OUP user menu

Tyrosine hydroxylase deficiency: a treatable disorder of brain catecholamine biosynthesis

Michèl A. Willemsen, Marcel M. Verbeek, Erik-Jan Kamsteeg, Johanneke F. de Rijk-van Andel, Alec Aeby, Nenad Blau, Alberto Burlina, Maria A. Donati, Ben Geurtz, Padraic J. Grattan-Smith, Martin Haeussler, Georg F. Hoffmann, Hans Jung, Johannis B. de Klerk, Marjo S. van der Knaap, Fernando Kok, Vincenzo Leuzzi, Pascale de Lonlay, Andre Megarbane, Hugh Monaghan, Willy O. Renier, Pierre Rondot, Monique M. Ryan, Jürgen Seeger, Jan A. Smeitink, Gerry C. Steenbergen-Spanjers, Evangeline Wassmer, Bernhard Weschke, Frits A. Wijburg, Bridget Wilcken, Dimitrios I. Zafeiriou, Ron A. Wevers
DOI: http://dx.doi.org/10.1093/brain/awq087 1810-1822 First published online: 29 April 2010


Tyrosine hydroxylase deficiency is an autosomal recessive disorder resulting from cerebral catecholamine deficiency. Tyrosine hydroxylase deficiency has been reported in fewer than 40 patients worldwide. To recapitulate all available evidence on clinical phenotypes and rational diagnostic and therapeutic approaches for this devastating, but treatable, neurometabolic disorder, we studied 36 patients with tyrosine hydroxylase deficiency and reviewed the literature. Based on the presenting neurological features, tyrosine hydroxylase deficiency can be divided in two phenotypes: an infantile onset, progressive, hypokinetic-rigid syndrome with dystonia (type A), and a complex encephalopathy with neonatal onset (type B). Decreased cerebrospinal fluid concentrations of homovanillic acid and 3-methoxy-4-hydroxyphenylethylene glycol, with normal 5-hydroxyindoleacetic acid cerebrospinal fluid concentrations, are the biochemical hallmark of tyrosine hydroxylase deficiency. The homovanillic acid concentrations and homovanillic acid/5-hydroxyindoleacetic acid ratio in cerebrospinal fluid correlate with the severity of the phenotype. Tyrosine hydroxylase deficiency is almost exclusively caused by missense mutations in the TH gene and its promoter region, suggesting that mutations with more deleterious effects on the protein are incompatible with life. Genotype–phenotype correlations do not exist for the common c.698G>A and c.707T>C mutations. Carriership of at least one promotor mutation, however, apparently predicts type A tyrosine hydroxylase deficiency. Most patients with tyrosine hydroxylase deficiency can be successfully treated with l-dopa.

  • tyrosine hydroxylase
  • neurotransmitters
  • cerebrospinal fluid
  • dystonia
  • l-dopa


The enzyme tyrosine hydroxylase (EC catalyzes the conversion of l-tyrosine to l-dihydroxyphenylalanine (l-dopa), which is the rate-limiting step in the biosynthesis of the catecholamines dopamine, norepinephrine and epinephrine (Fig. 1). Catecholamines are produced in the brain and adrenal medulla, but also in non-neuronal, e.g. renal, intestinal and lymphoid tissues. Their vital functions as neurotransmitters and hormones, and the crucial role of tyrosine hydroxylase in their biosynthesis are demonstrated by the observation that complete loss of tyrosine hydroxylase activity is lethal in knock-out mice (Zhou et al., 1995).

Figure 1

Simplified scheme of the biosynthesis and catabolism of serotonin and the catecholamines dopamine, norepinephrine and epinephrine. TPH = tryptophan hydroxylase; AADC = aromatic amino acid decarboxylase; PAH = phenylalanine hydroxylase; TH = tyrosine hydroxylase; BH4 = tetrahydrobiopterin; DOPAC = 3,4-dihydroxyphenylacetic acid.

Human tyrosine hydroxylase deficiency [THD; Online Mendelian Inheritance in Man (OMIM) number 191290] is an autosomal recessive neurometabolic disorder due to mutations in the tyrosine hydroxylase (TH) gene on chromosome 11p15.5. The first reports of THD described patients with an early onset, progressive l-dopa-responsive dystonia (Castaigne et al., 1971; Rondot and Ziegler, 1983; Rondot et al., 1992). Later, neonates were recognized with a more severe phenotype described as progressive, l-dopa-non-responsive encephalopathy (Hoffmann et al., 2003). THD can be diagnosed by demonstrating decreased CSF levels of the down-stream metabolites of the catecholamine degradation pathway (Fig. 1), i.e. homovanillic acid (HVA) and 3-methoxy-4-hydroxyphenylethylene glycol (MHPG) and by mutation analysis of the TH gene.

THD has been reported in fewer than 40 patients worldwide, reviewed in this article. This disorder is known under different names in the literature, namely ‘Segawa syndrome’, ‘infantile parkinsonism’ and ‘l-dopa-responsive dystonia’. ‘Segawa syndrome’, however, is also used to indicate another defect in neurotransmitter biosynthesis, caused by GTP cyclohydrolase I mutations. ‘Infantile parkinsonism’ and l-dopa-responsiveness are not found in all patients with THD. Furthermore, the l-dopa-responsive dystonias encompass a heterogeneous class of movement disorders, THD being only one of them (Muller et al., 1998; Albanese et al., 2006; Tarsy and Simon, 2006; Muller, 2009). Finally, the phenotype of THD can be so complex that it is not simply associated with an extrapyramidal movement disorder (Hoffmann et al., 2003). Altogether, we prefer to name the disorder after its underlying enzymatic defect, as is common practice for inborn errors of metabolism.

In this article, we summarize the medical literature on human THD, and include detailed novel clinical, biochemical and genetic data on thus far unpublished patients. This large review on THD recapitulates all available evidence on clinical phenotypes and rational diagnostic and therapeutic approaches of this severe, but potentially treatable disorder.


Our laboratory has a longstanding tradition of neurotransmitter analysis, and we have been performing TH gene mutation analysis since the genetic basis of THD was elucidated. This history supplies us with a unique database including patients with biochemically and genetically proven THD from many different countries.

The laboratory methodologies used for CSF neurotransmitter analysis have previously been reported in detail (Brautigam et al., 1998; Verbeek et al., 2008) Reference values for HVA and 5-hydroxyindoleacetic acid (5HIAA; end-product of serotonin degradation) in CSF decrease with age, and there is a rostrocaudal gradient for the concentrations of both metabolites, necessitating analysis of a standardized CSF fraction (Brautigam et al., 1998) Mutation analysis of the TH gene was performed as previously described (van den Heuvel et al., 1998) Numbering of coding sequence mutations was according to GenBank reference sequence NM_199292.1 (tyrosine hydroxylase isoform A) in which the A of the ATG transcription initiation codon is designated position 1. Mutations were named according to the guidelines of the Human Genome Variation Society (www.hgvs.org).

A questionnaire was sent to all physicians who have referred (samples from) THD patients to our centre. In this way we collected detailed information on demographic data, pregnancy and perinatal period, presenting clinical features, mode of treatment, follow-up during treatment and results of cerebral imaging studies. The results of CSF and mutation analysis were available in our database. In those few patients in whom CSF analysis was performed elsewhere, the results were obtained together with the appropriate reference values.

A Pubmed search was performed for reports in English, using the terms: ‘tyrosine hydroxylase recessive’ and ‘tyrosine hydroxylase dystonia’. The reference lists of all relevant papers were checked for other citations, especially those reports from the era before THD was recognized as a separate disease entity.

This study was approved by the ethics committee of the Radboud University Nijmegen Medical Centre, The Netherlands. The requirement for additional local ethical approval differed between participating countries and was obtained if required.


We had the names of 36 patients with THD in our database. Questionnaires were completed by the referring physicians of all patients.


Besides reports concerning patients who had been diagnosed in our laboratory (Table 1) (Castaigne et al., 1971; Rondot and Ziegler, 1983; Rondot et al., 1992; van den Heuvel et al., 1998; Brautigam et al., 1999; Wevers et al., 1999; de Lonlay et al., 2000; de Rijk-Van Andel et al., 2000; Dionisi-Vici et al., 2000; Janssen et al., 2000; Swaans et al., 2000; Haussler et al., 2001; Grattan-Smith et al., 2002; Hoffmann et al., 2003; Schiller et al., 2004; Verbeek et al., 2007; Zafeiriou et al., 2009), we only found 14 other THD patients from 12 families in whom the diagnosis was genetically proven (Ludecke et al., 1995, 1996; Knappskog et al., 1995; Surtees and Clayton, 1998; Furukawa et al., 2001; Diepold et al., 2005; Moller et al., 2005; Yeung et al., 2006; Giovanniello et al., 2007; Ribases et al., 2007; Wu et al., 2008; Clot et al., 2009; Doummar et al., 2009). The key data on these patients, and type A/B classification (see next paragraph) based on available clinical descriptions, are summarized in Table 2.

View this table:
Table 1

Clinical characteristics, demographic data and results of mutation analysis in 36 patients with THD

Patient (Family)Year of birthOriginPhenotypeAge at onsetl-dopa- responseNoteAllele 1Allele 2Referencesa
1 (1)1962FrenchA5 yearsGoodc.1010G>A/p.Arg337Hisc.1481C>T/p.Thr494MetCastaigne et al., 1971 (Case 2); Rondot and Ziegler, 1983 (Case 1); Rondot et al., 1992 (Case 2); Swaans et al., 2000 (Case 2)
2 (2)1950SwissA3 yearsGoodc.1127C>T/p.Ala376Valc.1493A>G/p.Asp498GlySchiller et al., 2004 (Case 1)
3 (2)1961SwissA3 yearsGoodc.1127C>T/p.Ala376Valc.1493A>G/p.Asp498GlySchiller et al., 2004 (Case 2)
4 (1)1965FrenchA2 yearsGoodc.1010G>A/p.Arg337Hisc.1481C>T/p.Thr494MetCastaigne et al., 1971 (Case 1); Rondot and Ziegler, 1983 (Case 2); Rondot et al., 1992 (Case 1); Swaans et al., 2000 (Case 1)
5 (3)1965BelgianA1.5 yearGoodc.826A>C/p.Thr276Proc.941C>T/p.Thr314MetSwaans et al., 2000 (Case 3)
6 (4)1989TurkishA> 1 yearGoodc.−70G>Ac.−70G>AVerbeek et al., 2007
7 (5)2002LebaneseA<12 monthsGoodc.−70G>Ac.−70G>AVerbeek et al., 2007
8 (6)2004GermanA9 monthsGoodc.680A>G/p.Asp227Glyc.698G>A/p.Arg233His
9 (7)1990LebaneseA8 monthsModeratec.698G>A/p.Arg233Hisc.698G>A/p.Arg233His
10 (8)1993ItalianA8 monthsGoodc.−69T>Ac.−69T>AVerbeek et al., 2007
11 (9)2004PakistanA8 monthsGoodSudden onsetc.1181T>C/p.Ile394Thrc.1181T>C/p.Ile394Thr
12 (4)1999TurkishA6 monthsGoodc.−70G>Ac.−70G>AVerbeek et al., 2007
13 (10)1997IrishA6 monthsGoodc.620G>A/p.Cys207Tyrc.698G>A/p.Arg233His
14 (11)1992DutchA6 monthsGoodc.295delC/p.Leu99fsbc.698G>A/p.Arg233HisBrautigam et al., 1998; Wevers et al., 1999; de Rijk-Van Andel et al., 2000
15 (12)1989DutchA6 monthsGoodc.−71C>Tc.1159C>A/p.Leu387MetVerbeek et al., 2007
16 (13)1993DutchA5 monthsGoodc.698G>A/p.Arg233Hisc.698G>A/p.Arg233HisBrautigam et al., 1998; van den Heuvel et al., 1998; Wevers et al., 1999; de Rijk-Van Andel et al., 2000
17 (7)1987LebaneseA4 monthsNonec.698G>A/p.Arg233Hisc.698G>A/p.Arg233His
18 (14)2003FijianA4 monthsGoodc.−70G>Ac.1475C>T/p.Pro492LeuVerbeek et al., 2007
19 (5)2004LebaneseA4 monthsGoodc.−70G>Ac.−70G>AVerbeek et al., 2007
20 (15)2000GreekA3 monthsUnknownLost from follow-upc.707T>C/p.Leu236Proc.707T>C/p.Leu236Pro
21 (16)2003BrazilianA3 monthsGoodc.698G>A/p.Arg233Hisc.721G>A/p.Ala241Thr
22 (17)1992DutchA3 monthsGoodc.698G>A/p.Arg233Hisc.698G>A/p.Arg233HisBrautigam et al., 1998; van den Heuvel et al., 1998; Wevers et al., 1999; de Rijk-Van Andel et al., 2000
23 (18)1993DutchA3 monthsGoodc.698G>A/p.Arg233Hisc.698G>A/p.Arg233HisBrautigam et al., 1998; van den Heuvel et al., 1998; Wevers et al., 1999; de Rijk-Van Andel et al., 2000
24 (19)1998LebaneseA2 monthsGoodc.698G>A/p.Arg233Hisc.698G>A/p.Arg233HisGrattan-Smith et al., 2002
25 (20)2005ItalianA2 monthsModeratec.776A>G/p.Glu259Glyc.1529T>A/p.Leu510Gln
26 (21)1997DutchB3 monthsGoodSudden onset at 3 months and sudden deterioration at 22 months during infectionc.698G>A/p.Arg233Hisc.698G>A/p.Arg233His
27 (16)1990BrazilianB3 monthsModeratec.698G>A/p.Arg233Hisc.721G>A/p.Ala241Thr
28 (22)2001DutchB2 monthsGoodc.698G>A/p.Arg233Hisc.698G>A/p.Arg233His
29 (23)1984GermanBNeon (5 months)NoneDied at 9 years of agec.1198–24T>A / p.?c.698G>A/p.Arg233HisHoffmann et al., 2003 (Case II)
30 (24)2000GreekBNeon (4 months)Moderatec.1375C>T/p.Gln459Xc.1475C>T/p.Pro492Leu
31 (25)1995ItalianBNeon (4 months)Moderatec.1076G>T/p.Cys359Phec.1076G>T/p.Cys359PheBrautigam et al., 1999; Hoffmann et al., 2003 (Case I); Dionisi-Vici et al., 2000
32 (23)1990GermanBNeon (4 months)Moderatec.1198-24T>A/p.?c.698G>A/p.Arg233HisJanssen et al., 2000; Haussler et al., 2001; Hoffmann et al., 2003 (Case III)
33 (26)1994FrenchBNeon (3 months)NoneDied at 2.5 years of agec.707T>C/p.Leu236Proc.707T>C/p.Leu236ProHoffmann et al., 2003 (Case IV)
34 (27)2001GreekBNeon (2 months)Nonec.707T>C/p.Leu236Proc.707T>C/p.Leu236ProZafeiriou et al., 2009
35 (28)2004BelgianBNeonModeratec.698G>A/p.Arg233Hisc.698G>A/p.Arg233His
36 (29)1994TurkishBNeonNonec.926T>C/p.Phe309Serc.926T>C/p.Phe309SerDe Lonlay et al., 2000
  • Phenotype A = ‘progressive extrapyramidal movement disorder (hypokinetic-rigid syndrome with dystonia) with onset in infancy or childhood’; B = ‘complex encephalopathy with onset in the neonatal period or early infancy’; Neon = neonatal. Response to l-dopa: ‘none’ means that there was no beneficial response at all, often reflecting the occurrence of such severe dyskinesia that treatment was impossible.

  • a References: papers in which (clinical, biochemical or genetic) data of the patient have been published previously.

  • b The c.295delC mutation was previously designated c.291delC, but renamed according to the guidelines of the Human Genome Variation Society (www.hgvs.org).

View this table:
Table 2

Patients reported in the literature (not diagnosed in our laboratory)

Family, PatientPheno- typeAge at onsetCognitive impairmentl-dopa- responsePresenting clinical featuresAllele 1Allele 2References
IA5 yearsN.r.GoodSpastic paraplegiac.296delTc.1493A>GFurukawa et al., 2001
IIA3 yearsN.r.GoodDelayed walking and rigidity, wheelchair bound at 10 yearsc.698G>Ac.739G>AWu et al., 2008
IIIAChildhoodMildGoodProgressive dystoniac.956G>Cc.1240G>AClot et al., 2009
IVA2 yearsMildGoodSpastic paraplegia from age 1 year, deterioration at 11 yearsc.1240G>Ac.1529T>AGiovanniello et al., 2007
VA14 monthsMildGoodNormal until 14 months, then deteriorationc.736C>Tc.1493A>GDiepold et al., 2005
VI-1AN.r.N.r.N.r.Progressive l-dopa-responsive dystoniac.1141C>Ac.1141C>ALudecke et al., 1995; Knappskog et al., 1995
VI-2AN.r.N.r.N.r.Progressive l-dopa-responsive dystoniac.1141C>Ac.1141C>ALudecke et al., 1995; Knappskog et al., 1995
VIIA6 monthsN.r.Moderatel-dopa-responsive encephalopathyc.−71C>Tc.−71C>TRibases et al., 2007
VIII-1A4 monthsYesModerateMental retardation, ‘mixed type cerebral palsy’, galactorrhea(p.Arg169X)bc.698G>AYeung et al., 2006
VIII-2AN.r.YesModerateAs sibling but without galactorrhea(p.Arg169X)bc.698G>AYeung et al., 2006
IXB9 monthsNoGoodHypokinesia, hypotonia, dystonia, ptosis; adopted childc.1125C>Gc.1399A>GClot et al., 2009; Doummar et al., 2009
XB5 monthsYesGoodComplex movement disorder, abnormal eye movements, diurmal fluctuationc.901C>Gc.901C>GClot et al., 2009
XIB<5 monthsYesModerateComplex movement disorder, ptosis, irritabilityc.982C>Tc.1196C>TMoller et al., 2005
XIIB3 monthsNoGoodComplex movement disorder, ptosisc.614T>Cc.614T>CLudecke et al., 1996; Surtees and Clayton, 1998
  • a References: papers in which (clinical, biochemical or genetic) data of the patient have been published previously.

  • b The cDNA mutation was not reported by the authors, only the protein change was given. N.r. = not reported.

Clinical features

After careful evaluation of the detailed case histories in the literature and the questionnaires used in this study, it was possible to class the different phenotypes at presentation into two major groups. Most patients (n = 25) suffered from a disorder that can be summarized as a progressive hypokinetic-rigid syndrome with dystonia. The onset of symptoms was generally in the first year of life (age range: 2 months and 5 years). This is the first phenotype described in the literature, further referred to as type A in this article (Castaigne et al., 1971; Rondot and Ziegler, 1983; Rondot et al., 1992; Knappskog et al., 1995; Ludecke et al., 1995; Swaans et al., 2000; Furukawa et al., 2001; Diepold et al., 2005; Yeung et al., 2006; Giovanniello et al., 2007; Ribases et al., 2007; Wu et al., 2008; Clot et al., 2009). The other eleven patients suffered from a more ‘complex encephalopathy’ with earlier onset (age range: 0–3 months), as described by Hoffmann et al. (2003), designated type B. These two phenotypes are defined in detail in Boxes 1 and 2. Although, we found no difficulties in designating individual patients into type A or B THD, it was obvious that the phenotype of THD is a spectrum with overlap of clinical features between both groups. Patient 24, previously described by Grattan-Smith et al. (2002), is a good example of a type A patient with a phenotype very close to type B.

Box 1. THD type A: ‘Progressive extrapyramidal movement disorder (hypokinetic-rigid syndrome with dystonia) with onset in infancy or childhood’

These patients are born after uncomplicated pregnancies and develop normally during the first months of life. In rather exceptional cases (including the first patients described with THD), psychomotor development is even normal or only slightly delayed during the first 2 to 5 years of life (Castaigne et al., 1971; Rondot and Ziegler, 1983; Rondot et al., 1992). Thereafter, however, progressive motor signs appear. Affected individuals become hypokinetic and rigid, and dystonia develops. In early stages, generally only one leg is involved, but with time both legs and also the arms, trunk, face and oropharyngeal musculature become affected. Initial complaints thus encompass abnormal posturing and walking difficulties, or frequent falls in those who already learned to walk before onset of symptoms. These children become wheelchair bound within some years.

Most patients with this type A THD are younger than those described above. In these infants, hypokinesia, bradykinesia and rigidity may dominate the neurological picture while dystonia tends to be less prominent. Initial motor symptoms are generally symmetric and involve arms as well as legs. The ability to walk is not achieved unless children are treated. Severity of dystonia may fluctuate during the day (generally worse in the afternoon), but can also fluctuate within days, giving the impression of a paroxysmal dystonia especially in the early stages of the disease. Mild, non-progressive mental retardation can be found in patients with relatively early onset of motor symptoms, while cognitive functions appear unaffected in patients who develop symptoms after the first year of life. Besides the hypokinetic-rigid syndrome with dystonia, other features like tremor, chorea, oculogyric crises and ptosis, as well as behavioural or autonomic disturbances are absent or—if present—are found as a mild feature and in a minority of patients.

In almost all patients with type A THD, treatment with l-dopa results in an excellent response, sometimes even a miraculous improvement of the neurological condition. During follow-up, all patients continue to be asymptomatic or display only mild motor or cognitive impairment while on a low dose of l-dopa. They show no evidence of progressive disease, and tolerate l-dopa well during many years. Extensive, very readable clinical descriptions of type A THD were, for example, provided by Castaigne et al. (1971), Rondot and Ziegler, (1983), Rondot et al. (1992), de Rijk-Van Andel et al. (2000) and Schiller et al. (2004).

Box 2. THD type B: ‘Complex encephalopathy with onset in the neonatal period or early infancy’

Immediately after birth, or after a symptom-free interval of only weeks, these patients rapidly develop a complex disorder. In most patients, the presenting signs are initially contributed to their complicated perinatal history, which makes estimation of age of onset difficult (Table 1; Patients 29–34). The initial signs may differ between infants, but they all develop a varied neurological disorder that generally includes marked hypokinesia, bradykinesia and hypotonia, mixed with focal or generalized dystonic features and (often excessive) jerky movements like tremor and myoclonus, and that can also encompass bilateral ptosis and oculogyric crises. Diurnal fluctuation of symptoms may be present to a minor degree but is generally absent. However, especially in the most severely affected infants dystonic crises occur within regular intervals of 4–5 days. Mental retardation is generally present, but—as far as can be judged in these severely handicapped children—cognitive functions seem stable over time. Autonomic functions are often disturbed, especially during periods of dystonia or so-called ‘lethargy-irritability crises’, leading to excessive drooling, sweating, body temperature instability and marked periods of ‘pyrexia of unknown origin’. True epileptic seizures and non-epileptic paroxysms may further complicate the clinical picture. l-dopa treatment does not improve all signs equally, and it may take months before all effects of treatment become clear. Hypersensitivity to l-dopa is an important management problem in many of these patients, necessitating (extremely) low l-dopa doses at start, divided over four to six doses per day, and only increased over periods of weeks or months. Compared to type A THD, prognosis with regard to final outcome is worse for motor as well as cognitive functions. Very readable case histories and videotapes have been provided on patients with this type B phenotype, for example by Surtees and Clayton (1998), de Lonlay et al. (2000), Hoffmann et al. (2003) and Zafeiriou et al. (2009).

Table 1 summarizes the key demographic data, and clinical characteristics, as well as the results of mutation analysis. The 36 patients came from 29 families with their roots in 13 different countries. THD is a movement disorder with very early onset: all type B patients had an age at onset within the first months of life, and 19 out of 25 (76%) of type A patients presented in the first year of life. The majority of patients (69%) suffered from type A THD. While responsiveness to l-dopa was absent (36%), moderate (45%) or good (18%) in the type B patients, it was good in 84% of the type A patients. Table 3 provides further details on the prevalence of other clinical features as well as the response to l-dopa treatment in type A and B patients as a group. Most additional clinical features are not unique to either type A or B patients. Nevertheless, there is a clear predominance of extra features in type B patients, especially with regard to perinatal abnormalities, diurnal fluctuations, autonomic disturbances and body length and weight at presentation. None of the patients in this series had clinical features suggesting systemic deficiency of catecholamines, such as abnormalities in the maintenance of blood pressure. These systemic phenomena were not, however, formally studied.

View this table:
Table 3

Prevalence of additional clinical features and response to treatment in type A and B patients as a group

Clinical dataType AType BP-value
Sex, pregnancy, delivery, neonatal periodn = 25n = 11
    Preterm birth (<37 weeks)04<0.01
    Foetal distress (meconium, heart rate abnormalities)06<0.01
    Perinatal asphyxia03<0.05
Presenting symptoms/signsn = 25n = 11
    Age at onset2 month–5 yearsNeonatal–3 months
    Age at onset >12 months60N.S.
    Diurnal fluctuation25<0.05
    Oculogyric crisis86N.S.
    Autonomic disturbances36<0.05
    Lethargy-irritability crises34N.S.
    Sleep disturbances33N.S.
    Length <−2SD14<0.05
    Weight <−2SD03<0.05
    Head circumference <−2SD34N.S.
Follow-up with medicationn = 24n = 11
    Age at start l-dopa6 months–15 years6 months–7 years
    Chronic l-dopa dose (mg/kg/dy)3.0–100.5–20
    l-dopa response: good/moderate/none21/2/12/5/4
    l-dopa response: within 1 week/within 2 weeks/after 2 weeks12/2/90/0/7
    Selegeline trial15
    Effects of selegilinePositivePositive but limited
    Other drugs (dopamine agonists) triedNoneBromocriptine, pramipexole
Long term follow-upn = 24n = 11
    Mental retardation (IQ < 70)a810<0.01
    Independent walking223<0.001
  • The differences in the occurrence of clinical features in the two patient categories were studied by using Fisher’s exact test. N.S. = not significant.

  • a See ‘Discussion’ section for details.

Type A patients generally showed a beneficial l-dopa response within the first 2 weeks of treatment, while positive effects in type B patients always occurred later. Selegiline, an inhibitor of dopamine degradation, as well as the dopamine-agonists bromocriptine and pramipexole, were prescribed in one type A and five type B patients, with some, although limited, additional effects. Treatment with l-dopa dramatically improved motor outcome in type A patients: one patient was lost from follow-up, Patients 17 and 25 remained wheelchair bound, and all others (22 out of 25, 88%) were able to walk independently during follow-up. Type B patients 26, 28 and 35 finally learned to walk independently, although the initial response to l-dopa had been less impressive than in most type A cases. The majority (67%) of type A patients had normal cognitive capacities during follow-up, while 10 out of 11 (91%) type B patients in our series were mentally retarded. Two type B patients (cases 29 and 33) died during follow-up due to infectious and respiratory complications.

Cerebral imaging

Most subjects (29 out of 36) underwent cerebral MRI studies. The images were not systematically reviewed for the purpose of this study. Non-specific, mild white matter signal changes and increased volume of extra-cerebral CSF spaces were reported in nine patients, while the majority of patients (n = 20) had normal images. The number of patients with abnormalities on MRI was 4 out of 21 (19%) and 5 out of 8 (63%) in Types A and B patients, respectively. Importantly, gross structural abnormalities of the brain and signal changes in the basal ganglia were never observed.

Biochemical findings

Based on the metabolic pathway involved, it can be predicted that CSF levels of the down-stream metabolites of the dopamine degradation pathway will be low, while the serotonin pathway is not affected (Fig. 1). Age-dependent reference values for the different metabolites have been published in detail (Brautigam et al., 1998). As an example, we give here the reference values (P2.5–P97.5) in childhood (2–5 years): HVA 384–769 nmol/l, 5HIAA 110–265 nmol/l, MHPG 35–64 nmol/l and HVA/5HIAA ratio 1.8–4.4.

Pre-treatment CSF results were lacking from seven patients [1–5, 17 (Family 7) and 29 (Family 23)]. Patients 1–5 were initially clinically diagnosed as having a l-dopa-responsive movement disorder in the 1970s, and never underwent lumbar puncture. In Families 7 and 23, a lumbar puncture was performed in only one of the two affected siblings. Decreased CSF concentrations of HVA and MHPG, with decreased HVA/5HIAA ratios were demonstrated in all patients in whom CSF was analysed (Fig. 2). To enable comparison of patients of different ages, with different references values, CSF HVA concentrations and HVA/5HIAA ratios were expressed as percentage of the lower reference limit (2.5th percentile) (Brautigam et al., 1998). The mean CSF HVA concentration in type B patients was significantly lower (P < 0.005) than in type A patients, namely 8.8% (SD 8.9) and 32.7% (SD 19.8), respectively. As a group, type B patients also had a significantly lower (P < 0.005) mean CSF HVA/5HIAA ratio than type A patients, namely 5.6 (SD 4.7) and 36.7 (SD 18.9), respectively. After ranking the patients in order of age of onset of the disease (as in Table 1), we were able to show a positive correlation between age of onset and CSF HVA concentrations as well as HVA/5HIAA ratio (Fig. 3). Urinary concentrations of catecholamines and their degradation products were available in a minority of the patients (data not shown) and were generally found to be non-informative.

Figure 2

Concentrations of HVA (left panel) and MHPG (middle panel), and HVA/5HIAA ratio (right panel) in CSF of patients with THD at diagnosis (i.e. without treatment) according to THD subtype. Concentrations are given in percentage of the lower reference limit (2.5th percentile) in controls (see text). For comparisons between the two groups the Student’s t-test was used. ***P < 0.005.

Figure 3

HVA concentration (left panel) and HVA/5HIAA ratio (right panel) concentrations in CSF of THD patients at diagnosis (i.e. without treatment), in relation to age at onset of disease. Concentrations are given in percentage of the lower reference limit (2.5th percentile) in controls. Regression coefficients are: HVA r2= 0.52 (P < 0.0001), HVA/5HIAA ratio r2 = 0.66 (P < 0.0001).

Mutation analysis

We identified 24 different TH gene mutations (Tables 1 and 4; Fig. 4), located in the promoter sequence, exons 3, 5–14, and in intron 11. Six mutations were not reported previously. The novel c.1375C>T mutation was predicted to lead to a stop codon (p.Gln459X). The other five novel mutations (c.620G>A, c.680A>G, c.721G>A, c.776A>G and c.1181T>C) were all considered pathogenic since they were (i) not encountered in 200 control alleles; (ii) not reported as polymorphisms in databases and the literature (Haavik et al., 2008); and (iii) affecting amino acids within the tyrosine hydroxylase protein that are highly conserved among various species. Additionally, the program SIFT (Sorting Intolerant From Tolerant) (Ng and Henikoff, 2001) predicted ‘not tolerated’, i.e. a deleterious effect for all five mutations.

Figure 4

Overview of all known pathogenic mutations in the TH gene. The cyclic adenosine monophosphate response element of the TH promotor resides between residues −67 and −74 upstream of the ATG initiation codon.

The total number of mutated alleles reported in THD is 100 (Table 4). Five out of these 100 alleles harboured (four different) mutations that lead to protein truncation [c.295delC, c.296delT, p.Arg169X (c.DNA change not reported) and c.1375C>T], while all other 95 alleles were affected by less deleterious missense mutations. Homozygosity for the common c.698G>A mutation was found in six type A and three type B patients. Homozygosity for the c.707T>C mutation occurred in one type A and two type B patients. Promoter mutations were only encountered in type A THD with good l-dopa responsiveness.

View this table:
Table 4

Mutations in the tyrosine hydroxylase gene that lead to THD

NumberExonMutationProtein changeNumber of alleles affected
30Intron 11c.1198–24T>Ap.?2
  • This table includes all mutations reported to date in THD, and six novel mutations (in bold) reported in this article. In case of affected sib-pairs, both patients are included. The total number of mutated alleles is 100. For all mutations: see Tables 1 (our series of patients) and 3 (all patients reported by others). n.r. = not reported.


In general, the results of the present study and available data in the literature are in perfect agreement. In total, reports on 50 THD patients (from 41 families) are now available in the literature, and are reviewed in this article. Type A THD (n = 35) is more often diagnosed than type B THD (n = 15).

Clinical features at presentation

The many different neurological features of THD (hypokinesia, bradykinesia, rigidity, dystonia, chorea, tremor, oculogyric crises, ptosis and hypersalivation, among others) are caused by cerebral dopamine and norepinephrine deficiency, as nicely explained and discussed previously (Grattan-Smith et al., 2002). THD leads to symptoms early in life, generally in infancy, but sometimes as early as the neonatal period. Presentation in childhood was very rare, and no patients identified to date have presented in adolescence or adulthood.

Since THD is rare and its features overlap with many other neurological disorders, the diagnosis will generally not be made on clinical grounds alone. The differential diagnosis of THD in neonates or very young infants with type B presentation initially encompasses a long list of progressive as well as stable, hereditary as well as acquired disorders. Type B THD is often accompanied by perinatal complications (Table 3), which may further distract the attention in the direction of common infectious or hypoxic-ischaemic encephalopathies. Type B THD can also mimic genetic disorders like catastrophic epileptic encephalopathies or mitochondrial disorders. Only extensive work-up, including cerebral imaging and screening for inborn errors of metabolism, including CSF analysis, will lead to the correct diagnosis. In type A patients on the other end of the spectrum, the children with ‘parkinsonian’ features and l-dopa-responsive dystonia, clinical recognition of the diagnosis might be easier. Importantly, THD with a relatively mild course can strongly mimic cerebral palsy, which may lead to serious diagnostic delay. Besides cerebral palsy, the differential diagnosis of ‘juvenile parkinsonism’ also includes various other acquired as well as genetic disorders, among which GTP cyclohydrolase deficiency and other defects (like sepiapterine reductase deficiency) in the synthesis of the tyrosine hydroxylase co-factor tetrahydrobiopterin (BH4) (Muller et al., 1998; Albanese et al., 2006; Tarsy and Simon, 2006; Muller, 2009). The lack of abnormalities on cerebral imaging studies and a marked responsiveness to l-dopa are clues to the disorders of neurotransmitter biosynthesis as a group.

In GTP cyclohydrolase deficiency, the most common defect of tetrahydrobiopterin biosynthesis, diurnal fluctuation of dystonia, can be a prominent hallmark. Both THD and GTP cyclohydrolase deficiency are considered l-dopa-responsive dystonias, and have been named Segawa syndrome and DYT5 in the past (see ‘Introduction’) (Muller et al., 1998; Albanese et al., 2006; Tarsy and Simon, 2006; Muller, 2009). THD, however, is generally more severe and characterized by an earlier onset of symptoms. Furthermore, the CSF profile of neurotransmitter and pterin metabolites discriminates between the two disorders. It has recently been proposed to designate these conditions into GTP cyclohydrolase deficiency and THD ‘dystonia 5a’ and ‘dystonia 5b’, respectively, illustrating their clinical and biochemical relationship and reflecting their different molecular basis (Muller et al., 1998).

Dopamine plays an important regulatory role in the neuro-endocrine system. Since dopamine suppresses the release of prolactin, THD may lead to hyperprolactinaemia. Indeed, serum prolactin may be increased in THD. In the literature, one patient with THD has been described who presented with galactorrhoea due to hyperprolactinaemia before the neurological features appeared (Yeung et al., 2006). Dopamine is also known to play an important role in growth hormone secretion. Nevertheless, height and pubertal development are generally normal in THD.

Treatment and clinical course

The natural course of THD is—in sensu strictu—unknown. However, the observations in all patients reported here and in the literature indicate that the severe neurological features will never reverse as long as patients are not treated properly.

Since THD leads to dopamine deficiency in the CNS, treatment with l-dopa is by far the strategy of first choice. In fact, the lacking metabolite can simply be supplemented (Fig. 1). Drugs containing l-dopa generally also contain a peripheral l-dopa decarboxylase inhibitor (benserazide or carbidopa) to prevent loss of l-dopa in the circulation. Without further specification, we write ‘treatment with l-dopa’, meaning l-dopa combined with a decarboxylase inhibitor.

l-dopa dosages commonly used in paediatric neurology range from 3 to 10 mg/kg bodyweight per day, given in three doses. Although l-dopa-responsive THD patients were usually treated with an ongoing l-dopa dose in this range, many types A and B patients started with lower doses, tolerating only very gradual increases in dosage over weeks or even months (see below). Type B patients especially were often extremely sensitive to l-dopa. This hypersensitivity necessitated initial dosages below 0.5 mg/kg bodyweight per day, in four (or even six) divided doses, and impeded some patients from being treated with l-dopa at all (Brautigam et al., 1999; de Lonlay et al., 2000; Dionisi-Vici et al., 2000; Janssen et al., 2000; Haussler et al., 2001; Grattan-Smith et al., 2002; Hoffmann et al., 2003; Zafeiriou et al., 2009).

Alternatively and in addition to l-dopa, THD patients can rationally be treated with inhibitors of dopamine degradation like selegiline, assuming that some dopamine is still formed in tyrosine hydroxylase-deficient neurons. Selegiline as well as dopamine agonists, anticholinergic drugs and benzodiazepines available for the treatment of dystonia (Albanese et al., 2006; Tarsy and Simon, 2006) were, however, hardly ever used in this series.

Patients who tolerated l-dopa well generally showed a good or at least moderate response with dramatic improvement of movements due to disappearance of hypokinesia, tremor, rigidity and dystonia, and subsequent impressive gain in motor functions. Children who had been in a wheelchair for years started to walk again. Good responders generally also responded early (Table 3). These patients only developed l-dopa induced dyskinesia when dosages were increased too quickly or in too great increments. To date, only one subject (Patient 23), developed dyskinesia probably as the consequence of the long-term, i.e. 13 year, treatment with l-dopa. All others tolerated l-dopa very well and remained without serious side effects after more than one (Cases 6, 9, 10, 13, 14, 16, 22, 24, 25, 27, 31) or even two or three (Cases 1–5) decades of treatment. In a limited number of l-dopa-responsive THD patients, lumbar punctures were performed to monitor treatment effects at the biochemical level. In those cases, clinical responses were good irrespective to the fact that HVA never reached normal levels.

Despite appropriate treatment, long-term cognitive development was subnormal in many patients with THD. We did not systematically study cognitive profiles in this series but based on available test reports, school performance, social functioning and clinical impression by experienced neurologists, it was concluded that mild to moderate mental retardation occurred in 33 and 91% of Types A and B patients, respectively (Table 3). In none of the patients we encountered signs of cognitive decline under treatment.

Based on shared experiences, but impossible to prove due to the low number of patients, we think that early diagnosis and treatment of THD improves the final outcome with regard to motor as well as cognitive functions. On the other hand, it has extensively been demonstrated in animal studies that dopamine plays a vital role in foetal brain development (Zhou et al., 1995; Araki et al., 2007). Dopamine deficiency might therefore cause, already before birth, irreversible structural brain abnormalities at the microscopic level and leading to mental retardation.

Taking all data together, THD can best be treated with an initial l-dopa dose of 0.5–1 (for type B patients) to 3 (for type A patients) mg/kg bodyweight per day, divided over three or four doses. When tolerated well, the dose can slowly be increased until the desired clinical response is observed or the occurrence of adverse effects forces dose reduction. It should be kept in mind that some patients respond only in the course of months: one should thus wait long enough to see the final effects in apparent moderate or non-responders. We think that type B patients with extreme l-dopa hypersensitivity might be ideal candidates for continuous duodenal administration of a soluble formulation of l-dopa (duodopa), although we did not encounter this approach in our series and the literature.

Biochemistry and genetics

Cathecholamine biosynthesis was long thought to occur only in specific neuronal cell populations in the CNS, sympathetic peripheral nervous system, adrenal glands and the kidneys. Interestingly, however, the presence of the tyrosine hydroxylase protein, its mRNA and its enzymatic activity, have been demonstrated in other non-neuronal tissues capable of catecholamine biosynthesis, such as lymphoid tissues, exocrine pancreas and the gastrointestinal tract (Mezey et al., 1996; Eisenhofer et al., 1997). Human THD leads to a neurological disorder, apparently leaving the other organs unaffected. An explanation for this observation might be that the brain is the most vulnerable organ, already severely affected by relatively minor changes in the TH gene (see below). A second explanation might be found in other enzymes, expressed in non-neuronal tissues, which can hydroxylate tyrosine. Tyrosinase (E.C., for example, is expressed in the epidermis and plays a crucial role in melanin biosynthesis from tyrosine (Rios et al., 1999). In humans, tyrosinase deficiency leads to oculocutaneous albinism type 1 (OMIM 606933) (Oetting et al., 2003). In mice, tyrosinase substantially contributes to peripheral, tyrosine hydroxylase-independent dopamine production (Eisenhofer et al., 2003).

The biochemical diagnosis of THD would ideally rely on direct measurement of enzyme activity in tissue samples, blood cells or cultured fibroblasts. Since it was traditionally thought that tyrosine hydroxylase was only expressed in tissues that would not be available for enzymatic analyses, we and others have always relied on CSF metabolites (Fig. 1) and TH gene mutation analysis to diagnose THD. The collected data of the large series of patients described here clearly demonstrate that the most severely affected patients have the lowest concentrations of neurotransmitter metabolites in CSF (Fig. 3). The HVA concentration and HVA/5HIAA ratio in CSF showed overlap between Types A and B patients, however this overlap was only minor for the HVA/5HIAA ratio (Fig. 2). These parameters (especially the HVA/5HIAA ratio in CSF), most probably reflecting the degree of residual tyrosine hydroxylase activity in the brain, may thus be used to predict l-dopa responsiveness and overall outcome.

Measurements of phenylalanine and tyrosine in body fluids, and urinary concentrations of catecholamines, HVA and MHPG, are non-informative in patients with THD (Brautigam et al., 1998; Wevers et al., 1999; Hoffmann et al., 2003). The surprisingly often normal urinary dopamine excretion in THD is hypothetically attributed to residual tyrosine hydroxylase enzyme activity in peripheral non-neuronal tissues or alternative enzymes with the capacity to hydroxylate tyrosine, as discussed above. The fact that profound cerebral dopamine deficiency due to defective dopamine biosynthesis can be accompanied by normal urinary dopamine excretion makes analysis of urine unreliable, even potentially misleading, in the diagnostic work-up for dopamine biosynthesis disorders.

The human TH gene (mRNA type 1) contains 14 exons with an open reading frame of 1491 bp encoding for a protein with 497 amino acids that is highly conserved among various species (Nagatsu and Ichinose, 1991) Tables 1, 2 and 4, and Fig. 4, give an overview of the 37 different TH mutations in THD patients. Out of 100 alleles, 95 were affected by missense mutations leading to amino acid substitutions in the protein with subsequent partial loss of enzyme activity. The pathogenic effects of some missense mutations on the protein have been confirmed, (Ludecke et al., 1995, 1996; Knappskog et al., 1995; Royo et al., 2005; Haavik et al., 2008) but most can only be designated pathogenic based on indirect evidence. Interestingly, we and others have recently identified pathogenic mutations in the promoter region of the TH gene as a rare underlying genetic mechanism (Ribases et al., 2007; Verbeek et al., 2007). A founder effect has been described for the c.698G>A mutation in the Dutch population (van den Heuvel et al., 1998). Except for this and two other mutations (c.−70G>A and c.707T>C), almost all other pathogenic changes in the TH gene are so-called ‘private’ mutations (Table 4). Based on the latter finding, it can be expected that THD occurs in all parts of the world, irrespective of the fact that up to now most patients were diagnosed in Western Europe.

The relatively high number of patients in this paper allows us to discuss possible genotype–phenotype correlations in THD. Only five patients, from four families, harboured deleterious mutations that lead to protein truncation: namely c.295delC and c.296delT, which both lead to p.Leu99fs; p.Arg169X (the c.DNA change was not reported in the two affected siblings); and c.1375C>T, which causes p.Gln459X. Heterozygosity for each of these mutations was found in type A as well as type B patients. Until now, no patient has been reported with homozygosity or compound heterozygosity for two truncating TH mutations. This observation most probably indicates that complete loss of tyrosine hydroxylase activity is incompatible with human life, comparable with the findings in knockout mice (Zhou et al., 1995). Homozygosity for the common c.698G>A mutation and homozygosity for the c.707T>C mutation was found in both THD phenotypes (Table 1), leaving these genotypes without predictive value with regard to THD phenotype. Importantly, however, all patients carrying at least one promoter mutation (n = 8, Tables 1 and 2) suffered from type A THD with good l-dopa responsiveness. As discussed previously, reduced TH gene transcription apparently leaves significant residual tyrosine hydroxylase enzyme activity, likely corresponding to a relatively mild type A phenotype (Ribases et al., 2007; Verbeek et al., 2007).

The differential diagnosis in children and adults with movement disorders is essentially different. This explains the remarkable diversity in the diagnostic approach depending on the patient’s age. Adult patients with dystonia can be investigated at the genetic level after dedicated neurological classification (Muller et al., 1998; Albanese et al., 2006; Tarsy and Simon, 2006; Wu et al., 2008; Clot et al., 2009; Muller, 2009), while children often have to be exposed to more extensive and invasive diagnostic procedures to illuminate the underlying cause of their disorder (Assmann et al., 2003). In contrast to what is advocated for adult patients, we strongly advise performance of a lumbar puncture in children with otherwise unexplained (simple and complex) movement disorders, to diagnose or rule out potentially treatable conditions like THD.

In conclusion, THD is a severe but often very treatable neurometabolic disorder resulting from cerebral catecholamine deficiency. The diagnosis of THD relies on clinical suspicion and the analysis of CSF metabolites. Importantly, the CSF concentration of HVA and HVA/5HIAA ratio correlate with the severity of the clinical phenotype. THD can be proven by demonstrating mutations in the TH gene or its promoter region. The disorder is almost exclusively caused by missense mutations, suggesting that mutations with more deleterious effects on the protein will be incompatible with life. Genotype–phenotype correlations do not exist for the common c.698G>A and c.707T>C mutations. Carriership of at least one promotor mutation, however, apparently predicts type A THD. Most patients with THD, but not all, can successfully be treated with l-dopa.


The Swiss National Science Foundation (Grant No. 3100A0-119982/1) (to N.B.); Zon-MW Innovational Research Grant (Vidi Program No. 917.46.331 to M.M.V.).


We thank Prof. Robert Surtees, London, UK, who sadly died during the preparation of this article. Prof. Robert Surtees was well-known for his expertise in the field of paediatric movement disorders, and contributed to the manuscript in its early phases by sharing his clinical experiences with us, and by offering helpful suggestions.


  • Abbreviations:
    5-hydroxyindoleacetic acid
    homovanillic acid
    3-methoxy-4-hydroxyphenylethylene glycol
    tyrosine hydroxylase deficiency


View Abstract