The PKU group consisted of 95 patients (41 males, 54 females, mean age 21.6 years ± 10.2, range 7.0–42.8 years). All patients were born in or after 1974, when neonatal screening was introduced in The Netherlands, and were treated since diagnosis early after birth. All patients paid regular follow-up visits to one of six Dutch university medical centres (and were treated according to National Instituted of Health (NIH) guidelines, published in 2000). The healthy matched control group included 95 subjects (30 males, 65 females, mean age 19.6 years ± 8.7, range 7.2–40.8 years). Seven controls were siblings or cousins of the patients. The PKU and control groups did not differ significantly in age and gender distribution (Table 1). No differences were observed regarding SES-related measures (e.g. mean family income or achieved education levels) either. Although in the normal range, the PKU group had a lower IQ (101 ± 12) than controls (108 ± 13): t(186) = 3.8, p < 0.001.
Blood samples were taken from the patients preprandially at home on the day of neuropsychological testing, which took place at the patients’ clinics. Historical Phe-levels were collected from hospital databases. Lifetime Phe was calculated as the mean of half-year median Phe-levels, from birth until the day of testing. Also, Phe-levels between 0–7, 8–12, 13–17 and ≥18 years were calculated when applicable, to gain more insight into Phe-levels during particular developmental stages. PKU-patients had a mean concurrent Phe-level of 556 μmol/L ± 328. Mean lifetime Phe was 399 μmol/L ± 154 (n = 92; lifetime Phe of three patients was for the larger part incomplete and could therefore not be calculated) (Table 1). Concurrent and lifetime Phe were correlated with each other (r = 0.67, p < 0.001). Age was significantly related to concurrent (r = 0.37, p < 0.001) and lifetime Phe (r = 0.55, p < 0.001), indicating higher Phe-levels for older patients. Historical Phe-levels between 0 and 7 years were higher for adults with PKU than for children and adolescents with PKU (F(2,88) = 3.2, p = 0.046).
Thirty patients used tetrahydrobiopterin (BH4-) doses up to 20 mg/kg with a maximum of 1400 mg/day. The BH4 group was younger (mean age 15.8 ± 8.5) than the non-BH4 group (mean age 24.4 ± 9.9): t(65) = 4.4, p < 0.001. Lifetime (t(90) = 6.3, p < 0.001) and concurrent Phe (t(90) = 5.0, p < 0.001) were lower for the BH4 group.
Children performed three tasks to measure social-cognitive abilities; adolescents and adults performed four tasks. In order to assess social skills, for children and adolescents the Social Skills Rating System (SSRS; Gresham and Elliot 1990) was filled out by parents, whereas adults filled out the Social Skills Checklist-Self Report (SSC-SR; Novotni 2000). Subtests Vocabulary and Block Design from the Wechsler Intelligence Scale for Children (WISC-III, Wechsler 1991) or the Wechsler Adult Intelligence Scale (WAIS-III, Wechsler 1997) were used to estimate verbal and performance IQ.
Two tasks were performed by all participants: the computerized Face Recognition and (FR) and Identification of Facial Emotions (IFE) tasks from the Amsterdam Neuropsychological Tasks (ANT; De Sonneville 2014). FR examines how fast and how well participants are able to recognize neutral faces and consists of three parts (40 trials each): faces were presented from the front, ‘en profile’, and upside down. Each trial is preceded by a target face, after which participants have to decide whether the subsequent signal, containing four faces, includes the target face. In the IFE, participants identify facial emotions as quickly and accurately as possible. The task has four parts (40 trials each), where consecutively happy, sad, angry and frightened faces have to be identified. The mean of age-corrected z-scores for accuracy and RT on the FR and IFE was calculated, and converted so that a higher score indicated a better performance, leading to one overall z-score representing social cognition as measured by computerized tasks.
One task was performed only by children up to the age of 12. This paper-and-pencil (P&P-) task, the Social Cognitive Skills Test (Van Manen et al 2007), examines ToM-skills. The task consists of seven stories with associated pictures. After each story, questions were asked representing eight aspects of ToM, leading to a total score with four levels. The standardized total score was used as dependent variable. A high score indicates better ToM. Two tasks were performed only by adolescents and adults. These P&P-tasks, the Faux-Pas Recognition Test (FPT; Baron-Cohen et al 1999) and the Reading the Mind in the Eyes (RME-) task (Baron-Cohen et al 2001) measure ToM. FPT measures how one interprets social situations that are potentially awkward (‘faux pas’). It consists of five stories containing a faux pas and four control stories, followed by questions about the social occurrence in the story. A higher score indicates a better ability to interpret social situations. This score was again standardized for statistical analyses. RME measures how well mental states of others are understood based solely on information from the eyes. The test consists of 36 photographs of eye regions of individuals, of which participants have to identify the best fitting emotion from four possible answers. A higher (standardized) score indicates a better ability to recognize emotions. Mean z-score for accuracy of the three P&P-tasks was used in statistical analyses.
Regarding the questionnaires, the SSRS consists of 38 items, measuring social skills on four different dimensions (cooperation, assertiveness, responsibility and self-control), thereby using 3-point rating scales. Higher scores represent better social skills. The SSC consists of 81 items, measuring social skills on eight different dimensions (basic manners, nonverbal and verbal communication, communication roadblocks, organizational skills, self-control, knowledge, relationships and self-care), using 3-point rating scales. Higher scores indicate better social skills. Standardized mean score for the SSRS and SSC-R was used to represent the level of social skilfulness.
As noted, standardized scores, calculated separately for the computerized tasks, P&P-tasks and questionnaires were used in statistical analyses (performed with IBM SPSS Statistics 22nd version). For all measurements, higher z-scores indicated better performance on the social cognition tasks or better social skills.
General linear model (GLM-) multivariate analyses of variance were conducted for all participants, then separately for participants <12 years, 12–17 years and ≥18 years. Within the separate age-group analyses further statistical control for age was exerted (by including it as a covariate) when applicable, as within each age group developmental influences may still be present. Social cognition-computerized, social cognition-P&P and social skills (all z-scores) were used as dependent variables. Analyses were repeated with IQ as a covariate. Comparisons between BH4-treated and non-BH4-treated patients were also made using t-tests. Associations between social outcomes and lifetime/concurrent Phe were investigated using Pearson’s correlations.