Twenty-one patients (31% of the original cohort; 6 male, 15 female) completed a neuropsychological assessment at two time points. The remaining 69% of the original cohort was lost to follow up. Patients were not under treatment anymore and/or were out of sight from physicians (n = 40) or were not willing to participate (n = 6), usually due to time constraints. IDC at T1 (t(62) = 1.1, p = 0.29) and the pre-treatment Phe (t(64) = 0.4, p = 0.72) did not significantly differ between the 21 versus 46 patients, but conPhe at T1 was higher for the 69% that was lost (t(49) = 2.0, p = 0.027). However, data regarding executive functioning was gathered for both groups at T1. The two groups did not differ on inhibitory control (FL and SSV), cognitive flexibility (SSV) and executive motor control (PU), indicating that the two groups had similar executive functioning at T1.
For those 21 patients who participated at T2, mean age at T1 was 10.4 years (SD = 2.0) and 25.8 years (SD = 2.3) at T2 (see Table 1 for descriptive statistics). According to the traditional classification for biochemical subtypes within the hyperphenylalaninemias (Blau et al. 2011), four patients were classified as having hyperphenylalaninemia (HPA, i.e. a pre-treatment Phe level below 600 μmol/L), eight as mild PKU (pre-treatment Phe between 600 and 1200 μmol/L) and nine had classical PKU (pre-treatment Phe ≥1200 μmol/L). Mean conPhe at T1 was 346 μmol/L (SD = 204) while at T2 this was 719 μmol/L (SD = 351). Mean IDC at T1 was 312 μmol/L (SD = 96) and IDC at T2 was 464 μmol/L (SD = 138). Phe levels between 0 and 12 years, between 13 and 17 and ≥18 years are reported in Table 1. Correlations between Phe variables are displayed in Table 2. Phe concentrations significantly increased with age, as shown with repeated measures including Phe 0–12, 13–17 and ≥18 years (Wilks’Λ = 0.35, F(2,19) = 17.30, p < 0.001, η²p = 0.65). Within T1 and within T2, age did not significantly correlate with Phe. At T1, BH4 was not yet prescribed in the Netherlands. At T2, five patients used BH4 doses up to 20 mg/kg with a max of 1400 mg/day.
Regarding the ‘low–high’ groups, three patients were allocated to the ‘low–low’ group (mean Phe <360 μmol/L when 0–12 years and from age 13 onwards), 11 patients were in the ‘low–high’ group (Phe <360 μmol/L until age 12 and ≥360 μmol/L from age 13) and seven patients were in the ‘high–high’ group (Phe ≥360 μmol/L in childhood and onwards). None of the patients had high values as a child and low values as an adult. Descriptive statistics and metabolic measurements for these groups are displayed in Table 1. Because the first group was too small, only the ‘low–high’ and ‘high-high’ groups were included in group-based statistical analyses. The ‘low–high’ group consisted of one patient with hyperphenylalaninemia, six patients with mild PKU and four patients with classical PKU, while the ‘high-high’ group included two mild PKU patients and five patients with classical PKU. The pre-treatment Phe concentration (t(16)= −1.5, p = 0.17) and the PKU classification (χ
2 = 2.3, p = 0.31) did not differ between the two groups. The socio-economic status, i.e. education and yearly income, of these two groups was similar, respectively χ
2 = 3.1, p = 0.37 and χ
2 = 1.7, p = 0.89.
Participant Characteristics: Relationships, Education, Occupation, Income
Sixteen out of 21 patients (76%) had a long-term, romantic relationship, 13 have had two or more long-term, romantic relationships in the past. All patients completed high school. Eight patients (38%) followed or completed higher vocational education and six patients (29%) completed higher education (bachelor’s or master’s degree), which is comparable to the healthy Dutch population, according to the Dutch Central Bureau for Statistics (CBS 2016a). All patients had an occupation at T2, working 12–40 h/week. Seven patients (33%) had a higher than average income and 12 out of 21 (57%) were house owners of private property, similar to the Dutch population (CBS 2016b).
Associations Between Metabolic Control, Executive Functioning and Mental Health
Regarding inhibitory control/interference suppression (FL-task), partial correlations (controlling for Phe 0–12 years) showed that Phe 13–17 years, the IDC difference score and IDC2 were significantly associated with percentage errors at T2 (see Table 3 for associations between Phe and ANT-tasks). The positive correlation between the IDC difference score and error percentage indicated that with a larger increase in Phe between childhood and adulthood, poorer performance was observed at T2.
Inhibition of prepotent responding (SSV-task) was not associated with any of the Phe indices. When measuring cognitive flexibility, the percentage of errors at T2 was significantly correlated with Phe 13–17 years while the correlation with the IDC difference score just failed to reach significance after controlling for Phe 0–12 years. RT at T2 was associated with IDC1 and with Phe 0–12 years.
Regarding executive motor control (PU-task), IDC1 and Phe 0–12 years were significantly related to accuracy and stability of executive motor control at T2. These correlations indicated that high Phe in childhood were related to poorer executive motor control in adulthood.
The Behavioral Regulation Index (BRI) of the BRIEF-A which was only measured at T2, was significantly associated with IDC1, Phe 0–12 years and IDC2 (see Table 4 for correlations between Phe and questionnaires). When controlling for Phe 0–12 years, the correlation with IDC2 became non-significant. The Global Executive Composite (GEC) was also related to Phe 0–12 years. These results demonstrate that childhood Phe is important for executive functioning in daily adult life.
With respect to mental health at T2, Depressive problems were related to Phe 0–12 years and to IDC2. Somatic problems had a significant relation with IDC1, Phe 0–12 years and IDC2. IDC1 and Phe 0–12 years were associated with Attention Deficit/Hyperactivity problems. Antisocial Personality problems were related to IDC1, Phe 0–12 years, Phe ≥18 years and IDC2. Overall internalizing problems were associated with IDC1 and Phe 0–12 years. The externalizing problem scale was related to IDC1, Phe 0–12 years and IDC2. However, when controlling for Phe 0–12 years, the correlations with Phe indices after childhood became non-significant. Finally, the overall total problem scale was again associated with IDC1 and Phe 0–12 years. The IDC difference score was not significantly related to the ASR scores.
Regarding the neuropsychological test outcomes, correction for multiple testing (seven IDCs times eight outcome measures) would result in non-significant correlations only, as a p value of 0.0009 would be the criterion. We opted for calculating all these correlations as the different IDCs may provide (subtly) different information. Still, the IDCs were often strongly correlated (see Table 2). If one would retain only IDC at T1 and the IDC difference score, as they represent childhood Phe and Phe-change thereafter, and therefore seem the most meaningful indices, no significant correlations are retained either, as a p-value of 0.003 would be the criterion. Some could be retained by reducing the number of outcome measures, e.g. one per task. However, there is no evidence to support one should choose either RT, error rate, or stability of performance as main outcome measures of the tasks.
Questionnaire data showed stronger correlations with IDCs, with IDC1 and Phe0–12 correlations with the BRI of the BRIEF almost remaining significant after applying p-level corrections with a factor 4, i.e. two indices of metabolic control (i.e. IDC1 and IDC difference score or Phe 0–12 and Phe >12), and two outcome measures (MCI and BRI indices). A number of results obtained on the ASR actually did remain significant after applying corrections with a factor 12 (i.e. two IDCs, six ASR-dimensions): somatic problems, attention deficit/hyperactivity problems, or 4 (i.e. two IDCs, two ASR-dimensions): internalizing and externalizing problems. The pattern of correlations indicated that especially those with IDC1 or Phe 0–12 were robust (Table 4). Further group comparisons were made (see below) in order to link results to upper target Phe levels (for different age periods) according to current treatment guidelines, and to gain more insight into development of cognition and mental health in PKU patients relative to healthy controls.
Comparison ‘Low–High’ and ‘High–High’ Groups
For inhibitory control/interference suppression (FL-task), repeated measures analyses of variance did not show significant effects for time or group. Also, there was no interaction effect. On the SSV-task, there were no significant effects for inhibition of prepotent responding. Concerning cognitive flexibility, the ‘low–high’ group was significantly faster than the ‘high-high’ group at both time points (F(1,15) = 8.0, p = 0.013, n
= 0.349) (see Fig. 1). Although descriptive statistics showed better Z-scores on cognitive flexibility for all patients at T2 compared to the age-appropriate norm, time and interaction effects were not significant. For executive motor control (PU-task) there was a significant effect for time. All participants improved over time compared to the age-appropriate norm: they had a more accurate (F(1,16) = 5.5, p = 0.033, n
= 0.255) and stable (F(1,16) = 4.5, p = 0.050, n
= 0.219) executive motor control at T2. Also the group effect was significant for stability of movement: the ‘low–high’ group had a more stable executive motor control than the ‘high–high’ group (F(1,16) = 5.6, p = 0.031, n
= 0.259) (see Fig. 2).
Regarding the BRIEF scales on executive functioning in daily life, there were no statistically significant differences between the ‘low–high’ (n = 11) and ‘high–high’ (n = 7) groups. Finally, with respect to mental health as measured by the ASR scales, the ‘high-high’ group reported significantly more Somatic problems than the ‘low–high’ group (t(14) = −3.8, p = 0.002). When examining the descriptive statistics, two patients from the ‘low–high’ group (18%), scored in the borderline range of the internalizing scale and had a lifetime Phe of 420 and 514 µmol/L respectively. Two patients (29%) from the ‘high-high’ group were in the clinical range with lifetime Phe of 446 and 678 µmol/L. The latter patient with the highest Phe also scored in the clinical range of the externalizing and overall total problem scale. This patient also scored in the clinical range of the BRIEF. The patients in the normal range had similar lifetime and childhood Phe as those in the borderline and clinical range.