Introduction
Epidemiological studies have shown that aspects of fetal growth such as birth weight and gestational age are positively correlated with cognitive abilities in adults [
1‐
6]. The brain morphological mechanisms underlying these effects are not yet fully understood.
The fetus is constantly exposed to environmental effects [
7]. For example, maternal stress [
8,
9], anxiety [
10,
11], smoking [
12], malnutrition [
13] and social status [
14] are associated with lower birth weight and gestational age. Therefore, birth weight and gestational age served as surrogate markers for a large variety of environmental influences on prenatal development in many studies [
1].
Gyrification is an index measuring cortical folding which can be extracted from MRI images. This measure has been utilized in imaging studies investigating the course of cortical development in utero, in new-borns, and throughout life [
15,
16]. Reduced birth weight or shortened gestational age leads to increases and decreases in gyrification distributed over large parts of the cortex, but especially in fronto-temporo-parietal regions [
17,
18]. Consistent with the current conceptualization of gyrification as an index largely determined during intrauterine development [
19‐
21], these structural alterations persist over the whole life span and can still be found in adults. Further, some of these gyrification alterations associated with prenatal growth have been linked to neurocognitive performance in infants and adults [
17,
18,
22]: Full-scale IQ has been associated with the left fusiform gyrus and lateral orbitofrontal cortex, the right superior parietal gyrus [
18], the bihemispheric lateral and anterior temporal cortices and the occipitotemporal junction [
17]. Regional variations in gyrification have repeatedly been associated with general neurocognitive performance in humans [
23,
24] and are considered as a potential neurocellular correlate of cognitive abilities.
While these studies have advanced our knowledge considerably, they leave several questions unanswered: 1. Many of the previous studies used only group comparisons (with arbitrary criteria for the definition of cut-offs for group divisions that were not based on neurobiological measures), rather than dimensional approaches, resulting in a loss of variance. 2. Most studies used only moderate sample sizes leading to little statistical power and limited generalizability. 3. Cortical folding takes place in a non-linear course primarily during the second half of intrauterine development but also non-linearly ex utero in infants [
1,
25‐
27]. However, previous studies did not test for non-linear associations between variables of prenatal growth and brain morphology. 4. Some studies included participants with adverse events during their gestation such as birth complications, maternal infection, alcohol or drug abuse, medication, or maternal malnutrition [
15,
17,
18,
28,
29], which all confound birth weight/gestational age as well as brain structural alterations. 5. Lastly, most studies investigated only intelligence by using abbreviated tests and refrained from measuring a broad range of cognitive domains. Some also used merely subjective parental reports as a proxy of child cognitive abilities. As a result, effects on specific cognitive domains, e.g., working, short- and long-term memory, attention or language have yet to be investigated.
In our study, we aimed to address these points. We recruited a large sample of healthy participants but excluded all subjects born after a high-risk pregnancy who had been exposed to gross harmful environmental influences, e.g., maternal infections, drug use, malnutrition or birth complications. Further, we used a comprehensive neuropsychological test battery to investigate associations between gyrification, fetal growth and specific neuropsychological domains. We hypothesized: 1. Gestational age as well as 2. birth weight are associated with reduced gyrification in healthy adults. 3. These relationships are non-linear. 4. Finally, we test whether gyrification is related to neuropsychological performance in adults, thus bridging the explanatory gap between prenatal development and adult neuropsychological outcomes.
Discussion
We investigated associations between birth weight, gestational age and gyrification in a large sample of adults, excluding high risk pregnancies and births. We then explored the impact of these associations on neurocognition. We found a positive linear association between gestational age and gyrification bilaterally in the superior frontal cortex, the left supramarginal cortex and in the lingual cortex bilaterally. The association between gyrification clusters and the neuropsychological factors language and working memory/attention was moderated by gestational age. These findings provide an important basis for understanding prenatal influences on brain morphology and their relations to cognitive functions in healthy adults.
We demonstrate positive, linear associations between duration of pregnancy and gyrification in healthy adults. We extracted gyrification values using an absolute mean curvature approach which is positively correlated with increases in total surface area [
64]. A methodological strength of this study is the exclusion of participants whose mothers had high-risk pregnancies (e.g., infections, malnutrition, drug abuse, etc.; see above) or pregnancies that are known to be associated with aberrant cortical folding in the fetuses (i.e., multiple pregnancy). Due to this approach, we can conclude that strong associations between rather subtle environmental influences (measured as gestational age) and cortical formation evidently persist in adults even when excluding more severe environmental effects during pregnancy that could have confounded these results. These findings emphasize the formative character from gestational age on brain anatomy which can already be observed in new-borns [
87] and shed new light on the importance of preserving maternal health during pregnancy.
Our findings can be interpreted in several ways. First, prenatal cortical growth that is interrupted due to early birth might postnatally not be completed because cortical maturation could potentially only be completed in utero. Second, postnatal compensatory mechanisms of cortical maturation could fail because the harming cause (e.g., maternal stress [
15,
88] or smoking [
89,
90]) that lead to reduced gestational age still persists in postnatal environment. The third explanation comes from a core assumption of the concept of Developmental Origins of Health and Disease [
7,
91,
92]. It posits prenatal mechanisms of permanent fetal programming that are triggered by in utero environmental effects that result in persistent modifications on the epigenome of the differentiating brain cells, which can lead to permanent physiological modifications in the offspring [
15]. The consequences of this in utero adaption are twofold: On the one hand, these altered cortical formations are due to the plasticity of the fetal brain which prepares the unborn for its postnatal environment—assuming it is similar to its environment in utero [
93]. Thereby, chances of immediate survival after birth and potential reproduction are maximized [
94]. On the other hand, these effects can also result in non-adaptive physiology when there is a mismatch between prenatal adaptive brain development and the biological demands that the postnatal environment places on the new-born child. This maladjustment could then lead to vulnerability for later mental disorders.
Our second main result is that gestational age moderates the relationship between localized gyrification and the neuropsychological factors language and working memory/attention. The association between a cluster mostly comprising the left superior frontal cortex and the factor working memory/attention was moderated by gestational age. The involvement of the superior frontal gyrus in working memory has been demonstrated consistently [
95]. Additionally, relationships between a gyrification cluster in the left supramarginal gyrus and the factor language as well as the factor working memory/attention were both again moderated by gestational duration. The supramarginal gyrus, as part of Wernicke’s area, has repeatedly been linked to phonological decisions [
96,
97], syntax [
98] and semantics [
99,
100]. Another line of research showed that the supramarginal gyrus is involved in verbal/auditory working memory [
101,
102]. Therefore, our results could demonstrate that variations in working memory and language performance are to a certain extent a product of the interactional effect of gestational age and gyrification alterations that are linked to prenatal cortical development. These findings are an important contribution to the identification of neurobiological pathways involved in the association between preterm birth and lower cognitive performance in adults that have been reported in epidemiological studies [
4].
Since we found only one direct bivariate association between a gyrification cluster significantly associated with gestational age and one neuropsychological factor but moderated effects it should be pointed out that alterations in gyrification alone might not be sufficient to explain some aspects of poorer neuropsychological outcome. Rather, it is an interplay between variation in gyrification and other neurobiological factors that are influenced by gestational age. This highlights again the importance of determining relationships between prenatal and early-life factors influencing cognitive ability in adulthood.
We found no brain morphological associations with birth weight when we applied a strict statistic (FWE in SPM), although birth weight and gestational age were correlated (
r = 0.378;
p < 0.001). Gestational age might be a more valid proxy for brain maturation because birth weight is affected by many variables, such as maternal ethnicity, infant gender, maternal smoking and maternal diabetes [
103].
Our findings also have implications for psychopathology. Cohort studies have shown that low gestational age is associated with a higher chance to develop a mental disorder in adulthood [
104,
105]. Individuals who are at high risk of developing schizophrenia or who will later develop schizophrenia show general cognitive deficits in various domains before the onset of the disorder, including language, processing speed, working memory, executive functioning and intelligence [
106,
107]. On a brain morphological level, there are gyrification differences between individuals with higher risk for schizophrenia and control subjects that partly overlap with cortical areas that were associated with gestational age in our study [
108,
109]. We suggest that the interaction effect between gestational age and gyrification alterations associated with shortened gestational length may constitute a risk phenotype for mental disorders clinically characterized by reduced cognitive performance.