Eur J Pediatr (2000) 159 [Suppl 2]: S121±S125
Ó Springer-Verlag 2000
REVIEW
Harald E. MoÈller á Kurt Ullrich á Josef Weglage
In vivo proton magnetic resonance spectroscopy in phenylketonuria
Abstract In vivo nuclear magnetic resonance spectroscopy permits the non-invasive examination of metabolic characteristics of the human brain in a clinical environment. Methods to detect elevated phenylalanine (Phe) in patients with phenylketonuria (PKU) using dierence spectroscopy and to estimate absolute brain Phe concentrations, [Phe]brain, have been developed. In patients with classical PKU, [Phe]brain typically varied between 0.14 and 0.78 mmol/l depending upon actual blood Phe concentrations, [Phe]blood , between 0.47 and 2.30 mmol/l. Dynamic investigations can be used to extract information about Phe transport at the human blood-brain barrier, which may be described by a symmetric Michaelis-Menten model. Carrier saturation and competitive inhibition of the in¯ux of other large neutral amino acids can be expected at blood levels usually found in PKU patients. In single cases of untreated, normal intelligent patients, abnormally low [Phe]brain £ 0.15 mmol/l were observed despite high stationary Phe levels ([Phe]blood 1.15 0.10 mmol/l). Conclusion Signi®cant variations in phenylalanine transport parameters in untreated, normal intelligent patients indicated that blood-brain barrier transport or intracerebral phenylalanine consumption are causative factors for the individual vulnerability to phenylketonuria. Key words Blood-brain barrier á Nuclear magnetic resonance spectroscopy á Phenylalanine transport á Phenylalanine turnover á Phenylketonuria Abbreviations BBB blood-brain barrier á Cr creatine á NMR nuclear magnetic resonance á Phe phenylalanine á [Phe]blood blood Phe concentration á [Phe]brain intracerebral Phe concentration á PKU phenylketonuria á PRESS point resolved spectroscopy á STEAM stimulated echo acquisition mode á T2 app apparent spin-spin relaxation time á TE echo time á VOI volume-of-interest
H. E. MoÈller (&) Institute for Physical Chemistry, University of MuÈnster, Schloûplatz 4/7, 48149 MuÈnster, Germany e-mail:
[email protected] Tel.: +49-251-8356140; Fax: +49-251-8347312 H. E. MoÈller Department of Clinical Radiology,
University of MuÈnster, MuÈnster, Germany K. Ullrich Department of Paediatrics, University Hospital Eppendorf, Hamburg, Germany J. Weglage Department of Paediatrics, University of MuÈnster, MuÈnster, Germany
S122
Introduction
Localised nuclear magnetic resonance (NMR) spectroscopy is a non-invasive technique, which permits the examination of metabolic tissue characteristics in a clinical environment. Avison et al. [1] were the ®rst to detect a phenylalanine (Phe) NMR signal in the brain of rabbits made hyperphenylalaninaemic showing that Phe does not achieve equal concentrations on both sides of the blood-brain barrier (BBB). The same group reported the detection of an elevated Phe signal in the brain of ®ve patients with phenylketonuria (PKU) o diet [19]. More recently, researchers at the Universities of Bern [11], Yale [20], and MuÈnster [15, 25] demonstrated independently that this technique oers a novel strategy for quantifying intracerebral Phe concentrations ([Phe]brain). The present review focuses on ®ndings using 1H-NMR of the brain in PKU patients. Besides elevated [Phe]brain, the proton spectra indicated normal levels of the routinely observed brain metabolites in the majority of studies [2, 11, 12, 15, 20, 22, 26]. This was corroborated by Pietz et al. [23] in a quantitative examination, demonstrating normal concentrations of N-acetyl-l-aspartate, total creatine (Cr), cholines, and myo-inositol in occipital grey matter and in parietal and frontal white matter. We may therefore exclude consistent biochemical alterations secondary to PKU concerning these metabolites. Nuclear magnetic resonance methods for measuring intracerebral phenylalanine concentrations
Investigation of [Phe]brain can be performed on clinical routine MRI scanners operating at a magnetic ®eld strength of 1.5 T or higher. Using the standard head coil, the spectroscopy scan may be combined conveniently with an MRI examination of potential white matter abnormalities [2]. For detection of low-concentration metabolites, such as Phe, single-voxel techniques, which acquire high-resolution spectra from a well-de®ned volume-of-interest (VOI), were proven to be particularly robust methods. Currently, two sequences known as point resolved spectroscopy (PRESS) [3], and stimulated echo acquisition mode (STEAM) [5], are most widely used. While stimulated echoes are especially suited to realise ultra-short echo times, PRESS oers the advantage of a two-fold gain in the signal-to-noise ratio. In some earlier experiments [2, 6, 8, 12], [Phe]brain could not be observed, and careful optimisation of the data acquisition procedure under consideration of the following characteristics is recommended. (1) At 1.5 T, the a and b protons of Phe give rise to complex multiplets of low intensity in the region between 3 and 4 ppm, which cannot be distinguished from overlying strong signals of abundant metabolites under in vivo conditions [11]. In contrast, all signals of the chemically and magnetically inequivalent phenyl protons collapse into a single peak
at 7.36 ppm of sucient intensity for quantitation. However, also the high-frequency region of the normal brain spectrum is composed of several overlapping peaks between 6.5 and 8.5 ppm, most of them not reliably assigned. Dierence spectroscopy should therefore be utilised for the unequivocal identi®cation of elevated Phe. Most ecient for removing the background signal is the use of spectra of normal healthy controls as baseline [11, 15, 20]. This method leads to absolute concentrations if the results are corrected by the normal level, which is approximately 0.05 mmol/l based upon biopsy data [13]. (2) As [Phe]brain is typically below 1 mmol/l in classical PKU patients under free nutrition, selection of a VOI in the order of ca. 25 ml or larger is required to obtain a sucient signal-to-noise ratio for reliable quantitation. (3) Another characteristic of the Phe peak at 7.36 ppm is a rapid decay of the echo amplitude, described by an apparent relaxation time (T2 app), which includes both eects from pure spin-spin relaxation and complex J-modulation. Estimates of T2 app from previous in vivo studies in a hyper-Phe rabbit at 4.7 T [1] and in a PKU patient at 1.5 T [11] were 40 ms and 65 ms, respectively. Choice of a short echo time (TE) £ 20 ms, is therefore necessary to minimise eects from T2-weighting of the spectrum. A detailed description of a data acquisition procedure using PRESS was published by Kreis et al. [11]. Molar metabolite concentrations were determined with reference to the brain tissue water signal. Alternatively, a STEAM sequence was employed by MoÈller et al. [15] using the Cr methyl resonance at 3.02 ppm as an internal concentration standard. The same quantitation method was also used by Novotny et al. [20], however, combined with the image-selected in vivo spectroscopy localisation technique [21]. Note that [Phe]brain estimates via the Cr peak require an assumption about the intracerebral concentration of total Cr. Such data are available from investigations of normal adult subjects [10, 14]. While STEAM and PRESS achieve localisation in a single-shot fashion, image-selected in vivo spectroscopy uses an add-subtract process, which makes it more susceptible to errors due to instabilities or motion. Model of phenylalanine transport at the blood-brain barrier
The cellular supply of essential amino acids is a function of their plasma concentrations and membrane transport processes. The rate-limiting transport step for Phe uptake lies at the BBB, and we may assume fast equilibration between interstitial and intracellular spaces resulting in a single kinetic pool within the brain as shown in Figure 1. Large neutral amino acids are transported across the BBB by a common saturable carrier. As a consequence of competition eects, only an apparent Michaelis transport constant, Kt,app, is measured under in vivo conditions. Recently, it was demonstrated that the relationship between [Phe]brain and [Phe]blood can be described by a symmetric Michaelis-
S123
Fig. 1 Model for Phe transport from blood into the brain. Tin and Tout are velocities for Phe transport into and out of the brain and Vmet is the intracerebral metabolic rate of Phe. A Two saturable carrier systems, one at the BBB and the other in the plasma membranes, mediate Phe transport from blood into interstitial space and from there into neurons and astroglia cells, respectively. B Assumption of carrier-mediated Phe transport at the BBB being rate-limiting and equilibration between interstitial and intracellular spaces being fast resulting in a single kinetic Phe pool within the brain. (Reprinted from Brain Research [16] with permission)
Menten model, which assumes identical kinetic parameters, Kt,app and Tmax, for Phe transport into and out of the brain and a constant Phe consumption velocity, Vmet, in the brain cells [16]: dPhebrain Tmax Pheblood Tmax Phebrain mmet : dt Kt;app Pheblood Kt;app Phebrain
concentrations with ratios [Phe]brain/[Phe]blood varying between 0.21 and 0.74 under steady-state conditions [15, 16, 20, 22]. Note that dierences in the quanti®cation procedures may contribute to variations in [Phe]brain/ [Phe]blood if data from dierent researchers are compared (MoÈller et al. [16]: [Phe]brain 0.20±0.76 mmol/l at [Phe]blood 0.47±2.24 mmol/l; Novotny et al. [20]: [Phe]brain 0.48±0.78 mmol/l at [Phe]blood 0.65±2.30 mmol/l; Pietz et al. [22]: [Phe]brain 0.13±0.41 mmol/kg wet weight at [Phe]blood 0.70±1.39 mmol/l). McKean [13] reported [Phe]brain 0.85 0.015 mmol/kg wet weight obtained biochemically at autopsy ([Phe]blood not available). Pietz et al. [22] reported a linear correlation between brain and blood Phe for plasma concentrations ranging from 0.70 to 1.39 mmol/l which is consistent with observations by MoÈller et al. [15, 16] for the same concentration regime. Deviations from linearity were reported for higher plasma levels, suggesting saturation kinetics of the transport system as predicted by the Michaelis-Menten model [16]. Qualitatively, this behaviour is similar to the in vivo NMR results in hyperphenylalaninaemic rabbits [1]. Data recorded from nine PKU patients are shown in Figure 2. Fits to the pooled data yielded Kt,app 0.16 0.11 mmol/l and Tmax/ Vmet 9.0 4.1. The apparent Michaelis constant for BBB Phe transport derived from NMR data compares well with previous animal and human results. Estimates obtained with the in situ brain perfusion technique in rats yielded 0.218 0.009 mmol/l [18]. In three healthy volunteers, Knudsen et al. [9] measured Kt,app between 0.03 and 0.58 mmol/l and Tmax between 14.4 and 94.3 nmol/g per min using the double-indicator method. This technique needs intravenous tracer injection with input detection at a peripheral artery and output detection at the internal jugular vein. Due to the poor signal-to-noise ratio of the Phe signal, the estimates of [Phe]brain derived from NMR spectroscopy are associated with relatively large errors, which contribute to the scatter in the data. Little is
1 Tmax is the maximal transport velocity. A similar model has been used by others in computer simulations of amino acid uptake and metabolic pathways in the brain [7]. Under steady-state conditions, d[Phe]brain/dt equals zero, and equation 1 can be rewritten as: f
Tmax =mmet 1gPheblood Kt;app steady-state Phebrain Kt;app : f
Tmax =mmet 1gKt;app Pheblood
2 A discussion of limitations inherent in equations 1 and 2 can be found in MoÈller et al. [16, 17]. Investigation of intracerebral phenylalanine under steady-state conditions
Intracerebral Phe levels in 44 examinations of 21 patients with classical PKU clearly remained below blood
Fig. 2 Plot of parieto-occipital brain Phe concentrations (31 examinations in nine patients) versus corresponding blood levels and result from non-linear least-squares ®tting (r 0.81) to equation 2. (Reprinted from Brain Research [16] with permission)
S124
known so far about the range of biological variations in [Phe]brain and their relation to PKU. Such variations would also contribute to the standard deviations of the kinetic parameters extracted from pooled data. Nuclear medicine ®ndings [9] indicate that substantial interindividual dierences in Phe transport characteristics cannot be excluded. Further studies are, however, needed to answer this question reliably. Individual vulnerability in phenylketonuria and correlation with intracerebral phenylalanine
An individual vulnerability of patients to elevated blood Phe is well known. Single cases of untreated adult patients with classical PKU and normal intelligence have been reported [24]. One explanation could be interindividual variations in [Phe]brain [26]. To investigate the relation between brain Phe and individual vulnerability in PKU, 11 patients with genotypes and blood Phe levels typical for classical PKU were grouped in a recent study according to their diet status and clinical outcome, and the blood-brain correlation of Phe concentrations was investigated [17]. Although patients of group 1 (n 3) had never received any dietary treatment, they were almost unaected clinically and reached normal or nearly normal intelligence scores. Patients of group 2a (n 4) were untreated in early infancy and retarded. Group 2b consisted of early treated adults (n 4) who had stopped diet 7 to 20 years ago. Intelligence quotients were within or close to the normal range in this group. Changes upon T2-weighted MRI were mildest in the untreated, ``atypical'' group 1. Patients of groups 2a and 2b had abnormalities of cerebral white matter to a variable extent depending upon their current diet status. In steady-state NMR experiments, [Phe]brain was investigated in all patients while having stationary [Phe]blood around 1.2 mmol/l [17]. No dierences were observed between groups 2a ([Phe]blood 1.28 0.06 mmol/l, [Phe]brain 0.60 0.04 mmol/l) and 2b ([Phe]blood 1.27 0.09 mmol/l, [Phe]brain 0.59 0.15 mmol/l) with typical outcomes depending on the individual diet status. In contrast, brain Phe was hardly detectable in the ``atypical'' patients ([Phe]blood 1.15 0.10 mmol/l, [Phe]brain £ 0.15 mmol/l). This strong correlation (P < 0.001) with the clinical parameters underlines the importance of brain Phe levels for the outcome in PKU. To test the hypothesis that the depression in [Phe]brain seen in group 1 was related to BBB transport characteristics, dynamic investigations of [Phe]blood and [Phe]brain were performed in three patients of group 1 and two patients of group 2b after oral loading tests with L-phenylalanine (100 mg/kg body weight) [17]. Similar to the results of Pietz et al. [22], [Phe]brain reached a maximum 16±23 h post-load in group 2b. In group 1, both blood and brain Phe relaxed quicker toward the baseline level after the load. Kinetic parameters obtained with the Michaelis-Menten model were
Kt,app 0.81 0.33 mmol/l and Tmax/Vmet 2.87 0.33 for group 1, and Kt,app 0.10 0.04 mmol/l and Tmax/Vmet 11.3 4.4 for group 2b. The kinetic parameters recorded from group 2b are consistent with the above-mentioned results from steady-state NMR experiments [16] and from the double-indicator measurements [9]. Compared with these data, Kt,app appeared to be signi®cantly larger (P < 0.04) in the ``atypical'' PKU patients. High BBB transport Michaelis constants make the brain uptake of amino acids in these individuals less sensitive to the eects of competition [4], which correlates well with their almost normal clinical outcome. The kinetic data provide further evidence that the excess of the in¯ux rate over the metabolic rate is substantially reduced (P < 0.03) in group 1. This could indicate increased brain Phe consumption rates in the ``atypical'' patients. Alternatively, an asymmetric transport at the BBB might also explain the observed dierences. Note that 1H-NMR cannot distinguish between individual processes contributing to the drain of the intracerebral Phe level. Conclusions
In vivo 1H-NMR spectroscopy can be used to quantitate intracerebral Phe concentrations non-invasively in PKU patients. This provides a basis to study competition effects in the transport of large neutral amino acids across the human BBB. Preliminary ®ndings suggest that interindividual variations in the kinetics of Phe uptake and metabolism do exist, leading to dierent brain concentrations of the neurotoxin Phe at comparable blood levels. Such variations seem to be causative factors in the pathogenesis of PKU. Therefore, dynamic NMR investigations may have a potential for optimising individual treatment strategies. For further veri®cation, a broader data base is, however, needed to assess the range of biological variations and to investigate the extent by which BBB amino acid transport contributes to the vulnerability of the brain in PKU. Acknowledgements The authors are grateful to Dr. Ulrich Bick, Dr. Peter Vermathen, and Dr. Dirk Wiedermann for their creativity and contributions over the years in spectroscopic investigations of phenylketonuria. Appreciation is extended to the Round Table, Germany and the Deutsche Interessengemeinschaft Phenylketonurie (PKU) e.V. for generous ®nancial support.
References 1. Avison MJ, Herschkowitz N, Novotny EJ, Petro OAC, Rothman DL, Colombo JP, Bachmann C, Shulman RG, Prichard JW (1990) Proton NMR observation of phenylalanine and an aromatic metabolite in the rabbit brain in vivo. Pediatr Res 27: 566±570 2. Bick U, Ullrich K, StoÈber U, MoÈller H, Schuierer G, Ludolph AC, Oberwittler C, Weglage J, Wendel U (1993) White matter abnormalities in patients with treated hyperphenylalaninemia: magnetic resonance relaxometry and proton spectroscopy ®ndings. Eur J Pediatr 152: 1012±1020
S125 3. Bottomley PA (1984) PRESS sequence. U.S. patent 4 480 228 4. Choi TB, Pardridge WM (1986) Phenylalanine transport at the human blood-brain barrier. J Biol Chem 261: 6536±6541 5. Frahm J, Merboldt KD, HaÈnicke W (1987) Localized proton spectroscopy using stimulated echoes. J Magn Reson 72: 502±508 6. HaÂjek M, Hejcmanova L, PraÂdny J (1993) Proton in vivo spectroscopy of patients with hyperphenylalaninemia. Neuropediatrics 24: 111±112 7. Hommes FA, Lee JS (1990) The control of 5-hydroxytryptamine and dopamine synthesis in the brain: a theoretical approach. J Inherit Metab Dis 13: 37±57 8. Johannik K, Van Hecke P, FrancËois B, Marchal G, Smet MH, Jaeken J, Breysem L, Wilms G, Baert AL (1994) Localized brain proton NMR spectroscopy in young adult phenylketonuria patients. Magn Reson Med 31: 53±57 9. Knudsen GM, Hasselbalch S, Toft PB, Christensen E, Paulson OB, Lou H (1995) Blood-brain barrier transport of amino acids in healthy controls and in patients with phenylketonuria. J Inherit Metab Dis 18: 653±664 10. Kreis R, Ernst T, Ross B (1993) Absolute quantitation of water and metabolites in the human brain. II. Metabolite concentrations. J Magn Reson B 102: 9±19 11. Kreis R, Pietz J, Penzien J, Herschkowitz N, Boesch C (1995) Identi®cation and quanti®cation of phenylalanine in the brain of patients with phenylketonuria by means of localized in vivo 1 H magnetic-resonance spectroscopy. J Magn Reson B 107: 242±251 12. Lou HC, Toft PB, Andresen J, Mikkelsen I, Olsen B, GuÈttler F, Wieslander S, Hendriksen O (1992) An occipito-temporal syndrome in adolescents with optimally controlled hyperphenylalaninemia. J Inherit Metab Dis 15: 687±695 13. McKean CM (1972) The eects of high phenylalanine concentrations on serotonin and catecholamine metabolism in the human brain. Brain Res 47: 469±476 14. Michaelis T, Merboldt KD, Bruhn H, HaÈnicke W, Frahm J (1993) Absolute concentrations of metabolites in the adult human brain in vivo: quanti®cation of localized proton MR spectra. Radiology 187: 219±227 15. MoÈller HE, Vermathen P, Ullrich K, Weglage J, Koch HG, Peters PE (1995) In-vivo NMR spectroscopy in patients with phenylketonuria: changes of cerebral phenylalanine levels under dietary treatment. Neuropediatrics 26: 199±202 16. MoÈller HE, Weglage J, Wiedermann D, Vermathen P, Bick U, Ullrich K (1997) Kinetics of phenylalanine transport at the
17.
18.
19.
20.
21. 22.
23.
24. 25.
26.
human blood-brain barrier investigated in vivo. Brain Res 778: 329±337 MoÈller HE, Weglage J, Wiedermann D, Ullrich K (1998) Blood-brain barrier phenylalanine transport and individual vulnerability in phenylketonuria. J Cereb Blood Flow Metab 18: 1184±1191 Momma S, Aoyagi M, Rapoport SI, Smith QR (1987) Phenylalanine transport across the blood-brain barrier as studied with the in situ brain perfusion technique. J Neurochem 48: 1291±1300 Novotny EJ, Avison MJ, Rothman DL, Seashore MR, Petro OAC, Herschkowitz N, Prichard JW, Shulman RG (1989) Detection of phenylalanine in the human brain. Society of Magnetic Resonance in Medicine, 8th Annual Meeting, Amsterdam, P 441 Novotny EJ, Avison MJ, Herschkowitz N, Petro OAC, Prichard JW, Seashore MA, Rothman DL (1995) In vivo measurement of phenylalanine in human brain by proton nuclear magnetic resonance spectroscopy. Pediatr Res 37: 244± 249 Ordidge RJ, Connelly A, Lohman JAB (1986) Image-selected in vivo spectroscopy (ISIS). A new technique for spatially resolved NMR spectroscopy. J Magn Reson 66: 283±294 Pietz J, Kreis R, Boesch C, Penzien J, Rating D, Herschkowitz N (1995) The dynamics of brain concentrations of phenylalanine and its clinical signi®cance in patients with phenylketonuria determined by in vivo 1H magnetic resonance spectroscopy. Pediatr Res 38: 657±663 Pietz J, Kreis R, Schmidt H, Meyding-Lamade UK, Rupp A, Boesch C (1996) Phenylketonuria: ®ndings at MR imaging and localized in vivo H-1 MR spectroscopy of the brain in patients with early treatment. Radiology 201: 413±420 Primrose DA (1983) Phenylketonuria with normal intelligence. J Ment De®c Res 27: 239±246 Ullrich K, MoÈller H, Weglage J, Schuierer G, Bick U, Ludolph A, Hahn-Ullrich H, FuÈnders B, Koch HG (1994) White matter abnormalities in phenylketonuria: results of magnetic resonance measurements. Acta Paediatr 407[Suppl]: 78±82 Weglage J, MoÈller HE, Wiedermann D, Cipcic-Schmidt S, Zschocke J, Ullrich K (1998) In vivo NMR spectroscopy in patients with phenylketonuria: clinical signi®cance of interindividual dierences in brain phenylalanine concentrations. J Inherit Metab Dis 21: 81±82