We did not find differences in age or sex between the PTPN11 and TD groups. We found a difference in age (t(13.7)=-2.58, p = 0.022), but not sex between the SOS1 and TD groups (Table 1). For the PTPN11 group, TBV (t(69)=-4.11, p = 0.00011, d=-0.98), total SA (t(69)=-2.38, p = 0.020, d=-0.55) and mean CT (t(69)=-4.19, p = 0.00010, d=-1.04) were smaller than those of the TD group. For the SOS1 group, TBV (t(49)=-1.46, p = 0.17, d=-0.49), total SA (t(49)=-0.80, p = 0.44, d=-0.44), and mean CT (t(49)=-1.91, p = 0.078, d=-0.70) were not significantly smaller compared to the TD group. However, we observed relatively large effect sizes (all d’s<-0.44) in the SOS1 group, indicating an overall effect of SOS1 in the same direction as PTPN11. All images survived Euler number cut off − 217 assessing the FreeSurfer-compatible quality of images before manual editing26. However, we detected significant differences (t(78) = 2.244, p = 0.028) in cumulative Euler number between the NS (mean − 178.25±-73.57) and TD (mean − 145.25±-56.90) groups. To address group differences in cumulative Euler number, we conducted gold-standard manual edits on all images in FreeSurfer (see Methods and Materials).
Noonan Syndrome Is Associated With Smaller Subcortical Volumes
To evaluate the effect of NS on brain anatomy, we first examined its effect on GMV. For PTPN11, we found GMV reductions in bilateral striatal structures, specifically in the caudate, putamen, and pallidum. We also detected smaller right hippocampal GMV relative to the TD group (Fig. 1). Similarly, in the SOS1 group, we found bilateral reduction in pallidum GMV compared to the TD group (Fig. 1). Using effect sizes to evaluate the clinical effect of PTPN11 and SOS1 variants on subcortical structures, we detected reduced subcortical volumes in both NS groups in a step-wise decrease pattern, with smaller effect (All d’s<-0.25) on striatal structures of SOS1 and larger effect (All d’s<-0.7) of PTPN11 compared to controls (Fig. 1).
Noonan syndrome affects gray matter volumes of precentral gyri and medial aspect of the occipital lobe
We detected smaller regional GMV in PTPN11 compared to the TD group, most prominently in the bilateral precentral gyri and medial aspect of the occipital lobe (All FDR p-values < 0.05) (Table S2; Fig. 2). Given that the SOS1 group was smaller (n = 10) than the PTPN11 group (n = 30), we lacked power to detect differences between these groups across the cortex. Therefore, to estimate the clinical effect of NS variants on the brain and compare these effects between NS groups, we used effect sizes. In general, we detected GMV decreases in a similar regional distribution for PTPN11 and SOS1 compared to controls (Fig. 3). For both NS groups, we observed large negative (All d’s<-0.8) effect sizes for GMV in the left caudal middle frontal gyri and medial aspect of the occipital lobe, compared to the TD group (Fig. 3).
Noonan syndrome is associated with surface area expansions in limbic regions and decreases in the frontal lobe
Next, we aimed to test whether the two determinants of cortical volume, SA and CT, are affected by NS. We observed SA decreases in bilateral entorhinal and left superior parietal cortices and SA expansion in the right frontal and bilateral temporal lobes in PTPN11 compared to the TD group (All FDR p-values < 0.05) (Table S2; Fig. 2). In both NS groups, we observed large positive (All d’s > 0.8) effect sizes indicating SA expansion in the left parahippocampal gyrus and large negative (All d’s<-0.8) effect sizes indicating SA decreases in left caudal middle frontal gyrus, relative to the TD group (Fig. 3).
Noonan Syndrome Is Linked To Cortical Thickness Reductions In The Precentral Gyrus And Parahippocampal Regions
We observed reductions in CT in bilateral precentral gyri and parahippocampal regions and increases in CT in lateral aspects of the occipital and frontal lobes in PTPN11 relative to the TD group (All FDR p-values < 0.05) (Table S2; Fig. 2). We observed large positive effect sizes (All d’s > 0.7) in the left lateral occipital cortex, indicating increases in CT in both NS groups relative to the TD group. Conversely, we observed medium to large negative effect sizes (All d’s<-0.5) in the left parsopercularis and superior temporal gyrus as well as in the medial aspect of the temporal lobe, bilaterally, in both NS groups compared to the TD group. In affected regions, SOS1 displays larger effect sizes relative to PTPN11, indicating that SOS1 might have a more pronounced effect on CT (Fig. 3).
Due to the group differences detected in cumulative Euler number means across groups, we conducted a sensitivity analysis. After repeating the GMV, CT, and SA analyses between the PTPN11 and TD groups with cumulative Euler number as a covariate, we observed overwhelmingly similar results relative to our initial analyses without this covariate. A subset of CT measures (in left inferior parietal, left middle temporal, left superior parietal, left supramarginal, right parahippocampal, and right superior temporal regions) and SA measures (in left precentral, left superior temporal, left insula, and right superior parietal regions) remained different (nominal p < 0.05) but did not survive after FDR correction.
Convergence Effect Of Ns Subgroups On Brain Anatomy
To test whether NS subgroups have converging effects on subcortical and cortical measures, we compared maps of brain changes in PTPN11 vs. SOS1. First, we calculated effect sizes and confidence intervals for each ROI and visually contrasted them (Fig. 1b; Fig. 3a). The results display the more extensive effect of PTPN11, relative to SOS1, on neuroanatomy. Next, we tested PTPN11 and SOS1 effect size relationships for subcortical and cortical GMV, SA, and CT25. Pearson correlations indicated statistically significant spatial coherence between PTPN11 and SOS1 on subcortical (r = 0.75, p < 0.05) and cortical GMV (r = 0.57, p < 0.001), SA (r = 0.63, p < 0.001), and CT (r = 0.45, p < 0.001). Permutation testing confirmed that observed correlations between NS subgroups (All permutation p’s < 0.001; Fig. 3c) are significantly greater than null expectations, indicating converging effects of NS subtypes on neuroanatomy.
Higher gene expressions of PTPN11 are related to larger effects of NS on surface area
To explore the relationship between genetics and neuroanatomy in the PTPN11 group, we correlated PTPN11 gene expression and SA effect size (Fig. 5a). PTPN11 expression positively correlated with SA effect size at the whole-brain level (r = 0.32, p = 0.0086). Following correction for spatial autocorrelation with the null-spatial model, we detected a significant difference between the null model and observed correlation (pnull−spatial=0.010), indicating that this relationship displays spatial specificity22,27. We also found significant differences between the null-random-gene and null-brain-gene models and the observed correlation (pnull−random−gene=0.026, pnull−brain−gene=0.027), suggesting that the association between SA effect size and PTPN11 expression association is unique to the PTPN11 gene in comparison to genes randomly selected from either the entire Allen Human Brain Atlas (AHBA) gene set or subset of these genes that are overexpressed in the brain22. In a subsequent lobe-wise analysis, we found that the temporal lobe (r = 0.54, p = 0.022), in particular, is driving these results (Fig. 5b-c). Furthermore, since SA effect size describes differences in SA between PTPN11 and TD groups in specific regions, these results suggest that the higher PTPN11 expression is in a given region, the larger the SA in children with PTPN11 compared to the TD group. Hence, in NS, higher PTPN11 expression is associated with larger SA. We did not find significant correlations between PTPN11 expression and effect sizes of GMV or CT in both the PTPN11 and TD groups.
Striatal volumes correlate with attentional measures in the PTPN11 group
To test whether NS and TD groups differ in brain-behavioral correlations, we focused on the PTPN11 group, given that the larger cohort provides greater power to detect differences. We tested cognitive-behavioral measures involving attention and memory as difficulties in these domains have previously been implicated in NS24,28. For neuroanatomical measures, we examined volumes of the striatum, which is involved in ADHD pathophysiology through the fronto-striatal pathway, and of the hippocampus, which is involved in memory29. Compared to the TD group, we found that PTPN11 performed worse on Auditory Attention (t(54.7)=-2.59, p = 0.012), Inhibition (t(56.4)=-3.81, p = 0.00034), Memory for Faces Delayed (t(49.9)=-3.34, p = 0.0016), Narrative Memory Free Recall (t(53.2)=-3.11, p = 0.0030), and Narrative Memory Free and Cued Recall (t(55.1)=-3.23, p = 0.0021) domains (Table 1). Given that a sizeable proportion of the PTPN11 (20%) and SOS1 (50%) groups were taking stimulant medications and that stimulants have been shown to improve performance on attentional measures in children with ADHD30, it is possible that differences between NS and TD groups on Auditory Attention and Inhibition scores may be even larger without the effect of stimulants. Bilateral striatal volumes negatively correlated with Inhibition scores in PTPN11 with its correlation coefficient differing significantly from the TD group’s (Left: r=-0.42, p = 0.022, TD: r = 0.23, p = 0.15, Fisher test: z=-2.69, p = 0.016; Right: r=-0.37, p = 0.047, TD: r = 0.18, p = 0.28, Fisher test: z=-2.22, p = 0.026) (Fig. 4). Finally, in both the PTPN11 and TD groups, there were no significant correlations between bilateral striatal volumes and Auditory Attention scores or bilateral hippocampal volumes and memory measures. Finally, no correlations were found after repeating these analyses without controlling for TBV. These findings confirm that brain-behavioral relationships are not driven by TBV differences between groups.