Main

Phenylketonuria (PKU) is the most frequent inborn disorder of amino acid metabolism. PKU is caused by an autosomal recessively transmitted deficiency of hepatic phenylalanine (Phe) hydroxylase (PAH; enzyme commission 1.14.16.1). In untreated PKU patients, Phe accumulates to plasma levels exceeding 1500 μM; normal plasma Phe in fasting adults is ∼60 μM. The characteristics of untreated PKU result from disturbed brain development and include microcephaly, epilepsy, severe mental retardation, and behavior problems (1).

Through the early start of a Phe-restricted diet, the phenotype of untreated PKU can be widely prevented. However, elevated plasma concentrations of Phe, which usually occur in adolescence because of a relaxed diet, are known to cause acute impairments of brain function. Reduced attentional abilities, slow information processing, and typical changes in the spectral pattern of brain electrical activity (EEG) have been observed (2). Magnetic resonance imaging (MRI) revealed white matter abnormalities also in neurologically normal early-treated patients (3).

Despite progress in the description of the molecular variability of PKU, the biochemical and cellular mechanisms underlying impaired brain development and cerebral dysfunction by acute Phe “toxicity” remain unclear. Besides several other important aspects, inhibition of enzymes involved in cerebral energy production has been shown for Phe in in vitro studies (4–7). However, no clear evidence could be established to prove that impaired cerebral energy metabolism represents a key explanatory factor for PKU-related brain damage or dysfunction in vivo(8).

Quantitative phosphorus-31 magnetic resonance spectroscopy (31P-MRS) can serve as a noninvasive tool for examination of energy status in vivo. This approach is routinely used to study muscle energetics (9, 10) but has also been used to investigate disturbances of brain function in many cerebral and metabolic diseases (11). The aim of the present study was to determine the concentrations of cerebral metabolites detected by 31P-MRS in early-treated PKU patients under steady-state conditions, to compare them with a control group, and to relate the results to blood and brain concentrations of Phe. In addition, 31P-MRS was done before and 24 h after an oral challenge with L-Phe, while brain activity was monitored using EEG spectral analysis (12).

METHODS

Patients and control subjects.

Eleven early-treated patients with PKU (four women, seven men, mean age 24.9 ± 3.9 y) participated in this study. Diet onset ranged from 11 to 63 d (Table 1). 31P-MRS data from 10 healthy volunteers (five women, five men; 26.2 ± 5.1 y) were acquired to define normal ranges. This study was approved by the Ethics Committee of the University of Heidelberg. Before the study, the experimental protocol was explained to all subjects and their written informed consent according to the declaration of Helsinki 1996 was obtained.

Table 1 Genetic and clinical data of 11 patients with phenylketonuria
Full size table

Laboratory and clinical examination.

Plasma Phe concentrations concurrent with 31P-MRS were determined with an enzymatic assay (Quantase Phe assay, Porton Cambridge, UK). An index of dietary control (IDC) was derived from the patients' plasma Phe values from diet onset (Table 1) until the time of the study to characterize long-term biochemical control. Further clinical evaluation (Table 1) included the identification of PAH mutations. Mutations were classified according to Guldberg et al.(13). All patients were examined with a standardized neurologic investigation, standard MRI, and intelligence testing (Wechsler Adult Intelligence Scale–Revised). MRI-visible white matter abnormalities were graded according to the scheme described earlier (3).

For determining brain tissue concentrations of Phe, localized 1H-MRS was performed after the 31P-MRS session on the same scanner as described in detail elsewhere (12, 14, 15). In brief, spectra were recorded from a large volume of ∼60 cm3 above the ventricular system involving white and gray matter using a spin-echo localization sequence (echo time 20 ms, repetition period 3 s). The full data processing scheme is detailed elsewhere (12).

31P-MRS.

All 31P MR spectra were recorded on a clinical MRI scanner (1.5T Signa, General Electric Medical Systems, Milwaukee, WI, U.S.A.) using a double-tuned bird cage 1H/31P head coil (quadrature for 31P, linear for 1H, GE Medical Systems). A supraventricular region of interest was selected using an ISIS sequence with adiabatic inversion, excitation, and saturation pulses (16). Nominal region of interest size was 70 cm3. Repetition time was 4.0 s. Four spectra of 128 scans were acquired per subject. Total examination time was 75 min. Absolute quantification was based on measurement of the local B1 field, the reciprocity principle, and calibration with solutions of known concentration, as well as constrained model fitting, as described in detail recently (17).

Intracellular pH was calculated from the chemical shift of inorganic phosphate (Pi) in relation to phosphocreatine (PCr) using the calibration of Petroff et al.(18). Free Mg2+ content was estimated from the frequency difference of β-ATP and PCr using the parameters of Iotti et al.(19). An overall cytosolic ADP concentration (20) was calculated on the basis of the creatine kinase equilibrium: MATH where the in vitro value of 1.66 109 M−1 is used for the creatine kinase equilibrium constant KCK(20, 21). [Cr] is calculated from the measured PCr concentration, assuming a constant total creatine content of 6.3 mmol/kgww(22), and [H] is the proton concentration derived from the measured pH (10−pH M). The effective KCK depends on local pH, Mg2+, and K+ contents, all of which can be assumed to be unchanged in PKU. For pH and Mg2+, this assumption is confirmed by the current results. For total creatine, no significant changes have been found for PKU in 1H-MRS studies. In addition, the phosphorylation potential (PP), an index of free available energy per ATP, was derived (21): MATH

The velocity of oxidative metabolism relative to its maximum (V/ Vmax) was calculated on the basis of a steady-state Michaelis-Menten formulation: MATH where [ADP] and [Pi] are in units of mM (23) such that V/ Vmax is dimensionless.

Oral load with L-phenylalanine in five patients.

Five patients (3, 4, 5, 8, and 11) underwent an oral challenge with L-Phe. They were administered a single oral dose of 100 mg of purified L-Phe per kilogram of body weight. 31P-MRS was performed preload and 22 h postload. For detecting Phe-related changes of brain activity, EEG in the relaxed awake state with eyes closed were derived preload and 24 h postload. Electrodes were placed according to the 10/20 system. Spectral analysis of EEG activity at rest was analyzed for occipital leads according to procedures described earlier (12). Because EEG spectra show systematic diurnal variations, comparison of EEG parameters was restricted to preload and 24 h postload measurements, which both were recorded during the morning hours.

Statistical analysis.

Statistical analysis was carried out using the SAS statistical software package version 8.0 (SAS Institute, Cary, NC, U.S.A.) on Windows-NT. For testing for differences between groups, the Kruskal-Wallis test was used. Correlation analysis was done by calculating Spearman rank correlation coefficients. The t test for repeated measurements was used for within-group retest comparisons during the Phe loading part of the study. P < 0.05 was taken to be significant;p values between 0.05 and 0.10 were reported as trends.

RESULTS

Clinical Data

Clinical data are summarized in Table 1. Most mutations were typical for European PKU patients. Five patients corresponded to severe PKU, four patients had a moderate type, and two patients had a mild or mild to moderate type (13). Other aspects of the PKU genotypes of this group have recently been discussed (15).

Concurrent plasma Phe values ranged between 799 and 1784 μM, which is in the typical range for adult PKU patients (Table 1). Plasma and brain concentrations of Phe were closely associated (r = 0.95, p < 0.0001), as previously reported for these patients as part of a larger 1H-MRS study (15). Mean IQ in adult age was 100.5 (61–120). Neurologic status was normal except for mild tremor, which was present in two patients (2 and 6). Standard MRI in all 11 patients showed supra- and, less frequently, infratentorial nonspace-occupying white matter abnormalities.

31P-MR Spectroscopy

Comparison of PKU with control group under steady-state conditions.

The cerebral 31P MR spectrum of a typical PKU patient is presented in Figure 1, together with the best fitting model spectrum. In addition, the fitted spectrum was decomposed into three subspectra to demonstrate the contributions of the individual metabolites. The 31P resonances can be attributed to nucleotide triphosphates, primarily ATP (with its subcomponents α-, β-, and γ-ATP), PCr, phosphodiesters (PDE), phosphomonoesters (PME), and Pi. The apparent signal from α-ATP has a shoulder that is attributed to NAD. Underlying these sharper resonances is a broad baseline signal that originates mostly from less mobile membrane-bound phospholipids (PL) and possibly to a smaller extent nucleic acids or other macromolecules (24). The steady-state tissue contents of these metabolites; the derived concentrations of ADP and Mg2+; and cerebral pH, PP, and V/ Vmax are listed in Table 2. Figure 2 contains the spectra of all PKU patients and control subjects;Figure 3 depicts the corresponding average spectra.

Figure 1
figure 1

Cerebral 31P-MR spectrum from a 24-year-old male PKU patient with a plasma Phe level of 1784 μM. This figure demonstrates the quality of individual spectra and the data modeling procedure. The raw spectrum is the sum of four separately recorded ISIS spectra. Below, the sum of the four separate model fits is displayed. This fitted spectrum is divided into three subspectra to demonstrate the contributions of the individual metabolites. The last row is the difference between recorded and modeled spectra. The spectra are displayed after 4 Hz Gauss apodization and 0 order phasing.

Full size image
Table 2 Steady-state results from quantitative 31P-MRS in 11 adult PKU patients and a control group (mmol/kgww)
Full size table
Figure 2
figure 2

31P-MR spectra of all PKU patients at steady state in comparison with healthy controls. The mean signal-to-noise ratio with respect to PCr is 44 with relatively little variation between spectra, because neither coil loading nor line widths or voxel size changed much between different investigations. The Cramer-Rao minimum variance bounds, which take prior knowledge into account and, hence, are the relevant error estimates for the fitting results, were slightly below 3% for PCr, as well as ATP for each subject. They are somewhat smaller than the interindividual variations found for the control group (Table 2), which are codetermined by further fluctuations introduced by absolute quantitation procedures, exact placement of voxel, and true biologic variations.

Full size image
Figure 3
figure 3

Averaged 31P-MR spectra of PKU patients in comparison with healthy controls. The averaged raw spectrum of 11 PKU patients at steady state is compared with the average control spectrum. The spectra reflect the changes in concentrations found by quantitative analysis only partly, because they are not corrected for coil loading and different line widths. However, the peak area of Pi, which was significantly reduced in quantitative analysis, is found to be reduced also in visual evaluation. The broad baseline component PL was reduced quantitatively in the PKU patients. This is also notable in the spectra. There are no obvious differences for ATP and PME.

Full size image

Concentrations of ATP, PCr, NAD, PME, and PDE, as well as Mg2+ and pH were indistinguishable for PKU patients and controls. In contrast, ADP was significantly elevated by ∼10% in the PKU patients. This difference corresponds roughly to the measurement variation for ADP expected for single subjects based on the fitting uncertainties in single spectra. Five of the 11 PKU patients (1, 6, 7, 8, and 10) were on or above 2 SD of the ADP control mean. Pi was decreased by ∼20% in PKU. Six PKU patients (1, 2, 3, 4, 7, and 8) had Pi values that were >2 SD below the Pi control mean. The baseline PL component was ∼8% lower in PKU compared with the controls.

ADP concentrations were correlated with plasma phenylalanine (Phe; r = 0.65, p < 0.05). There was a trend for an association of ADP with brain Phe (r = 0.55, p < 0.1), as well as IDC (r = 0.53, p < 0.1). The PP was significantly negatively correlated with IDC (r = −0.62, p < 0.05), whereas correlations with concurrent plasma (r = −0.50) and brain (r = −0.38) Phe concentrations were not significant. V/ Vmax, which is usually ∼50% in normocapnia, was comparable for the control subjects and the PKU patients under steady-state conditions (Table 2). None of the parameters derived from 31P-MRS was associated with the patients' IQ.

Dynamic data during an oral Phe challenge.

After the Phe load, plasma Phe concentrations increased steeply from preload values (1073 ± 315; 799–1465 μM Phe) to a maximum ∼1 h postload and slowly decreased thereafter. Twenty-four-hour postload plasma Phe concentrations (1749 ± 156; 1549–1929 μM Phe) were still substantially elevated in comparison with preload values. Concomitantly, Phe concentrations in brain tissue increased from baseline 180 ± 140 μmol/kgww to 420 ± 70 μmol/kgww 24 h postload.

Under the Phe load, PCr decreased in four of the five patients (3, 4, 5, and 11) and remained unchanged in the subject with lowest preload PCr (Fig. 4A). In addition, ATP decreased in the same four patients (Fig. 4B). Statistically, this yielded a trend for lower postload mean PCr and ATP concentrations (Table 3). Postload PCr values of the five PKU patients were significantly lower in comparison with the control mean (χ2 = 4.3, p < 0.05). ADP concentrations increased during the Phe load (Fig. 4C) in all five PKU patients, and postload ADP values were significantly higher in comparison with preload values and the control mean (χ2 = 4.2, p < 0.05). Postload Pi was not significantly different from preload Pi values, even though it had increased in four of five patients (Fig. 4D). PP, which was increased in comparison with the control group under steady-state conditions, decreased during the Phe load (Fig. 4E). Postload PP values were comparable to the control group (χ2 = 0.02, ns). During Phe loading, V/ Vmax (Fig. 4F) increased significantly by ∼7%. NAD, PME, PDE, and the membrane-bound PL compounds as well as pH and Mg2+ concentrations were not affected by Phe loading.

Figure 4
figure 4

31P-MRS data of PKU patients (▪) and control subjects (♦) plotted against the concurrent plasma level of Phe. In the control subjects, plasma Phe was set to 60 μM. Normal ranges (±2 SD) are indicated in dotted lines. In five PKU patients, an oral load with 100 mg/kg body wt L-Phe was performed. Postload 31P-MRS results are given as triangles (▴). A, PCr showed a significant decrease postload. B, ATP, which was comparable to control values under steady-state conditions, showed a decreasing trend with Phe loading. C, ADP, which was already elevated during steady state, increased further under the Phe challenge. D, Pi was lowered in comparison with control but showed no significant changes during Phe loading. E, PP was elevated at steady state but decreased concomitant to the acute rise in Phe concentration under the Phe load. F, The relative velocity of oxidative metabolism (V/Vmax), which was comparable for PKU and control values but increased with Phe loading.

Full size image
Table 3 31P-MRS and EEG findings before and 24 h after an oral load with L-Phe in 5 PKU patients (mean ± SD)
Full size table

EEG results showed that whereas the individual profiles of the power spectra were reproduced, a shift of the dominant peak of background activity to lower frequency had occurred postload (example from a single subject in Fig. 5). An increase of θ activity was accompanied by a simultaneous decrease of α2 activity. There was a trend for a lower α/θ quotient. Peak frequency and mean power frequency (MPF) decreased (Table 2). These changes were noted in all five patients. Visual evaluations of all EEG recordings revealed no abnormalities. The amount of MPF changes between preload and postload measurements (ΔMPF) was not associated with the amount of Phe increase from preload steady-state measurements to postload values (ΔPhe) in plasma (r = 0.15, ns) or brain tissue (r = −0.45). ΔMPF correlated significantly with the increase of cerebral ADP concentrations (ΔMPF * ΔADP r = −0.93, p < 0.05).

Figure 5
figure 5

Power spectra (lead Oz) of EEG activity at rest derived from two adult PKU patients (left, patient 5;right, patient 11) before (- - -) and 24 h after (———) an oral load with a single dose of 100 mg/kg body wt L-Phe. Whereas the individual profiles of the power spectra were reproduced under both conditions, the peak of the dominant background activity moved to lower frequency.

Full size image

During the Phe load, blood glucose levels were above 4.4 mM, and standard laboratory evaluations (red and white blood cell counts, electrolytes, transaminases) remained in the normal range. After 10 days, Phe levels had always returned to preload values.

DISCUSSION

In this study, energy status of the brain of patients with PKU was investigated in vivo by 31P-MRS. In general, 31P-MRS is an excellent technique to monitor cerebral energy metabolism in vivo. Moreover, it provides information on PL metabolism in the brain. Both aspects are important in understanding impaired brain development and function in PKU. In comparing quantitative 31P-MRS data of adult PKU patients and healthy controls, significant differences were found under steady-state conditions, thereby sggesting an influence of the specific biochemical properties of PKU on cerebral energy status.

The challenge of five PKU patients with high doses of oral L-Phe corroborated the impact of Phe on the balance of high-energy phosphorus metabolites. Changes in the phosphorus compounds were accompanied by slowing of EEG background activity, which has been found to be a specific sign of impaired brain function (2, 12).

Given the relatively small magnitude of the changes and the limited number of investigated subjects, one should keep alternative explanations in mind. In particular, it could be discussed if changes in compartment sizes or osmotic effects could induce the observed alterations. After the oral load, brain tissue concentrations of Phe increased from 180 to 420 μM/kgww, which is not sufficient to cause unspecific effects, such as disturbed balances of intra- and extracellular water or other aspects of cellular homeostasis. The elevation of ADP levels could be interpreted as being caused by decreased total creatine levels in PKU (Eq. 1). However, that no significant concentration changes have been reported for total creatine and N-acetylaspartate (3, 25) and that the changes in Phe content are small render these interpretations implausible. In addition, the acute rise in ADP with Phe loading cannot be easily explained by lowered creatine levels. Thus, it is most likely that the observed changes in cerebral energy status and brain function are caused by specific effects of Phe.

Energy metabolism in PKU.

Comparisons between PKU patients and controls showed that there are no significant group differences for the high-energy phosphate metabolites PCr and ATP. However, Phe loading yielded trends for decreasing ATP and PCr levels with increasing Phe concentrations (Fig. 4A). ADP, which is calculated from PCr, ATP, and pH, was increased (as a result of additive nonsignificant but unidirectional trends in these three variables). An increase of ADP concentration, which is considered to be the major driving force for mitochondrial energy production, suggests that energy production, as well as consumption, could be increased in PKU. However, normal PCr and ATP levels at steady-state confirm that the system is compensated. In a decompensated state, e.g. in mitochondrial disease (26), PCr and PP are typically reduced, whereas Pi is increased.

Pi was found to be lower in PKU than in controls under steady-state conditions. It is presumed that the Pi signal in 31P-MRS represents <50% of the total intracellular levels, because Pi that is bound to the mitochondrial inner membrane may not be mobile enough and thus is MR-invisible (27, 28). This suggests that there is less cerebral Pi or that more of it is in a bound state. Lowered Pi was already observed in an animal model of PKU (29). In this study, brain tissue of hyperphenylalaninemic rats was investigated with 31P-MRS and biochemical and histologic methods. Decreased Pi was one of the significant 31P-MRS findings, whereas PCr was normal. It could be speculated that an enhanced synthesis of PCr might be involved in the mechanisms responsible for low Pi levels in PKU (9).

Additional evidence for involvement of cerebral energy metabolism in pathophysiological processes in PKU comes from other animal and in vitro studies that demonstrated that L-Phe itself, as well as its metabolites, inhibited brain energy metabolism (4, 5, 30, 31). Inhibition of hexokinase [EC 2.7.1.1.] and possibly also pyruvate kinase [EC 2.7.1.40] activity was discussed as a major cause of these effects (32). Recently, new results corroborating our clinical findings have been reported. Inhibition of the mitochondrial chain and increased oxidative stress by elevated Phe concentrations were reported for rat brain (33, 34).

The link between acute impairments of brain function and imbalances of high-energy phosphates, as shown in this 31P-MRS study, adds a new aspect to the hitherto unresolved puzzle of Phe neurotoxicity. Interestingly, the amount of EEG changes as expressed by ΔMPF correlated significantly with the amount of ADP increase in these five patients. However, as the number of subjects involved was low, this result requires confirmation. As a result of normal PAH activity, healthy controls show different kinetics after standardized oral L-Phe loading with steeply declining plasma Phe immediately after reaching peak concentrations. Thus, comparable investigations unfortunately cannot easily be performed in control subjects.

PL components in 31P-MRS.

MRI studies on early-, late-, and untreated PKU patients show a high frequency of white matter abnormalities in neurologically normal patients. The nature of these white matter abnormalities remains unclear but could relate to mobile and membrane-bound PL, which constitute essential parts of cell membranes and myelin and which can partly be observed by 31P-MRS.

The PME multicomponent comprises predominantly phosphoethanolamine and phosphocholine, as well as precursor and breakdown products of membrane PL. It is regarded as representing mainly the anabolic activity of PL (35). PDE consists of glycerol-3-phosphoethanolamine and -choline, as well as mobile PL, phosphorylized glycolipids, and glycoproteins. Thus, this peak represents relatively mobile PDE and at the same time metabolites involved in the catabolic pathway of membrane PL. The 31P signals that are evaluated as broad PL baseline primarily originate from less mobile PL in membranes, but there is a smooth transition between PDE and PL baseline signals. The segregation between the two is method dependent. Although our study revealed no abnormalities in the PME and the narrow PDE compounds, it clearly showed a reduction of the broad baseline signal (Table 2). This is in accordance with observations made by 31P-MRS in hyperphenylalaninemic rats (29). In these animals, brain myelinogenesis was disturbed, because they had lower brain weights, cerebrosides, sulfatides, myelin basic protein, reduced cerebroside sulfotransferase, and 2′3′-cyclic nucleotide-3′-phosphohydrolase activities. Biochemical analysis revealed no changes in total proteins, lipids, PL, phosphatidylethanolamine, and phosphorylethanolamine. The baseline PL were not evaluated.

In our study, we observed neither lower PDE concentrations nor any correlation of PDE or PL with the amount of MRI-visible white matter changes in the PKU group. As the mobile part of the PDE decreases with increasing severity of demyelination (35), normal PDE in this study together with normal choline concentrations observed by means of 1H MRS (3) contradicts demyelination in adults with PKU and high-signal white matter abnormalities.

CONCLUSION

In summary, in vivo 31P-MRS demonstrated changes of cerebral energy metabolism in adult PKU patients. Under steady-state conditions, the balance of energy production and consumption seemed to be undisturbed, and ATP concentrations were normal. Both further impairment of energy status and typical Phe-related changes in EEG background activity were observed with an acute rise of plasma and brain Phe, induced by an oral Phe challenge. Thus, acute Phe neurotoxicity may, at least partly, be attributed to imbalances of the high-energy phosphates. It remains unclear whether the impact of elevated Phe levels on cerebral energy metabolism also contributes to the severe disturbances of brain development in untreated PKU subjects. However, our first clinical results conjoined with the results of earlier and recent in vitro studies might serve to stimulate new attempts to investigate the role of energy metabolism in PKU pathophysiology.