Protective Effects of Early Neonatal Methylxanthine Treatment on Cognitive and Language Outcomes in Premature Infants with and without High-Risk Perinatal Factors

Caffeine and other methylxanthines are drugs used in the care of premature infants [1]. Caffeine and theophylline are effective methylxanthines (“xanthines”) for the treatment of apnea of prematurity [1, 2]. Newborns can cross-metabolize these xanthines via methylation/demethylation, and both forms increase metabolic rate and oxygen consumption of preterm infants [3]. Caffeine is now used almost exclusively due to a longer half-life and fewer side effects [4].

Cumulative evidence suggests that xanthines also provide neural protection against hypoxic-ischemic (HI) injuries in newborns. HI reflects a reduction in oxygen and/or blood flow to metabolically active brain tissue, a common occurrence in preterm infants [5]. HI can be the result of fetal distress in the perinatal period, intraventricular hemorrhage (IVH), reperfusion failure, and/or intermittent hypoxia associated with respiratory distress and/or apnea [6, 7].

While xanthines are not approved for the treatment of HI in preterm human newborns, cumulative animal data suggests that they may offer an important therapeutic option [8‒10]. One comprehensive longitudinal study of randomized premature infants treated with caffeine or placebo (“CAP” trial) reported that caffeine treatment led to a lower incidence of cognitive delay at 18–21 months of age [11, 12], with benefits still evident at 11 years [13]. This important study did not offer mechanism(s) of caffeine protection nor a comprehensive evaluation of perinatal factors that might lead to caffeine being more beneficial in a subgroup of the patient population. Specifically, though subgroup analyses were performed in the CAP trial, subanalyses focused on differences in cognitive outcomes by association with postnatal and discharge conditions such as BPD rather than perinatal conditions (e.g., chorioamnionitis [chorio] and preeclampsia that are inflammatory risk factors) and also used limited outcome measures (i.e., developmental delay, cerebral palsy [CP], severe disability, and death) [14]. Other studies have also shown motor, sensory, cognitive, and language benefits of xanthines when administered less than 2–3 days postdelivery in very low birth weight infants [15, 16]. Interestingly, one study found that benefits were most robust for very low birth weight infants with chorio, postulating specific anti-inflammatory mechanisms of protection [16].

Preterm animal studies have also documented the neuroprotective effects of caffeine following HI injury when neuropathological and behavioral outcomes are evaluated. Multiple studies from our lab and others show that impairments in motor, cognitive, memory, and sensory processing tasks in preterm-equivalent rat pups with neonatal HI can be prevented via immediate caffeine treatment, with concomitant brain tissue preservation relative to untreated HI counterparts [8‒10]. Importantly, many of the behavioral tests used in these animal studies of neonatal HI (e.g., rapid auditory processing) are the same or similar to those used in assessing cognitive and language function in human infants and children [17]. Neuroprotection has also been shown in rabbit and mouse models of HIE following caffeine treatment [18, 19]. Many of these studies revealed better outcomes following immediate versus delayed treatment [20].

With regard to mechanism of action, both caffeine and theophylline are nonselective adenosine antagonists and can bind to all four types of adenosine receptors in the brain (A1, A2A, A2B, and A3). Adenosine levels rise in response to cerebral hypoxia, leading to increased cerebral inflammation and heightened immune response. Adenosine receptors serve as important modulators of this postinjury cascade [21], with postinjury caffeine treatment reducing plasma inflammatory markers measured within 24 h of HI (Fitch, unpublished data) [22].

Following HI, neuroprotective xanthine effects are thought to be mediated by antagonism at either A1 and/or A2A receptor sites. On neurons, antagonism of A2A receptors (A2AR) prevents excess calcium influx and subsequent release of protein-bound intracellular calcium, thus reducing cell death [23]. The A2AR is also present in microglia. Antagonism here prevents microglial activation, reduces inflammation, and mitigates cell death and tissue loss [24]. Caffeine decreases microglial activation after HI injury only in male rats [10], yet both male and female preterm rat pups show comparable neuroprotection, suggesting that caffeine exerts protective effects through different adenosine receptor sites and mechanisms in females (e.g., neuronal A2AR). Although all the mechanisms of protective xanthine action following HI injury remain unknown, the net effect is the preservation of tissue and brain function.

In the current study, we sought to explore the importance of timing in neonatal xanthine treatment. Timing has been shown as a critical mediator of postinjury rescue efficacy for many therapies. To this end, we compared outcomes among preterm infants treated with methylxanthine (theophylline or caffeine) at <48 h after delivery versus >48 h after delivery. The cutoff of 48 h was based on prior human and animal data as described above [8, 15, 16, 20, 25, 26]. An improved understanding of the role of timing could greatly impact how newborn infants are treated at birth, including recommendations for routine caffeine treatment in at-risk infants as soon as possible after delivery.

In addition to timing, other critically important variables should be considered in assessing outcomes with xanthine treatment in preterm infants. For example, it is well known that preterm female neonates have better outcomes than males under comparable conditions [27‒29]. Our work has also shown that the mechanism of xanthine protection differs by sex [10]. This could mean that one sex may benefit more from caffeine treatment than the other. Such a sex difference would not be unprecedented. For example, neonatal rat HI studies show that females exhibit superior overall outcomes than males and that they benefit more following comparable hypothermia treatment [30, 31].

Infants born to mothers with preeclampsia show a greater risk for antenatal hypoxia and inflammation [32, 33], with poorer neurodevelopmental outcomes. This is likely due to inflammatory proteins from the placenta entering the fetal circulation and fetal brain [34]. “Chorio” can also lead to perinatal neuroinflammation [35, 36] and is clinically associated with periventricular leukomalacia, CP, and neurodevelopmental disorders. Given these inflammatory risks, preeclampsia and chorio were included in our study as perinatal inflammatory risk factors.

In brief, the current study assessed neurodevelopmental outcomes of preterm infants between 23 and 30 weeks’ gestation born at the University of Connecticut Health Center (UCHC) and cared for at the UCHC/Connecticut Children’s Medical Center (CCMC) NICU from 1991 to 2017 (initial n = 858). Our purpose was to determine the efficacy of xanthines based on treatment timing and explore possible interactions with sex and perinatal inflammatory risk status (preeclampsia/chorio). The vast majority of preterm infants admitted to UCHC received xanthine treatment at some point during hospitalization (>91%). However, substantial variability in timing allowed us to assess the impact of hours until treatment (<48 h or >48 h postdelivery), along with sex and perinatal inflammatory conditions (preeclampsia/chorio), using behavioral outcomes obtained at the Neurodevelopmental Follow-up Clinic at UCHC and CCMC at 18 months of age. Theophylline and caffeine treatment were combined in the xanthine group based on pharmacological and clinical comparability as well as the fact that infants readily metabolize the drugs interchangeably [4, 37].

Our main hypothesis was that the timing of xanthine treatment would influence cognitive and linguistic outcomes, with infants treated within 48 h of birth showing better outcomes. A secondary hypothesis was that treatment timing and perinatal inflammation risk would interact, with early xanthine treatment providing particular benefits to infants with high inflammatory exposure.

All methods were approved by the institution’s Institutional Review Board (IRB; CCMC IRB #19-112 and UCHC IRB 19X-211-1). Consent was waived given the retrospective study design.

Subjects were inborn infants at the University of Connecticut Health Center (UCHC) and admitted to the UCHC/Connecticut Children’s Medical Center (CCMC) Neonatal Intensive Care Unit between January 1, 1991 and December 31, 2017. All admitted subjects who met study criteria (born at 23–30 weeks gestational age [GA], and without IVH/PVH grade >III) were initially included (n = 858). Infants who had an IVH or PVH >III were excluded given well-established deleterious impacts on cognitive and language outcomes that could interact with xanthine timing and mask or confound other findings. Only those infants with an ∼18-month neurodevelopmental follow-up were included in order to assess outcomes. We initially included infants who did not receive xanthine in analyses. However, a few (n = 86) infants born at 23–30 weeks over this period who did not receive xanthines were found to be older, heavier, and with a shorter hospital stays. Since several of these indices differed prior to any xanthine treatment, the subset was excluded from further analyses of xanthine effects (Table 1). Some infants were excluded from treatment timing analysis and treated infant calculations due to an error in treatment timing coding (n = 2). For the final n, the 81% who both had a neurodevelopmental follow-up and were treated with xanthine were retained for the final analysis (n = 696).

Medical data collected during the NICU stay (admission through discharge) were stored in a computerized clinical database called the Neonatal Information System (NIS) (Medical Data Systems, Philadelphia, PA). Patient data were retrieved from the database. To account for changes in practices and care over time, data were categorized into 3-decade groups – 1991–2000, 2001–2010, and 2011–2017. Original data were extracted from patient charts in real time and entered into the database by NICU nurses trained in data entry, following consistent medical definitions. There were over 400 variables in the core dataset. Variables included demographic information, birth history, vital signs, laboratory results, treatment data, comorbid diagnoses and feeding requirements among others. Only the variables of interest for this study were extracted and entered into a study database (GA, preeclampsia, chorio, magnesium sulfate treatment, mode of delivery, birth year, sex, birth weight, Apgar score, cord pH, Snap score, IVH grade, length of stay (LOS), timing of xanthine treatment, duration of treatment). Follow-up behavioral data were extracted from the UCHC NIS High-Risk Follow-up database and the CCMC Neurodevelopmental Follow-up Clinic database and paired with data from the NICU stay. Subjects were de-identified and assigned a new subject number at the time of data collection.

We were able to gather some information about the general insurance usage in the UCHC NICU, which can relate to SES. Between 1990 and 2000, 56% of infants (23–30 weeks’ GA) used private insurance and 44% used public. Between 2000 and 2006, 52% of infants (23–30 weeks’ GA) used private insurance and 48% used public. This trend has remained fairly consistent in the UCHC NICU and gives us little reason to believe that SES is playing a large driver of outcomes, though we cannot confirm this in the current study, as SES or insurance type was not collected before de-identification.

The time elapsed between birth and methylxanthine administration was calculated by taking the difference between the date and time of the first xanthine dose, and the date and time of birth. Infants with a greater than 48-h administration time were put into the >48 h xanthine group (n = 261; mean initiation of treatment = 16,927 days, SE = 0.854), while those treated within 48 h were placed in the <48 h xanthine group (n = 435; mean initiation of treatment = 0.609 days; SE = 0.0298). Infants whose mothers had chorio and/or preeclampsia were assigned to a group termed “high prenatal risk” for further analysis (n = 212 high-risk infants, including both early and late xanthine subjects).

Behavioral and cognitive outcomes were collected at follow-up visits at approximately 18 months of age (corrected for GA at birth). Outcomes were measured using the Bayley and/or CAT/CLAMS assessments. Only infants for whom adequate follow-up measures were obtained were included in this study (n = 696; retention rate = 81%). We note that multiple editions of the Bayley scales (II–III) were in use during the period of the study. Some infants received the Bayley III (n = 177), the Bayley II (n = 97), the CAT/CLAMS (n = 411), and both the Bayley II and CAT/CLAMS (n = 11). To accommodate the use of multiple instruments of varying scales, Z-scoring of subscores was performed (1) within decade and (2) within subtest (either cognitive/CAT or cognitive/Bayley III, etc.), prior to consolidation into a mean “cognitive” or “language” score. For the Bayley III, this included Gross Motor, Fine Motor, Problem Solving, Expressive Language, Receptive Language, Language Articulation, Self-Help, and Relationships with Others or the Motor, Cognitive, and Language Composite Scores; for the Bayley II, language and cognitive subtest components, which are used to calculate the mental developmental index, were recorded; for the CAT/CLAMS (the Cognitive Adaptive Test [CAT] and the Clinical Linguistic Auditory Milestone Scale [CLAMS]), subcomponents included Cognitive and Language scores. Raw scores from these subtests were converted to z-scores for each subject, using the formula $z=x−μ/σ$⁠, where x is the raw score, μ is the population mean, and σ is the population standard deviation. This calculation was performed independently for each assessment subtest within each of the 3 decades (1991–2000, 2001–2010, and 2011–2017), accounting for changes in practices and/or testing over the 27 years. Subtest Z-scores were then categorized into 4 domains: Social (comprised of the Self-help and Relationships with others Z-scores), Cognitive (comprised of Problem Solving, Cognitive Composite score, and the CAT Z-scores), Language (comprised of Expressive Language, Receptive Language, Language Articulation, Language Composite score, and the CLAMS Z-scores), and Motor (comprised of Gross Motor, Fine Motor, and Motor Composite score Z-scores). Scores were averaged within-category for any subject with multiple measures. This increased the pool of infants who had at least one measure in Cognitive and Language domains. However, very few subjects received raw test scores that allowed us to calculate scores for Motor and Social. Thus, only Cognitive and Language outcomes are reported. There were some infants for whom an ∼18-month Cognitive or Language z-score could not be calculated (n = 9).

The use of all available test data was necessary because not all toddlers could complete all the tasks, and none had scores for all assessments. Some score components were also missing due to test version (Bayley II vs. Bayley III). By performing Z-scoring within subtasks across all infants who had a score (per decade), we were able to standardize individual performance to the available data without bias toward any group of infants. Such normalization is common and necessary when using large retrospective clinical infant datasets, given well-established constraints in capturing infant/toddler follow-up data. Finally, the literature shows that CAT/CLAMS and Bayley scores yield highly intercorrelated measures [38], further supporting our normalization.

After data collection, de-identification, and consolidation, SPSS software was used for analysis. Initially, the early (<48 h), late xanthine (>48 h), and no treatment groups were compared on demographic and medical indices to ascertain possible differences in health status that might confound results. Evidence of a priori GA and weight differences led to exclusion of the no-treatment group. We saw no significant differences between the two remaining groups (early/late xanthine; Table 1) on GA, distribution of sex, weight, or length of hospital stay. Pearson correlations were also run on some of the continuous measures. We also assessed sex differences for the entire study sample, including nontreated infants, and the treated sample to replicate previously described sex differences in cognitive outcome.

Our primary aim was to understand how timing of xanthine treatment affected cognitive and language outcomes. To address this, we assessed outcomes in Cognitive and Language domains (Bayley and CAT/CLAMS averaged subtest Z-scores, 2 composite domains) via multivariate ANCOVA (see below for covariates and fixed factors) for Early versus Late groups. We also analyzed the interaction between sex and treatment timing using multivariate ANCOVA.

A secondary aim was to examine the effect of perinatal inflammation risk on 18-month outcomes. To address this, Early and Late groups were split based on prenatal inflammatory risk condition (without sex). The new 4 groups (prenatal risk xanthine <48 h; no prenatal risk xanthine <48 h; prenatal risk, xanthine >48 h; no prenatal risk, xanthine >48 h) were compared for Cognitive and Language scores using multivariate ANCOVA. In any analysis where key measures were not included as a between-variable (decade, sex, prenatal risk [preeclampsia and/or chorio]), these values were included as covariates or fixed factors.

Table 1 shows total n by subgroups (sex, timing, prenatal risk), as well as means and standard error for weight, GA, and LOS for all groups. The no-treatment group was excluded from timing analysis due to significant group differences on GA, weight, and LOS (p < 0.001) on a Tukey post hoc analysis, indicating better health (older, heavier, shorter stays). These a priori health differences introduced bias that would prevent any interpretation of timing group outcome differences. There were no significant a priori differences, however, between the Early and Late xanthine groups on a Tukey post hoc analysis. The likelihood of ventilation or other respiratory support is directly related to GA and weight [39], as well as with LOS [40], and these relationships are particularly strong within a single NICU. LOS is also highly related to incidence of BPD and other chronic respiratory issues [40]. The fact that none of these variables differed significantly between the Early and Late xanthine groups (GA, p = 0.548; birthweight, p = 0.892, LOS, p = 0.069) makes it highly unlikely that differences in respiratory support led to observed differences in outcomes. We did however see a significant difference in birth weight by sex (p < 0.05), supporting the use of sex as a covariate.

We performed further validation analyses on our dataset by running correlations between outcomes and common measures of infant wellness (GA, birth weight, and LOS in the NICU). We found a weak but significant positive correlation between GA and cognitive outcomes (r = 0.103, p = 0.007), confirming an expected pattern of older GA infants showing better cognitive outcomes. We saw a similar expected correlation between birth weight and cognitive outcomes (r = 0.142, p < 0.001) as well as Language (r = 0.082, p = 0.031), with heavier babies also showing better outcomes.

There was a significant main effect of sex on cognitive outcomes (F(1, 775) = 9.114, p = 0.003, η² = 0.012) in the overall subject pool (including nontreated infants). This sex effect remained when nontreated infants were excluded (F(1, 687) = 8.821, p = 0.003, η² = 0.013). These results show higher scores among females, supporting the use of sex as a covariate in further analyses. This result also replicates the sex differences reported in other studies (27–29). Sex was not significant for Language outcomes in the subject pool including nontreated infants (F(1, 775) = 0.335, p = 0.563, η² = 0.000) or excluding nontreated infants (F(1, 687) = 1.362, p = 0.244, η² = 0.002). We also saw no interaction between Sex × Treatment timing on cognitive outcomes (F(1, 687) = 0.065, p = 0.799, η² = 0.000) or Language outcomes (F(1, 687) = 1.238, p = 0.266, η² = 0.002).

We next report effects of treatment timing to address our first aim (Fig. 1). We found significant main effects of xanthine timing on both Cognitive and Language measures. Infants treated <48 h after birth (Early) performing significantly better than those treated with xanthines >48 h after birth (Late) on both cognitive outcomes (F(1, 687) = 7.867, p = 0.005, η² = 0.012; Fig. 1a), and Language outcomes (F(1, 687) = 4.673, p = 0.031, η² = 0.007, Fig. 1b). Note that the multivariate ANCOVAs were run using Decade of birth (3 levels) as a covariate, sex (2 levels), magnesium sulfate (y/n), preeclampsia (y/n), and chorio (y/n) as fixed factors.

Next, the infants in the high prenatal risk group (perinatal chorio, preeclampsia, or both) had significantly lower cognitive scores compared to those with low prenatal risk (no chorio or preeclampsia) when xanthine treatment was Late (>48 h; F(1, 260) = 6.693, p = 0.010, η² = 0.026, Fig. 2). This difference was not seen in the Early (<48 h) group (F(1, 427) = 0.018, p = 0.893, η² = 0.000). Similarly, there was a significant difference between Early/Late among infants in the high prenatal risk group, with worse scores in the late-treated group (F(1, 210) = 4.356, p = 0.038, η² = 0.021 Fig. 2). Overall results show that infants with high prenatal inflammatory risk had substantially better cognitive outcomes if treated with xanthines within 48 h of birth.

Similar effects were seen for language outcomes at 18 months, with the high prenatal risk infants showing lower language scores if they were in the Late treatment group (F(1, 260) = 5.784, p = 0.017, η² = 0.023, Fig. 3), a difference not seen within the Early treatment group (F(1, 427) = 0.031, p = 0.860, η² = 0.000 Fig. 3). Again, multivariate ANCOVAs were performed using decade of birth as a covariate and magnesium sulfate treatment and sex as fixed factors.

The current dataset is validated by correlations consistent with well-established evidence that higher GA and birthweights as associated with better developmental scores [41]. We also found an overall sex difference in cognitive outcomes (females showing higher scores), consistent with prior findings [27, 28]. However, sex was not found to interact with the effects of xanthine treatment timing in the current study.

The primary finding from the current study is that xanthine treatment <48 h postdelivery offered a significant benefit to Cognitive and Language outcomes in a preterm population (23–30 GA), confirming prior work that xanthine treatment within the first 48 h of birth leads to significantly better developmental outcomes [20, 25]. The significant benefit of early (<48 h postdelivery) xanthine treatment on cognitive and language outcomes was seen across all infants in the current dataset, regardless of sex or prior prenatal conditions (Fig. 1). This result adds to a growing body of work, including the CAP Trial, showing that postdelivery treatment with xanthines is beneficial to the preterm infant [20, 25].

Identification of a neuroprotective intervention for this population is of utmost importance because preterm infants are at risk for repeated HI brain injury resulting from apnea of prematurity, immature cardiopulmonary systems, and inflammation due to prenatal factors [42‒46]. Since most instances of HI-related injury such as IVH occur within the first few days of life following preterm delivery, protective treatment as soon after delivery as feasible may be ideal. It should be noted that all of our results had a small effect size, and further research will be required to confirm our results. Caffeine is easily administered and is amenable to use in low-resource and underserved areas with limited medical resources. Based on all evidence, we suggest that the use of caffeine in preterm infants should be considered in ongoing clinical studies for inclusion in routine care [42‒46].

The second critical and unique finding in our current study was that in the presence of perinatal inflammation (chorio and/or preeclampsia) infants who received late (>48 h post-birth) xanthine treatment had poor Cognitive and Language outcomes (Fig. 2, 3). This is not an entirely unexpected result given substantial evidence that these conditions cause elevated inflammatory proteins and end-organ damage, especially in the brain [32, 35, 47, 48]. Inflammation is associated with poor outcomes following brain injury (both in the preterm brain and in the adult traumatic brain injury) [49‒51]. Infection leading to sepsis can itself lead to brain injury in adults (sepsis-associated encephalopathy), as a result of the breakdown of the blood-brain barrier [52]. Potentially similar events occur in our high-risk infant group following perinatal exposure to inflammatory proteins in the bloodstream. In support of this view, preterm newborn lambs exposed to lipopolysaccharide (a component of Gram-negative bacteria commonly seen in chorio) showed increased inflammation and reductions in white matter at 15 days following exposure as compared to untreated lambs [53].

Our third critical and unique finding was the effect of the timing of xanthine treatment and prenatal exposure to inflammation (chorio and/or preeclampsia). Xanthine treatment within the first 48 h after birth significantly improved outcomes and generated enough protection to counteract the highly deleterious consequences of prenatal inflammation as seen for cognitive outcomes in the late-treated high-risk group (Fig. 2). Such effects have been suggested in a prior study [54]. Importantly, this result does not mean that caffeine offers no benefit to preterms without prenatal risk but rather, that the importance of immediate treatment is heightened in the high-risk group. This protective effect of early xanthine therapy in the presence of inflammation may have driven the positive effect of xanthines on the entire study population.

Our results also suggest that the mechanism for neuroprotection in this population is mediated via the blockade of adenosine receptors in the brain, though more research is needed. It would appear that this blockade both diminishes the production of inflammatory proteins and blocks their effect on the brain tissue.

Implications of the current results for future xanthine treatment of infants with prenatal risk are substantial. Indeed, the data warrant additional studies on long-term outcomes for term infants with prenatal exposure to chorio [55] or preeclampsia. Current literature shows an increased risk of brain injury and CP in term infants exposed to chorio [56, 57]. As such, it would be important to know if early xanthine administration could ameliorate brain inflammation and brain injury in this population. Future studies and clinical trials should explore these perinatal influences further.

We acknowledge several limitations to the current study. First, we could not access information about the mother’s socioeconomic status or education level, which can certainly influence developmental outcomes. We were able to gather some information about insurance usage, although this information (collected after subjects had been de-identified) was not able to be used as a covariate and controlled for. Future studies of a similar nature should collect and control for these data. Second, we were only able to report developmental results at 18 months and not later (due to high levels of subject attrition at 24-month and 3-year follow-ups). Third, we had insufficient data points within the Social and Motor subtests to evaluate the beneficial effects of xanthine in these domains. Fourth, we were unable to compare outcomes for preterm infants treated with xanthines at any time versus those not treated, due to the very small number of preterm infants across these 27 years who did not receive xanthines, as well as the fact that this subgroup was significantly older and healthier than the overall sample. It is important to highlight this point because the fact that low-prenatal-risk infants show little effect of xanthine timing in our sample does not imply that xanthine treatment per se is not beneficial (as has been shown repeatedly elsewhere), only that the importance of speedy treatment is not as relevant as in the high prenatal risk subgroup. It is also important to acknowledge that due to the methodology of this study, we cannot account for immortal time bias. Future studies should take this bias into account in their design and replication of these findings.

Fifth, we were unable to gather consistent data on the reason for xanthine treatment (including ventilation), limiting the interpretation of results. We attempt to address this by showing comparable a priori heath indices in Early and Late groups, and by using decade of birth as a covariate (to address any concurrent changes in ventilation/xanthine practices). Sixth, the ratios of early to late treatment were not proportional for each decade, which limits the interpretation of our results without bias. We did attempt to address this concern using decade of birth as a covariate, but future studies should control for this change of practice. Finally, we acknowledge that this protocol was not registered online before it began. Future studies using comprehensive datasets that include these additional variables and protocol methods could further refine and expand the current results.

In summary, the current results show that early initiation of xanthine treatment is associated with significantly better Cognitive and Language outcomes in infants born at 23–30 weeks’ GA, with the greatest impact in the subgroup of infants exposed to perinatal inflammation (preeclampsia and chorio). Our findings support a growing body of knowledge about xanthines and their ability to modulate inflammation in the brain and other organs in the preterm infant and may have profound implications for expanded therapeutic use in the at-risk neonatal population, with further research.

This study protocol was reviewed and approved by the Connecticut Children’s Medical Center Institutional Review Board (Approval No. 19-112) and the University of Connecticut Health Center Institutional Review Board (Approval No. 19X-211-1). This study has been granted an exemption from requiring written informed consent by the above institutional review boards (CCMC IRB #19-112 and UCHC IRB 19X-211-1).

The authors have no conflicts of interest to declare.

This material is based upon work supported by the National Science Foundation under Grant DGE-1747486. The funder had no role in the design, data collection, data analysis, and reporting of this study.

R.M. Mcleod was responsible for the design of the work, data acquisition, analysis, interpretation, and writing of the manuscript. T.S. Rosenkrantz was responsible for the design of the work, data interpretation, and editing of the manuscript. R.H. Fitch was responsible for the design of the work, data analysis, interpretation, and editing of the manuscript. All authors agree to be responsible for the work, its accuracy, and integrity.

The dataset used in this study is not publicly available due to containing information that could compromise the privacy of participants. Requests to access the dataset should be directed to the corresponding author, R.M.M.