from aortic chemoreceptors, which are activated during severe hypoxaemia in near-term fetal sheep (Jensen & Hanson, 1995). Our previous findings suggest that vagotomy might impair adrenal catecholamine release during brief UCOs J Physiol 601.10 (Cheung, 1990;, which might contribute to the attenuated fall in FVC during the first UCO in the present study. Thereafter, there was no difference in FVC between the groups, suggesting that any such effect was short lived.

Conclusions
The mechanisms of intrapartum decelerations have been controversial for decades, but the present findings should help to guide the physiological interpretation of intrapartum decelerations. Here, we show that the peripheral chemoreflex is the dominant mechanism triggering and sustaining FHR decelerations during the brief repeated hypoxaemia characteristic of labour, at a time when fetuses showed effective cardiovascular adaptation and maintenance of arterial pressure. In contrast, myocardial hypoxia contributed to decelerations in a delayed fashion and became more important with progressive fetal compromise. Even when the fetal cardiovascular adaptation was failing, with severe acidaemia and hypotension, decelerations were still faster in intact control fetuses, indicating that the peripheral chemoreflex was still being reactivated and probably still acting to prolong fetal survival during hypoxaemia. Nonetheless, during fetal compromise myocardial hypoxia progressively became the overall dominant regulator of FHR during decelerations and contributed to progressively deeper and later deceleration nadirs. Biomarkers of this transition towards a greater influence of myocardial hypoxia during decelerations would help to predict fetal compromise. Furthermore, our findings illustrate that impairment of peripheral chemoreflex-mediated decelerations accelerated the development of hypotension, failing ventricular contractility and metabolic acidaemia, providing evidence that intrapartum decelerations help to protect the fetus from cardiovascular compromise and hypoxia-ischaemia.