Alpha‐adrenergic mediated changes in blood pressure variability after hypoxia‐ischaemia in preterm fetal sheep

Abstract

Abstract

Brain injury in preterm infants is common but often difficult to detect in the early stages. The present study investigated whether evolving preterm brain injury after exposure to hypoxia‐ischaemia (HI) is associated with dynamic changes in arterial blood pressure variability (BPV) and tested the hypothesis that it was mediated by alpha‐adrenoreceptor activity. Chronically instrumented 0.7 gestation fetal sheep received sham‐HI (n = 8) or HI induced by 25 min of umbilical cord occlusion, followed by i.v. infusion of either the alpha‐adrenergic blocker phentolamine (n = 6) or saline (n = 8) from 15 min to 8 h post‐HI. HI was associated with acutely reduced beat‐to‐beat dispersion, sequential and spectral BPV measures, which normalised by 3 h post‐HI, followed by increased low‐frequency BPV from 3 h to 4 h. From 8 h to 13 h post‐HI, there was sustained reduction in all BPV measures throughout the secondary phase of injury. Alpha‐adrenoreceptor blockade abolished the increase in low‐frequency power during the latent phase and partly influenced secondary phase BPV. The HI group also showed greater 3 hourly reading‐to‐reading BPV during the first 24 h post‐HI, which was completely suppressed with alpha‐adrenoreceptor inhibition. Based on these initial findings, in a subsequent validation cohort [sham‐HI (n = 10) and HI‐saline (n = 10)] a short‐term BPV threshold had 90% sensitivity and 80% specificity for HI. This study highlights the evolving, phase‐dependent changes in BPV after HI, that they are partly mediated by alpha‐adrenoreceptor activation, and suggests that BPV could be a potential early biomarker for brain injury.

Key points

• Brain injury in preterm infants is common but often difficult to detect clinically, partly because of their normal low muscle tone and that seizures are typically electrographic.

• Blood pressure variation (BPV) is mediated mainly by autonomic activity, and so may be a marker for neural injury.

• In preterm fetal sheep, hypoxia‐ischaemia was associated with evolving, phase‐dependent changes in BPV.

• Changes in BPV were partly mediated by alpha‐adrenoreceptor activation.

• These findings support the concept that BPV could be a useful, pragmatic tool to help screen for brain injury.

Abstract figure legend Clinical detection of brain injury in preterm infantsremains a significant challenge. We investigated whether evolving brain injuryafter severe hypoxia‐ischaemia (HI) in chronically instrumented preterm fetalsheep is associated with dynamic changes in arterial blood pressure variability(BPV) and tested the hypothesis that post‐HI BPV changes are mediated byalpha‐adrenoreceptor activation. HI was associated with increased low‐frequencyBPV during the latent phase, followed by sustained reduction during thesecondary phase of injury. The HI group also had a greater reading‐to‐readingcoefficient of variation during the first day. Importantly, changes in thelatent phase beat‐to‐beat and short‐term BPV were suppressed by non‐selectivealpha‐adrenoceptor inhibition. These findings suggest that alpha‐adrenergicreceptor‐mediated phase‐dependent changes in BPV could serve as an earlybiomarker of brain injury. Created with Biorender.com.

Introduction

Preterm infants are at a significant risk of brain injury and severe, life‐long neurodevelopmental disabilities (Sarda et al., 2021). Exposure to hypoxia‐ischaemia is a significant contributor to the multifactorial aetiology of preterm brain injury, and rates of hypoxic‐ischaemic encephalopathy (HIE) in preterm infants are much higher than at term (Manuck et al., 2016). It is now well established that HI brain injury evolves over time, with transient recovery of mitochondrial function and metabolism during a latent phase, which lasts for ∼6–8 h after moderate to severe acute HI, followed by secondary deterioration of oxidative metabolism and bulk cell death lasting over ∼72 h (Dhillon et al., 2022). After this, there can be long‐term evolving injury and neuroregeneration/repair during a tertiary phase that lasts for weeks to months (Lear, Lear et al., 2023). A precise understanding of the stage of injury is key to developing potential interventions.

A major challenge is that in preterm infants the manifestations of brain injury can be relatively subtle and difficult to diagnose because their tone is normally much lower than at term, and seizures are frequently electrographic only (Gopagondanahalli et al., 2016; Pisani et al., 2020). HI injury can affect the integrity of the autonomic nervous system and so impair regulation of cardiovascular function (Andersen et al., 2019). Thus, heart rate variability has been assessed as a non‐invasive biomarker for HI injury in multiple preclinical and clinical studies (Beacom et al., 2025). It is moderately predictive of injury outcomes in term neonates with HIE, although the strength of evidence is low (Bersani et al., 2021). Similarly, fluctuations in blood pressure [blood pressure variability (BPV)] over defined time periods are effectively a composite variable that reflect interactions between haemodynamic, neural and humoral control and behavioural factors (Sheikh et al., 2023). In principle, it could provide an easily monitored biomarker of the stage of evolution of brain injury.

In small cohort studies of preterm neonates, increased BPV in the first few days after birth was associated with fluctuations in cerebral oxygenation and predicted the risk for developing brain injury associated with intraventricular haemorrhage (Jiang et al., 2023; Wong et al., 2012). In a cohort study in term neonates with HIE (n = 91), time and frequency domain measures of BPV from 18 h to 27 h after birth were additive with perinatal factors for predicting an adverse EEG profile at 48 h (Flower et al., 2022). Furthermore, there is evidence that BPV can have prognostic significance after ischaemic stroke and neurodegenerative disorders (Appiah et al., 2021; Ma et al., 2021; Xiao et al., 2023). However, the relationship between changes in BPV and the evolution of HI injury over time is unclear.

We have previously shown that moderate–severe HI in preterm fetal sheep is associated with characteristic changes in arterial blood pressure and blood flow in the peripheral and central beds and that alpha‐adrenergic receptor activation is a key mediator of changes in perfusion after HI (Dhillon et al., 2023; Quaedackers et al., 2004). In the present study, we investigated whether evolving brain injury after HI induced by 25 min of umbilical cord occlusion in 0.7 gestation preterm fetal sheep was associated with dynamic changes in BPV. We then tested the hypothesis that alpha‐adrenergic receptor activation contributes to changes in BPV after HI and examined the effect of non‐specific alpha blockade with phentolamine (Dhillon et al., 2023; Quaedackers et al., 2004).

Methods

Ethical approval

All procedures were approved by the Animal Ethics Committee of the University of Auckland following the New Zealand Animal Welfare Act 1999 and carried out in accordance with the code of Animal ethical conduct established by the Ministry of Primary Industries of New Zealand for the use of animals for teaching and research (AEC approval numbers 1942 and 22 069). The experiments are reported in accordance with the ARRIVE guidelines for reporting animal research (Percie du Sert et al., 2020).

Fetal surgery

Forty‐two Romney‐Suffolk cross fetal sheep were sourced from the University farm and instrumented on gestation days 99–100, as previously described (Bennet et al., 2018). For twin pregnancies, only one fetus was surgically instrumented. Ewes were acclimatised to laboratory conditions for 1 week before the surgery, during which time regular veterinary and welfare checks were performed. Food, but not water, was withdrawn 12–18 h before the surgery to reduce the risk of aspiration during surgery. Ewes were given an i.m. injection of the antibiotic oxytetracycline (20 mg kg⁻¹; Phoenix Pharm, Auckland, New Zealand), 30 min before surgery for prophylaxis. Anaesthesia was induced by an i.v. injection of propofol (5 mg kg⁻¹; AstraZeneca, Auckland, New Zealand) and maintained with 2%–3% isoflurane in oxygen after intubation. The depth of anaesthesia, maternal respiration and heart rate were constantly monitored during surgery by trained anaesthetic staff. Maternal fluid balance was maintained by a continuous i.v . infusion of plasma‐lyte 148 (∼250 mL h⁻¹) (Baxter, Auckland, New Zealand).

All surgical procedures were performed using aseptic techniques, as previously described (Bennet et al., 2018). A maternal midline incision was made to expose the uterus, and the fetus was partially exteriorised for instrumentation. A femoral and brachial artery, and one brachial vein were catheterised with saline‐filled polyvinyl catheters (SteriHealth, Dandenong South, VIC, Australia) for blood pressure measurement, pre‐ductal fetal blood sampling and drug infusion, respectively. An additional catheter was secured to the fetus to measure amniotic fluid pressure.

A pair of electrodes (AS633‐3SSF wire; Cooner Wire, Chatsworth, CA, USA) was placed s.c. across the fetal chest to measure the fetal ECG. Two pairs of electrodes (AS633‐7SSF; Cooner Wire) were placed through burr holes on the dura over the parietal parasagittal cortex bilaterally, 5 and 10 mm anterior, 5 mm lateral to the bregma to measure fetal EEG activity. The burr holes were sealed with surgical wax and electrodes fixed in place using cyanoacrylate glue. A reference electrode was placed over the occiput. EMG electrodes were sewn into the nuchal muscle about 10 mm apart. An inflatable silicone occluder (OC16HD, 16 mm; In Vivo Metric, Healdsburg, CA, USA) was loosely placed around the umbilical cord to allow post‐surgical occlusion to induce fetal HI.

At the end of surgery, fetal catheters were heparinised (20 U mL⁻¹ heparin in saline), the fetus was returned to the uterus, and the amniotic fluid lost during surgery was replaced with sterile 0.9% saline (∼500 mL at 39°C). The uterus was closed and the antibiotic gentamicin (80 mg; Pfizer, Auckland, New Zealand) was administered into the amniotic sac. All the fetal leads were exteriorised through the maternal flank. The maternal midline skin incision was repaired and infiltrated with a local analgesic: 10 mL of 0.5% bupivacaine plus adrenaline (AstraZeneca). A maternal long saphenous vein was catheterised for post‐operative care.

After surgery, animals were housed together in individual metabolic cages with access to concentrated pelleted food and water ad libitum. Rooms were temperature‐controlled (16 ± 1°C, humidity 50 ± 10%) under a 12:12 h light/dark photocycle, with lights on at 06:00 h. A period of 4–5 days of recovery was allowed before the commencement of experiments. Antibiotics were given i.v . to the ewe daily for 4 days; 600 mg of benzylpenicillin sodium (Novartis, Auckland, New Zealand) and 80 mg of gentamicin (Pfizer, Auckland, New Zealand). Fetal and maternal vascular catheters were maintained patent by continuous infusion of heparinised saline (20 U mL⁻¹ at a rate of 0.2 mL h⁻¹). The fetal condition was assessed via continuous computer recordings of all fetal physiological variables (LabVIEW for Windows, National Instruments, Austin, TX, USA) and daily fetal arterial samples were taken to monitor pH and blood gases (ABL800 Flex analyser; Radiometer, Auckland, New Zealand), glucose and lactate (YSI 2300 Analyser; YSI Ltd, Yellow Springs, Ohio, USA).

Physiological recordings

Fetal mean arterial blood pressure (MAP) (Novatrans II, MX860; Medex, Hilliard, OH, USA), corrected for maternal movement by subtraction of amniotic fluid pressure, fetal heart rate (FHR) derived from the ECG and EEG were recorded continuously from 24 h before until 72 h after HI. Data were stored for offline analysis using custom data acquisition software (LabVIEW for Windows). All pressure signals were low‐pass filtered with a fifth‐order Butterworth filter with a cutoff frequency at 20 Hz and then digitised at a sampling rate of 512 Hz. The raw ECG signal was filtered with an analogue first‐order high‐pass filter with a cutoff frequency set at 0.05 Hz and a fifth‐order low‐pass Bessel filter with a cutoff at 100 Hz and digitised at a sampling rate of 1024 Hz. R‐R intervals were extracted from this signal to calculate heart rate.

The EEG signals were amplified 10,000× and then processed with a first‐order high‐pass filter at 1.6 Hz and an analogue fifth‐order low‐pass Butterworth filter with a cutoff frequency set at 500 Hz and digitised at 4096 Hz. The signal was then filtered by a low‐pass filter with a digital IIR type 2 Chebyshev filter with a cutoff frequency of 256 Hz for analysis of raw EEG waveforms for seizures. The real‐time intensity spectra and associated parameters were extracted from 4 s epochs. Total EEG power (in microvolts squared (µV²)) was calculated on the power spectrum between 1 and 20 Hz. For data presentation, the EEG power was log‐transformed [EEG power (dB), 10 × log (power)] to give a better approximation of a normal distribution (Williams & Gluckman, 1990). The nuchal EMG signal was amplified 4000× and filtered with a sixth‐order low‐pass Butterworth filter, with a cutoff frequency of 2 kHz. Signals were band‐pass filtered between 100 and 1000 Hz and integrated using a time constant of 0.1 s and digitised at 512 Hz. The data were stored as 1 min averages.

Experimental design

Experiments were conducted at 104–105 days of gestation. The study groups are shown in Fig. 1. Fetuses were randomly assigned to the following groups: sham‐HI (n = 18), HI‐saline (n = 18) or HI‐phentolamine (n = 6). A subset of sham‐HI (n = 8) and HI‐saline (n = 8) animal groups were included in the primary BPV analysis (Primary group) and the other subset (sham‐HI (n = 10) and HI‐saline (n = 10)) were used for validation analysis (Validation group). Sham occlusion fetuses received no occlusion. HI was induced by inflating the umbilical cord occluder with a known volume of saline to completely occlude the cord for 25 min. The occluder was completely deflated at the end of occlusion period. This duration of occlusion in 0.7 gestation fetal sheep results in severe cerebral HI, resulting in moderate subcortical neuronal loss and diffuse white matter injury in periventricular and intragyral areas (Dhillon et al., 2023; Wassink et al., 2017). All of the occlusions were undertaken between 09:00 h and 09:30 h. Fetuses either received a continuous intravenous infusion of saline or the non‐selective alpha‐adrenergic antagonist phentolamine (10 mg loading bolus, followed by continuous infusion at 10 mg h⁻¹) from 15 min to 8 h post‐HI. Fetal arterial blood was taken at 30 min prior to umbilical cord occlusion, at 5 and 17 min during occlusion, and then at 1, 2, 4, 6, 24, 48 and 72 h after occlusion for pH and blood gas determination and for glucose and lactate measurements. 72 h after umbilical cord occlusion, ewes and fetuses were killed with an overdose of pentobarbitone sodium, i.v. administered to the ewe (9 g; Pentobarb 300; Chemstock International, Christchurch, New Zealand). This method is consistent with the Animal Welfare Act of New Zealand. Fetuses were delivered by Caesarean section and fetal weight and sex were determined, and then fetal organ weights measured.

Figure 1. Study design flowchart.

Flowchart showing the animal groups analysed to investigate the mechanisms underlying blood pressure variability after HI and the separate validation group to assess the reproducibility of the primary group analysis.

Data analysis

All physiological analysis was performed blinded to the experimental group using codes for experimental protocols. Offline analysis of fetal physiological parameters was performed with customised LabVIEW‐based programs. Data were assessed as hourly averages, with the 24 h period before the experiment used as a reference to evaluate acute physiological changes associated with the experiment. Log transformed EEG power was normalised to baseline for analysis. EEG seizures were identified visually on the raw EEG. Seizure activity was defined as the appearance of sudden, repetitive, evolving stereotypic waveforms, lasting more than 10 s with an amplitude greater than 20 µV (Bennet et al., 2018; Lear et al., 2025).

For beat‐to‐beat BPV analysis, systolic arterial blood pressure peak amplitudes were extracted using LabVIEW software and imported into a customised physiological data analysis toolbox using MATLAB (MATLAB 2023A; The MathWorks, Inc., Natick, MA, USA) to calculate features of beat‐to‐beat variability. This toolbox was created by one of the co‐authors (Michael J. Beacom). All metrics were assessed in continuous, non‐overlapping 1 min windows unless otherwise specified. Dispersion measures were calculated as SD and coefficient of variation of systolic pressure peak amplitude, and the sequential measures were calculated as average real variability and successive variation in systolic pressure (Parati et al., 2018). Frequency‐domain measures were obtained using the Lomb–Scargle periodogram, selected because of the irregular nature of the raw blood pressure signals. Unlike traditional periodograms, which assume evenly sampled time series, the Lomb–Scargle method can accurately analyse frequency content in unevenly sampled datasets (Laguna et al., 1998; Scargle, 1982). The frequency‐domain boundaries used in this study were adapted from standard fetal heart rate variability bands: very‐low frequency 0.0033–0.04 Hz, low frequency 0.04–0.15 Hz, high frequency 0.15–0.4 Hz and very‐high frequency 0.4–1.5 Hz.

For short‐term BPV analysis, 1 min systolic blood pressure readings taken every 3 h were exported (readings taken at 1, 4, 7, 10, 13, 16, 19, 22 and 25 h post‐HI, that is, 10.00 h, 13.00 h, 16.00 h, 19.00 h, and 20.00 h on the day of occlusion, and 01.00 h, 04.00 h, 07.00 h and 10.00 h the following day). The dispersion and sequential changes between these readings were then calculated within each 24 h interval. Additionally, for comparison, BPV was also evaluated over 24 h periods using 3 h averaged systolic blood pressure values. Using the BPV between the consecutive 3 hourly readings, a threshold was established at 1 SD above the mean baseline coefficient of variation within the primary analysis group (Fig. 1). A blinded observer (Simerdeep K. Dhillon) applied this coefficient of variation threshold to identify animals with HI injury in the validation group. From this, the sensitivity, specificity and accuracy of the systolic pressure coefficient of variation as a biomarker to predict HI injury was evaluated.

Statistical analysis

Statistical analysis was performed using SPSS, version 25 (IBM Corp., Armonk, NY, USA). There was no mortality in the experimental groups. For the continuously recorded physiological data and fetal biochemical parameters, between group comparisons were performed by two‐factor mixed‐methods ANOVA, with time as repeated measures and group as independent variables. Post hoc tests were performed when a significant overall effect of group or interaction between group and time was found. Between and within‐group comparisons were performed by univariate analysis with Tukey's post hoc test. P < 0.05 was considered statistically significanct. Data are presented as the mean ± SD.

Results

Baseline parameters

The sex distribution was comparable between groups (female:male ratio: primary analysis dataset – Sham‐HI 4:4 and HI‐saline 4:4; validation dataset – Sham‐HI 6:4 and HI‐saline 6:4, HI‐phentolamine 4:2). The distribution of singleton and twin pregnancies was also similar between groups, with no twins in the primary analysis dataset and one twin pregnancy recorded in both the Sham‐HI and HI‐saline groups in the validation dataset. There was no difference in postmortem body weights between the groups. During the pre‐HI baseline, there was no difference between the groups in any physiological parameters.

Cardiovascular and biochemical parameters during HI

HI was associated with immediate bradycardia [average nadir FHR: primary analysis dataset – Sham‐HI 191.4 ± 10.9 beats min⁻¹ and HI‐saline 62.0 ± 7.7 beats min⁻¹; validation dataset – Sham‐HI 186.5 ± 6.8 beats min⁻¹ and HI‐saline 66.6 ± 11.1 beats min⁻¹, HI‐phentolamine 65.6 ± 15.2 beats min⁻¹], progressive hypotension [average nadir of blood pressure: primary analysis dataset – Sham‐HI 37.1 ± 3.1 mmHg and HI‐saline 13.4 ± 3.1 mmHg; validation dataset – Sham‐HI 36.4 ± 1.8 mmHg and HI‐saline 11.2 ± 3.5 mmHg, HI‐phentolamine 12.6 ± 3.0 mmHg], hypoxia, hypercapnia and mixed acidosis with no difference between the HI groups (Fig. 2 and Table 1). The time course of post‐HI recovery of blood gases and acid‐base status after HI was comparable between the groups. Glucose and lactate concentrations were higher in the HI‐saline group than the sham‐HI group until 6 h post‐HI but recovered to baseline levels by 2 h in the HI‐phentolamine group (Table 1).

Figure 2. Cardiovascular changes during umbilical cord occlusion.

Time sequence of changes in mean arterial pressure (mmHg, top) and fetal heart rate (beats min⁻¹, bottom) for 30 min pre‐occlusion baseline, 25 min of umbilical cord occlusion and 30 min of post‐HI recovery in the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. The shaded area shows the period of occlusion. Data are 1 min averages presented as the mean ± SD (SD shown as bars) and were analysed using mixed‐design ANOVA with time as a repeated measure and HI and phentolamine as independent factors. Between group comparisons were performed using Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05, (b) sham‐HI vs. HI‐phentolamine P < 0.05.

Table 1.

Fetal arterial blood pH, partial pressures of oxygen and carbon dioxide, glucose and lactate values at 1 h before 25 min occlusion, 5 and 17 min during occlusion, and 1, 2, 4, 6, 24, 48 and 72 h after the end of occlusion in sham‐HI (n = 8), HI‐saline (n = 8) and HI‐phentolamine (n = 6) groups.

	Baseline	5 min	17 min	1 hour	2 h	4 h	6 h	24 h	48 h	72 h
pH
Sham‐HI	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0	7.37 ± 0.0
HI‐saline	7.38 ± 0.0	7.04 ± 0.0a	6.85 ± 0.0a	7.29 ± 0.0	7.33 ± 0.1	7.39 ± 0.0	7.39 ± 0.0	7.36 ± 0.0	7.36 ± 0.0	7.35 ± 0.0
HI‐phentolamine	7.39 ± 0.0	7.09 ± 0.1b	6.89 ± 0.0b	7.29 ± 0.0	7.39 ± 0.0	7.40 ± 0.0	7.39 ± 0.0	7.38 ± 0.0	7.39 ± 0.0	7.38 ± 0.0
$P_{{\mathrm{O}}_2}$ (mmHg)
Sham‐HI	25.0 ± 3.3	24.5 ± 3.0	24.2 ± 2.2	24.6 ± 1.8	23.5 ± 2.7	24.1 ± 2.7	24.9 ± 4.0	23.7 ± 3.1	24.3 ± 3.0	23.5 ± 3.9
HI‐saline	24.3 ± 3.6	8.4 ± 1.6a	11.1 ± 1.7a	28.5 ± 5.5	25.4 ± 4.1	23.0 ± 5.1	23.3 ± 5.0	25.9 ± 5.7	27.0 ± 5.6	27.0 ± 5.6
HI‐phentolamine	24.8 ± 2.0	8.0 ± 2.0b	10.0 ± 1.8b	24.5 ± 4.3	22.3 ± 3.3	24.5 ± 2.8	25.8 ± 3.9	27.3 ± 3.5	28.2 ± 2.6	27.6 ± 3.2
$P_{\mathrm{C}{\mathrm{O}}_2}$ (mmHg)
Sham‐HI	47.6 ± 1.9	47.5 ± 4.2	45.5 ± 3.2	48.1 ± 2.8	50.0 ± 2.6	49.3 ± 2.9	50.1 ± 2.6	50.3 ± 2.6	49.7 ± 3.9	49.4 ± 3.8
HI‐saline	49.3 ± 3.9	100.1 ± 8.4a	133.7 ± 9.4a	51.2 ± 9.4	51.5 ± 8.5	47.6 ± 5.5	49.4 ± 4.5	48.3 ± 3.5	46.3 ± 5.0	49.1 ± 4.8
HI‐phentolamine	47.7 ± 6.3	93.6 ± 7.6b	132.8 ± 7.6b	46.4 ± 4.8	46.8 ± 5.5	46.9 ± 4.0	46.6 ± 4.4	46.1 ± 3.6	46.3 ± 2.5	44.8 ± 3.2
Glucose (mmol L−1)
Sham‐HI	1.0 ± 0.4	1.2 ± 0.3	1.1 ± 0.3	1.1 ± 0.3	1.1 ± 0.3	1.1 ± 0.3	1.2 ± 0.3	1.2 ± 0.2	1.2 ± 0.2	1.1 ± 0.2
HI‐saline	1.0 ± 0.2	0.3 ± 0.1a	0.7 ± 0.2a	1.3 ± 0.3	1.4 ± 0.4a	1.4 ± 0.3a	1.4 ± 0.2	1.1 ± 0.2	1.2 ± 0.2	1.1 ± 0.2
HI‐phentolamine	1.0 ± 0.3	0.3 ± 0.1b	0.6 ± 0.3b	1.5 ± 0.7	1.0 ± 0.1c	1.0 ± 0.1c	1.0 ± 0.1	1.1 ± 0.3	1.2 ± 0.2	1.2 ± 0.3
Lactate (mmol L−1)
Sham‐HI	0.7 ± 0.2	0.8 ± 0.1	0.9 ± 0.2	0.9 ± 0.2	0.8 ± 0.2	0.9 ± 0.2	0.9 ± 0.2	0.8 ± 0.1	0.8 ± 0.2	0.8 ± 0.2
HI‐saline	0.9 ± 0.3	4.3 ± 0.9a	7.1 ± 0.9a	4.4 ± 1.6a	4.1 ± 2.4a	2.9 ± 1.9a	2.5 ± 1.6a	1.1 ± 0.6	0.9 ± 0.3	0.8 ± 0.2
HI‐phentolamine	0.6 ± 0.2	3.7 ± 0.5b	6.1 ± 0.5b	3.1 ± 1.1b	1.6 ± 0.8c	1.0 ± 0.8c	1.1 ± 0.7c	0.7 ± 0.2	0.7 ± 0.2	0.7 ± 0.1

Data are presented as the mean ± SD. Between group comparisons are shown as (a) P < 0.05 Sham‐HI vs. HI‐saline, (b) P < 0.05 HI‐saline vs. HI‐phentolamine and (c) P < 0.05 HI‐saline vs. HI‐phentolamine group.

Post‐occlusion cardiovascular recovery

HI was associated with transient tachycardia from 3 h to 4 h post‐HI, followed by a reduction in heart rate from 12 h to 24 h. There was both a significant main effect of time (P = 0.001) and a time and group interaction (P = 0.001), indicating that the pattern of change differed between the groups. However, post hoc comparisons showed no significant differences in heart rate between the groups at any individual time point (Fig. 3). There was a moderate increase in mean arterial pressure from 1 to 2 h post‐HI (P = 0.016, Sham‐HI vs. HI‐saline) (Fig. 3). Phentolamine infusion was associated with lower MAP during early recovery (P = 0.001, HI‐saline vs. HI‐phentolamine, 1–3 h post‐HI) and from 12 h to 24 h after HI (P = 0.019).

Figure 3. Cardiovascular, EEG and Nuchal EMG changes after umbilical cord occlusion.

Time sequence of changes in fetal heart rate (beats min⁻¹) (A), mean arterial pressure (mmHg) (B), EEG power (ΔdB) (C) and nuchal EMG (µV) (D) during 24 h before umbilical cord occlusion and 72 h of post‐HI recovery in the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. Data during umbilical cord occlusion are not shown. The shaded area shows the period of latent phase (1–6 h). Data are hourly averages presented as the mean ± SD (SD shown as bars) and were analysed using a mixed‐design ANOVA with time as a repeated measure and HI and phentolamine as independent factors. Between group comparisons were performed using the Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05, (b) sham‐HI vs. HI‐phentolamine P < 0.05 and (c) HI‐saline vs. HI‐phentolamine P < 0.05.

Post‐occlusion neurophysiological and EMG recovery

HI was associated with reduced EEG power throughout the recovery period with no difference between the occlusion groups (P = 0.001, Sham‐HI vs. HI‐saline; P = 0.002, Sham‐HI vs. HI‐phentolamine) (Fig. 3). There was a complex pattern of nuchal EMG activity recovery post‐HI, with suppression during the first hour, recovery to sham‐HI level between 2 h and 5 h in the HI‐saline group. The recovery in the HI‐phentolamine group was slower but there was no significant difference between groups between 3 h and 5 h post‐HI (P = 0.325 Sham‐HI vs. HI‐saline, P = 0.077 Sham‐HI vs. HI‐phentolamine). This was followed by a secondary reduction from 6 h to 12 h in both the HI groups (P = 0.015 Sham‐HI vs. HI‐saline, P = 0.001 Sham‐HI vs. HI‐phentolamine) (Fig. 3). There was no significant difference between the HI groups.

In both HI groups, high‐amplitude stereotypic seizures developed 7–10 h after HI. There was no difference between the occlusion groups in the average onset time (HI‐saline 13.3 ± 9.2 h; HI‐phentolamine 7.3 ± 2.4 h), total time having seizures (HI‐saline 30.2 ± 13.1 h; HI‐phentolamine 31.8 ± 17.3 h), average seizure burden (HI‐saline 172.4 ± 36.8 seizures h⁻¹; HI‐phentolamine 146.1 ± 47.9 seizures h⁻¹), seizure amplitude (HI‐saline 161.3 ± 37.6 µV, HI‐phentolamine 140.7 ± 39.7 µV), duration (HI‐saline 93.9 ± 25.0 s vs. HI‐phentolamine 74.3 ± 27.0 s) and total seizure count (HI‐saline 58.8 ± 28.1 vs, HI‐phentolamine 66.3 ± 43.4).

Post‐occlusion beat‐to‐beat blood pressure variability

Latent phase (1–6 h post‐HI)

The HI‐saline group showed significant changes over time in measures of beat‐to‐beat dispersion, sequential and spectral systolic BPV. There was an acute reduction in the SD (P = 0.000), coefficient of variation (P = 0.000), average real variability (P = 0.001), successive variation (P = 0.001), low‐frequency (P = 0.000), high‐frequency (P = 0.003) and very‐high frequency (P = 0.002) absolute power compared to controls during the first 2 h after HI (Figs 4 and 5). After the initial suppression, low‐frequency power transiently increased above control levels between 3 h and 4 h post‐HI (P = 0.044) (Fig. 5), whereas all other measures recovered to sham‐HI levels from 3 h to 6 h.

Figure 4. Changes in dispersion and sequential measures of blood pressure variability after umbilical cord occlusion.

Time sequence of changes in beat‐to‐beat standard deviation (mmHg) (A), coefficient of variation (B), average real variability (mmHg) (C) and successive variation (D) of arterial systolic pressure during 24 h before umbilical cord occlusion and 72 h of post‐HI recovery in the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. Data during umbilical cord occlusion are not shown here. The shaded area shows the period of latent phase (1–6 h). Data are hourly averages presented as the mean ± SD (SD shown as bars) and were analysed using mixed‐design ANOVA with time as a repeated measure and HI and phentolamine as independent factors. Between group comparisons were performed using the Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05, (b) sham‐HI vs. HI‐phentolamine P < 0.05 and (c) HI‐saline vs. HI‐phentolamine P < 0.05.

Figure 5. Changes in spectral measures of systolic blood pressure variability after umbilical cord occlusion.

Time sequence of changes in beat‐to‐beat absolute very‐low frequency (A), low‐frequency (B), high‐frequency (C) and very‐high frequency (D) arterial systolic pressure variability during 24 h before umbilical cord occlusion and 72 h of post‐HI recovery in the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. Data during umbilical cord occlusion are not shown. The shaded area shows the period of latent phase (1–6 h). Data are hourly averages presented as the mean ± SD (SD shown as bars) and were analysed using mixed‐design ANOVA with time as a repeated measure and HI and phentolamine as independent factors. Between group comparisons were performed using the Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05, (b) sham‐HI vs. HI‐phentolamine P < 0.05 and (c) HI‐saline vs. HI‐phentolamine P < 0.05.

Alpha‐adrenergic receptor blockade with phentolamine did not significantly affect dispersion and sequential BPV (Fig. 4). However, phentolamine was associated with frequency specific effects in spectral BPV measures (Fig. 5), particularly a sustained reduction in very‐low frequency power during the latent phase (P = 0.045 Sham‐HI vs. HI‐phentolamine, P = 0.034 HI‐saline vs. HI‐phentolamine) with complete suppression of the acute rise in low‐frequency power from 3 h to 4 h post‐HI (P = 0.014 HI‐saline vs. HI‐phentolamine).

Secondary phase (7–72 h post‐HI)

The transient recovery of beat‐to‐beat BPV measures in the latent phase was followed by a marked delayed reduction starting between 8 h and 13 h post‐HI (Figs  4 and 5). There was sustained suppression of SD (P = 0.019), coefficient of variation (P = 0.035) from 8 h to 60 h, average real variability (P = 0.043), successive variation (P = 0.021), high‐frequency power (P = 0.034) and very‐high frequency power (P = 0.033) throughout the recovery phase from 7 h to 72 h. By contrast, there was only a transient reduction in low‐frequency power from 23 h to 42 h (P = 0.042) and no change in very‐low frequency power (P = 0.842). A similar pattern of acute reduction, transient recovery in the latent phase, followed by a secondary decrease was also observed in the dispersion, sequential and spectral diastolic BPV measures after HI (Figs 6 and 7).

Figure 6. Changes in dispersion and sequential measures of diastolic blood pressure variability after umbilical cord occlusion.

Time sequence of changes in beat‐to‐beat standard deviation (mmHg) (A), coefficient of variation (B), average real variability (mmHg) (C) and successive variation (D) of arterial diastolic pressure during 24 h before umbilical cord occlusion and 72 h of post‐HI recovery in the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. Data during umbilical cord occlusion are not shown here. The shaded area shows the period of latent phase (1–6 h). Data are hourly averages presented as the mean ± SD (SD shown as bars) and were analysed using mixed‐design ANOVA with time as a repeated measure and HI and phentolamine as independent factors. Between group comparisons were performed using the Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05 and (b) sham‐HI vs. HI‐phentolamine P < 0.05.

Figure 7. Changes in spectral measures of diastolic blood pressure variability after umbilical cord occlusion.

Time sequence of changes in beat‐to‐beat absolute very‐low frequency (A), low‐frequency (B), high‐frequency (C) and very‐high frequency (D) arterial diastolic pressure variability during 24 h before umbilical cord occlusion and 72 h of post‐HI recovery in the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. Data during umbilical cord occlusion are not shown here. The shaded area shows the period of latent phase (1–6 h). Data are hourly averages presented as the mean ± SD (SD shown as bars) and were analysed using mixed‐design ANOVA with time as a repeated measure and HI and phentolamine as independent factors. Between group comparisons were performed using the Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05, (b) sham‐HI vs. HI‐phentolamine P < 0.05 and (c) HI‐saline vs. HI‐phentolamine P < 0.05.

A broadly similar delayed reduction in BPV measures during the secondary phase was observed in the HI‐phentolamine group. However, alpha‐adrenergic receptor inhibition was associated with an earlier decline in low‐frequency power (P = 0.021 HI‐saline vs. HI‐phentolamine, 13 h to 18 h post‐HI) and a greater reduction in SD (P = 0.039 HI‐saline vs. HI‐phentolamine) and coefficient of variation (P = 0.045 HI‐saline vs. HI‐phentolamine) of systolic blood pressure from 13 h to 20 h post‐HI.

Post‐occlusion short‐term (24 h) blood pressure variability

HI injury was associated with an increased systolic BPV between consecutive readings taken every 3 h during the first 2 days following the injury (Fig. 8). The HI group had higher SD (P = 0.029), coefficient of variation (P = 0.030), successive variation (P = 0.000) and variation independent of the mean (P = 0.001) compared to sham‐HI during the first 24 h post‐occlusion. Phentolamine infusion completely inhibited the increase in short‐term BPV measures after HI. When assessed using 3 h averaged systolic pressure values, a comparable pattern of dispersion and sequential BPV changes was observed, although with a relatively smaller magnitude (Fig. 9), confirming that consecutive systolic blood pressure measurements every 3 h offer a reliable assessment of BPV.

Figure 8. Short‐term (24 h) systolic pressure variability in consecutive 3‐ hourly systolic blood pressure recordings after umbilical cord occlusion.

Daily standard deviation (A), coefficient of variation (B), successive variation (C) and variation independent of the mean (D) of consecutive 3 hourly systolic pressure readings during 24 h of baseline before umbilical cord occlusion and 3 days post‐HI recovery the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. Data are presented as individual animals (the central bar mean ± SD) and analysed using mixed design two‐way ANOVA with time as repeated measure and HI and phentolamine as independent variables. Comparisons between the groups were performed using the Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05 and (c) HI‐saline vs. HI‐phentolamine.

Figure 9. Short‐term (24 h) systolic pressure variability in 3 h averaged systolic blood pressure recordings after umbilical cord occlusion.

Daily standard deviation (A), coefficient of variation (B), successive variation (C) and variation independent of the mean (D) of 3 hourly averages of systolic pressure recordings during 24 h of baseline before umbilical cord occlusion and 3 days post‐HI recovery the sham‐HI (black circles, n = 8), HI‐saline (red triangles, n = 8) and HI‐phentolamine (blue squares, n = 6) groups. Data are presented as individual animals (the central bar mean ± SD) and analysed using mixed design two‐way ANOVA with time as repeated measure and HI and phentolamine as independent variables. Comparisons between the groups were performed using the Tukey's post hoc test. Figure symbols: (a) sham‐HI vs. HI‐saline P < 0.05 and (b) sham‐HI vs. HI‐phentolamine.

Viability of short BPV as a biomarker for severe HI

A threshold was set at one standard deviation above the average baseline coefficient of variation between 3 hourly consecutive systolic blood pressure readings in the primary analysis subgroup (0.043) as a first pass statistical measure of change from the baseline variability. In the primary analysis group, the coefficient of variation in the HI‐saline group was greater than one standard deviation above that observed in the sham‐HI group during the first 24 h post‐HI. This confirmed that the threshold was a reasonable approach for distinguishing variability associated with HI injury from normal physiological fluctuation. Applying this set threshold, in the validation dataset animals with severe HI were identified with 90% sensitivity and 80% specificity (Fig. 10).

Figure 10. Coefficient of variation 24 h after umbilical cord occlusion in the validation subgroup.

Coefficient of variation of 3 hourly systolic pressure readings during 24 h of post‐HI recovery the validation subgroup sham‐HI (black circles, n = 10) and HI‐saline (red triangles, n = 10). Data points are individual fetuses. The horizontal bar represents a +1 SD threshold coefficient of variation for predicting the presence of severe HI injury.

Discussion

The present study demonstrates that moderate to severe HI in preterm fetal sheep is associated with complex changes over time in the variability of arterial systolic pressure. There was a transient early reduction in variation followed by recovery of most BPV measures to pre‐HI baseline values after the first 2 h. Interestingly, there was a distinct increase in low‐frequency BPV during the latent phase. Infusion of the non‐selective alpha‐adrenergic antagonist phentolamine after HI abolished the acute increase in low‐frequency power with a reduction in very‐low frequency power. These data support the hypothesis that low‐frequency changes in beat‐to‐beat BPV during the latent phase are primarily mediated by alpha‐adrenergic receptor activation. During the secondary phase, there was a delayed and sustained reduction in dispersion, as well as in sequential and spectral measures of beat‐to‐beat BPV, and alpha‐adrenoceptor inhibition exerted a partial influence on these changes. Additionally, there was an overall increase in dispersion and sequential BPV between 3 hourly consecutive systolic blood pressure readings during the first day after HI, which was mediated by alpha‐adrenergic receptor activation. Importantly, a +1 SD threshold for short‐term variability was effective in identifying the HI group during the first 24 h of recovery. Collectively, these findings suggest that short‐term and beat‐to‐beat measures have the potential to be useful biomarkers of severe HI injury and to help identify the stage of evolution of injury.

Beat‐to‐beat BPV is determined by modulation of vascular tone by neuronal and humoral factors (Bakkar et al., 2021) and probably by fetal behaviours such as body movements (Dalton et al., 1977). The early transient suppression of beat‐to‐beat BPV for 1–2 h post‐HI coincided with suppression of EEG power and fetal body movements. This finding is consistent with our previous reports of acute reduction in fetal heart rate variability shortly after HI (Kasai et al., 2019; Lear, Maeda et al., 2023). This suggests that the early reduction in BPV is primarily a function of the overall suppression of brain activity, and autonomic outflow immediately after profound HI. By contrast, the secondary endogenous neuroinhibition after HI is actively mediated by factors such as the upregulation of neurosteroids and sympathetic nervous system activation and is neuroprotective (Dhillon et al., 2023; Yawno et al., 2007).

During the latter part of the latent phase from 3 h to 6 h post‐HI, even though fetal EEG activity remained suppressed, most BPV measures recovered to sham HI levels. Interestingly, during this phase, there was increased low‐frequency BPV. The present study demonstrates that alpha‐adrenergic receptor activation mediates the increase in low‐frequency power during the latent phase. These findings are supported by the evidence from studies in adult animals and humans showing that changes in vascular tone elicited by sympathetic tone modulation mainly affect low‐frequency BPV (Stauss et al., 1997, 1998). Phentolamine is a non‐specific alpha‐adrenergic receptor antagonist, and so the present study did not investigate the relative contribution of adrenoreceptor subtypes. Responsiveness of postjunctional alpha‐1 and 2 adrenergic receptors to sympathetic activity is considered a major and direct determinant of blood pressure fluctuations (Dinenno et al., 2002; Guerrero et al., 2025). By contrast, central alpha‐2 receptors may exert an indirect influence by modulating the activity of sympathetic postganglionic neurons (Klassen et al., 2024).

Notably, low‐frequency power did not decrease below sham‐HI levels during the phentolamine infusion. This suggests that non‐neuronal factors such as vascular myogenic tone, endothelial‐derived factors and the renin‐angiotensin system also contribute to changes in vasomotor activity (Lohman et al., 2024; Stauss, 2007). Plasma levels of pressor hormones are elevated after severe HI in fetal sheep (Lear et al., 2020; Lumbers et al., 2001; Summanen et al., 2017). We also observed an overall increase in high‐frequency components of BPV during the later latent phase. High‐frequency BPV can be specifically modulated by endothelial‐dependent nitric oxide synthesis (Nafz et al., 1997), which is suppressed during the latent phase after HI (Barrett et al., 2014).

In the present study, low‐frequency power showed the strongest correlation between BPV and heart rate variability (HRV), presumptively mediated by common modulation by baroreflex and autonomic tone (Goldstein et al., 2011; Yoshino & Matsuoka, 2005).  It should be noted that the magnitude of change and temporal profile of low‐frequency BPV observed in the present study were very different from those of low‐frequency HRV previously reported in this paradigm (Lear, Maeda et al., 2023). This suggests that HRV probably only partially explains the fluctuations in low‐frequency BPV. Further investigation of the coherence between HRV and BPV will be needed to assess whether it may provide a more comprehensive measure of autonomic dysregulation after HI. Pragmatically, the complexity of changes in BPV during the latent phase probably limits their viability as a biomarker for identifying the latent phase of injury.

Subsequent delayed, progressive falls in sequential and frequency measures of beat‐to‐beat BPV were associated with sustained suppression of EEG and nuchal EMG activity. The timing of the onset of this delayed reduction in BPV measures corresponds with the phase of secondary deterioration of oxidative metabolism and bulk cell loss after severe HI in the present paradigm (Bennet et al., 2006; Dhillon et al., 2022). Most of the changes in BPV during the secondary phase occur after the end of phentolamine infusion in the present study. However, alpha‐adrenoreceptor inhibition was associated with either an earlier or a more rapid decline in specific BPV measures during the secondary phase. Previously, we demonstrated using near‐infrared spectroscopy that phentolamine infusion following HI accelerates the progression of secondary mitochondrial failure and is associated with a more severe neural injury (Dhillon et al., 2023). Previous studies employing the same model as that used in the present study have demonstrated that there is also profound suppression of FHRV measures during the secondary phase of injury (Lear, Maeda et al., 2023; Yamaguchi et al., 2018). Overall, low beat‐to‐beat variability in both heart rate and blood pressure suggest reduced autonomic outflow and altered baroreflex sensitivity (Tian et al., 2019), probably related to evolving brainstem injury in this phase (George et al., 2004). Speculatively, impaired autonomic flexibility during the secondary phase may affect responsiveness to subsequent cardiovascular challenges.

A sustained reduction in beat‐to‐beat BPV could be a potential marker for the progression of severe injury to the secondary phase. In principle, such secondary phase biomarkers might help to identify infants who might benefit from delayed treatments to prevent tertiary phase injury (Lear et al., 2024). Limited clinical data in term neonates with moderate‐severe HIE also shows an association between lower variability in time and frequency domain measures of BPV during 18–27 h after birth and abnormal EEG outcomes (Flower et al., 2022). However, the need for continuous arterial blood pressure monitoring may limit its applicability as a diagnostic marker because continuous blood pressure recordings in neonates are typically acquired using indwelling arterial catheters and pose a risk of infection (Rao et al., 2023). Continuous non‐invasive blood pressure monitoring using photoplethysmography, machine learning and wearable sensors is currently being investigated for neonates (Baker et al., 2023).

To assess whether intermittent measurements of variability may be practical, to avoid needing continuous blood pressure monitoring at high data acquisition rates, we also analysed short‐term BPV by examining changes between consecutive systolic pressure readings taken every 3 h. Interestingly, we also observed greater short‐term BPV during the first day after HI. Elevated coefficient of variation in BPV and variation independent of the mean suggest that systolic pressure in the HI group fluctuated significantly even after accounting for the higher average blood pressure in the early post‐HI period. Such fluctuations in blood pressure over hours reflect dynamic physiological interactions not only determined mainly by autonomic nervous system activity, but also affected by physical activity, circadian rhythm and sleep states (Sheikh et al., 2023). Increased short‐term systolic BPV after HI was associated with time‐dependent changes in blood pressure, peripheral blood flow and vascular resistance. We have previously demonstrated in preterm fetal sheep that post‐HI changes in vascular tone are mediated by sympathetic activation and are associated with underlying organ metabolism (Dhillon et al., 2023; Quaedackers et al., 2004). It is postulated that increased vascular resistance after HI preferentially supports blood pressure reducing cardiac work and allows the heart to repair reversible injury (Quaedackers et al., 2004). Therefore, changes in sympathetic tone are probably the major determinant of systolic BPV in short‐term (3 hourly) intervals. Supporting this hypothesis, alpha‐adrenergic inhibition in the present study effectively attenuated the post‐HI increase in mean arterial pressure and short‐term systolic BPV during the first 24 h after HI.

The early period after HI in the same paradigm as the present study is associated with an overall suppression of fetal body movements and disruption of cardiovascular circadian rhythmicity (Lear, Maeda et al., 2023), and, at 0.7 gestation, fetuses have not yet developed mature sleep‐state related cardiovascular patterns. Thus, these factors probably do not influence post‐HI short‐term BPV. The period of high short‐term BPV also partly coincided with the secondary phase of stereotypic seizure activity. We have recently shown that individual post‐HI seizures in preterm fetal sheep are consistently associated with increased mean arterial pressures, mediated by increased peripheral vascular resistance, and minute‐to‐minute variation in mean arterial pressure was predictive of EEG seizure activity (Lear et al., 2025). Although analysis of the effects of seizures was outside the scope of the present study, it is plausible that seizure‐related blood pressure fluctuations may have partially contributed to increased short‐term BPV. Importantly, elevated short‐term BPV during the first 24 h post‐insult was highly associated with exposure to HI, indicating that monitoring variability between consecutive blood pressure readings taken every 3 h could serve as a practical early biomarker for detecting HI brain injury.

Some limitations of the present study should be considered. The studies were undertaken in stable preterm fetal sheep, whereas transitioning asphyxiated neonates experience significant cardio‐respiratory disturbances, and vasopressor use is routine practice to support arterial blood pressure and might also affect BPV (Giesinger et al., 2021). We evaluated BPV measures for 72 h after severe HI. Although this subacute phase was the primary focus, the tertiary phase of injury may also be associated with characteristic BPV changes that reflect evolving pathophysiological processes, similar to the long‐term alterations in circadian rhythms of HRV measures (Lear, Maeda et al., 2023). The severity of insult must affect post‐HI cardiovascular adaptation (Yamaguchi et al., 2018). For example, the speed of onset and duration of alpha‐adrenergic receptor‐mediated secondary hypoperfusion after HI are broadly related to the severity of the initial insult (Bennet et al., 2012). Thus, an important area for future research will also be to investigate how less severe hypoxic insults affect BPV, to assess whether it can be used to detect mild HI injury. Finally, this study was conducted in very preterm equivalent (0.7 gestation) fetuses; future investigations should evaluate BPV in near‐term fetal sheep. Potentially, HI‐associated BPV changes may be even more pronounced at term gestation, given the progressive maturation of blood pressure control with advancing age, mediated by increasing baroreflex sensitivity, maturation of autonomic regulation and changes in vascular tone (Booth et al., 2011; Wassink et al., 2007).

Conclusions

The present study demonstrates that both beat‐to‐beat and short‐term systolic pressure variability consistently evolve over time after HI, and alpha‐adrenergic receptor activation contributes to these early BPV changes. These findings support the potential utility of short‐term and beat‐to‐beat systolic BPV measures as tools for detecting HI injury and monitoring its evolution over time. Robust biological markers are particularly relevant for preterm infants, where neurological signs may be subtle and difficult to detect clinically (Gopagondanahalli et al., 2016). BPV could provide a low‐cost and easily accessible biomarker for the assessment and prognosis of severe HI injury. Further studies should characterise BPV responses after milder insults and elucidate its association with other physiological parameters such as cerebral blood flow and oxygenation. Finally, and not least, for clinical translation, standardised definitions and normative reference ranges will be important.

Competing interests

The authors declare that they have no competing interests.

Author contributions

These experiments were conducted in the Foetal Physiology and Neuroscience Group laboratory, at the University of Auckland. S.K.D., A.J.G. and L.B. conceptualised and designed the study. S.K.D., M.J.B., V.J.K., M.N., J.O.D., O.J.L., B.A.L., C.A.L., J.O.D. and L.B. were responsible for data collection and analysis. All authors were involved in data interpretation and critically reviewed the manuscript and approved the final version of the manuscript submitted for publication.

Funding

This study was funded by the Health Research Council of New Zealand (22/559).

Supporting information

Peer Review History

Biography

Simerdeep Dhillon is a research fellow working with Professors Laura Bennet and Alistair Gunn in the Fetal Physiology and Neuroscience group, The University of Auckland, New Zealand. Her research interest is to understand the mechanisms of preterm brain injury, particularly associated with hypoxia‐ischaemia and inflammation and use this information to develop and refine neuroprotective strategies. Her focus at present is on clinically available treatments that would be most easily translatable.

Data availability statement

Data are available on reasonable request from the corresponding author.