Participants in this study were recruited from the Adult Inherited Metabolic Disorders Unit at the Hospital Clínic (Barcelona, Catalonia, Spain). The total sample comprised 57 participants, 35 of whom were early-treated PKU or hyperphenylalaninemia (HPA) patients; the remaining 22 participants were healthy controls (HC) matched according to age, sex, and body mass index (BMI).

The inclusion criteria for PKU patients were: (1) age above 18 years old and (2) genetic diagnosis of PKU or HPA. The exclusion criteria for patients were: (1) intelligence quotient estimation below 70 according to Wechsler Adult Intelligence Scale 4th edition (WAIS-IV) tests, (2) pregnancy or planning a pregnancy during the study period, (3) active cancer, (4) severe chronic hepatic disease, (5) acute cardiovascular event in the 6 months prior to study inclusion, (6) common MRI contraindications, (7) claustrophobia, (8) pathological MRI findings other than mild white matter hyperintensities in long-TR sequences, and (9) MRI artifacts. The exclusion criteria for HC were the same as those applied to the PKU group.

This study was approved by the Bioethics Committee of the University of Barcelona (IRB00003099) and Hospital Clínic of Barcelona (HCB/2020/0552) and was conducted in accordance with the basic principles of the Declaration of Helsinki, among other relevant regulations and guidelines. All participants in this study provided signed written informed consent, after a complete explanation of the procedures involved.

Sociodemographic information and clinical features of participants were also collected, including the date of PKU or HPA diagnosis, Phe monitoring using DBS, venous Phe, previous/current pharmacological treatments, use of adapted formulas (e.g., protein substitutes), consumption of dietetic supplements, subjective cognitive complaints through a clinical interview, and other medical diagnoses. Specific information regarding protein intake was also available, differentiating between natural protein intake and total (natural plus formula-derived protein) intake.

The Phe analysis was performed in DBS via tandem mass spectrometry (MS/MS) using the NeoBase™ 2 Non-derivatized kit (Revvity, Inc; Waltham, Massachusetts, U.S.). Briefly, to extract Phe from 3.2 mm of DBS, an organic compound solution that includes the deuterated Phe-d3 (internal standard) was added. Subsequently, 10 µL of this solution was directly injected into the MS/MS (Xevo-TQD; Waters Corp; Milford, Massachusetts, U.S.). Acquisition was performed in positive ionization and Multiple Reaction Monitoring modes using Masslynx software (Waters Corp). The run time was 2.5 min. The concentration of Phe was calculated based on the area relative to its internal standard, which has a known concentration, using Neolynx software (Waters Corp). The results are expressed in µmol/L.

Values of Phe obtained from DBS conformed to the index of dietary control (IDC), which was approximated using the median Phe levels measured in DBS recorded in the year prior to study inclusion (approximately 6–12 measurements per year for each patient) [30, 32].

The participants included in this study (both HC and patients with PKU) underwent a comprehensive neuropsychological assessment selected according to the most affected domains reported in previous literature [33, 34].The neuropsychological battery included (1) the Vocabulary subtest (WAIS-IV) used to estimate premorbid functioning, (2) the Similarities subtest (WAIS-IV), (3) the Arithmetic subtest (WAIS-IV), (4) the Digit Span subtest (WAIS-IV), which includes Digit Forward, Digit Backward and Digit Sequencing, (5) Letter-Number Sequencing (WAIS-IV), (6) Block Design (WAIS-IV), (7) Matrix Reasoning (WAIS-IV), (8) Digit-Symbol Coding (WAIS-IV), (9) Symbol Search (WAIS-IV), (10) Rey’s Auditory Verbal Learning Test (RAVLT), including immediate, delayed recall after 20 min, and recognition, (11) parts A and B of the Trail Making Test (TMT), (12) semantic (animals) and phonemic (letter “P”) fluency tests, and (13) the Rey-Osterrieth Complex Figure (ROCF), including copy and immediate recall. Behavior and executive functioning were also assessed through the self-reported Behavior Rating Inventory of Executive Function for Adults (BRIEF-A).

High-resolution three-dimensional T1-weighted images were acquired in the sagittal plane (TR = 2400 ms, TE = 2.22 ms, TI = 1000 ms, flip angle 8°, 208 slices, FOV = 256 mm; 0.8 mm isotropic voxel) using a Siemens Magnetom Prisma 3 T scanner located at the Centre de Diagnostic per la Imatge of the Hospital Clínic de Barcelona (Catalonia, Spain). The scanning protocol used in this study also included T2-weighted images in axial orientation (TR = 3200 ms, TE = 563 ms; 512x307 matrix, flip angle 120°, slice thickness 0.8 mm with a 1.5 mm interslice gap) and an axial FLAIR sequence (TR = 9000 ms, TE = 125 ms; TI = 2500; 250×171 matrix, flip angle 150°, slice thickness 4 mm).

Cortical thickness was estimated using the automated FreeSurfer stream version 7.1 available at https://surfer.nmr.mgh.harvard.edu/. The procedures in FreeSurfer included removal of non-brain data, registration to Talairach space, intensity normalization, and tessellation of the gray matter and white matter boundaries, automated topology correction, and accurate surface deformation following intensity gradients to identify tissue borders. Cortical thickness was calculated as the distance between the white and gray matter surfaces at each vertex of the reconstructed cortical mantle (https://freesurfer.net/fswiki/FreeSurferMethodsCitation). Results for each subject were visually inspected, and the appropriate manual corrections were performed to ensure the accuracy of registration, skull stripping, segmentation, and cortical surface reconstruction (https://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/TroubleshootingData).

Deep gray matter mean volumes (e.g., in the thalamus, putamen, pallidum, caudate, hippocampus, amygdala, accumbens, and brainstem) and total cortical and subcortical gray matter were extracted through FreeSurfer version 7.1 [35].

The volumetric measures are presented in ratios, dividing the brain structure volume by the estimated total intracranial volume (ICV) and multiplying the result by 100.

Analyses of sociodemographic, neuropsychological, volumetric, and clinical data were performed using IBM SPSS Statistics 27.0.1.0. Continuous variables were analyzed using Student t-tests or Mann–Whitney U tests to accommodate both normal and non-normal distributions, while categorical variables were assessed through Pearson’s chi-squared test. The Mann–Whitney U test specifically addressed non-normal variables. However, significance in neuropsychological and volume data was reported exclusively for values that passed multiple comparison corrections that had survived family-wise error rate (FWE) correction and the false discovery rate (FDR) through MATLAB (v.R2020b). In all cases the significance threshold was set at a p-value < 0.05.

The study also explored relationships between volume variables, neuropsychological performance and metabolic control assessed by IDC, using pairwise Spearman’s rank correlation analyses, with a significance threshold of p < 0.05.

Intergroup cortical thickness analyses were conducted through a vertex-by-vertex general linear model (GLM) with FreeSurfer version 7.1, including cortical thickness as a dependent factor and group as an independent factor. Cluster-extent correction for multiple comparisons was tested using the Monte-Carlo simulation with 10,000 iterations implemented in FreeSurfer, in order to prevent false positives. Only the clusters that survived FDR with the statistical significance threshold set at p < 0.05 were reported.

The “ggseg” package in R was used for the graphical visualization of volumetric data.