Poster
Essay
Abstract
Infancy is known to be an important developmental stage, and the developmental rate is especially high for newborns. Suggested by the brain-sparing effect, which happens under fetal growth restriction (FGR), the body prioritise metabolic resources to the brain over the rest of the body during this critical developmental stage. An existing study suggests that REM sleep is important for brain functions and has established a model predicting the “REM-to-total sleep ratio” from brain mass. However, since the existing study does not have newborn-subjects and developmental rates usually differ in newborn, whether the existing model can apply to newborns is questionable. Thus, this study focuses on three research questions. Firstly, does the body of human fetuses prioritise metabolic resources to the brain as suggested by the brain-sparing effect? Secondly, is the existing model that predicts “REM-to-total sleep ratio” from brain mass applicable to healthy newborns? Lastly, is the existing model applicable to FGR newborns that underwent the brain-sparing effect?
Highlights
-
There is no significant difference between FGR and non-FGR newborns in brain development but significant difference in body development. This suggests that human fetuses prioritise brain development under metabolic economy.
-
There is a significant difference between the existing model and the “ln(REM-to- total sleep ratio) vs. ln(brain mass)” relationship generated from data of non-FGR newborns. This suggests that the existing model might underestimate REM sleep time required by healthy newborns.
-
The “ln(REM-to-total sleep ratio) vs. ln(brain mass)” relationship generated from data of FGR newborns is statistically insignificant.
Introduction
Human development, a process that starts from the embryonic stage, has its rate consistently changing throughout life (Borsani et al., 2019; Tan and Lewandowski, 2020). Infancy is known to have faster developmental rates (Lejarraga, 2012). According to the growth charts published by the World Health Organization (WHO) in 2009, infant developmental rates are highest in newborns: the stage from birth to 12-week (approximately 3-month) old as defined by the National Health Service (NHS).
How does the body of human fetuses allocate metabolic resources?
Metabolic resources are important for fetal development. However, under conditions like fetal growth restriction (FGR), in which metabolic resources are limited (Miller et al., 2016), the body must allocate the scarce resources. In fact, the brain-sparing effect—an adaptation of directing blood flow to the brain to secure essential life functions that is observed —in many FGR fetuses (Fleiss et al., 2019). This suggests that the body prioritise the brain over the rest of the body (Scherjon et al., 1992; Baschat, 2004) and leads to the first research question of whether the fetal body prioritise brain development, especially under resource limitations. To investigate this question, we compared the brain and body growth curves of FGR and non-FGR newborns to see if there are significant differences. We hypothesised that because most FGR fetuses have brain-sparing as an adaptive mechanism, there would be no significant difference in brain growth but significant difference in body growth between FGR and non-FGR newborns.
How does this allocation of metabolic resources apply to sleep?
Sleep, especially rapid eye movement (REM) sleep, is known to be responsible for cognitive functions, which are highly associated with brain development (Roffwarg et al., 1966; Boyce et al., 2016; Qin et al., 2022). Cao et al. (2020) identified that REM sleep is responsible for neural reorganisation in humans during early ontogeny (from birth to approximately 2.4 years old). As shown in Figure 3-b. of Cao et al. (2020), the study established a model predicting the “REM-to-total sleep ratio (tR/tS)” from brain mass. This model suggests that brain mass (which increases with age) is a proxy of brain metabolic rate and its increase results in decreased proportion of sleep devoted to REM sleep.
However, since Cao et al. (2020) studied childhood in general and is unlikely to have newborn-subjects (in order words, it is unlikely for the study to have subjects younger than 3 months old), whether the established model predicting the “REM-to-total sleep ratio (tR/tS)” from brain mass is accurate for newborns is unclear. This leads to the second research question: ‘Is the existing model predicting the “REM-to-total sleep ratio (tR/tS)” from brain mass applicable to healthy newborns?’ To explore this question, we replicate Cao et al. (2020)’s study by investigating the “ln(tR/tS) vs. ln(brain mass)” relationship in non-FGR newborns and perform a simple linear regression. Then, we juxtapose this relationship to the existing model to see if there is a significant difference. Since newborns usually have faster developmental rates and higher proportion of sleep devoted into REM sleep compared to later infancy and childhood (Roffwarg et al., 1966), we hypothesised that “ln(tR/tS) vs. ln(brain mass)” relationship in non-FGR newborns would be different from Cao et al. (2020)’s model.
Is the “REM-to-total sleep ratio (tR/tS)” dependent on relative or absolute brain size?
Neural reorganisation, which REM sleep is responsible for, is the process of transferring stimuli perceived during awake time into synaptic changes (Cao et al., 2020). Because stimuli can be perceived by the entire body surface area (for example, the somatosensory sensation) and the transferring of stimuli into synaptic changes is primarily done by the brain, it is questionable whether the rate of neural reorganisation and therefore the time required for REM sleep is dependent on the absolute brain size or the relative brain size compared to the overall body size. This leads to the third research question: ‘Is Cao et al. (2020)’s model predicting the “REM-to-total sleep ratio (tR/tS)” from brain mass accurate for FGR newborns?’
Because most FGR newborns have undergone the brain-sparing effect, 70~80% of them have relatively large brains compared to the overall body (Fleiss et al., 2019). If the rate of neural reorganisation is dependent on the relative rather than the absolute brain size, and that relatively large brain would increases the neural reorganisation rate, we expect that FGR newborns to require less REM sleep compared to healthy newborns (See Figure 1. for numerical examples from the actual data). Because the REM sleep requirement is expected to differ in FGR newborns due to the abnormal brain-to-body scaling, we hypothesised that Cao et al. (2020)’s model does not apply to FGR newborns. To test this hypothesis, we investigate the “ln(tR/tS) vs. ln(brain mass)” relationship in FGR newborns, perform a simple linear regression, then juxtapose this relationship to the existing model to see if there is a significant difference.
Figure 1. Numerical example of hypothesised effect of relative brain mass on neural reorganisation rate and REM sleep time required. The birth weight and estimated birth brain mass values of FGR and non-FGR newborn are each from one newborn of the actual dataset. The birth brain mass is estimated from birth head circumference using Cooke et al., (1977)’s model. Because FGR newborns have a lower body-to-brain mass ratio, which in theory is proportional to the amount of information needed to be process by each neuron, we expect neural reorganisation rate to be higher in FGR newborns, and therefore less REM sleep would be required.
Method
Data Sources
In this study, we investigated three research questions by analysing an existing dataset collected by my supervisor Dr Whitehead, and partly published in Georgoulas et al., (2021).
The dataset includes the sleep-staging data, gestational age (GA), postnatal age (PNA), postmenstrual age (PMA) at the time of study. Sleep-staging data, which is available for 187 subjects, was recorded by electroencephalogram (EEG), which detects the stage of sleep the newborns were in by 30-second epochs. The dataset also includes newborns’ birth weight, birth head circumference, and their body weight and head circumferences when they were discharged home.
Research Question 1
To investigate if human fetal body prioritises brain development, we compared the body and brain growth curves of FGR and non-FGR newborns. Because it is difficult to directly measure the brain mass of newborns in clinical settings, we estimated brain mass from head circumference using Cooke et al., (1977)’s model. Combining body weight and estimated brain mass measured at three different time points (at the time of sleep-staging recording, at birth, and at the time when newborns are discharged home) of 187 subjects; we ended up with a dataset with 716 samples in which 363 are FGR and 353 are non-FGR.
We calculated the real “body” mass by subtracting the “total” body weight by the estimated brain mass. Then, we plotted the “total”, “body”, and “brain” masses against the postmenstrual age (PMA) separately for FGR and non-FGR newborns to see if the developmental curves are different between the groups. To test whether the observed differences are statistically significant, we created separate graphs for the developmental curves of “body” and “brain” mass (we neglected “total” mass as it is the sum of “body” and “brain” mass) and plotted the curves with 95% confidence interval of both FGR and non-FGR newborns to see if the confidence interval overlay.
Research Question 2 & 3
To test if the existing model predicting the “REM-to-total sleep ratio (tR/tS)” from brain mass applies to healthy newborns or to FGR newborns, we replicate Cao et al. (2020)’s study by plotting the ln(tR/tS) against ln(brain mass) separately for FGR and non-FGR newborns, then compared these relationship to Cao et al. (2020)’s model. First, we calculated the (tR/tS) values for the 187 subjects (20 FGR; 167 non-FGR) with available sleep-staging data by dividing the number of epochs the newborns were in REM sleep by the number of epochs the newborns were in REM or nREM sleep.
Result
The “Body” and “Brain” Growth Curves
To investigate if human fetal body prioritises brain development as suggested by the brain- sparing effect in FGR, we compared the “total”, “body”, and “brain” masses vs. postmenstrual age (PMA) relationships between FGR and non-FGR newborns. As shown in Figure 2., in both FGR and non-FGR newborns, the growth of “total”, “body”, and “brain” decrease in rate as PMA increases, and all developmental curves show sigmoidal patterns (which has the lowest residual standard error compared to other models). As expected by the hypothesis, the brain development seems to be the same for FGR and non-FGR newborns and the body development seems to be slower in FGR newborns.
Figure 2. “Total”, “body”, and “brain” developmental curves of (a) FGR and (b) non- FGR newborns. “Total mass” refers to the mass of the entire body, whereas “body mass” refers to the mass of the body excluding brain mass. In simpler terms, “total = body + brain.” The dotted lines show the sigmoidal models fitted.
To test if there is statistically significant difference in body development between FGR and non-FGR newborns, we compared the two body growth curves by plotting them on the same graph, adding 95% confidence intervals, and see if the confidence intervals overlap. As shown in Figure 3-a, because the confidence intervals do not overlap, the difference in body development between FGR and non-FGR newborns is statistically significant. We perform the same procedure for brain growth curves to test if there is no statistically significant difference in brain development. As shown in Figure 3-b, since the confidence intervals heavily overlap, there is no statistically significant difference in brain development between FGR and non-FGR newborns.
Figure 3. “Body” and “brain” growth curves comparison between FGR and non-FGR newborns. (a) Body growth curves of FGR and non-FGR newborns with 95% confidence intervals and (b) Brain growth curves of FGR and non-FGR newborns with 95% confidence intervals. The dotted lines represent the upper and lower of the 95% confidence intervals. Sigmoidal models were fitted, with all
Comparison to the existing “ln(tR/tS) vs. ln(brain mass)” model
To test if the existing model predicting the “REM-to-total sleep ratio (tR/tS)” from brain mass is accurate for healthy newborns, we replicate Cao et al. (2020)’s study by plotting the ln(tR/tS) against ln(brain mass) and perform a simple linear regression for FGR, then we compared the relationship to Cao et al. (2020)’s model. We performed the same procedure to test if the model applies to FGR newborns, who have relatively large brains compared to the overall body due to the brain-sparing effect, to investigate whether it is the “absolute” or “relative” brain mass that affects the REM sleep time required by newborns.
As shown in Figure 4-a, the “ln(tR/tS) vs. ln(brain mass)” relationship of healthy newborns is significantly different from Cao et al. (2020)’s model in that neither the 95% confidence interval of the slope nor the intercept of the linear model of healthy newborns includes the corresponding value of the existing model. Here, we excluded an outlier (reference number: 255101) with unusually high leverage not only because it violates the assumptions of linear regression but also because it has tR/tS of 0, and it is impossible to take natural log of its tR/tS for further analysis.
The two models intersect at a point where ln(brain mass)= -1.62. Since Cao et al. (2020) suggested that there is a correlation between age and brain mass, using the linear model of “ln(brain mass) vs. ln(PMA)” of our sample (p<0.001, R2=0.90), we estimated that the intersection correspond to approximately PMA = 31 week. Despite the linear model being statistically significant (p<0.05), the R2 value is very low, suggesting that the correlation between the proportion of sleep devoted to REM sleep and brain mass is very weak in healthy newborns.
As shown in Figure 4-b, we found an insignificant “ln(tR/tS) vs. ln(brain mass)” relationship in FGR newborns. As observed in the “ln(tR/tS) vs. ln(brain mass)” model of healthy newborns, the linear model of FGR newborns has a very low R2 value, suggesting a weak correlation between the proportion of sleep devoted to REM sleep and brain mass.
Figure 4. The comparison of the “ln(tR/tS) vs. ln(brain mass)” relationship of newborns to Cao et al. (2020)’s model. (a) compares the relationship of healthy newborns (in blue, n=166, y~ (-0.180.10)*x - (0.540.08)) to the existing model (in black, y= -0.60x - 1.22). (b) compares the relationship of FGR newborns (in red, n=20, y~ (-0.190.20)*x - (0.520.25)) to the existing model (in black, y= -0.60x - 1.22). “tR/tS” refers to the proportion of sleep devoted to REM sleep, which is known to be responsible for neural reorganisation.
Summary
Our results to the three research questions are as follow:
-
There is significant difference in body development but not in brain development between FGR (which underwent metabolic resource limitation) and non-FGR newborns.
-
The existing model predicting the proportion of sleep devoted to REM sleep from brain mass does not apply to healthy newborns.
-
There is no statistically significant relationship between the proportion of sleep devoted to REM sleep and brain mass in FGR newborns.
Discussion
How are metabolic resources applied to human fetuses?
Since development rates are faster in newborns, and it is likely that human fetal body prioritises brain development rather than the rest of the body, we investigate if the body allocate more metabolic resources to the brain under limited resources by comparing the body and brain growth curves of FGR and non-FGR newborns. By doing so, we found a significant difference in body development but not in brain development between the groups. This supports our hypothesis that human fetal body prioritises brain development, especially under metabolic economy.
Our results suggest that because human fetal body prioritising resources to the brain, there is no difference in brain development even when fetuses undergo nutrition limitation. However, this contradicts the actual cognitive developments of individuals that experienced FGR as a fetus. Studies have found correlations between FGR and various developmental abnormalities, including cognitive abnormalities (Gotlieb et al, 1988; Scherjon et al., 1992; Miller et al, 2016). This contradiction might be reasonable if cognitive development does not solely depend on brain mass but more complicated factors like structural and functional connectivity, or other factors that FGR newborns are disadvantaged by, e.g. since FGR newborns usually stay in the hospital for postnatal care, they might be less stimulated by social factors, a factor of cognitive development (Weisglas-Kuperus et al., 1993).
How does this apply to sleep?
Applying this finding to REM sleep, a process responsible for brain development that is highly demanded in newborns, we wonder if the existing model predicting the proportion of sleep devoted to REM sleep from brain mass (that was generated from data of mostly children) applies to healthy newborns. We investigate this by comparing the “ln(tR/tS) vs. ln(brain mass)” linear model of non-FGR newborns to the existing model. By doing so, we found the models to be significantly different and therefore a separate model is needed for estimating REM sleep time required by newborns from brain mass.
As shown in mentioned in the results section, the two models intersect at the point where PMA is approximately 31 weeks, and the model of non-FGR newborns is above the existing model after the intersection. Given that typical gestation period is about 40 weeks (Lynch and Zhang, 2007), this suggests that the existing model would underestimate the REM sleep time required by healthy newborns.
We also wonder if it is the “relative” brain mass compared to the body rather than the absolute brain mass that affects REM sleep time needed in newborns, and we investigated this by comparing the “ln(tR/tS) vs. ln(brain mass)” linear model of FGR newborns to the existing model. However, we found no significant relationship in FGR newborns. This is likely due to the limited sample size (n=20).
In addition, we found the model of both non-FGR and FGR newborns to have very weak correlation. This suggests that REM sleep requirement might be high in newborns regardless of brain mass. Furthermore, we found that the ln(brain mass) values of our data to be lower than the 0-year-old ln(brain mass) value, roughly –1.0, in Cao et al. (2020)’s study. This is likely because newborns have faster brain growth, and Cao et al. (2020) established the ln(brain mass) vs. age relationship using data mostly from children.
Limitations
This study has several limitations since data were collected from newborns in clinical settings. Firstly, we could only use head circumference as a proxy for brain mass. Secondly, the sleep recordings were short (lasting only a few hours) because parents are unlikely to consent to have electrodes on babies for long periods. Thus, the precision of the tR/tS values is likely limited.
Conclusion
In conclusion, we verified that the human fetal body prioritises brain development as suggested by the brain-sparing effect. Applying this to REM sleep, an important process for brain development, our data suggests that REM sleep is required more in newborns, and this is likely to be independent of brain mass, a dependent variable of REM sleep requirements suggested by Cao et al. (2020). Moreover, even if brain mass does affect the REM sleep time needed in newborns, the existing model that predicts the proportion of sleep devoted to REM sleep from brain mass is likely to underestimate REM sleep time in newborns.
References
Baschat, D.A.A. (2004) Fetal responses to placental insufficiency: an update. BJOG: An International Journal of Obstetrics & Gynaecology. 111 (10), 1031–1041. doi:10.1111/j.1471-0528.2004.00273.x.
Borsani, E., Della Vedova, A.M., Rezzani, R., Rodella, L.F. & Cristini, C. (2019) Correlation between human nervous system development and acquisition of fetal skills: An overview. Brain and Development. 41 (3), 225–233. doi:10.1016/j.braindev.2018.10.009.
Boyce, R., Glasgow, S.D., Williams, S. & Adamantidis, A. (2016) Causal evidence for the role of REM sleep theta rhythm in contextual memory consolidation. Science. 352 (6287), 812–816. doi:10.1126/science.aad5252.
Cao, J., Herman, A.B., West, G.B., Poe, G. & Savage, V.M. (2020) Unraveling why we sleep: Quantitative analysis reveals abrupt transition from neural reorganization to repair in early development. Science Advances. doi:10.1126/sciadv.aba0398.
Cooke, R.W.I., Lucas, A., Yudkin, P.L.N. & Pryse-Davies, J. (1977) Head circumference as an index of brain weight in the fetus and newborn. Early Human Development. 1 (2), 145–
149. doi:10.1016/0378-3782(77)90015-9.
Georgoulas, A., Jones, L., Laudiano-Dray, M.P., Meek, J., Fabrizi, L. & Whitehead, K. (2021) Sleep–wake regulation in preterm and term infants. Sleep. 44 (1), zsaa148. doi:10.1093/sleep/zsaa148.
Lejarraga, H. (2012) Growth in Infancy and Childhood. In: Human Growth and Development. Elsevier. pp. 23–56. doi:10.1016/B978-0-12-383882-7.00002-7.
Lynch, C.D. & Zhang, J. (2007) The research implications of the selection of a gestational age estimation method. https://onlinelibrary.wiley.com/doi/10.1111/j.1365- 3016.2007.00865.x.
Miller, S.L., Huppi, P.S. & Mallard, C. (2016) The consequences of fetal growth restriction on brain structure and neurodevelopmental outcome. The Journal of Physiology. 594 (4), 807–823. doi:10.1113/JP271402.
Qin, H., Fu, L., Jian, T., Jin, W., Liang, M., et al. (2022) REM sleep-active hypothalamic neurons may contribute to hippocampal social-memory consolidation. Neuron. 110 (23), 4000-4014.e6. doi:10.1016/j.neuron.2022.09.004.
Roffwarg, H.P., Muzio, J.N. & Dement, W.C. (1966) Ontogenetic Development of the Human Sleep-Dream Cycle. Science. 152 (3722), 604–619. doi:10.1126/science.152.3722.604.
Scherjon, S., Briët, J., Oosting, H. & Kok, J. (2000) The Discrepancy Between Maturation of Visual-Evoked Potentials and Cognitive Outcome at Five Years in Very Preterm Infants With and Without Hemodynamic Signs of Fetal Brain-Sparing. Pediatrics. 105 (2), 385–391. doi:10.1542/peds.105.2.385.
Tan, C.M.J. & Lewandowski, A.J. (2020) The Transitional Heart: From Early Embryonic and Fetal Development to Neonatal Life. Fetal Diagnosis and Therapy. 47 (5), 373–386. doi:10.1159/000501906.
Weisglas-Kuperus, N., Baerts, W., Smrkovsky, M. & Sauer, P.J.J. (1993) Effects of Biological and Social Factors on the Cognitive Development of Very Low Birth Weight Children.
Pediatrics. 92 (5), 658–665. doi:10.1542/peds.92.5.658.
Please sign in
If you are a registered user on Laidlaw Scholars Network, please sign in