Timely recognition of fetal distress, during pregnancy and labour, in order to intervene adequately is of major importance to avoid neonatal morbidity and mortality. As discussed in chapter 1, the cardiotocogram (CTG) might be a useful screening test for fetal monitoring but it has insufficient specificity and requires additional diagnostic tests in case of suspected fetal compromise to avoid unnecessary operative deliveries. Potential additional techniques used in clinical practice are fetal scalp blood sampling (FBS) and ST-waveform analysis of the fetal electrocardiogram (ECG; STAN®). However, publications on these techniques provide limited support for the use of these methods in the presence of a non-reassuring CTG for reducing caesarean sections. In addition, these techniques are invasive and can therefore only be used during labour at the term or the near term period. Consequently, it is of great clinical importance that additional methods are developed that contribute to more reliable assessment of fetal condition. Preferably, this information is obtained non-invasively. Valuable additional information on the fetal condition can possibly be obtained by spectral analysis of fetal heart rate variability (HRV). The fetal heart rate fluctuates under the control of the autonomic part of the central nervous system. The autonomic cardiac modulation is discussed in chapter 2. The sympathetic and parasympathetic nervous systems typically operate on partly different timescales. Time-frequency analysis (spectral analysis) of fetal beat-to-beat HRV can hence quantify sympathetic and parasympathetic modulation and characterise autonomic cardiac control . The low frequency (LF) component of HRV is associated with both sympathetic and parasympathetic modulation while the high frequency (HF) component is associated with parasympathetic modulation alone2. Spectral estimates of HRV might indirectly reflect fetal wellbeing and increase insight in the human fetal autonomic cardiac response. In chapter 3, technical details for retrieving fetal beat-to-beat heart rate and its spectral estimates are provided. In this thesis spectral analysis of fetal HRV is investigated. The first objective is to study the value of spectral analysis of fetal HRV as a tool to assess fetal wellbeing during labour at term. The second objective is to monitor spectral estimates of fetal HRV, non-invasively, during gestation to increase insight in the development of human fetal autonomic cardiac control. Since Akselrod reported the relation between autonomic nervous system modulation and LF and HF peaks in frequency domain1, frequency analysis of RR interval fluctuations is widely performed . For human adults, standards for HRV measurement and physiological interpretation have been developed2. Although HRV parameters are reported to be highly prognostic in human adults in case of cardiac disease, little research is done towards the value of these parameters in assessing fetal distress in the human fetus, as shown in chapter 4. In this chapter, the literature about time-frequency analysis of human fetal HRV is reviewed in order to determine the value of spectral estimates for fetal surveillance. Articles that described spectral analysis of human fetal HRV and compared the energy in spectral bands with fetal bloodgas values were included. Only six studies met our inclusion criteria. One study found an initial increase in LF power during the first stage of fetal compromise, which was thought to point to stress-induced sympathetic hyperactivity3. Five out of six studies showed a decrease in LF power in case of fetal distress , , , , ,. This decrease in LF power in case of severe fetal compromise was thought to be the result of immaturity or decompensation of the fetal autonomic nervous system. These findings support the hypothesis that spectral analysis of fetal HRV might be a promising method for fetal surveillance. All studies included in the literature review used absolute values of LF and HF power. Although absolute LF and HF power of HRV provide useful information on autonomic modulation, especially when considering fetal autonomic development, LF and HF power may also be measured in normalised units. Normalised LF (LFn) and normalised HF power (HFn) of HRV represent the relative value of each power component in proportion to the total power2. Adrenergic stimulation can cause a sympathetically-modulated increase in fetal heart rate . A negative correlation however exists between heart rate and HRV . As a result, the sympathetic stimulation can decrease the total power of HRV and even the absolute LF power. When normalising the absolute LF (and HF) with respect to the total power, a shift in activity from HFn to LFn might become visible, revealing the expected underlying sympathetic activity. Thus, because changes in total power influence absolute spectral estimates in the same direction, normalised values of LF and HF power seem more suitable for fetal monitoring. In other words, normalised spectral estimates detect relative changes that are no longer masked by changes in total power2. LFn and HFn power are calculated by dividing LF and HF power, respectively, by total power and represent the controlled and balanced behaviour of the two branches of the autonomic nervous system2. In chapter 5 we hypothesised that the autonomic cardiovascular control is functional in fetuses at term, and that LFn power increases in case of distress due to increased sympathetic modulation. During labour at term, ten acidaemic fetuses were compared with ten healthy fetuses. During the last 30 minutes of labour, acidaemic fetuses had significantly higher LFn power and lower HFn power than control fetuses, which points to increased sympathetic modulation. No differences in absolute LF or HF power were found between both groups. The observed differences in normalised spectral estimates of HRV were not observed earlier in labour. In conclusion, it seems that the autonomic nervous system of human fetuses at term responds adequately to severe stress during labour. Normalised spectral estimates of HRV might be able to discriminate between normal and abnormal fetal condition. Although we found significant differences in normalised spectral estimates between healthy and acidaemic fetuses, we wondered whether spectral power of HRV is also related to fetal distress in an earlier stage. The next step in chapter 6 was therefore, to investigate whether spectral estimates are related to fetal scalp blood pH during labour. Term fetuses during labour, in cephalic presentation, that underwent one or more scalp blood samples were studied. Beat-to-beat fetal heart rate segments, preceding the scalp blood measurement, were used to calculate spectral estimates. In total 39 FBS from 30 patients were studied. We found that normalised spectral estimates are related to fetal scalp blood pH while absolute spectral estimates are not related to fetal pH. It was further demonstrated that LFn power is negatively related and HFn power is positively related to fetal pH. These findings point to increased sympathetic and decreased parasympathetic cardiac modulation in human fetuses at term upon decrease of their pH value. This study confirms the hypothesis that normalised spectral values of fetal HRV are related to fetal distress in an early stage. Previous studies showed that absolute LF and HF power increase as pregnancy progresses, which is attributed to fetal autonomic maturation , . Since it is yet unclear how LFn and HFn evolve with progressing pregnancy, before using spectral analysis for fetal monitoring, it has to be determined whether gestational age has to be corrected for. In addition, fetal autonomic fluctuations, and thus spectral estimates of HRV, are influenced by fetal behavioural state . Since these states continue to change during labour , thorough understanding of the way in which these changes in state influence spectral power is necessary for the interpretation of spectral values during labour at term. Therefore, in chapter 7, we examined whether differences in spectral estimates exist between healthy near term and post term fetuses during labour. In case such differences do exist, they should be taken into consideration for fetal monitoring. The quiet and active sleep states were studied separately to determine the influence of fetal behavioural state on spectral estimates of HRV during labour around term. No significant differences in spectral estimates were found between near term and post term fetuses during active sleep. During quiet sleep, LFn power was lower and HF and HFn power were higher in post term compared to near term fetuses, no significant differences in LF power were observed between both groups. LF, HF and LFn power were higher and HFn power was lower during active sleep compared to quiet sleep in both groups. This seems to point to sympathetic predominance during the active state in fetuses around term. In addition, post term parasympathetic modulation during rest seems increased compared to near term. In conclusion, fetal behavioural state and gestational age cause a considerable variability in spectral estimates in fetuses during labour, around term, which should be taken into consideration when using spectral estimates for fetal monitoring. In chapters 4 to 6, spectral estimates of beat-to-beat fetal HRV were studied using fetal ECG recordings that were obtained directly from the fetal scalp during labour. However, the second objective of this thesis is to obtain spectral estimates non-invasively during gestation to increase insight in the development of human fetal autonomic cardiac control. The fetal ECG is also present on the maternal abdomen, although much smaller in amplitude and obscured by the maternal ECG and noise. Chapter 8 focused on non-invasive measurement of the fetal ECG from the maternal abdomen. These measurements allow for obtaining beat-to-beat fetal heart rate non-invasively. Therefore, this method can be used to obtain spectral estimates of fetal HRV throughout gestation. Although abdominal recording of the fetal ECG may offer valuable additional information, it is troubled by poor signal-to-noise ratios (SNR) during certain parts of pregnancy, e.g. during the immature period and during the vernix period. To increase the usability of abdominal fetal ECG recordings, an algorithm was developed that uses a priori knowledge on the physiology of the fetal heart to enhance the fetal ECG components in multi-lead abdominal fetal ECG recordings, before QRS-detection. Evaluation of the method on generated fetal ECG recordings with controlled SNR showed excellent results. The method for non-invasive fetal ECG and beat-to-beat heart rate detection presented in chapter 8 was used for analysis in chapter 9. The feasibility of this method in a longitudinal patient study was investigated. In addition, changes in spectral estimates of HRV during pregnancy were studied and related to fetal rest-activity state to study the development of fetal autonomic cardiac control. We found that approximately 3% of non-invasive fetal ECG recordings could be used for spectral analysis. Therefore, improvement of both equipment and algorithms is still needed to obtain more good-quality data. The percentage of successfully retrieved data depends on gestational age. Before 18 and between 30 and 34 weeks no good-quality beat-to-beat heart rate data were available. We found an increase in LF and HF power of fetal HRV with increasing gestational age, between 21 to 30 weeks of gestation. This increase in LF and HF power is probably due to increased sympathetic and parasympathetic modulation and might be a sign of autonomic development. Furthermore, we found sympathetic predominance during the active state compared to the quiet state in near term fetuses (34 to 41 weeks of gestation), comparable to the results observed during labour around term. During 34 to 41 weeks a (non-significant) decrease in LF and LFn power and a (non-significant) increase in HF and HFn power were observed. These non-significant changes in spectral estimates in near term fetuses might be associated with changes in fetal rest-activity state and increased parasympathetic modulation as pregnancy progresses. However, more research is needed to confirm this.
|Qualification||Doctor of Philosophy|
|Award date||31 Oct 2012|
|Place of Publication||Eindhoven|
|Publication status||Published - 2012|