Embracing the chaos Fetal heart rate biomarkers to detect evolving brain injury before birth

Abstract

Background. Hypoxia-ischaemia (HI) during fetal life is a major cause of life-long neurodevelopmental disability. Detection during pregnancy is difficult, limiting clinical management and potential treatment. Fetal heart rate variability (FHRV) is one potential diagnostic biomarker, but the long-term effects of HI on FHRV are poorly understood, and current analytical approaches do not account for the chaotic, non-stationary nature of the signal. My specific thesis objectives were to examine whether FHRV indices can be used as biomarkers for HI injury in utero and, if so, whether non-linear FHRV measures were better than traditional linear measures. Secondary objectives were to assess if FHRV measures could be used to determine the presence of seizures and whether HI impaired the maturation of sleep state development. Methods. Chronically instrumented preterm fetal sheep (0.7 gestation) underwent 25 minutes of complete umbilical cord occlusion (n=8) or sham occlusion (n=8). FHRV and a wide range of physiological parameters were recorded for 21 days after HI. To facilitate the assessment of FHRV, I developed a suite of tools to calculate a wide range of time, frequency and non-linear FHRV measures. Results. The latent phase of injury (0-6h) was associated with a broad reduction in magnitude of time and frequency-domain measures, with reduced complexity but increased irregularity of FHRV. The early secondary phase (6-18h) was associated with increased fractal complexity. The peak of the secondary phase (18-48h) saw profound suppression of time and frequency-domain measures, reduced entropy-based measures of complexity but persisting increase in fractal complexity. Seizures occurred between ~6-48 hours. FHRV measures did not change during individual seizures in the secondary phase but may be linked to the overall increase of fractal complexity. The tertiary phase (48h onwards) saw partial recovery of most FHRV measures, but many remained suppressed until 14 days. Notably, very low frequency activity, Distribution-Entropy and Deceleration-Reserve remained suppressed even at 21 days. Circadian rhythmicity in FHRV was lost or suppressed for 4-5 days post-HI. When day:night cycling became re-established, it was characterised by exaggerated cycling, especially from 14 days onwards, with a significantly greater day-time nadir, which persisted until the end of the experiment. At 21 days, HI impaired fetal sleep state with impaired electroencephalographic power and greater time in rapid-eye-movement sleep; however, HI had only a modest effect on FHRV during sleep states. Discussion. My thesis has demonstrated that both linear non-linear measures of FHRV can be used as a biomarker to show that a severe HI insult has occurred in utero and can map to different phases of injury. Combined use of linear and non-linear measures provided the best diagnostic capacity. Key discoveries include being able to determine the onset and duration of the secondary loss of mitochondrial function (secondary phase) with non-linear measures and the finding that circadian rhythmicity is a key feature of brain-injured fetuses in nearly all FHRV measures, with a characteristic lower nadir during the day. Future work is needed to examine changes in FHRV in response to different severities of brain injuries, responses to treatments, and mechanisms mediating the changes in circadian patterns.