Experience of early-life pain in premature infants is associated with atypical cerebellar development and later neurodevelopmental deficits

Nearly 1 out of 10 babies worldwide are born premature, with minimal change over the past decade [1‐4], although survival rates have increased significantly [5, 6]. Despite increasing survival, infants born very preterm (VPT) < 32 weeks postmenstrual age (PMA) and extremely preterm (EPT) < 28 weeks PMA [7] remain at heightened risk for developmental deficits even in the absence of any detectable brain injury, especially sensorimotor [8‐10], language [11, 12], and cognitive [10, 12, 13] impairments. Moreover, they experience higher rates of neurodevelopmental disorders such as autism spectrum disorder (ASD) [14‐17].

While the association between prematurity and developmental outcomes is well established, mechanisms driving the association remain unclear. One possibility is alterations in brain development arising from earlier-than-typical exposure to the extrauterine environment. The late second through third trimester is a critical period for brain development, seeing rapid increases in global and regional brain volumes [18‐22], strengthening of white matter tracts [23, 24], and organizing of functional brain networks [25, 26]. While V/EPT infant’s brains exhibit volumetric increases [27], it is less than term-born infants, especially in the cerebellum and brainstem [22, 28‐30]. Furthermore, functional network development in preterm infants is disrupted compared to term-born infants [31, 32].

Importantly, not all infants born V/EPT will experience neurodevelopmental impairments. It is likely the qualities of experiences during this critical period play a role in modifying trajectories. Neonatal intensive care units (NICU) are significantly more sensorially intense relative to the intrauterine environment, with bright lights and loud noises that reach as high as 85 decibels [33, 34]. Exposure to these intense sensory experiences have been linked to altered development (for review see Ream & Lehwald (2018)) [35]. In fact, one study found that the rates of neurodevelopmental disorders such as ASD are higher following a NICU stay even in term-born infants, suggesting NICU exposure itself may play a role [36].

Lengthy NICU stays for V/EPT infants include exposures to various noxious stimuli, especially repeated painful exposures. These infants are regularly exposed to painful procedures over the course of their clinical care, including heel sticks, venipunctures, and catheterizations. Infants born VPT experience an average of 115 painful procedures during their NICU stay, with some reporting over 600 [37] and most occur in the absence of any analgesia [38‐40]. Pain pathways are vulnerable to perturbations in V/EPT infants, because of asynchronicity in the earlier developing afferent nociceptive pathways and later maturating efferent modulatory responses [41], suggesting the development of pain sensations precedes the ability to moderate the experience of pain. Research into the effects of painful exposures in the NICU on functional brain development is relatively limited; however, greater NICU pain exposures have been associated with other health and neurodevelopmental alterations including decreased infant weight [42], lower white matter fractional anisotropy, and lower IQ at school age [43], as well as decreased cerebellar volume at school age [44]. Moreover, animal models have linked early pain exposures to increased neuroinflammation and hippocampal development [45], as well as greater neural sensitivity to future noxious exposures in the somatosensory and prefrontal cortex [46]. Most of this research focuses on differences later into childhood and emphasizes structural abnormalities. This results in a significant gap in our understanding of how these experiences change the functional organization of the brain and how early in development these differences can be detected—an essential step in identifying periods of early intervention.

This study aims to examine the association between early pain exposures in both near-term development of functional connections and later neurodevelopment. Using a cohort of V/EPT infants who underwent resting state fMRI at term-equivalent age (TEA), we examined differences in early functional connectivity from term-born infants. We then assessed whether NICU skin break procedures prior to TEA were associated with changes in inter-regional functional brain connectivity. Thereafter, we evaluated whether breaks were associated with risk for neurodevelopmental difficulties at 18 months, and whether the connectivity in regions sensitive to pain were associated with deficits at 18 months.

Our findings provide four major advances in the understanding of early-life painful exposures in preterm infants. First, exposures are associated with alterations in functional connectivity as early as TEA. Second, exposures are associated with worse neurodevelopmental outcomes including motor and language deficits and autism risk. Third, there may be an early critical period of vulnerability to pain. And fourth, connectivity within pain-sensitive connections may be predictive of future neurodevelopmental impairments.

Our cerebellar results harmonize with previous research showing atypical development in V/EPT infants, with smaller cerebellar volumes at TEA [22, 28‐30]. This may be due to relative early cerebellar development, which experiences rapid growth in the second trimester [63] and can quadruple in volume prior to birth [30]. Between rapid structural growth and acting as an early functional hub [58], the cerebellum is a logical focus for understanding early disruptions in brain development.

The differences in cerebellum connectivity may stem from its role in the brain’s pain matrix. The cerebellum, together with the primary somatosensory cortex, limbic and paralimbic regions, including the amygdala, ventral striatum, and insula [64‐67] make up the backbone of the mature, adult pain system [68]. While this network is fully intact in childhood, most functional networks are just beginning to organize during late gestation [26, 69‐71]. While not directly causative, we may be observing early disruptions in this network’s formation with co-activations through repeated painful exposures. This interpretation would strongly comport with findings in adults, where chronic pain is associated with hyper-connectivity between pain regions, including the cerebellum, brainstem, and prefrontal regions [72, 73] (for review see Mayer et al. [74]). Our findings mirror chronic pain in adults, where repeated experiences of pain are associated with increases in connectivity. It should be noted, however, that other reports emphasize changes in the medial and inferior frontal lobes in both humans [72, 73] and rodents [46]. Conversely, our results largely implicate only the inferior frontal lobe. While not clear, this may be due to the relative underdevelopment of the region at later gestational ages (GA), making these more immature connections easier to influence than more established regions. Although more work is required to directly assess causal mechanisms between painful experiences and alterations in connectivity, greater connectivity is believed to be a proxy for the degree of synchronicity between two regions, derived from consistent and repeated coactivation. Therefore, functional connections are generally thought to form and strengthen in an experience-driven manner [75]. Importantly, the observed differences are not simply brain-wide increases in connectivity, but instead appear to be heavily localized within pain regions, further suggesting the changes are likely associated the repeated coactivation of the pain matrix when exposed to painful experiences. Importantly, this pattern extends beyond the broad, functional dysmaturation often observed in premature infants compared to term born both prior to TEA [22, 76] as well as at and beyond TEA [77], and instead suggest that pain exposures are associated with greater connectivity between selective regions that far outpaces the deficits seen between term and preterm born infants across other regions.

In our study cerebellar alterations at TEA were associated with development at 18 months. Altered cerebellar volume and connectivity are a consistently reported feature in ASD [78‐80], including hyperconnectivity between the cerebellum and inferior frontal regions [81‐83]. In addition, disrupted limbic connectivity has been associated with deficits in ASD including social and executive function [84, 85]. Therefore, pronounced alterations in the connections of these regions may be mimicking autistic-like differences. Additionally, our findings in expressive language are bolstered by growing evidence in the importance of the cerebellum in language, especially language production [86] and has similarly been linked to ASD [87]. While this cohort will continue to be assessed throughout toddlerhood and early childhood, these findings represent important associations to be followed as more infants reach assessment ages to validate the significance of these findings in the larger cohort.

Finally, there may be a particularly vulnerable period for EPT infants during which pain exposure may elevate the risk of neurodevelopmental deficits. Our findings corroborate prior work on the worsening impacts of prematurity on structural brain development [88]. The most likely reason being the lagging development of pain modulation pathways relative to afferent nociceptive pathways. Additionally, due to the rostral to caudal development of the brain in third trimester [89, 90], the most rostral frontal regions may be the most vulnerable to insult as they are still more actively developing than more posterior regions.

Despite the strengths of this study, there are some limitations. First, there is no standard way to measure NICU pain, with highly variable ratings across raters and between subjective, physiological, and behavioral measures [91‐93]. Skin breaks serve as a common procedure that is widely accepted to cause pain/discomfort that occur frequently as part of routine care and have been linked with development of pain pathways in both human and animal studies [49, 88, 94]. Given the known discomfort associated with skin breaks, routine care often employees nonpharmacological interventions such as kangaroo care [95] swaddling or pacifier use to temper discomfort [96, 97]. Such interventions are difficult to quantify and are unaccounted for in this work. Furthermore, pharmacologic interventions, such as morphine, are also known to adversely influence brain development, and as such seldom employed as routine interventions prior to common skin break procedures. Similarly, we utilized caregiver report and questionnaires at 18 months rather than clinician-driven neurodevelopmental assessments. Future prospective studies should collect and control for both forms of pain management [95] as well as utilize clinical neurodevelopmental assessments. Second, in some cases multiple skin breaks may be required to successfully complete a single procedure, such as requiring a second heel stick to successfully draw blood. Multiple attempts for skin break procedures are variably recorded (routine for central line and intravenous placement, but not for heel lance) and, as such, could not be fully accounted for in our analyses. Third, the neurodevelopmental outcomes reported likely reflect both the effects of skin break procedures, as well as illness severity. Infants with higher disease burden require more consistent monitoring including more heel sticks and receive more interventions. This significant collinearity cannot be adequately controlled in the presented study design due to the tight coupling between the variables. As a result, we are unable to fully disentangle the impact of disease burden and the frequent experience of invasive skin break procedures. However, there is evidence that painful experiences are associated with alterations in brain connectivity [72, 73]. As a result, it is likely a contributor to altered neurodevelopmental trajectories that, unlike disease severity, is a more easily modifiable experience that can be targeted by interventions to optimally support neurodevelopment even in cases of significant illness. Fourth, pain is only one of several sensory experiences that are heightened in the extrauterine versus intrauterine environment. While pain is an intuitive mechanism driving these changes, extreme auditory and visual stimuli likely also impact development but were not accounted for here. Future studies should explore their impact on development. Finally, our subsample analyses linking connectivity to neurodevelopment are likely underpowered to detect more subtle relationships, especially as many of our observed correlations are between 0.2 and 0.4. Brain-behavior connections are often complex even when not assessed with 18-month lag and report correlation values within the observed range [98‐100]. More work is required, however, to properly assess the longitudinal effects of painful exposures and interventions at all stages of development. Finally, social and societal risk factors are thought to play a significant role in neurodevelopment. While our findings remain durable when controlling for a subset of sociodemographic, a more robust collection of a wide range of socioeconomic data should be collected in the future to assess its interaction with pain and long-term development.

These findings provide an important step in understanding the impacts of earlier-than-typical exposure to the extrauterine environment. Our results suggest that early, repeated exposures to pain have a significant impact on the developing functional connectome and neurodevelopmental outcomes. These observed differences have important implications for our wider understanding of the effect of painful NICU procedures, which are often performed in the absence of analgesia, and how we conceptualize the long-term risks of these procedures. Finally, these findings emphasize the need to anticipate and develop novel strategies to treat pain given the known adverse effects of traditional, opioid-based pharmacotherapies.