The ethical committee of the Beijing Children's Hospital approved this study, and informed consent was obtained from all subjects and their legal guardian. A total of 42 children (31 boys and 11 girls) with NSCLP (NSCLP group) and 42 age- and gender-matched typical developing healthy controls were recruited for the current study from the Beijing Children's Hospital. The inclusion criteria were as follows: (1) NSCLP patients (ranging from 6 to 16 years) who started the speech therapy 3–6 months after a successful pharyngeal closure surgery, by the frequency of 30 min/day and 3 times/week, lasting for half a year, till they reached the basic point 86 according to the Chinese language clear degree scale (CLCDS) scores; (2) normal vision and hearing (by auditory brainstem response examination below 30 dB nHL); (3) average intelligence [more than 90 scores using the Chinese Wechsler Intelligence Scale for Children-IV (CWISC-IV)]; and (4) right-handed subjects. The exclusion criteria of the patients were as follows: NSCLP patients with clinic diagnoses of (1) dysarthria; (2) velopharyngeal insufficiency; (3) hearing and/or vision impairments; (4) developmental delays; (5) congenital disorders; (6) other syndromes or possible; and (7) other chronic health diseases.

All data were acquired on a 3.0 T GE MRI system at the Department of Radiology (Beijing Children's Hospital). For each subject, high resolution 3D T1-weighted gradient-echo images were acquired with the following parameters: repetition time (TR) = 8.2 ms, echo time (TE) = 3.5 ms, inversion time (TI) = 450 ms, flip angle (FA) = 13°, matrix = 256 × 256, and FOV = 256 × 256, 164 continuous sagittal slices were scanned to cover the whole brain, and slice thickness = 1 mm.

Cortical thickness and gyrification index are widely used in the SBM analysis of structural plasticity. Cortical thickness is a sensitive metric of dynamic brain alterations across development stemming from maturational pruning and experiential neuroplasticity (15), and the gyrification index provides a new method for analyzing brain structure. It focuses on the frequency of gyrus folding, describing either structurally constrained cognitive characteristics or acquired long-term experience-dependent plasticity (16). Moreover, because the gyrification index is characterized by relative lifetime stability, it is suitable as an index of aberrant neurodevelopment (17).

For the SBM analysis, the CAT12 (Computational Anatomy Toolbox, http://dbm.neuro.uni-jena.de/cat/) and SPM12 (statistical parametric mapping, http://www.fil.ion.ucl.ac.uk/spm/) were used within the MATLAB environment. We used this software to reconstruct the cortical surface of each subject and calculate multiple morphometric parameters (18, 19). During the procedure, cortical thickness estimation and the central surface were carried out in one step based on the projection-based thickness (PBT) approach (19). Then, the topology correction, spherical mapping, and spherical registration were performed for cortical thickness and central surface. After that, the gyrification index was extracted from the central surface. Finally, the surface of cortical thickness and gyrification index was resampled and smoothed with a 15-mm filter for each hemisphere. For further statistical analysis, the mean cortical thickness and gyrification index were extracted for 68 ROIs defined by the Desikan-Killiany atlas (20) with CAT12.

Statistical analyses were conducted using the statistical module CAT12/SPM12. The age and gender were taken as covariates, and the between-group differences in the mean regional value of the two morphometric parameters were tested with two-sample t-tests (p < 0.05) and permutation tests 5,000 times (p < 0.05). In addition, the mean values of between-group different regions were calculated for the correlations between the metrics and CLCDS in the NSCLP group using Pearson's correlation analyses (p < 0.05).

Table 1 shows that it was not significant in the age (t = −0.46, p = 0.96), CWISC-IV scores (t = −1.18, p = 0.24), and education years (t = −1.12, p = 0.28) of children between the NSCLP group and healthy controls. The CLCDS scores were 91.45 ± 3.92 in the NSCLP children. The number of boys and girls showed no between-group significant differences in the two groups (χ² = 0, p = 1).

Figure 1A depicts a whole-brain cortical thickness analysis comparing healthy controls to the NSCLP group, controlling for age and gender with permutation tests (p < 0.05). Compared with healthy controls, the cortical thickness in the bilateral lingual gyrus was lower in the NSCLP group. The thinner cortex was found in additional left hemisphere areas, including the superior/middle frontal and parahippocampal gyrus. In the right hemisphere, the regions of the thinner cortex in the NSCLP group included pericalcarine and cuneus cortices. The size (mm²) of each cluster and the location coordinated at Montreal Neurological Institute (MNI) are shown in Table 2. Compared to healthy controls, the bar plot graph of each cluster with thinner cortex in the NSCLP group was demonstrated in Figure 2.

Figures 1B,C depicts whole-brain regional gyrification index analysis comparing healthy controls to the NSCLP group, controlling for age and gender with permutation tests (p < 0.05). The gyrification index of the bilateral insula gyrus, the right parahippocampal gyrus, and the left hemisphere (lateral orbitofrontal, pars opercularis, pars triangularis, superior parietal, and cingulate gyrus) were higher in the NSCLP group than in healthy controls. Compared to healthy controls, a lower gyrification index was shown in the NSCLP group on the right hemisphere transverse temporal and caudal superior temporal gyrus (p < 0.05). The size (mm²) of each cluster and the location coordinated at Montreal Neurological Institute (MNI) were shown in Table 1. Compared to healthy controls, the bar plot graph of each cluster with a different gyrification index in the NSCLP group was demonstrated in Figure 3.

The correlations between the cortical morphological parameters and CLCDS scores were calculated, finding a positive correlation between the cortical thickness of the left parahippocampal gyrus and CLCDS scores (r = 0.370, p = 0.017) in the NSCLP group (Figure 4).

Consistent with our hypothesis, we identified that the residual speech errors in NSCLP children after speech rehabilitation demonstrated a Thinner cortex than healthy controls in language-related brain areas, including the left superior frontal gyrus, left caudal middle frontal, left pars orbitalis, left parahippocampal, right pericalcarine, right cuneus, and bilateral lingual. The left superior frontal and middle frontal gyrus were crucial areas of the dorsal phonological processing stream (21). At the same time, the cuneus, pericalcarine cortices, and lingual gyrus were essential brain regions in the ventral visual pathway (22) and the reading circuitry (23). The parahippocampal gyrus is involved in increasing expertise (24, 25). Previous studies confirmed that expertise training, e.g., simultaneous interpretation learning (26) and diving training (25), could increase cortical thickness. Our result found that the cortical thickness of the language-related brain areas in NSCLP children after speech therapy still did not reach an average level, which indicated that the speech-rehabilitated children with NSCLP also had structural brain differences.

The declining period of the gyrification index describes the fine-tuning process during the school-aged child period for cortical maturation (27). In this period, a previous study confirmed that developmental dyslexia could lead to an increased gyrification index (28). Moreover, expertise training induces the deceased gyrification index, such as diving training (29).

The higher gyrification indexes were detected in the insula, lateral orbitofrontal cortices, pars opercularis, pars triangularis, superior parietal lobule, and cingulate gyrus of the left hemisphere and in the insula and parahippocampal gyrus of the right hemisphere. The pars triangularis and pars opercularis belong to Broca's area, which was related to the motor function of language (30). In our study, all subjects were at age 6–16. Thus, they were in the declining period of the gyrification index. The higher gyrification index of these language-related brain regions suggested immature neural circuitry, which might result from the speech dysfunctions of NSCLP children. Our result showed that the higher gyrification index did not develop to normal levels in NSCLP children after speech therapy, indicating structural-developmental disorders.

The insular cortex, lateral orbitofrontal cortices, pars opercularis, pars triangularis, and superior parietal lobule are associated with phonological processing (21, 31). The cingulate gyrus is concerned with assessing the spatial, context, and personal relevance of sensory information (32). Therefore, this abnormal brain structural development may be closely related to speech disorders. A study found increased gyrification of parietal, frontal, and temporal lobes in adults with NSCLP after articulation rehabilitation (14), similar to our research.

Moreover, the lower gyrification indexes were located in the right transverse and caudal superior temporal gyrus. The caudal superior temporal belongs to Wernicke's area, the sensory language center (33). The transverse temporal gyrus is involved in acoustic signal perception (34). Besides, the caudal superior temporal gyrus is demonstrated to transfer language sound into phonemes from the transverse temporal gyrus (21). The decreased gyrification index suggested that neural connections in these regions were strengthened for cortical maturation (27). Therefore, we could infer that not only the earliest sound process of the right transverse temporal and causal superior temporal gyrus was not impaired, but also more sound information of speech therapy through these brain regions led to the neuroplasticity, describing the lower gyrification index.

Additionally, the correlation analysis showed that the mean cortical thickness of the left parahippocampal gyrus was positively correlated with the CLCDS scores. The parahippocampus, an important memory brain region, receives input from other brain regions and processes several types of sensory information (35). Li et al. (14) found that changed attention and language networks and working-memory brain structures (hippocampal and parahippocampal gyrus) were related to impaired language function. Our results suggested that the cortical thickness in the left parahippocampal gyrus was closely related to pronunciation function in the NSCLP children, indicating the state of pronunciation function. Furthermore, the mean cortical thickness of the left parahippocampal gyrus was lower in the NSCLP group than in healthy controls, which suggested that the damage to the parahippocampal gyrus might prevent the pronunciation function recovery. The lower cortical thickness of the left parahippocampal gyrus might be underlying neural mechanisms in NSCLP children after speech therapy, which might be a biomarker for evaluating pronunciation function.