In the current study, we wanted to elucidate if NF1 transmission has an impact on the wide variability on cognitive phenotype in NF1 children. The only difference from a broad battery of neuropsychological tests -including psychometric, reading (text and word), phonological, visual-spatial, reading comprehension (sentences and text), receptive language, attention and psychosocial assessments- was in the IQ scores.
4.1. IQ differences between sporadic and familial form of NF1
We detected a highly significant difference between sporadic and familial NF1 cases in all index scores -except Processing Speed Index (PSI)-, Full Scale IQ (FSIQ) and three subtests: Matrix Reasoning, Block Design and Letter-Number Sequencing.
For the FSIQ, PSI and WMI, children affected by the familial variant of NF1 were lower than average (standard scores around 80 to 85 for indexes and between 6 and 7 for subtests) while children affected by the sporadic variant of NF1 were average (90 to 95 for indexes and around 9 for subtests) suggesting that patients with sporadic NF1 adapt better to the disease than familial cases.
Our results were in accordance to those of Coutinho et al. [91] that have found better scores in FSIQ, VCI and WMI (the details of the subtests has not been carried out) in children with sporadic NF1 than children with familial NF1. However, our results differ from those of Lehtonen et al. [92], Hyman et al. [10] and Ferner et al. [20] who found similar IQ scores between sporadic and familial variants. Several reasons can be put forward to explain such discrepancies in the results. First, these three studies were designed to compare cognitive profiles between NF1 patients and controls and not between the sporadic and familial variants of NF1. Thus, the authors did not specify if both groups were taken from comparable populations (if number, percentage, age, sex, number of borderline IQ is comparable between groups, if the main confounders were identified and taken into account in the design and analyses to minimise the risk of bias, etc.). Secondly, differences can be due to the sample age. In our study, children are between 8 and 12 years old, while in Hyman et al. [10] and Lehtonen et al. [92], children were older (8 to 16.75 years and 6 to 16 years, respectively) and in Ferner et al. [20], the 103 patients with NF1 (51 sporadic NF1 cases and 52 familial NF1 cases) are between 6 and 75 years (mean age 27.6 y/o; SD 18.2). Genetic influences –that explain significant parts of the observed variation in cognitive functioning, both for children and adults [93,94]– tend to increase in significance with age, while environmental influences decrease in significance across development [95-97]. Brant et al. [95] especially show that the environmental factors that have an influence on variance in intelligence are very minor from age twelve onwards. There is a great similarity of the pattern of contributing factors from between ages twelve and sixteen, suggesting that the aetiology of individual changes in intelligence development is extremely constant by early adolescence. Another and final explanation to such a discrepancy is that, in these three previous studies, mental retardation is not excluded and 6.2% of children with NF1 in Hyman et al. [10], 6% in Lehtonen et al. [92] and 8% in Ferner et al. [20] have an FSIQ less than 70. The inclusion of extreme cognitive profiles is probably a bias (controlled in our study) that leads to different results.
Over the last two decades, there have been a number of studies, summarised in the systematic review of Lehtonen [12], that have studied the general intellectual functioning of children with NF1. The majority of studies have shown that, although children with NF1 have IQs in the normal range, their IQ is often lower (around 90s) than their peers or than their unaffected siblings [10,19,20,98]. However, some studies failed to demonstrate significant differences in IQ between children with NF1 and norms [29,99-101). In addition, Lehtonen et al. [12] pointed out a disagreement with regards to the exact IQ profile of children with NF1: while some studies demonstrated that children with NF1 scored less on all subtests of the WISC, some others detected some significant differences in only some subtests (Block design or Digit span, for example.) between children with NF1 and their siblings. Difference in proportion of transmission (proportion of sporadic vs. familial NF1 in the final sample) variant could perhaps explain these contradictory results.
We found a difference between sporadic and familial NF1 children regarding Block Design (8.3 vs. 6.7, respectively), Matrix Reasoning (9 vs. 7.4) and Letter-Number Sequencing (8.9 vs. 7.1). Block designed[1] and Matrix reasoning[2] (moderately correlate each other; r = 0.55) are known to be a good measure of general and fluid intelligence abilities. They measure non-verbal reasoning, visual processing and abstract, visual perception and organisation, visual-spatial ability (and visual-constructional ability for Block Design). Letter-Number Sequencing[3] measures attention span, short-term auditory memory processing, sequential processing and mental manipulation [102-105]. Those three subtests may be influenced by concentration and attention.
Differences between sporadic and familial groups of NF1 children in those three subtests are very interesting. All three are considered to be the hallmark phenotypic characteristics of patients with NF1: children with NF1 are known to have serious difficulties in visual-spatial abilities, memory and attention [11,12,23,25,81,82,92,106].
Altogether, it is therefore legitimate to ask whether IQ difference -largely previously proved between NF1 children and peers or unaffected siblings- persist if the modality of transmission is taken into account. Are differences maintained between children affected by a sporadic variant of NF1 and peers and siblings? Do children with the familial variant of NF1 constitute a ‘bias’ or an explanation to the wide variability in cognitive profile of NF1? More research is needed to detail this specific topic. The mode of transmission of NF1 also seems essential to be taken into account in future studies about the cognitive profile of NF1 subjects.
- What is the role of socioeconomic status (SES) for such IQ differences in NF1?
Today, the concept that the cognitive performance of an individual depends approximately equally on his/her genetic heritage and his/her environment is a consensus. Recent genome-wide meta-analyses and research studies have identified genomic loci and genes linked to variation in intelligence [107-112]. However, it is also known that the socio-economic background of the child places constraints on their IQ [96,113]. First, indices of the families' SES (education, occupation and income of parents) have been proved to moderate the heritability of their children's intelligence [114,115,116]. The heritability of IQ is generally higher for children who are raised in high SES environments [116]. The results of Turkheimer et al. [115] especially demonstrate that the proportion of IQ variance attributable to genes and environment vary non-linearly with SES: in disadvantaged families, the contribution of genes is close to zero and the environment (SES) explains 60% of the IQ variance, whereas is it the reverse in wealthy families. Secondly, in the general literature, the SES environment has been shown to account for variance in cognitive functioning in childhood in many studies. The effects of the environment on IQ, especially the link between the socio-economic level of parents (socioeconomic status and parental education) and the cognitive performance of children is therefore well-established [97,116]. Of course, the level of education of the child's family environment is involved (especially that of the mother): parents from high SES environments indeed offer more occasions for activities and learning experiences to boost and encourage children’s intellectual development [116]. But differences in intellectual outcomes could also be attributable to the family income, nutrition, sleep, stress, availability of parents, maternal and paternal involvement, etc. -that have a direct impact on the child's cognitive development and that is directly connected to the child’s environment. For example, concerning income, Noble et al. [117] followed a cohort of 1,099 individuals aged 3 to 20 years. Authors highlighted that income relates most strongly to brain structure (especially in regions supporting language, reading, executive functions and spatial skills) among the most disadvantaged children where small differences in income were associated with large differences in brain surface area (whereas in higher income families, similar income increments were associated with smaller differences in surface area). Thirdly, some recent studies tend to highlight the link between IQ and epigenetic mechanisms (temporary (or not) genetic changes supported by environment). For example, in times of high stress, physiological changes in the organism can modify genes. In turn, changes can influence a range of characteristics that can have knock-on effects altering the course of child development. Kaminski et al. [118] have especially found a strong relationship between the epigenetic modifications of one particular gene and general IQ, suggesting our experiences have an impact on the genetic mechanisms involved in complex processes such as intelligence. Authors thus show that individual differences in IQ are linked to differences in brain activity and epigenetic changes, which are both under environmental influences.
Altogether, studies have found a strong relationship between IQ and SES in the general population, suggesting our experiences/environment not only affect our quality of life, the wiring of our brain, but the very way our cognitive function evolves.
In the NF1 children population, Hyman, Lehtonen and Ferner’s studies have examined predictors of the lowering of general cognitive ability and have only found an association with socioeconomic status. SES has also been found to significantly correlate with general intelligence in Lorenzo et al. [119,120].
In our study, we have found a strong link between (1) sporadic/familial form, (2) IQ and (3) SES family background, especially the mother’s education level.
Firstly, children with familial NF1 had a significantly lower SES than children with sporadic NF1, which is consistent with other NF1 studies [91,92,119]. Lorenzo et al. [119] especially found in a population of 43 children with NF1 (25 sporadic cases and 18 familial cases) that 68% had mothers who completed a university or postgraduate degree in sporadic cases group compared to 28% in the familial cases group. Coutinho et al. [91] similarly found that SES tended to be lower in children with the familial transmission (41% had a low SES) than in those with a sporadic transmission (19%). Such distribution does not appear to be an unexpected outcome: the sporadic vs. familial NF1 variant has an impact on the social level in which the child evolves [92,119]. NF1 frequently leads to learning disabilities, poor school academic performances [23], lower education level (less likely to graduate from school, less likely to complete tertiary education), and restrict individuals choice and their professional future (individuals with NF1 are and thus fall into lower socio-economic groups) [121].
Secondly, and as previously shown in the general population [116,122] and in the NF1 population [10,91,119], we found that SES was, in turn, associated with IQ achievement. NF1 children from greater SES backgrounds (here children affected by the sporadic variant) had greater cognition scores than children who came from lower SES backgrounds (here children affected by the inherited variant). Our results are in accordance with Lorenzo et al. [119] and Hyman et al. [10]. However, both studies have addressed the issue of the relationship between NF1 in its entirety, IQ and SES [10,119,120], without addressing the specific question of the relationship between NF1 variants, IQ and SES. Our results are also in line with Coutinho et al. [91], who found that children with the familial transmission had a lower FSIQ and tended to have a lower SES compared to those with sporadic NF1. However, the authors did not discuss this association (Cause and effect? Consequence? etc.). Our findings therefore increase those of these four previous studies in the comprehension of this trend, highlighting that the disparity recognised between the sporadic and familial variants is likely due to the impact that the NF1 transmission modality has on the SES environment of the family.
In addition, we also demonstrated that there is no significant interaction between group (transmission forms: sporadic vs. familial) and the relationship between the mother’s education level and the IQ of the children. In other words, the mother’s education level has an impact on the IQ of the NF1 child, irrespective of the transmission mode (sporadic or familial). Having a low SES has a snowball effect on other variables -as cognitive variables- but effect is irrespective of inherited variant. However, as familial NF1 leads more frequently to a low SES, familial NF1 children are most often affected.
- Toward a more complex and multi-factorial approach to explaining specific cognitive phenotypes in NF1
Another important finding is the absence of differences for tests exploring the usually affected cognitive domains in NF1 (language, visual-spatial domain, executive functions, attention) between the two forms of NF1. We indeed used 10 tests leading to 49 measures, completed by four SES measures. We only found a single test and only seven measures out of 49 where there is a difference between the sporadic and familial NF1 variants. The majority of cognitive functions are therefore not different between the two groups. Consequently, we can argue that transmission (sporadic vs. familial) alone failed to explain the wide variability in phenotype NF1 expression.
Our results were consistent with those of Coutinho et al. [91]. Although authors found that children with sporadic NF1 performed better than those with familial NF1 in a set of neuropsychological tests (Rey Complex Figure Copy, JLO, Imitation of Hand Positions, Spatial Memory, and Reading Comprehension tasks), differences were cancelled when FSIQ and SES were taken into account (except for JLO). Our results were also consistent with those of Lehtonen et al. [92] and Erdogan-Bakar et al. [123] where the heritability status of NF1 did not lead to any differences in the performance of the children with NF1 (sporadic vs. familial NF1 groups) on any of the measures (visual-spatial, working memory, spatial memory, executive function, attention, etc.). Note, however, that these two studies were not designed to observe this effect (this is here an ancillary result), so the groups were not controlled and adjusted in terms of number, age, sex, IQ, SES, etc. Our results therefore reinforce, confirm and extend these previous ones with equivalent groups (no bias), and a study especially designed to reply to this question.
The causes of NF1 cognitive phenotype and its variability have been explored with genetic, brain imaging or histological studies [27] but none have successfully explained them until now [10,18,21,27 for eg.]. Snippets of explanation are sometimes pointed (UBOs, visual-spatial abilities, etc.) but findings are inconsistent across studies. The IQ variability could be explained by the transmission (sporadic vs. familial) and SES status, while another variability typology (motor impairment, social deficit, executive function impairment, etc.) could be explained by another cause. It is therefore possible that the wide variability in NF1 can be explained by a multitude of causes and not just one, which would partly explain why studies fail to explain phenotype variability in NF1 when they address this question from just one perspective.
Overall, cause-and-effect relationships to explain phenotype variability in NF1 are not always easy to establish and more global approaches are probably needed. Multi-causality is also a possible explanation that should be investigated: either as interrelated causes that interact in a particular order to produce the effect; or as the interaction of multiple risk factors, including environmental, economic, lifestyle and genetic predisposition factors.
[1] BD (core Perceptual Reasoning subtest) require children to put together red-and-white specially designed blocks in a pattern according to a displayed model. The subtest is timed.
[2] MR (core Perceptual Reasoning subtest) require children to complete a matrix or serial reasoning problem by selecting a missing picture from five response choices. The subtest is untimed.
[3] LNS (core Working Memory Subtest) require children to repeat in a predetermined order to the examiner a series of numbers and letters that they just heard. The subtest is untimed.