Delayed cord clamping is associated with improved dynamic cerebral autoregulation and decreased incidence of intraventricular hemorrhage in preterm infants.

INTRODUCTION
Delayed cord clamping (DCC) improves neurologic outcomes in preterm infants through a reduction in intraventricular hemorrhage (IVH) incidence. The mechanism behind this neuroprotective effect is not known.


METHODS
Infants born <28 weeks gestation were recruited for longitudinal monitoring. All infants underwent 72 hours of synchronized near-infrared spectroscopy (NIRS) and mean arterial blood pressure (MABP) recording within 24 hours of birth. Infants with DCC were compared to control infants with immediate cord clamping (ICC), controlling for severity of illness (CRIB-II score), chorioamnionitis, antenatal steroids, sedation, inotropes, and delivery mode. Autoregulatory dampening was calculated as the transfer function gain coefficient between the MABP and NIRS signals.


RESULTS
45 infants were included (DCC n=15, paired 2:1 with ICC controls n=30). ICC and DCC groups were similar including gestational age (25.5 vs. 25.2 weeks, p=0.48), birth weight (852.3 vs. 816.6 grams, p=0.73), percent female (40% vs. 40%, p=0.75), and dopamine usage (27% vs 23%, p=1.00). There was a significant difference in IVH incidence between the DCC and ICC groups (20% vs. 50%, p=0.04). Mean MABP was not different (35.9 vs. 35.1 mmHg, p=0.44). Compared to the DCC group, the ICC group had diminished autoregulatory dampening capacity (-12.96 dB vs. -15.06 dB, p=0.01), that remained significant when controlling for confounders. Dampening capacity was, in turn, strongly associated with decreased risk of IVH (OR= 0.14, p<0.01).


DISCUSSION
The results of this pilot study demonstrate that DCC is associated with improved dynamic cerebral autoregulatory function and may be the mechanism behind the decreased incidence of IVH.


INTRODUCTION
At birth, infants undergo a dramatic physiological transition, driven by the shift in respiratory function from the placenta to the lungs. During the fetal period, there is significant pulmonary vasoconstriction; oxygenated blood returned from the placenta via the umbilical vein passes directly from the right atrium to the left atrium through a patent foramen ovale with minimal pulmonary blood flow. The onset of breathing introduces atmospheric oxygen to the pulmonary vascular bed, leading to rapid pulmonary vasodilation and increased right ventricular output. The sudden increase in venous capacitance leads to a significant transfusion of blood from the lowresistance placental circuit to the infant (17). Premature clamping of the umbilical cord, before respiration has been fully established, leaves a substantial amount of circulating blood volume in the placenta and may lead to inadequate preload to the left ventricle, as pulmonary blood flow is still limited (22). Delaying clamping of the umbilical cord provides additional time for pulmonary vascular resistance to drop and results in increased circulating blood volume.
Delayed clamping of the umbilical cord following birth has been a long-established practice by many obstetricians and midwives (5). It fell out of favor in the 20th century with advent of modern neonatology, with a focus instead on assessment and initiation of resuscitation. However, there has been a recent resurgence of interest in this practice after several randomized trials demonstrated potential benefits for premature infants, with little apparent downside (13,23,28), and it is now recommended for routine use in all preterm birth (3b).
One of the most provocative findings has been a reduction in rates of intraventricular hemorrhage (IVH) in preterm infants who have received delayed cord clamping (DCC). While this protective effect has been demonstrated in several different studies (8,19), the mechanism remains unclear, with speculation that it may confer this advantage by stabilizing blood pressure (9) or venous blood flow (20), a reduction in supplemental oxygen requirements (13), or stabilizing the cardiovascular system (2,30) or have yet some other effect from increased transmission of stem cells (33).
In a previous study we demonstrated that DCC is not associated with a difference in initiation of inotropic medications or mean arterial blood pressure (MABP) (36). We hypothesized that DCC may exert its effect through stabilization of cerebral hemodynamics, specifically the cerebrovascular autoregulatory system. Fluctuations in perfusion occur over two different time scales (slow, fast) and accordingly are regulated by different systems. Static autoregulation drives changes in large cerebral blood vessel diameter to maintain constant cerebral blood flow but responds too slowly to filter rapid fluctuations. In contrast, dynamic cerebral autoregulation rapidly compensates for sudden changes in perfusion, quickly returning cerebral blood flow to baseline levels. Failure or impaired function of dynamic cerebral autoregulation has been associated with increased risk of IVH in preterm infants (24,31,35). In this study, we examine dynamic cerebral autoregulatory function in a group of infants who received DCC in comparison to a 2:1 matched control group of infants who did not receive DCC to evaluate the magnitude of the effect of DCC on cerebral autoregulation.

Patient selection.
Infants admitted to the Neonatal Intensive Care Unit at St. Louis Children's Hospital, a tertiary care center serving urban, suburban, and rural populations, were recruited as soon as possible after birth for a prospective longitudinal monitoring study. Inclusion criteria were birth Յ28 completed weeks of gestation and placement of an umbilical or peripheral arterial catheter (needed for study methodology; no lines were placed solely for research purposes). Exclusion criteria included known congenital anomalies, lack of arterial access, or moribund status (not expected to survive beyond 24 h of life, as determined by primary medical team). Infants were not randomly assigned to DCC or immediate cord clamping (ICC) groups but rather were placed into groups by year of birth (before DCC protocol was initiated, after DCC protocol was initiated). Informed written consent was obtained from the parents before study initiation. The study protocol was reviewed and approved by the Washington University School of Medicine Institutional Review Board.
DCC protocol. A DCC protocol was initiated at the delivery hospital serving our Neonatal Intensive Care Unit in April 2014 for all infants born before 32 completed weeks of gestation. At delivery, the obstetric and pediatric teams discussed the protocol with the patient to ensure agreement on the inclusion criteria (excluding multiple gestation and placental abruption). Once the infant was delivered, he or she was held at the level of the perineum (vaginal delivery) or placed on the mother's legs (Cesarean delivery). The pediatric team operated the timer and called out the elapsed time in 15-s intervals. The goal delay time was between 45 and 60 s (standard clinical practice). Previous studies have indicated that the majority of placental transfusion occurs in the first 120 s after delivery (22). The goal delay time was chosen as a balance between maximal benefit from DCC and potential harm from a delay in the initiation of resuscitation.
Once the delay was complete, the umbilical cord was doubly clamped and then cut. The infant was handed over to the Pediatric team and resuscitation proceeded. During the delay period, the infants received the standard first step of the Neonatal Resuscitation Program (warm, dry, stimulate) to encourage respiration. Infants with immediate cord clamping received the same treatment. Temperature management strategies included raising the room temperature to 26°C and placement of the infant on the mother's abdomen or held at the level of the perineum, depending on delivery type.
Maternal and infant clinical characteristics. Perinatal factors were collected including antenatal steroid and/or magnesium sulfate administration, type of delivery (vaginal or Cesarean section), and Apgar score at 5 min. Infant clinical characteristics were also collected from the infant's medical record including the components of the controlling for severity of illness score [clinical risk index for babies (CRIB-II) score] (25), a composite value representing degree of prematurity and severity of illness (GA, sex, birthweight, admission temperature, base excess), race, dopamine, or fentanyl administration, mechanical ventilation at 72 h of life, and cranial ultrasound reports. A standardized approach to ultrasound screening was used for all infants, with at least one exam during days 1-3 and at least one more exam during days 7-10. During the study window, there were no major changes in treatment of premature infants at the institution (including respiratory, cardiovascular, and fluid management).
NIRS. Cerebral tissue oxygen saturation (SctO 2) was obtained using a NIRS device (FORE-SIGHT or FORE-SIGHT Elite; CAS Medical Systems, Branford, CT) with a transducer containing an LED-emitter and an optical detector located 25 mm from the light source. The nonadhesive optode was placed in a standardized location on the right frontoparietal scalp (chosen as site due to the limited amount of hair and ease of access to monitor for skin irritation), and recording was conducted over the first 72 h following birth. Recordings were briefly interrupted every 12 h to reposition the sensor to prevent skin bruising or breakdown.
Blood pressure. Per clinical practice, invasive umbilical blood pressure measurements were made using a pressure transducer (Tru-Wave; Edwards Lifesciences, Irvine, CA), which interfaces the umbilical arterial catheter (3.5-Fr Argyle single lumen umbilical vessel catheter; Medtronic, Minneapolis, MN) with the patient monitor (IntelliVue MP70 or MX800; Philips Medical, Andover, MA).
Data collection. Both NIRS and MABP data streams were captured simultaneously on a laptop computer using custom data acquisition software (CAS Medical Systems) in a time-synchronized fashion at a sampling rate of 0.5 Hz. Any interruptions in recording, such as "rest periods," withdrawal of blood through the catheter for laboratory testing, or infusion through the line, were noted by the bedside nurse or research assistant. The resulting data file was exported for offline analysis.
Cerebrovascular autoregulation analysis. The cerebrovascular autoregulatory system may best be thought of as a "black box," which transforms fluctuations in the blood pressure into smooth and stable cerebral blood flow. The degree to which this system dampens systemic blood flow can be modeled mathematically if the input and output are known. Although there has been much investigation into cerebral autoregulation, there has been substantially diversity in the approaches used previously, making it challenging to compare results. Transfer function analysis has emerged as a leading approach, mathematically modeling the degree to which dynamic cerebral autoregulation dampens oscillations in systemic blood flow using known input (arterial blood pressure) and output metrics (cerebral blood flow). This novel approach was validated and published in our previous study (35). With the use of this technique, the gain coefficient can be measured as a logarithmically scaled value where increasingly negative values represent greater degrees of dampening capacity. In this study, the input is the MABP and the output is SctO 2, a proxy of cerebral blood flow.
In an effort to standardize transfer function analysis of dynamic cerebral autoregulation, Claassen et al. (3a) published a white paper in 2016 outlining "best practices" for this approach. We have incorporated these concepts as much as possible into our approach.
Application of this approach to quantifying dynamic cerebral autoregulatory dampening capacity can be summarized as follows: 1) simultaneous MABP and cerebral saturation data were digitally captured in a beat-to-beat fashion with a sampling rate of 0.5 Hz, 2) recording was started as soon as possible after consent and continued uninterrupted until 72 h of life, 3) all infants were located in servocontrolled isolettes designed to maintain a constant skin temperature of 36.5°C, 4) after completion, the entire recording was partitioned into epochs of 20 min and inspected for errors [e.g., motion artifact, interrupted recording, probe not in contact with skin, desaturation (SpO 2 Ͻ 85%; Ref. 37)], 5) epochs containing any errors were rejected (no interpolation was performed), 6) remaining epochs were windowed using the Hanning anti-leakage window before spectral smoothing using Welch's method with 50% superposition, and 7) the normalized mean gain (%saturation/mmHg) and phase (degrees) coefficients as well as coherence were then calculated over the frequency range 0.02-0.25 Hz for each infant. Given the goal of identifying the frequency-specific effect of dynamic autoregulation, the gain coefficient was converted to decibels for evaluation of the frequency response of the this "autoregulatory filter." With the use of the coherence cut-off values defined by Claassen et al. (3a), a recording was rejected if the minimum coherence threshold was not reached given the number of windows of data in the recording. As the recordings were all longer than the 15-window maximum given in the table, a critical coherence threshold of 0.12 was used to achieve 95% confidence.
All signal processing was conducted using an in-house software package developed using MATLAB 9.4 (The MathWorks, Natick, MA).
Statistical approach. As study recruitment spanned the periods preand postimplementation of the DCC protocol, it served as a natural breakpoint between cohorts, providing unbiased groups for comparison without loss of clinical equipoise. The DCC group comprised all infants recruited to the study after protocol implementation, while the ICC control group was derived from infants recruited to the study before initiation of DCC. Given the small sample size, statistical power was increased by matching the control group 2:1 across key drivers of IVH risk: gestational, birthweight, and sex. A between- groups comparison was made for those with and without DCC by use of the Mann-Whitney U-test for continuous variables and a two-sided Fisher's exact test for categorical variables.
Although the inclusion and exclusion criteria were unchanged between groups, there are many factors that might influence autoregulatory function. The effect of DCC on dynamic autoregulation was assessed using a generalized linear modeling, controlling for CRIB-II score [a composite factor encompassing degree of prematurity and severity of illness (25)], medication exposure (including antenatal steroids, postnatal fentanyl, and dopamine), and IVH status. A similar binary logistic regression model was then constructed to assess the association between autoregulatory dampening capacity and risk of IVH using a similar set of confounders (CRIB-II score, antenatal steroids, and fentanyl or dopamine exposure).
Statistical analysis was conducted using R version 3.5.1 (R Project for Statistical Computing, Vienna, Austria).
Recording and data quality. Recording was started at a mean of 16.1 Ϯ 6.1 h of life and yielded a mean of 43.5 Ϯ 2.1 error-corrected hours per infant. The overall mean data rejection rate was 59% (comprised of desaturation 43%; loss of NIRS signal 44%; loss of MABP signal 8%; motion artifact 4.5%; break in recording 0.5%). Error rates for each class of error were similar between the ICC and DCC groups. Analytic output is shown for a single exemplar patient in Fig. 1.
Although none of the recordings were rejected on the basis of insufficient coherence, the mean coherence across the cohort (0.355) was somewhat less than the "optimal" values of coherence (0.4 -0.5) described in the past (4,38). This may be the result of a higher noise floor in the NIRS signal or a notable nonlinear relationship between the signals. In studies of adults, these issues can be overcome by intentional induction of high-amplitude blood pressure oscillations (e.g., Valsalva maneuver) to temporarily increase the signal-to-noise ratio. This option is not available in the neonatal population. To evaluate the impact of DCC on dampening and also on the likelihood of IVH, two regression models were generated. Controlling for severity of illness (CRIB-II score), antenatal steroid exposure, opioid sedation, and inotropic medication use, DCC was associated with an~2 dB increase in dynamic autoregulatory dampening (␤ ϭ Ϫ1.92, P ϭ 0.04) compared with those with ICC. Confirming previous studies, DCC was, in turn, protective of IVH with an odds ratio of 0.14. Variance inflation factor for all factors included in both models were Ͻ5. The complete output of the dampening and IVH regression models is shown in Tables 2 and 3.
Spectral power was nearly identical between the ICC and DCC groups across the entire frequency band from 0 to 0.25 Hz for MABP. Spectral power for the cerebral NIRS signal was somewhat lower across all frequencies for the DCC group, as would be expected from the stronger dampening noted by transfer function gain analysis (Fig. 5).

DISCUSSION
In this study of preterm infants with and without DCC, we have demonstrated that DCC is associated with significantly more robust dynamic cerebral autoregulatory function and a significantly lower rate of IVH. Although previous studies have shown a decrease in rates of IVH in the DCC population (27), the neuroprotective mechanism of DCC has remained unclear as it does not exert its effect through the most obvious pathway of a reduction in rates of hypotension requiring inotropic support (36). This study offers the first evidence of a potential mechanism by which DCC exerts a neuroprotective effect in preterm infants, namely stabilization of the cerebro-   IVH, intraventricular hemorrhage; VIF, variance inflation factor; CRIB-II clinical risk index for babies score. *Significance at P Ͻ 0.05. vascular system and improved resistance to fluctuations in systemic blood pressure.
Early investigation into DCC included speculation as to the underlying effect on the cardiovascular system, particularly that DCC provides cardiovascular stabilization. Sommers et al. (30) described notable improvement in right heart function, with greater superior vena cava flow and greater right ventricular output via larger stroke volumes. Bhatt et al. (2) describe similar cardiovascular effects of DCC in a lamb model, noting improved pulmonary blood flow and stable cardiac output, providing a smoother cardiovascular transition compared with those with ICC. Although there was mixed methodology across these studies, there is no easily identifiable impact of DCC any of these possible mediating pathways for IVH reduction.
Much has been written about the numerous forms of vascular control in the human body, including direct innervation by the autonomic nervous system, stretch receptors, metabolic demand, and cell-to-cell signaling via cytosolic pathways, all of which form an autoregulatory system that is designed to deliver sufficient blood flow to end organs (6,11,15,26). The cerebrovascular autoregulatory system is perhaps the most well studied of these vascular control mechanisms, owing the potentially devastating impact of brain injury when the system fails. In this study, we have mathematically modeled the dynamic autoregulatory system using two measurements, the arterial blood pressure from a central catheter and cerebral oxygen saturations, which serve as practical proxy measurements for the input and output of the cerebral autoregulatory system. Our data suggest that DCC enhances the function of this system, permitting greater dampening of the systemic blood pressure compared with those without DCC.
The mean transfer function gain of Ϫ15.06 dB in the DCC group corresponds to an attenuation of the flow amplitude to 17.7% of what would occur in a pressure-passive system, whereas the mean transfer gain of Ϫ12.96 dB in the ICC group corresponds an attenuation to 22.5%. Although this comparison represents 27% less dampening of amplitude and a 62% less dampening of transmitted energy in the ICC group, the 77.5% amplitude attenuation in the ICC group still indicates that a high degree of vasoreactivity is present in many of the ICC neonates. More research is needed to evaluate the impact of standard neonatal care on blood pressure regulation and/or cerebrovascular autoregulation (e.g., endotracheal intubation, blood transfusion, medication administration) to interpret this difference and to investigate possible interventions to manipulate autoregulatory function for improved neuroprotection.
In this study we have compared the ensemble average of dynamic autoregulatory function over a wide frequency band between infants with and without DCC. This measure provides an estimate of average autoregulatory function over the 72-h study period and is appropriate for the single outcome measure of IVH, most often obtained on ultrasound imaging around 72 h of life. As shown in Fig. 1, autoregulation is dynamic and likely changes with clinical circumstance and time. Future studies should incorporate more frequent imaging for closer identification of the timing of IVH, permitting investigation into autoregulatory function before and after the hemorrhage.
Estimates of the volume of transfused blood during DCC ranges between 5 and 20 ml/kg (1,18). We speculate that the increased volume may stabilize arterial baroreceptors. Baroreceptors are mechanoreceptors that are excited when increased intravascular volume distends the vessel and activates the autonomic nervous system, reducing cardiac output and alter- ing peripheral vascular resistance (7). This baroreflex response is responsible for maintaining the MABP within a set range and provides short-term correction during periods of hyper-and hypotension (32). The "smooth cardiovascular transition" associated with DCC may arise from the increased intravascular volume causing a time-dependent decrease in atrial stretch receptor firing, thus leading to substantially less cardiovascular volatility.
This combination of factors may reduce the risk of IVH by two mechanisms. First, dynamic autoregulation may be directly improved, as the sensitivity of the baroreceptor reflex is inversely related to the function of cerebral autoregulation (21,34). Alternatively, DCC may separately impact the cardiovascular system (decrease in perfusion fluctuation from stable heart rate and blood pressure) and the cerebrovascular system (improved dampening), the net effect of which is to reduce the likelihood of hemorrhage.
There are a number of limitations of this study, the most obvious of which is the sample size. This was not a prospective randomized trial, but rather a prospective observational study with fortuitous timing of a change in clinical practice (adoption of DCC). Although there were no other major changes in clinical practice during the time period of this study, other minor changes in practice difficult to quantify (including turnover of practitioners) may have occurred, potentially confounding the study. Owing to the study methodology, including the need for arterial access for at least 72 h following birth, future investigation is hampered by decreased placement of arterial catheters. This is likely the result of improved postbirth cardiorespiratory stability in an era where 70 -90% of all women giving preterm birth are given antenatal steroids (3), and perhaps even the effect of DCC itself, although this cannot be proven. Placental transfusion was conducted exclusively by DCC in this cohort, due to institutional practices. It remains unclear if cord milking has a similar effect, although it is not unreasonable to assume a similar outcome, given equivalent outcomes noted in randomized trials (12,16). While not a limitation, it is important to note the difference in fentanyl sedation use between the two groups of infants. We speculate that this difference is driven in large part by the greater rate of mechanical ventilation in the DCC group (86 vs. 79%).
The partial pressure of CO 2 in the blood has not been factored into this analysis but is an important factor in the function of the autoregulatory system. Current technology for continuous capture of CO 2 levels do not work well in the very preterm population. The small tidal volume of infants Ͻ1,000 g (Ͻ5 ml) is below the threshold for reliable measurement for the typical end-tidal CO 2 detector (14,29). Transcutaneous CO 2 detectors can be used in the very preterm population but have significant issues with reliability (10) and have the potential to cause burns on the skin. New technology will need to be developed for reliable, safe CO 2 monitoring and will be essential for incorporation into any model of cerebrovascular autoregulation.
Conclusions. DCC has offered a simple, low-cost method for improving neurologic outcomes in preterm infants. The results of this pilot study suggest that improved dynamic cerebral autoregulation is a possible mechanism for this neuroprotective effect. Further validation in a larger cohort should be performed.