Background
Over the years, very preterm born children (<32 weeks gestation) have better survival rates [
1] and improved outcome [
2]. It is, however, of concern that an increased risk of psychiatric problems has been reported in preterm-born individuals (<37 weeks gestation), especially anxiety symptoms and disorders, attention-deficit/hyperactivity disorder (ADHD) and autism spectrum traits and disorders (ASD) [
3‐
5].
The cerebellum is of particular interest in the preterm born due to its extensive development during the third trimester of gestation. Indeed, during this period it surpasses the growth rate of the cerebral hemispheres [
6]. For the very preterm born infant, the extensive development of the cerebellum takes place in an extra uterine environment, where respiratory problems, infections and nutritional challenges may influence cerebellar development. Cerebellar injuries (hemorrhage, infarction) and mal/underdevelopment following premature birth occur more frequently than previously thought [
7,
8]. It has been proposed that cerebellar involvement may play a central role in cognitive, mental health, and socialization deficits found later in life in this population [
9,
10]. In the general population, the cerebellum has been associated with psychiatric problems such us mood disorders, anxiety problems, schizophrenia, ASD and attention problems [
11]. The underlying pathophysiology still remains unknown. However, it has been proposed that the cerebellum might serve as modulatory [
12‐
14] and timing station [
15‐
17] for integrating [
18] brain processes, due to its extensive connections with the whole brain [
14,
19‐
22]. Projections from the cerebellum to the cerebral cortex constitute the cerebello-thalamo-cortical (CTC) pathway [
23,
24], and early disruption of the cerebellar circuitry development has been positively correlated with ASD, attention deficit and emotional problems [
25]. Injury to the immature cerebellum could affect neurologic function through mechanisms that interfere with later development of remote regions of the cerebral cortex [
26].
Some studies have shown cerebellar abnormalities to be associated with psychiatric symptoms in preterm children. In a retrospective, case–control study, preterm children who had perinatal cerebellar hemorrhage presented a higher prevalence of deficits in cognition, communication, and social and behavioral function at 2–3 years of age than preterm peers without cerebellar pathology [
10]. In another MRI study with very preterm participants, total cerebellar volume reduction from adolescence to adulthood was associated with having more psychiatric symptoms [
27]. However, these studies used questionnaires to assess mental health problems and did not differentiate between cerebellar gray (GM) and white matter (WM).
Our research group has studied preterm born adolescents with very low birth weight (VLBW ≤ 1500 g) and controls during adolescence. At 15 years of age, the VLBW children had smaller cerebellar WM volumes compared with controls [
28], and they had increased rates of psychiatric symptoms and diagnoses assessed with questionnaires and clinical interview [
3,
4]. At 19 years of age, the VLBW group still had smaller cerebellar WM volumes than term-born peers [
29] and more psychiatric problems [
5]. During adolescence they also displayed a trend towards increasing psychiatric morbidity [
5]. However, cerebellar growth rate did not differ from controls [
29].
Based on these findings, we aimed to study the relationship between cerebellar volumes and psychiatric symptoms and diagnoses at 15 and 19 years of age. Our main hypothesis was that reduced cerebellar volumes would be associated with higher rates of psychiatric symptoms and diagnoses at both 15 and 19 years. Furthermore, we hypothesized that small cerebellar volume was associated with increased risk of developing psychiatric problems during adolescence.
In this article, we report an association between persistent smaller cerebellar volumes and psychiatric symptoms during adolescence in children born preterm and with VLBW.
Discussion
We studied the relationship between cerebellar volumes and psychiatric symptoms and diagnoses at 15 and 19 years of age in adolescents born very preterm and with VLBW. Cerebellar growth trajectories from 15 to 19 years of age were equal between adolescents born with VLBW and controls, regardless of psychiatric morbidity (Fig.
2). However, VLBW adolescents with a persisting/developing diagnosis during adolescence had maintained smaller cerebellar WM and GM volumes compared with controls and compared with VLBW adolescents who were healthy or became healthy during this period. Moreover, cerebellar volumes did not differ between VLBW adolescents who were or became healthy from 15 to 19 years of age and controls (Fig.
3). At both 15 and 19 years of age, larger cerebellar WM and GM volumes correlated with better general psychosocial functioning in the VLBW group (Fig.
4).
Smaller cerebellar volumes have been consistently reported in preterm children compared with term-born peers from birth [
6,
48‐
51] to childhood [
52] and adolescence [
27‐
29,
53‐
55]. Even though overall smaller brain volumes are a common trait in children born very preterm and with VLBW [
55], some studies suggest that there are not brain growth differences, including the cerebellum, between individuals born preterm and term-born peers [
29,
55,
56]. Nonetheless, other studies have found differences in cerebellar trajectories between preterm and term-born children during adolescence. In an MRI longitudinal study, Parker et al. [
27] reported total cerebellar volume reduction from 15 to 18 years of age in a cohort of adolescents born very preterm compared with term-born peers. This reduction in cerebellar volume was associated with having more problems in several questionnaire items concerning concentration, feeling useful, decision-making capability, overcoming difficulties, feeling confident and feeling worthless. Abnormal cerebellar growth has also been reported to occur right after birth, even after normal cerebellar ultrasound [
6,
48]. Preterm children with the most deviant cerebellar development have higher rates of intraventricular hemorrhage and other associated complications like post-hemorrhagic hydrocephalus and neurosurgical interventions [
50,
57‐
59]. However, extreme prematurity has been noted as the most explicative factor for disruptive cerebellar development in VLBW neonates [
6,
51,
57,
59,
60]. The causes of deviant cerebellar development in the absence of apparent damage are unknown [
9]. We speculate that the smaller cerebellar volumes in VLBW adolescents with persistent/increasing mental health problems might be originated in the perinatal/neonatal period.
In order to properly understand the role of the cerebellum in the appearance and maintenance of psychiatric disorders in children born very preterm and with VLBW, it is important to study its anatomy and how premature birth affects its development. The cerebellum is connected with the whole brain, especially with the cerebral cortex [
24,
61,
62]. The vast development of the cerebellum occurs, mainly, in the third trimester of gestation, where the cerebellum increases its volume fourfold [
6]. Early disruption of the cerebellar circuitry development has been positively correlated with ASD, attention deficit and emotional problems [
25]. In fact, a meta-analysis pointed out cerebellar abnormalities as the most consistently reported structural finding for ADHD [
63], but the results whether total cerebellar volume, WM or GM or both are abnormal were inconclusive. ADHD symptom severity has been shown to correlate with overall cerebellar volume [
64]. Many studies have found cerebellar abnormalities in ADHD, especially in the vermis [
65‐
71]. Our finding of smaller cerebellar GM volume associated with higher inattention symptoms in the VLBW group points out GM as related to their symptoms. However, the lack of correlation with WM volumes might be due to low statistical power. Diffusion tensor imaging studies investigating WM differences in ADHD have reported reduced fractional anisotropy in the middle cerebellar peduncles [
72,
73], and WM in the left cerebellum [
73,
74]. A recent fMRI study has reported a link of abnormal cerebro-cerebellar connectivity with ADHD, and particularly for inattention, suggesting that ADHD symptoms are not the direct result of cerebellar abnormalities, but the result of damages in the whole network [
75].
Injury confined to the cerebellum has been associated with reduced growth in specific regions of the uninjured contralateral cerebral cortex in preterm born neonates (dorsolateral prefrontal, premotor, sensorimotor, and mid-temporal regions), and this secondary growth reduction has been associated with a higher risk of impaired cognitive, language, behavior, and motor performance [
49,
76]. This suggests that injury to the immature cerebellum could affect neurologic function through mechanisms that interfere with later development of remote regions of the cerebral cortex [
26]. Projections from the cerebellum to the cerebral cortex constitute the cerebello-thalamo-cortical (CTC) pathway, the main efferent cerebellar projection [
23,
24]. The CTC pathway has been linked with deficits in information processing in schizophrenia [
77,
78], and disruption in cortico-cerebellar connectivity has been proposed as a major neurobiological mechanism of emotional dysregulation [
79]. In preterm infants, thalamo-cortical projections measured with DTI were affected compared with born-at-term controls, suggesting that such pathways might be especially vulnerable to preterm birth [
80]. Different fMRI studies have also shown activation of the cerebellum in different emotional tasks, such us viewing pictures of deceased loved ones [
81] and viewing pictures and movies evoking both negative and positive feelings [
82,
83]. These results suggest that the cerebellum is directly involved in the appraisal of emotional stimuli. The closed-loop circuits between the cerebellum and the cerebral cortex serve as the anatomical substrate by which the cerebellum can modulate activation patterns in distal regions. Based on these interactions, it has been proposed that cerebellar dysfunction or disruption early in development could have major impacts on the structure and function of the cortical regions to which it projects [
2]. We hypothesize that deviant cerebellar development, probably occurring around the neonatal period, might cause smaller cerebellar volumes that are carried onto adolescence affecting the cerebellar circuitry and its extensive connections with the brain, and that this injury pattern probably plays an important role in the development and maintenance of psychiatric disorders in this population. Moreover, discriminating VLBW children with significant smaller cerebellar volumes than controls might help to identify which of them are at risk of developing psychiatric problems, and it might aid to take preventive actions at an earlier stage.
In order to clarify the relationship between psychiatric symptoms, general cognitive abilities and cerebellar volumes, we corrected our analyses for IQ. After this, all the results remained significant, except for differences in cerebellar GM volume at 15, the relationship between cerebellar volumes and inattention, and cerebellar WM volumes and psychosocial function at both ages and GM at 15. IQ summarizes psychological processes such as attention, executive functions, and general knowledge [
84], while social cognition measures are not properly covered. Having intellectual disability has been linked to higher prevalence of psychiatric disorders [
85], and recent research suggests that impaired executive functions (i.e. inhibition, working memory, and cognitive flexibility) is a core feature in many mental illnesses. If a deficit in executive functions is a core symptom of psychiatric disorders, correcting for IQ might be an overcorrection in brain imaging studies trying to find the neural basis of mental health illness [
86]. Still, the nature of this correlation is not fully understood yet [
85,
87]. A cohort study using data from over 900,000 individuals in Sweden has linked higher IQ scores during adolescence with higher risk of developing bipolar disorder later in life, controlling for socioeconomic group and parental education [
88]. At the same time, neurocognitive deficits in bipolar disorder have been highlighted as an important determinant of the disruptive nature of this disorder [
89]. In preterm born individuals, cognitive abilities might be affected by the same brain mechanisms that affect mental health problems. Thus, if IQ, which is also the result of brain characteristics [
90], is a mediating variable in mental health disorders, rather than a core feature, correcting for IQ might shadow the direct relationship between brain abnormalities and psychiatric symptoms [
46]. More research is needed to clarify this matter.
The strength of this study is the usage of both questionnaires and a semi-structured diagnostic interview for identifying psychiatric symptoms and disorders, thereby contributing to a comprehensive psychiatric evaluation. Another strength is the combined cross-sectional and longitudinal design with clinical and MRI assessments at both 15 and 19 years of age in this cohort. The participation rate was comparable to other follow-up studies with similar study groups [
91] and participants and non-participants did not differ in perinatal variables (gestational age, birth weight, maternal age at birth), making selection bias less likely. Another strength is that the groups did not differ in levels of socio-economic status.
We used a well-known and reliable automated MR segmentation method to measure cerebellar volumes (
http://surfer.nmr.mgh.harvard.edu/). However, this method only provides delineation of the total cerebellum, differentiating between WM and GM, but not accounting for the cerebellar vermis. FreeSurfer has been shown to underestimate the cerebellar region [
92] and problems to distinguish perfectly between the cerebellum and brain stem [
93]. We inspected all processed images for accuracy of the FreeSurfer segmentation and rejected structures with obvious segmentation errors. However, in order to avoid introducing bias and increasing variances into the data set of MRI images, no manual editing was performed. Further studies using MRI with higher field strength or improved algorithms for segmentation that allows more detailed cerebellar segmentations, may be required to support or reject our results.
Another limitation of the study is group size. Due to the relatively small sample, only large differences and strong associations could reach significant levels. Hence, negative findings should be interpreted with caution. We had longitudinal data for a smaller sample than the cross-sectional study groups, which reduced the statistical power and hence, the generalization of the longitudinal results. Studies with larger samples are certainly needed to confirm the findings.
The ASEBA self-report composite scales were not associated with cerebellar volumes, while results from diagnostic evaluation were. This can be explained by a discrepancy in our VLBW group between the ASEBA self-report and the clinical reports at both assessments, suggesting that the VLBW group might be under-reporting problems on questionnaires [
4,
94]. Similar discrepancies have been reported in other studies between self-reports and parent-reports in VLBW young adult populations [
95,
96]. Therefore, the lack of association with the ASEBA subscales should be interpreted with this in mind.