Main

PKU or hyperphenylalaninemia due to Phe hydroxylase deficiency (McKusik 26160) is the most frequent inborn error of amino acid metabolism(1). Without treatment or when treatment is delayed, severe irreversible intellectual and neurologic impairments are unavoidable, whereas early initiation of a strictly Phe-reduced diet results in development close to normal(2–7). Until recently the policy was to relax the diet after 8-10 y of age, but the current recommendations are to maintain strict Phe control as long as possible(8–12). However, there is ongoing debate on how long and how strict the patients should be maintained on diet(3, 10, 13–17).

Hyperphenylalaninemia can interfere with the development and function of the CNS by different ways. A classical theory is that high Phe levels in the brain disturb myelination. Derived from a mouse model, Hommes(18) and Hommes and Moss(19) postulated that increased myelin turnover is not compensated by an increased rate of myelin synthesis resulting in demyelination. Elevated Phe levels may also result in a decrease of neurotransmitter receptor density and therefore in a decrease of cell connectivity(18). Hyperphenylalaninemia does also competitively inhibit the synthesis of dopamine and serotonin, and may limit the transport of tyrosine and tryptophan and their availability to the brain, because Phe, tyrosine, and tryptophan share the same transport system across the blood-brain barrier(20–26). The structural and/or “toxic” hypotheses both might reasonably explain the clinical correlates of intellectual and neurologic impairments, but a complete model is still lacking.

Longitudinal studies have shown that the brain is most vulnerable to elevated Phe levels during the first years of life(6, 27–33). After the age of 8-10 y, it becomes difficult to demonstrate any independent influence of Phe control on measures of general ability(31). On the other hand, neuropsychologic studies have revealed correlations between Phe levels and variables of information processing in all age periods from late childhood to adulthood(26, 34–38).

In line with the hypothesis put forward by Woolf(39) that neuropsychologic measures could be early markers of long-term and aggravating effects of a “late onset Phe intoxication,” it was the aim of the present study to compare early treated patients from childhood to adulthood in a cross-sectional design. Effects of different treatment policies were investigated by a cross-national comparison of patients from Germany and from Paris. In Paris treatment is usually relaxed after 5 y of strict Phe control, whereas in Germany patients are recommended to maintain Phe levels below 360 μmol/L (6 mg/dL) until the age of 10 y, below 600 μmol/L (10 mg/dL) until the age of 15 y, and <900 μmol/L (15 mg/dL) afterward.

The study was designed to test four hypotheses. 1) Hypothesis of normal developmental trends: because there is no evidence that patients with PKU show regressive development, it was hypothesized that trends across age groups should not be different in the three treatment groups, reflecting qualitative similar developmental pathways of decreasing reaction times with age(40). 2) Recommendation hypothesis: recent publications claimed that only Phe levels below 360 μmol/L (6 mg/dL) can be regarded as safe(2, 24, 41). According to this allegation, only the German subsamples should not differ from a healthy control group as long as the mean Phe levels were below this threshold.3) Hypothesis of a continuous linear scale of treatment policies: beyond the threshold of 360 μmol/L (6 mg/dL) the influence of Phe levels on intellectual processes has been conceptualized as continuous and linear(2); according to this assumption for all tests and all age groups the data should represent an ordinal structure with control subjects scoring better than German patients on diet scoring better than French patients off diet. 4) Aggravation hypothesis: cumulating effects of long-term elevated Phe levels should result in increasing differences between patients on and off diet.

METHODS

In the Department of Pediatrics of the Hôpital des Enfants Malades in Paris, 22 early treated French patients were studied. In 20 patients, dietary treatment had been relaxed at a mean age of 5 y; one patient in the youngest age group (ID no. 2) still took his protein substitute but had no qualitative restriction of his diet; one adolescent (ID no. 9) still avoided high protein natural food and took his Phe-free amino acid mixture as well as special products; both had Phe levels of 900 μmol/L (15 mg/dL) at the day of the test. Twenty-three German patients were on a strict diet supplemented by Phe-free amino acid mixtures. Summary statistics of the patients and a control group of 21 healthy German subjects are given in Table 1.

Table 1 Sex, age, and IQ of PKU patients and healthy control subjects
Full size table

The study started with the investigation of the French patients. These patients were selected from the whole sample of patients treated by the second author at the Hôpital des Enfants Malades in Paris. Criteria for selection were diagnoses of classical or mild PKU, start of treatment before the 3rd mo of life, strict dietary control (Phe blood level <360 μmol/L) during the first 5 y of life. All patients from the treatment center who fulfilled these criteria and were willing to participate in the study were included in the sample. The German patients in the two younger age groups were selected from the sample of the German Collaborative Study of PKU treated in Heidelberg, those of the oldest age group were selected from the patients treated at the departments of pediatrics of the universities of Heidelberg and Münster. Control subjects, originally tested for other research purposes, and the German patients were matched to their French counterparts with regard to age, sex, and IQ. In addition, the Phe levels of the German patients should have been as low as possible compared with those of their French counterparts.

Phenotypic classification(42, 43) of the patients was done in relation to the estimated severity of the Phe hydroxylase deficiency. French patients were considered to have type I classical PKU when their plasma Phe level was >1200 μmol/L (20 mg/dL) before treatment, and their Phe tolerance at 5 y was below 0.12 mmol (20 mg)/kg body of weight/d. Type II PKU was defined by pretreatment Phe levels between 600 and 1200μmol/L (10-20 mg/dL), and a Phe tolerance at 5 y of 0.13-0.3 mmol (21-50 mg)/kg of body weight/d. For the German patients tolerances were not determined; instead, the Phe levels after a standardized protein load with 180 mg/kg of body weight Phe in the form of milk protein during three successive days were reported. DNA analysis was completed on 16 of the 22 French patients and on 10 of the 23 German patients (Table 2). As was recently shown(44) in vitro predicted residual enzyme activity measured by expression analysis in cultured mammalian cells and the threshold of 1200 μmol/L after a protein loading test are nearly equivalent for prediction of the type of PKU. The index of dietary control was computed by the mean of all yearly median Phe blood levels(Table 3).

Table 2 Genotypic and phenotypic classification of PKU patients
Full size table
Table 3 Means and standard deviations (in brackets) of treatment parameters of PKU patients
Full size table

Intelligence data resulted from previous testing done for the routine monitoring of the patients; no single test result was older than 1 y. For children and adolescents the German and French versions of the WISC-R and for adults the WAIS-R were administered. These tests are used in most PKU studies(4, 17, 45, 46) and allow a direct comparison of different samples.

All subjects were investigated with three subtests of the SVAT(47–49) implemented on a MS-DOS personal computer. Functions of information processing are hypothesized to be impaired when reaction times in the respective subtest are prolonged. The FMSE measures simple visuomotor reaction time. On the computer screen an exclamation mark is displayed which, after a random time interval, is substituted by a white square. The test requires the subject to press the button of a response panel as fast as possible after the square appears. The DPE measures sustained attention by a binary choice task. Six hundred random configurations of three, four, or five dots (200 each) are displayed on the computer screen. The subject is required to press the response button of the dominant hand as quickly as possible after stimuli of four dots and the button of the nondominant hand after stimuli of three or five dots. For statistical analysis the 600 reactions were divided into 50 series of 12 reactions each; variables of interest were mean ST as a measure of speed and the SD of the 50 mean series times (SDST) as a measure of attentional stability (fluctuation). The LPE requires the subject to scan stimulus sets of four letters displayed on the computer screen for one, two, or three target letters (k; k&r; k&r&d). The number of letters (load) models task difficulty, and the response key of the dominant hand has to be pressed when the stimulus fits the target set and the key of the nondominant hand when the stimulus does not match the target set. Variables of interest were the time for hits as a measure of the mental speed of the visual search process and the number of errors (misses and false alarms) as a measure of accuracy. The three tests examine functions of information processing which have been demonstrated to be sensitive to Phe level variations in all three age groups(38, 49). The French patients were tested by the first author in the Department of Pediatrics of the Hôpital des Enfants Malades in Paris. German patients and control subjects were tested in the Department of Pediatrics at the University of Heidelberg by neuropsychologists of the German Collaborative Study of PKU. Informed consent was obtained from all adult patients and parents or guardians of the children who participated in the study.

Analyses of variance were performed separately for the different tasks and age groups. To test the aggravation hypothesis, within each age group reaction times of the FMSE, DPE, and LPE were converted to SDS (= test score - mean of the respective control group/SD of the control group) allowing a direct comparison of the results of the three age groups. These SDS were averaged over all variables. Aggravation of the disease should be represented by mean SDS increasing with age. Correction of the α error was not necessary because only a priori hypotheses were tested.

RESULTS

No outliers defined as individual reaction times greater than 3 SD from the sample means could be identified for any test. Overall error rates were below 10% and did not differ significantly for the three samples and age groups, indicating correct understanding of the instructions and excluding that differences in reaction times resulted from variations in accuracy. Mean FMSE reaction times were significantly different between the two patient groups in childhood and between adolescent patients from Heidelberg and control subjects. German patients in the childhood group performed as fast as control subjects, and adult subjects were not different at all (Fig. 1,Table 4). DPE levels of performance (ST) showed the same pattern supplemented by a significant difference between adult Heidelberg patients and control subjects (Fig. 2,Table 4). DPE stability of performance (SDST) most clearly distinguished patients and control subjects in all age groups (Fig. 3). In childhood German patients were not different from control subjects, but both adolescent and adult patient groups performed less stably than did control subjects. Except from adolescents, German patients always scored better than French patients (Fig. 2,Table 4).

Figure 1
figure 1

FMSE simple visuomotor reaction times for the three subsamples and age groups (P, Paris; H, Heidelberg;C, control subjects).

Full size image
Figure 4
figure 4

LPE visual scanning. Mean reaction times for three loads.

Full size image
Figure 2
figure 2

Level of DPE sustained attention. ST for the three subsamples and age groups (P, Paris; H, Heidelberg;C, control subjects).

Full size image
Figure 3
figure 3

Stability of DPE sustained attention. SDST for three subsamples and age groups (P, Paris; H, Heidelberg;C, control subjects).

Full size image

LPE results (Fig. 4) were tested separately for each age group by three two-way analyses of variance for repeated measurements. One analysis with three groups (Paris, Heidelberg, controls) by three loads to test the whole design, and two analyses with two groups (Paris versus Heidelberg; Heidelberg versus controls) by three loads to test the ordinal structure of the treatment hypothesis(Table 5).

Table 5 Analysis of variance of LPE results for three age groups
Full size table

In all age groups reaction times increased significantly by load. For children there was no difference between German patients and control subjects, who both differed significantly from the French patients. German and French adolescent patients were not statistically different from each other and from the control group. The slopes of the loads of the two adolescent patient groups were different from the slope of the control subjects. The three adult groups were not different from each other.

Mean SDS of the FMSE, DPE, and LPE indicated that with increasing age the patients from Paris gradually approached the control subjects as well as the patients from Heidelberg. The distance between the Heidelberg patients and the control group increased from childhood to adolescence by 1 SD but again decreased by 0.7 SD from adolescence to adulthood (Table 6).

Table 6 Mean SDS for FMSE and DPE results
Full size table

French patients with type I and type II PKU were not significantly different in IQ [[horizontal bar over]Xtype I = 107; SD = 16.5; [horizontal bar over]Xtype II = 111; SD = 17.0;t (df = 20) = -0.6, NS], but they differed significantly in their Phe level at the day of testing [[horizontal bar over]Xtype I = 1410 μmol/L; SD = 252; [horizontal bar over]Xtype II = 1002; SD = 162; t (df = 20) =-4.5, p < 0.001] and in all neuropsychologic variables except from the FMSE. The average SDS (mean of FMSE, DPE, LPE) was 0.79 (SD = 0.6) for type II patients, and 2.83 (SD = 2.2) for patients with type I PKU[t (df = 19.6) = 3.4, p < 0.01]. Type II SDS of 0.79 was also significantly different from zero (t = 3.06,p < 0.05). The mean IQ of the three German patients with type II was 101 (SD = 12.5; [horizontal bar over]Xtype I = 108; SD = 9.5) and their mean Phe level at the day of testing was 678 μmol/L (SD = 348; [horizontal bar over]Xtype I = 582; SD = 372). Their mean SDS was 0.95 (SD = 0.6) compared with 0.7 (SD = 1.2) for the patients with type I PKU. Due to the small sample size, significance tests were not carried out.

DISCUSSION

Neuropsychologic studies of PKU are primarily designed for two purposes, to investigate the causes behind impairments measured by the psychometric approach (e.g. IQ tests), or to determine subtle deficits as markers for more severe late effects. A review of 21 published studies(49) reported motor speed, speech, language, memory, and basic logic to be generally unaffected. Deficits were primarily found in abstract reasoning, executive functioning (i.e. the guidance of actions by internal representations rather than by external stimulation), and the attention area of information processing. Children of all ages were equally likely to show problem solving deficits and slower reaction times, and performances were more closely related to the concurrent Phe level than to long-term dietary status. However, the authors concluded that “... samples available for study were generally small and heterogeneous with regards to age, treatment history, current dietary status and intellectual capabilities” (p.102).

The present study investigated three different age groups and the various treatment histories of the French and German patients that resulted from different treatment policies. At each age level the groups were small, and the authors prefer to stress the whole pattern of results more than the results of a single variable or age group. The design allowed to test four central hypotheses about the consequences of elevated Phe levels. 1) With the exception of the FMSE results of adult control subjects, mean reaction times always decreased from childhood to adulthood, confirming the hypothesis of normal developmental trends. 2) PKU children from Heidelberg had a mean concurrent Phe level less than 360 μmol/L (6 mg/dL) and were consistently not different from healthy controls, confirming the recommendation hypothesis. 3) The hypothesis of a continuous scale of treatment policies was only violated by the adult control subjects in the FMSE. The means of all other tests showed a clear ordinal structure with control subjects scoring the highest and patients from Paris scoring the lowest. 4) Age trends in SDS averaged over the test variables falsified the aggravation hypothesis. With increasing age, patients from Paris approached the control group as well as the patients from Heidelberg, without, however, reaching the same level of performance. From childhood to adulthood, differences between the Heidelberg patients and control subjects did not show a clear trend. The distance between the two groups increased from childhood to adolescence and decreased for adults. However, it cannot be decided whether the increase (decrease) in reaction time during adolescence (adulthood) was due to the (adaptation to) increased Phe levels of the German patients after childhood (adolescence) or due to sampling error in a cross-sectional design. There was also evidence that results may depend on different subtests. The results of the LPE, probably the test with the highest cognitive demand, completely converged in adulthood, whereas the stability of sustained attention (SDST), the performance with the greatest demand for vigilance, still significantly distinguished all three adult groups.

The mean SD scores of the neuropsychologic variables as well as the Phe level at the day of testing were statistically different for French patients with type I and type II PKU. The French patients with mild PKU had a mean SDS in the same range as the German patients with type I and type II, although their mean Phe level was about 50% higher than the mean of the German patients. Taking into account the small sample size of the type II patients, the whole pattern of the four groups raises the question, whether after childhood the treatment of type II patients has any therapeutic effect.

The results of this study again have demonstrated the influence of concurrent Phe levels on neuropsychologic test results. The pattern of our results indicates that effects will be the same whenever Phe levels increase after the period of strict treatment during childhood. On the other hand, there is also evidence that, even after longer periods of poor compliance, information processing speeds up when Phe levels decrease(38). This was probably the case for two pregnant French women with Phe levels of 240 μmol/L (4 mg/dL) who were tested parallel to this study and scored on the level of healthy control subjects.

The sampling criterion of very strict dietary control during treatment and the voluntariness of the test participation might have contributed to the fact that our patients represented a sample from the upper tail of the IQ distribution and we cannot exclude that high IQs are associated with self instructions for strategy use in neuropsychologic tests(51, 52), counterbalancing the possible adverse effects of elevated Phe levels. In particular, this might explain the nearly identical results of all adult subjects in the LPE, the test with the greatest cognitive demands in our battery. However, within the limitations of the cross-sectional design it could be shown that relaxation of dietary control does not necessarily result in a late onset Phe intoxication, impairing faculties of information processing, instead on the average the three groups seem to converge in older age. The corroboration of our results in a longitudinal design with a larger and more heterogeneous sample of patients in combination with results from longitudinal IQ studies(45, 46) would significantly contribute to answering the question of the duration of dietary treatment. There also remains the determination of the practical relevance of neuropsychologic effects resulting from Phe blood levels(53). There is insufficient knowledge whether the impairments found in neuropsychologic studies necessitate the burdensome diet necessary for low Phe levels. This ignorance is not reserved for neuropsychology but can also be addressed to studies using IQ data. Study designs with greater ecologic validity(54) and empirical as well as theoretical links to clinical phenomena would certainly improve our understanding of PKU in adolescence and adulthood. Finally early and strictly treated PKU is a condition too young to study the consequences of a long-term poor dietary control in middle or late adulthood. The present study could not prove long-term and aggravating effects of late onset Phe intoxication for the period of the first 15 y after diet relaxation in the French patients. However, the pattern of differences between the three groups in childhood makes it much more difficult to justify a policy of deliberately stopping the diet at least during the first 10 y of life, when the French patients off diet consistently scored lower than the German patients on diet.

To our knowledge early and strictly treated patients do not show absolute disability in performing any particular test as a consequence of elevated Phe levels. Instead, in comparison with healthy control subjects, their cognitive functions are most often impaired in terms of prolonged reaction times and less often by higher error rates. Therefore, in summary it seems to be more appropriate to characterize patients with early treated PKU as performing qualitatively normal on a reduced level than to perform qualitatively different from healthy subjects.

Table 4 Analysis of variance results and a priori one-tailed t-tests of the FMSE and DPE data
Full size table