Introduction: The Central Autonomic Network (CAN) is a hierarchy of brain structures that collectively influence cardiac autonomic input, mediating the majority of brain-heart interactions, but has never been studied in premature neonates. In this study, we use heart rate variability (HRV), which has been described as the “primary output” of the CAN, and resting-state functional MRI (rsfMRI) to characterize brain-heart relationships in premature neonates. Methods: We studied premature neonates who underwent rsfMRI at term (37-week postmenstrual age or above) and had HRV data recorded during the same week of their MRI. HRV was derived from continuous electrocardiogram data during the week of the rsfMRI scan. For rsfMRI, a seed-based approach was used to define regions of interest (ROIs) pertinent to the CAN, and blood oxygen level-dependent signal was correlated between each ROI as a measure of functional connectivity. HRV was correlated with CAN connectivity (CANconn) for each region, and subgroup analysis was performed based on sex and clinical comorbidities. Results: Forty-seven premature neonates were included in this study, with a mean gestational age at birth of 28.1 +/− 2.6 weeks. Term CANconn was found to be significantly correlated with HRV in approximately one-fifth of CAN connections. Two distinct patterns emerged among these HRV-CANconn relationships. In the first, increased HRV was associated with stronger CANconn of limbic regions. In the second pattern, stronger CANconn at the precuneus was associated with impaired HRV maturation. These patterns were especially pronounced in male premature neonates. Conclusion: We report for the first time evidence of brain-heart relationships in premature neonates and an emerging CAN, most striking in male neonates, suggesting that the brain-heart axis may be more vulnerable in male premature neonates. Signatures in the heart rate may eventually become an important noninvasive tool to identify premature males at highest risk for neurodevelopmental impairment.

The Central Autonomic Network (CAN) may be conceptualized as consisting of lower centers in the brainstem and a network of higher centers in the cerebrum (cortical and subcortical) and cerebellum, integrated through complex hierarchically organized pathways [1]. The brainstem CAN regulates homeostatic reflexes in response to ascending interoceptive stimuli from other systems such as the cardiorespiratory system. Descending pathways from the higher CAN centers converge on the brainstem CAN centers, transmitting exteroceptive, cognitive, and limbic system influences. In this manner, the descending CAN pathways influence brainstem autonomic outflow into the peripheral autonomic system via sympathetic and parasympathetic nerves, including those to the sinoatrial node of the heart [2, 3]. The most readily accessible and reliable measure of such autonomic output is heart rate variability (HRV), the “primary output” of the CAN [3]. While maturational changes in HRV have been studied in the newborn, including in the context of prematurity and critical illness [4‒7], the relationship between higher CAN centers and autonomic maturation remains relatively unknown. Prematurity poses an especially intriguing context to study this relationship since the immature autonomic nervous system output (as measured by HRV) undergoes significant maturation during the latter half of gestation.

Prior studies of the CAN have focused primarily on adults and have described the association of physiologic variables like HRV with functional connectivity of the CAN [8] using resting-state functional MRI (rsfMRI). The clinical significance of the CAN has been documented in multiple central neurologic pathologies where structural changes in a variety of conditions like traumatic injury [9], multiple sclerosis [10], and neurodegenerative conditions [11] are associated with HRV abnormalities. Abnormalities of HRV have also been reported in neuropsychiatric conditions without gross structural brain pathology, including attention deficit disorders [12], depression [13], anxiety [14], and autism [15], conditions which are significantly more common in children born prematurely compared to their term-born peers [16‒20]. Although studies of connectivity between the CAN centers have not been reported previously in the newborn, there is growing evidence of emerging central modulation of HRV at this early stage in brain development, in low-risk term newborns [21] and in pathologies like hypoxic-ischemic encephalopathy and neonatal seizures [22‒25].

Recently, abnormal neonatal heart rate patterns have been described in male children who develop autism [26], suggesting that HRV signatures in the newborn may, in the future, help risk-stratify infants for future neuropsychiatric morbidity. In this exploratory study, we aim to characterize maturational relationships in the CAN of premature neonates without major structural brain injury utilizing rsfMRI and HRV at term equivalent age (TEA). We hypothesize that increased HRV will be associated with stronger term CAN connectivity (CANconn) at TEA. To our knowledge, this is the first study to explore the development of CANconn using rsfMRI in a prematurely born cohort.

Participants

We studied premature neonates admitted to the neonatal intensive care unit (NICU) at Children’s National Hospital in Washington, DC who enrolled in a longitudinal observational study exploring brain development between 2017 to 2021. Inclusion criteria for this study included premature infants born prior to 35-week gestational age (GA) who underwent rsfMRI at TEA (37-week postmenstrual age [PMA] or above) and had HRV data recorded during the same week of their MRI. Infants with Kidokoro cerebral white matter score >2 [27], moderate-to-severe ventriculomegaly, brain malformations, and other congenital anomalies suggestive of an underlying genetic syndrome, were excluded from this study. The study was approved by the Institutional Review Board (IRB) of Children’s National, and all study components were performed within the rules and regulations of the IRB. Informed written consent was obtained from parents or guardians of each participant.

HRV Methods

For HRV data, continuous electrocardiogram (EKG) was collected on each infant from the time of NICU admission until time of discharge. EKG was recorded at a sampling rate of 250 Hz using the Philips bedside monitor (IntelliVue MP70, MA, USA), archived and then retrieved from a Philips Data Warehouse System (Philips PIIC IX, MA, USA). The EKG signal was bandpass-filtered between 0.5 – 60 Hz and R waves were identified. Beat-to-beat interval was calculated and divided into 10-min epochs. HRV metrics used in this analysis included alpha 1, alpha 2, RMS1, and RMS2, which were calculated using detrended fluctuation analysis. Deriving the time series HRV metrics using detrended fluctuation analysis involves first subtracting the mean value of the beat-to-beat interval from the interval series and calculating the cumulative sum, or profile function. The profile function is then partitioned into end-to-end windows containing “s” number of beats and fit to the profile in each window using a fourth degree polynomial. The local fluctuation function is calculated as the root mean square of the difference between the profile and best fits. Next, the fluctuation function is calculated as the median of the local fluctuation function from all the windows. These steps are repeated for different window sizes. In our calculation, we varied the window size from 6 beats to one-fourth of the number of RRi samples in the 10-min epoch. RMS1 (s) and RMS2 (s) were calculated as the maximum value of the detrended fluctuation function for “s” between 15 and 50 beats and 100–150 beats, respectively. We also calculated alpha 1 and alpha 2 as the slope of the fluctuation and “s” in double logarithmic plot from the regions of 15–30 beats (short-term scale) and 35–150 beats (long-term scale/ultralow frequency), respectively.

HRV metrics were extracted for each available 10-min epoch and averaged over weekly intervals from 27- to 41-week PMA. Weeks with fewer than 4 days of available EKG data were excluded from the analysis. HRV metrics from the week during which the fMRI scan was performed were used for the primary analysis. Alpha 1, RMS1, and RMS2 are metrics of sympathetic autonomic tone that increase as premature infants approach TEA but are known to remain low when compared to term-born neonates [4]. These metrics, particularly alpha 1, also have been shown to be influenced by morbidity [28]. Alpha 1 characterizes the short-term self-affinity of the heart rate, while alpha 2 reflects more long-term fluctuations. Alpha 2 tends to decrease slightly with increasing PMA and does not significantly differ between preterm and term neonates [29]. RMS1 and 2 characterize the variability in the beat-to-beat interval at short- and long-term scales, respectively. These indices were selected over frequency-domain metrics because they are less vulnerable to the non-stationarity of physiologic signals in intensive care settings [30]; in addition, previous work demonstrates effects of prematurity-related morbidities on the maturational trajectory of time-domain metrics [28]. However, for additional comparative value to other studies, LF/HF ratios were included in the overall HRV analysis as well. Given their established maturational trajectories, lower, impaired, or less mature HRV in this paper refers to lower alpha 1, RMS1, and RMS2, and higher alpha 2.

RsfMRI Methods

All infants underwent a non-sedated, blood oxygen level-dependent (BOLD)-based rsfMRI at TEA. RsfMRI detects changes in BOLD signals as markers of regional cerebral activation; connectivity refers to the degree of co-activation of different brain regions. While fMRI is often performed in conjunction with specific tasks or stimuli for the study of activity-specific cerebral networks (task-based fMRI), the premature infants involved in this study were not provided a specific stimulus and were therefore studied in a “resting state.” When medically appropriate, infants would be fed and swaddled prior to entering the MRI scanner. Ear protection was used to diminish auditory stimuli that may disturb the infant. We acquired sagittal, axial, and coronal anatomical T2-weighted images using single-shot fast spin echo sequences with 2 mm slice thickness on a 3 T MRI scanner (GE Discovery MR750, GE Healthcare, Milwaukee, WI). Gradient-echo planar images in premature infants were acquired with the following parameters: repetition time, 2,000; echo time, 35; voxel size = 3.125 × 3.125 × 3 mm, field of view, 200 mm, and matrix size 64 × 64.

Resting state data were preprocessed based on a previously published pipelines [31] using tools from the Analysis of Functional NeuroImages (AFNI) [31]. In brief, scans underwent within-volume motion correction, coregistration, slice-time correction, dropping the first four volumes. Following, images were despiked, bias-field corrected, and intensity outliers were censored. Functional images were then aligned anatomically and normalized to a 40-week GA template [32]. Signal was intensity scaled to a global mode of 1,000 and smoothed using a 5 mm full-width half-maximum blur. Motion, CSF, and white matter signal were regressed out, and the image underwent 0.009–0.08 bandpass filtering. To reduce the effect of motion, volumes with motion exceeding 0.2 mm were excluded from the analyses. After censoring, only images with at least 4 min of available data were included in the analyses. The Neonatal Automated Anatomical Labeling Atlas [33] was adapted to include segmentations for the midbrain, pons, medulla, and bilateral cerebellum. Additional regions of interest (ROIs) included known centers of the higher CAN network, including the bilateral anterior cingulate cortex, medial prefrontal cortex, precuneus, thalamus, hippocampus, amygdala, and insula (see Fig. 1). Using a seed-based approach, correlation of the BOLD signal was assessed between each of these 19 preselected ROIs.

Fig. 1.

a–d Three-dimensional models exhibiting whole brain atlas views with CAN ROIs are highlighted in color and are shown in a coronal view with labeled ROIs, a sagittal view, a superior axial view, and an inferior axial view.

Fig. 1.

a–d Three-dimensional models exhibiting whole brain atlas views with CAN ROIs are highlighted in color and are shown in a coronal view with labeled ROIs, a sagittal view, a superior axial view, and an inferior axial view.

Close modal

Statistical Analysis

Statistical analysis was performed using SPSS (IBM SPSS Statistics for Mac, Version 29.0). Each HRV metric was correlated with connectivity at each ROI-ROI connection for all subjects. Subgroup analysis was performed twice, with one grouping based on assigned sex and another grouping based on morbidity. To determine morbidity, subjects were assigned one point each for culture positive sepsis, a diagnosis of necrotizing enterocolitis (NEC), and moderate-to-severe bronchopulmonary dysplasia (BPD). An additional point was added if surgical intervention was needed for NEC and for either dependence on supplemental oxygen or feeding tube at time of discharge. Morbidity was then classified as high (score of 2 or greater) or low (score of 0–1) which allowed for relatively balanced subgroups. Baseline clinical characteristics and HRV were compared between the two groups using two sample t tests for continuous variables and χ2 tests for categorical variables.

Baseline Study Characteristics

Forty-seven premature neonates were included in this analysis. The mean GA at birth was 28.1 +/− 2.6 weeks, ranging from 23.3 weeks to 32.1 weeks. The mean PMA at the time of rsfMRI scan was 39.0 +/− 1.5 weeks, ranging from 37 to 42.75 weeks. Table 1 summarizes the baseline study characteristics of our study cohort. Notably, there were no major clinical differences between males and females, including GA at birth or PMA at time of term MRI. The high- and low-morbidity groups, as expected, demonstrated many significant differences including GA at birth, birth weight, prevalence of sepsis, NEC, surgical NEC, BPD, and need for medical support devices at discharge.

Table 1.

Clinical characteristics of all subjects

VariableAllMaleFemalep valueHigh morbidityLow morbidityp value
Total subjects 47 24 23  25 22  
Sex 
 Male 24    11 13 0.302 
 Female 23    14  
GA at birth, weeks 28.1±2.6 28.1±2.2 28.0±2.9 0.844 26.9±2.5 29.4±1.9 <0.001 
Maternal age at birth, years 30.6±4.8 30.2±5.6 31.0±3.9 0.599 31.4±4.2 29.6±5.4 0.194 
Birth weight, g 1,051±324 1,108±294 991±341 0.222 924±300 1,195±293 0.003 
Head circumference at birth, cm 25.0±3.1 25.5±2.5 24.5±3.5 0.261 23.7±2.8 26.5±2.8 0.001 
Apgar score 
 1 min 5±2.2 4±1.8 5±2.3 0.049 5±2.0 5±2.4 0.73 
 5 min 7±1.8 7±1.9 7±1.8 0.434 7.2±1.6 7±2.2 0.7 
Delivery type 
 Vaginal 18 11 0.278 12 0.032 
 Cesarean 29 13 16  19 10  
NEC 
 Diagnosis 20 11 0.642 19 <0.001 
 Requiring surgery 0.727 0.007 
BPD (moderate-to-severe) 31 16 15 0.917 21 10 0.005 
Sepsis (culture positive) 0.965 0.05 
Feeding at time of discharge 
 Ad lib oral feeding 37 21 16 0.286 20 17 0.436 
 Gastrostomy tube-dependent 
 Unknown, transferred prior to term   
Respiratory needs at discharge 
 Room air 34 20 14 0.085 12 22 <0.001 
 Supplemental oxygen 13 13 
 Tracheostomy-dependent   
Completed weeks at birth 
 34–36 weeks 0.091 0.022 
 32–34 weeks 
 28–32 weeks 22 14 15 
 Under 28 weeks 22 10 12  16  
Morbidity 
 Score 1.7±1.2 1.6±1.1 1.8±1.2 0.557 2.6±0.7 0.7±0.5 <0.001 
 Low-morbidity group 22 13 0.302    
 High-morbidity group 25 11 14     
PMA at time of term MRI, weeks 39.0±1.5 39.0±1.7 39.0±1.3 0.89 39.4±1.5 38.6±1.5 0.095 
Term HRV 
 Alpha 1 1.287±0.231 1.297±0.261 1.276±0.195 0.764 1.190±0.220 1.392±0.198 0.002 
 Alpha 2 1.269±0.083 1.274±0.085 1.263±0.080 0.652 1.297±0.088 1.239±0.065 0.016 
 RMS1 0.015±0.006 0.017±0.007 0.014±0.005 0.204 0.014±0.007 0.017±0.006 0.049 
 RMS2 0.069±0.022 0.075±0.025 0.063±0.018 0.089 0.063±0.023 0.076±0.021 0.041 
 LF/HF ratio 0.769±0.067 0.767±0.063 0.770±0.072 0.908 0.783±0.067 0.750±0.063 0.120 
VariableAllMaleFemalep valueHigh morbidityLow morbidityp value
Total subjects 47 24 23  25 22  
Sex 
 Male 24    11 13 0.302 
 Female 23    14  
GA at birth, weeks 28.1±2.6 28.1±2.2 28.0±2.9 0.844 26.9±2.5 29.4±1.9 <0.001 
Maternal age at birth, years 30.6±4.8 30.2±5.6 31.0±3.9 0.599 31.4±4.2 29.6±5.4 0.194 
Birth weight, g 1,051±324 1,108±294 991±341 0.222 924±300 1,195±293 0.003 
Head circumference at birth, cm 25.0±3.1 25.5±2.5 24.5±3.5 0.261 23.7±2.8 26.5±2.8 0.001 
Apgar score 
 1 min 5±2.2 4±1.8 5±2.3 0.049 5±2.0 5±2.4 0.73 
 5 min 7±1.8 7±1.9 7±1.8 0.434 7.2±1.6 7±2.2 0.7 
Delivery type 
 Vaginal 18 11 0.278 12 0.032 
 Cesarean 29 13 16  19 10  
NEC 
 Diagnosis 20 11 0.642 19 <0.001 
 Requiring surgery 0.727 0.007 
BPD (moderate-to-severe) 31 16 15 0.917 21 10 0.005 
Sepsis (culture positive) 0.965 0.05 
Feeding at time of discharge 
 Ad lib oral feeding 37 21 16 0.286 20 17 0.436 
 Gastrostomy tube-dependent 
 Unknown, transferred prior to term   
Respiratory needs at discharge 
 Room air 34 20 14 0.085 12 22 <0.001 
 Supplemental oxygen 13 13 
 Tracheostomy-dependent   
Completed weeks at birth 
 34–36 weeks 0.091 0.022 
 32–34 weeks 
 28–32 weeks 22 14 15 
 Under 28 weeks 22 10 12  16  
Morbidity 
 Score 1.7±1.2 1.6±1.1 1.8±1.2 0.557 2.6±0.7 0.7±0.5 <0.001 
 Low-morbidity group 22 13 0.302    
 High-morbidity group 25 11 14     
PMA at time of term MRI, weeks 39.0±1.5 39.0±1.7 39.0±1.3 0.89 39.4±1.5 38.6±1.5 0.095 
Term HRV 
 Alpha 1 1.287±0.231 1.297±0.261 1.276±0.195 0.764 1.190±0.220 1.392±0.198 0.002 
 Alpha 2 1.269±0.083 1.274±0.085 1.263±0.080 0.652 1.297±0.088 1.239±0.065 0.016 
 RMS1 0.015±0.006 0.017±0.007 0.014±0.005 0.204 0.014±0.007 0.017±0.006 0.049 
 RMS2 0.069±0.022 0.075±0.025 0.063±0.018 0.089 0.063±0.023 0.076±0.021 0.041 
 LF/HF ratio 0.769±0.067 0.767±0.063 0.770±0.072 0.908 0.783±0.067 0.750±0.063 0.120 

Subjects are divided into male and female subjects in columns three and four and into high- and low-morbidity groups in columns six and seven. Continuous variables were compared between groups using t tests, and categorical variables using χ2 tests. Significant differences are denoted in bold.

HRV Results

All available HRV data were compiled to show the trajectory of HRV maturation from 27-week PMA to 41-week PMA, as demonstrated in Figure 2. Specifically, maturation of HRV was associated with an increase at TEA in alpha 1, RMS1, and RMS2, and a decrease in alpha 2, which corroborates earlier studies mapping the maturation of HRV in premature infants [28]. High-morbidity subjects had reduced HRV (at time of TEA scan) than the low-morbidity group in all HRV metrics except LF/HF ratio, while there were no significant differences in any HRV metric between sexes. Birth GA had a significant inverse relationship with alpha 2 at TEA (r = −0.375, p = 0.01), suggesting that earlier exposure to the extrauterine environment had a negative influence on autonomic maturation. Birth GA was not associated with alpha 1 (r = 0.082, p = 0.586), RMS1 (r = −0.011, p = 0.942), or RMS2 (r = −0.117, p = 0.441) at TEA.

Fig. 2.

Trajectory of average weekly HRV based on sex (females in yellow, males in green) and morbidity (low morbidity in light blue, high morbidity in purple) for each HRV metric: alpha 1 (a, b), alpha 2 (c, d), RMS1 (e, f), RMS2 (g, h), and LF/HF ratio (i, j).

Fig. 2.

Trajectory of average weekly HRV based on sex (females in yellow, males in green) and morbidity (low morbidity in light blue, high morbidity in purple) for each HRV metric: alpha 1 (a, b), alpha 2 (c, d), RMS1 (e, f), RMS2 (g, h), and LF/HF ratio (i, j).

Close modal

HRV-CAN Relationships

We found that connectivity was associated with HRV at TEA in approximately one-fifth of all CAN connections (55 out of 171 connections). Connectivity was found to be significantly correlated with at least one HRV metric in 34 out of 171 connections, while 13 connections were correlated with multiple HRV metrics (Fig. 3). CAN regions with the highest number of significant HRV-CANconn associations included the brainstem (24), precuneus (19), amygdala (15), insula (11), hippocampus (9), and cerebellum (9). Two distinct patterns emerged among these HRV-CANconn relationships. In the first pattern, increased connectivity of the amygdala, hippocampus, insula, cerebellum, and brainstem was associated with increased HRV in alpha 1, RMS1, and RMS2 (Fig. 4a). The second pattern involved connectivity at the precuneus, where increased connectivity of the precuneus was associated with reduced HRV in alpha 1, RMS1, and RMS2 (Fig. 4b). This suggests that stronger connectivity at the precuneus was associated with impaired HRV maturation. No clear HRV-CANconn patterns emerged involving the medial prefrontal cortex or in the anterior cingulate.

Fig. 3.

Summary of all significant correlations between HRV and connectivity for all subjects, grouped by HRV metric. Pearson’s correlation coefficient and p value are reported. Columns with the correlation coefficient are shaded in the manner of a tricolor heat map where values fall on a spectrum between red and green based on their relation to the minimum (red), median (yellow), and maximum values (green). Green represents more positive R values, and red represents more negative R values.

Fig. 3.

Summary of all significant correlations between HRV and connectivity for all subjects, grouped by HRV metric. Pearson’s correlation coefficient and p value are reported. Columns with the correlation coefficient are shaded in the manner of a tricolor heat map where values fall on a spectrum between red and green based on their relation to the minimum (red), median (yellow), and maximum values (green). Green represents more positive R values, and red represents more negative R values.

Close modal
Fig. 4.

Subgroup analysis focused on HRV-CANconn involving limbic regions in (a) and precuneus in (b). There are several additional columns for subgroup analysis of males, females, low-morbidity groups, and high-morbidity groups. Pearson’s correlation coefficient is reported for each group followed by the associated p value. For each subgroup, significant correlations with p value <0.05 are bolded. Columns with the correlation coefficient are shaded in the manner of a tricolor heat map where values fall on a spectrum between red and green based on their relation to the minimum (red), median (yellow), and maximum values (green). Green represents more positive R values, and red represents more negative R values.

Fig. 4.

Subgroup analysis focused on HRV-CANconn involving limbic regions in (a) and precuneus in (b). There are several additional columns for subgroup analysis of males, females, low-morbidity groups, and high-morbidity groups. Pearson’s correlation coefficient is reported for each group followed by the associated p value. For each subgroup, significant correlations with p value <0.05 are bolded. Columns with the correlation coefficient are shaded in the manner of a tricolor heat map where values fall on a spectrum between red and green based on their relation to the minimum (red), median (yellow), and maximum values (green). Green represents more positive R values, and red represents more negative R values.

Close modal

Of all 55 significant HRV-CANconn relationships found, there were very few between-group differences in connectivity. Connectivity was significantly higher in males between the right insula and right cerebellum (p = 0.04), with no other significant differences between sexes. Connectivity differed between high- and low-morbidity groups in only 3 connections, involving the medulla and bilateral insula (right, p = 0.028; left, p = 0.024), and the left cerebellum and right precuneus (stronger in the high-morbidity group, p = 0.005).

We report for the first time the emerging maturational relationships within CAN centers and autonomic function of prematurely born infants. Our findings suggest that measures of autonomic maturation (as measured by HRV) are associated with evolving brain connectivity. Given the complex relationships between the CAN and limbic system, these findings suggest neonatal heart rate signatures may someday provide a noninvasive means to stratify those premature neonates at highest risk of neuropsychiatric impairment. We also report a striking sexual dimorphism in HRV-CAN relationships that is not apparent in other clinical factors in our cohort, suggesting that the brain-heart axis may be more vulnerable in male premature neonates. Brain regions with some of the strongest relationships to HRV maturation included the brainstem, insula, amygdala, hippocampus, thalamus, and the precuneus. Notably, these regions all play a role in processing stress and pain, which are critical inputs to the developing brain even in the neonatal period [34] that in the NICU take the form of painful procedures, illness, injury, discomfort, and limited parent-child interaction. Our findings suggest that stress and pain in this critical period may disrupt the natural trajectory of CAN development, resulting in abnormal HRV-CAN relationships.

Stress in the neonatal period has been shown to be associated with reduced connectivity at the insula, amygdala, and hippocampus [35, 36]. In our study, reduced CANconn at these and other important limbic regions was associated with less mature HRV. Although we did not have a direct measure of stress in our study, it is likely that some component of stress is reflected in HRV given that infants in the high-morbidity cohort exhibited immature HRV when compared to lower morbidity group. Interestingly, there was an inverse HRV-CANconn relationship at the precuneus, where higher CANconn at the precuneus was associated with impaired HRV. Activation at the precuneus is known to be active following painful stimuli in neonates [37], which may suggest that repetitive or prolonged early engagement of the precuneus due to painful stimuli may lead to unusual connectivity patterns that deviate from the normal developmental pathways of the brain.

Particularly at the precuneus but involving subcortical forebrain structures as well, these HRV-CANconn relationships were most striking in male neonates, suggesting biological factors may influence descending CAN pathways. Differences in brain connectivity based on sex have been described in premature neonates previously [38], and male sex in the setting of prematurity remains associated with poorer neurodevelopmental outcomes [39, 40]. Additionally, sex is known to influence pain sensitivity in individuals born prematurely [41, 42], suggesting sex differences in pain processing may begin early in brain development, altering the development of the CAN and its relationship to autonomic function. This study is not the first to identify possible sex influences in neonatal brain-heart relationships but adds to recent work by Blackard et al. [26], which identified abnormal heart rate patterns in the neonatal period in male children diagnosed with ASD. Taken together, this work suggests that brain-heart relationships may be more fragile in the developing male brain and that heart rate analysis may be a better indicator for developing brain connectivity in already vulnerable male neonates.

CANconn of the brainstem, subcortical forebrain, and cerebellum correlated most strongly with alpha 2 than other HRV metrics. This may be because alpha 2 reflects longer-term fluctuations in heart rate and may have captured more of the additive morbidity of infant’s NICU course than other metrics. Our findings also add to previous work from our institution demonstrating maturational trajectory of HRV in premature neonates in relation to morbidity [28], and work establishing a cortical topography for the CAN using electroencephalography in term neonates [21]. We augment this work by reporting the first study of the relationship between development of connectivity within the CAN (by rsfMRI) and development of autonomic pathways that control HRV in newborns prematurely developing in an ex-utero environment.

Limitations of this study include its exploratory nature and volume of data that may limit some of the interpretation of the findings. Given the exploratory nature of the analysis, we did not specifically correct for multiple comparisons. However, there were consistent trends revealed in these data, including sex differences in brain-heart relationships and recurring associations involving limbic regions and the precuneus. Additionally, we defined high or low morbidity using a non-standardized method based on selected clinical characteristics that were not all-encompassing of the potential morbidities encountered in the NICU, with the potential for some overlap in scoring criteria like home oxygen dependence and moderate-to-severe BPD. Lastly, we are limited in that this study can only comment on the existence and not causality of the association between CANconn and HRV; we suspect there is an especially unique bi-directional relationship between the CAN and HRV in premature neonates. While the CAN may modulate autonomic outflow, peripheral autonomic dysmaturity likely also contributes to chronic, repetitive brain changes that may impact brain connectivity and ongoing development of the CAN.

In this study, we examined the CAN in premature neonates and identified a relationship between term HRV and CANconn. Impaired brain-heart electrophysiological relationships may imply abnormal development of the CAN, which given its many components, may reflect abnormal overall brain development in premature neonates. Decoding heart rate patterns that are associated with abnormal brain development may help us better understand the disproportionate neuropsychiatric burden of these infants in childhood and beyond and someday lead to simple, noninvasive means for earlier detection and intervention for premature neonates at higher risk of neurodevelopmental impairment.

This study protocol was reviewed and approved by the Institutional Review Board (IRB) of Children’s National, approval number 2391. All study components were performed within the rules and regulations of the IRB. Informed written consent was obtained from parents or guardians of each participant.

The authors have no conflicts of interest to declare.

Extramural funding for this work included NIH T-32 fellowship support for KC (T32HD098066), DC-IDDRC grant number 1U54HD090257, and NICHD grant number 1RO1 HD099393.

The authors listed above meet authorship criteria based on their substantial contributions to the work. All authors were involved in designing the overall investigation and critical review of the manuscript. Kelsey Christoffel was involved in literature review, statistical analysis, and initial drafting of the manuscript. Josepheen De Asis-Cruz, Kevin Cook, Jung-Hoon Kim, Kushal Kapse, and Kelsey Christoffel were involved in image acquisition, processing, and analysis. Rathinaswamy Govindan extracted and analyzed the heart rate variability data. Emma Spoehr was involved in data collection. Nickie Andescavage, Sudeepta Basu, Catherine Limperopoulos, and Adre du Plessis oversaw the overarching study.

Data that support the findings of this study are not publicly available due to privacy reasons but are available through contacting the corresponding author upon reasonable request (ADP).

1.
Benarroch
EE
.
The central autonomic network: functional organization, dysfunction, and perspective
.
Mayo Clin Proc
.
1993
;
68
(
10
):
988
1001
. .
2.
Valenza
G
,
Sclocco
R
,
Duggento
A
,
Passamonti
L
,
Napadow
V
,
Barbieri
R
, et al
.
The central autonomic network at rest: uncovering functional MRI correlates of time-varying autonomic outflow
.
Neuroimage
.
2019
;
197
:
383
90
. .
3.
Silvani
A
,
Calandra-Buonaura
G
,
Dampney
RAL
,
Cortelli
P
.
Brain-heart interactions: physiology and clinical implications
.
Philos Trans A Math Phys Eng Sci
.
2016
;
374
(
2067
):
20150181
. .
4.
Mulkey
SB
,
Kota
S
,
Swisher
CB
,
Hitchings
L
,
Metzler
M
,
Wang
Y
, et al
.
Autonomic nervous system depression at term in neurologically normal premature infants
.
Early Hum Dev
.
2018
;
123
:
11
6
. .
5.
Thiriez
G
,
Mougey
C
,
Vermeylen
D
,
Wermenbol
V
,
Lanquart
JP
,
Lin
JS
, et al
.
Altered autonomic control in preterm newborns with impaired neurological outcomes
.
Clin Auton Res
.
2015
;
25
(
4
):
233
42
. .
6.
Patural
H
,
Barthelemy
JC
,
Pichot
V
,
Mazzocchi
C
,
Teyssier
G
,
Damon
G
, et al
.
Birth prematurity determines prolonged autonomic nervous system immaturity
.
Clin Auton Res
.
2004
;
14
(
6
):
391
5
. .
7.
Fyfe
KL
,
Yiallourou
SR
,
Wong
FY
,
Horne
RSC
.
The development of cardiovascular and cerebral vascular control in preterm infants
.
Sleep Med Rev
.
2014
;
18
(
4
):
299
310
. .
8.
Thayer
JF
,
Ahs
F
,
Fredrikson
M
,
Sollers
JJ
3rd
,
Wager
TD
.
A meta-analysis of heart rate variability and neuroimaging studies: implications for heart rate variability as a marker of stress and health
.
Neurosci Biobehav Rev
.
2012
;
36
(
2
):
747
56
. .
9.
Keren
O
,
Yupatov
S
,
Radai
MM
,
Elad-Yarum
R
,
Faraggi
D
,
Abboud
S
, et al
.
Heart rate variability (HRV) of patients with traumatic brain injury (TBI) during the post-insult sub-acute period
.
Brain Inj
.
2005
;
19
(
8
):
605
11
. .
10.
Findling
O
,
Hauer
L
,
Pezawas
T
,
Rommer
PS
,
Struhal
W
,
Sellner
J
.
Cardiac autonomic dysfunction in multiple sclerosis: a systematic review of current knowledge and impact of immunotherapies
.
J Clin Med
.
2020
;
9
(
2
):
335
. .
11.
Sambati
L
,
Calandra-Buonaura
G
,
Doria
A
,
Cortelli
P
.
Diagnosis and management of autonomic failure in neurodegenerative disorders
.
Eur Neurol
.
2015
;
73
(
1–2
):
126
33
. .
12.
Rukmani
MR
,
Seshadri
SP
,
Thennarasu
K
,
Raju
TR
,
Sathyaprabha
TN
.
Heart rate variability in children with attention-deficit/hyperactivity disorder: a pilot study
.
Ann Neurosci
.
2016
;
23
(
2
):
81
8
. .
13.
Wang
Y
,
Zhao
X
,
O’Neil
A
,
Turner
A
,
Liu
X
,
Berk
M
.
Altered cardiac autonomic nervous function in depression
.
BMC Psychiatry
.
2013
;
13
:
187
. .
14.
Chalmers
JA
,
Quintana
DS
,
Abbott
MJ
,
Kemp
AH
.
Anxiety disorders are associated with reduced heart rate variability: a meta-analysis
.
Front Psychiatry
.
2014
;
5
:
80
. .
15.
Thapa
R
,
Alvares
GA
,
Zaidi
TA
,
Thomas
EE
,
Hickie
IB
,
Park
SH
, et al
.
Reduced heart rate variability in adults with autism spectrum disorder
.
Autism Res
.
2019
;
12
(
6
):
922
30
. .
16.
Franz
AP
,
Bolat
GU
,
Bolat
H
,
Bolat
H
,
Matijasevich
A
,
Santos
IS
, et al
.
Attention-deficit/hyperactivity disorder and very preterm/very low birth weight: a meta-analysis
.
Pediatrics
.
2018
;
141
(
1
):
e20171645
. .
17.
Agrawal
S
,
Rao
SC
,
Bulsara
MK
,
Patole
SK
.
Prevalence of autism spectrum disorder in preterm infants: a meta-analysis
.
Pediatrics
.
2018
;
142
(
3
):
e20180134
. .
18.
Johnson
S
,
Wolke
D
,
Hennessy
E
,
Marlow
N
.
Educational outcomes in extremely preterm children: neuropsychological correlates and predictors of attainment
.
Dev Neuropsychol
.
2011
;
36
(
1
):
74
95
. .
19.
Treyvaud
K
,
Ure
A
,
Doyle
LW
,
Lee
KJ
,
Rogers
CE
,
Kidokoro
H
, et al
.
Psychiatric outcomes at age seven for very preterm children: rates and predictors
.
J Child Psychol Psychiatry
.
2013
;
54
(
7
):
772
9
. .
20.
Delobel-Ayoub
M
,
Arnaud
C
,
White-Koning
M
,
Casper
C
,
Pierrat
V
,
Garel
M
, et al
.
Behavioral problems and cognitive performance at 5 years of age after very preterm birth: the EPIPAGE study
.
Pediatrics
.
2009
;
123
(
6
):
1485
92
. .
21.
Mulkey
SB
,
Hitchings
L
,
Persaud
R
,
Kota
S
,
Maxwell
GL
,
Baker
R
, et al
.
Cerebral cortical autonomic connectivity in low-risk term newborns
.
Clin Auton Res
.
2021
;
31
(
3
):
415
24
. .
22.
Schlatterer
SD
,
Govindan
RB
,
Murnick
J
,
Barnett
SD
,
Lopez
C
,
Donofrio
MT
, et al
.
In infants with congenital heart disease autonomic dysfunction is associated with pre-operative brain injury
.
Pediatr Res
.
2022
;
91
(
7
):
1723
9
. .In
23.
Andersen
M
,
Andelius
TCK
,
Pedersen
MV
,
Kyng
KJ
,
Henriksen
TB
.
Severity of hypoxic ischemic encephalopathy and heart rate variability in neonates: a systematic review
.
BMC Pediatr
.
2019
;
19
(
1
):
242
. .
24.
Metzler
M
,
Govindan
R
,
Al-Shargabi
T
,
Vezina
G
,
Andescavage
N
,
Wang
Y
, et al
.
Pattern of brain injury and depressed heart rate variability in newborns with hypoxic ischemic encephalopathy
.
Pediatr Res
.
2017
;
82
(
3
):
438
43
. .
25.
Statello
R
,
Carnevali
L
,
Sgoifo
A
,
Miragoli
M
,
Pisani
F
.
Heart rate variability in neonatal seizures: investigation and implications for management
.
Neurophysiol Clin
.
2021
;
51
(
6
):
483
92
. .
26.
Kidokoro
H
,
Neil
JJ
,
Inder
TE
.
New MR imaging assessment tool to define brain abnormalities in very preterm infants at term
.
AJNR Am J Neuroradiol
.
2013
;
34
(
11
):
2208
14
. .
27.
Govindan
RB
,
Massaro
AN
,
Niforatos
N
,
du Plessis
A
.
Mitigating the effect of non-stationarity in spectral analysis-an application to neonate heart rate analysis
.
Comput Biol Med
.
2013
;
43
(
12
):
2001
6
. .
28.
Schlatterer
SD
,
Govindan
RB
,
Barnett
SD
,
Al-Shargabi
T
,
Reich
DA
,
Iyer
S
, et al
.
Autonomic development in preterm infants is associated with morbidity of prematurity
.
Pediatr Res
.
2022
;
91
(
1
):
171
7
. .
29.
De Asis-Cruz
J
,
Donofrio
MT
,
Vezina
G
,
Limperopoulos
C
.
Aberrant brain functional connectivity in newborns with congenital heart disease before cardiac surgery
.
Neuroimage Clin
.
2018
;
17
:
31
42
. .
30.
Cox
RW
.
AFNI: software for analysis and visualization of functional magnetic resonance neuroimages
.
Comput Biomed Res
.
1996
;
29
(
3
):
162
73
. .
31.
Shi
F
,
Yap
P
,
Wu
G
,
Jia
H
,
Gilmore
JH
,
Lin
W
, et al
.
Infant brain atlases from neonates to 1- and 2-year-olds
.
PLoS One
.
2011
;
6
(
4
):
e18746
. .
32.
Braun
K
.
The prefrontal-limbic system: development, neuroanatomy, function, and implications for socioemotional development
.
Clin Perinatol
.
2011
;
38
(
4
):
685
702
. .
33.
Lammertink
F
,
Benders
MJNL
,
Hermans
EJ
,
Tataranno
ML
,
Dudink
J
,
Vinkers
CH
, et al
.
Vulnerability of the neonatal connectome following postnatal stress
.
J Neurosci
.
2022
;
42
(
48
):
8948
59
. .
34.
Tortora
D
,
Severino
M
,
Di Biase
C
,
Malova
M
,
Parodi
A
,
Minghetti
D
, et al
.
Early pain exposure influences functional brain connectivity in very preterm neonates
.
Front Neurosci
.
2019
;
13
:
899
. .
35.
Goksan
S
,
Hartley
C
,
Emery
F
,
Cockrill
N
,
Poorun
R
,
Moultrie
F
, et al
.
fMRI reveals neural activity overlap between adult and infant pain
.
Elife
.
2015
;
4
:
e06356
. .
36.
Kozhemiako
N
,
Nunes
AS
,
Vakorin
VA
,
Chau
CMY
,
Moiseev
A
,
Ribary
U
, et al
.
Sex differences in brain connectivity and male vulnerability in very preterm children
.
Hum Brain Mapp
.
2020
;
41
(
2
):
388
400
. .
37.
Linsell
L
,
Johnson
S
,
Wolke
D
,
O’Reilly
H
,
Morris
JK
,
Kurinczuk
JJ
, et al
.
Cognitive trajectories from infancy to early adulthood following birth before 26 weeks of gestation: a prospective, population-based cohort study
.
Arch Dis Child
.
2018
;
103
(
4
):
363
70
. .
38.
Rysavy
MA
,
Marlow
N
,
Doyle
LW
,
Tyson
JE
,
Serenius
F
,
Iams
JD
, et al
.
Reporting outcomes of extremely preterm births
.
Pediatrics
.
2016
;
138
(
3
):
e20160689
. .
39.
Walker
SM
,
Melbourne
A
,
O’Reilly
H
,
Beckmann
J
,
Eaton-Rosen
Z
,
Ourselin
S
, et al
.
Somatosensory function and pain in extremely preterm young adults from the UK EPICure cohort: sex-dependent differences and impact of neonatal surgery
.
Br J Anaesth
.
2018
;
121
(
3
):
623
35
. .
40.
Mogil
JS
.
Sex differences in pain and pain inhibition: multiple explanations of a controversial phenomenon
.
Nat Rev Neurosci
.
2012
;
13
(
12
):
859
66
. .
41.
Blackard
KR
,
Krahn
KN
,
Andris
RT
,
Lake
DE
,
Fairchild
KD
.
Autism risk in neonatal intensive care unit patients associated with novel heart rate patterns
.
Pediatr Res
.
2021
;
90
(
6
):
1186
92
. .
42.
Mulkey
SB
,
Govindan
RB
,
Hitchings
L
,
Al-Shargabi
T
,
Herrera
N
,
Swisher
CB
, et al
.
Autonomic nervous system maturation in the premature extrauterine milieu
.
Pediatr Res
.
2021
;
89
(
4
):
863
8
. .