Treatable metabolic causes of early onset epilepsy (in the first few months of life) are uncommon but it is important to diagnose them because delay in specific treatment commonly results in poor neurological and cognitive outcome. Indeed, some of these epilepsies are fatal if left untreated. The disorders are listed in table 1 and are divided into vitamin-responsive and other metabolic epilepsies.

We are not going to discuss all of these disorders, but will confine ourselves to those presenting at or very close to birth. The investigation and treatment of the other epilepsies are summarised in table 2.

PYRIDOXINE-DEPENDENT EPILEPSY

Typical neonatal onset pyridoxine-dependent epilepsy (PDE) presents in the first few days of life with multiple seizure types which are intractable to anticonvulsant drug treatment. Experienced mothers may also recognise seizures in utero from about 20 weeks of gestation. Approximately one third of infants with PDE will also have features of a neonatal encephalopathy (hyperalertness, irritability and a stimular sensitive startle) which can be accompanied by systemic features. Neuroimaging may show structural brain abnormalities, hydrocephalus or haemorrhage. EEG is abnormal but has no diagnostic features.¹ The seizures and EEG abnormalities respond promptly (within minutes) to 100 mg of intravenous pyridoxine. However, in about 20% of infants with pyridoxine-dependent epilepsy the first dose of pyridoxine can also cause cerebral depression, which is more likely if the infant is receiving anticonvulsant drugs. Children with PDE require life-long treatment with pyridoxine at a dose of ∼15 mg/kg/day up to 500 mg/day. With prompt treatment prognosis is generally good, although there may be later learning difficulties, particularly with language. If treatment is delayed by months or years, children develop severe learning difficulties, four limb motor disorder and sensory disturbances.

The major cause of PDE has been recently identified² as being due to mutations in the ALDH 7A1 gene which encodes a central nervous system α-aminoadipic semialdehyde dehydrogenase. This results in accumulation of α-aminoadipic semialdehyde (α-AASA). α-AASA is in reversible equilibrium with piperideine-6-carboxylate which can condense with pyridoxal phosphate and inactivate its cofactor activity. The build-up of α-AASA and its spill-over into plasma and urine make it a specific and sensitive marker of PDE caused by α-aminoadipic semialdehyde dehydrogenase deficiency.² Another metabolite more proximal in this lysine degradation pathway is pipecolic acid, and this is also raised in PDE but is less specific and sensitive than α-AASA.³ Both markers remain raised after treatment with pyridoxine. This is demonstrated in this issue of ADC by Bok et al who have shown that all patients in the Netherlands with clinically definite PDE (according to Baxter’s criteria⁴) and six of eight patients with clinically probable or possible PDE had raised urinary and plasma α-AASA concentrations (one patient was not tested and one was negative). In these patients plasma, but not urinary, concentrations of pipecolic acid were also raised. Bok et al also calculate that PDE caused by α-aminoadipic semialdehyde dehydrogenase deficiency has a birth incidence in the Netherlands of at least 1:276 000.

PYRIDOX(AM)INE PHOSPHATE OXIDASE DEFICIENCY

Pyridox(am)ine phosphate oxidase (PNPO) deficiency presents as foetal distress in late pregnancy. Seizures start in the first few days of life after birth and are intractable to anticonvulsant drugs. EEG is abnormal with no normal background activity and bursts of spike-wave discharges. Neuroimaging is normal at the onset. There is no response to intravenous pyridoxine. However, there is a prompt and lasting response to oral pyridoxal phosphate (hence the alternative name of pyridoxal phosphate-responsive seizures) given at a dose of 10 mg/kg. Like PDE, treatment of PNPO deficiency with pyridoxal phosphate can cause transient cerebral depression. Again treatment is life-long, requiring pyridoxal phosphate at a dose of 30–50 mg/kg/day. With prompt treatment the prognosis is generally good. However, if left untreated, most cases of PNPO deficiency are fatal in the first year, and rare survivors have severe neurological handicap and profound brain atrophy.^5,6

Some biochemical markers of PNPO deficiency are evident in cerebrospinal fluid and plasma.⁵ Because vitamin B₆ crosses from cerebrospinal fluid into neural cells as phosphorylated esters of pyridoxine and pyridoxamine as well as pyridoxal, PNPO is necessary to convert the pyridoxine phosphate and pyridoxamine phosphate to the active cofactor pyridoxal phosphate. Pyridoxal phosphate is the cofactor for over 100 enzymes involved in amino acid and amine metabolism, three of which are L-aromatic amino acid decarboxylase (AAAD), the glycine cleavage system and threonine dehydratase. Dysfunction of the latter two enzymes results in a mild increase in the concentrations of glycine and threonine, respectively, in both plasma and cerebrospinal fluid. AAAD dysfunction causes a reduction in cerebrospinal fluid concentrations of homovanillic acid (the stable acidic metabolite of dopamine) and 5-hydroxyindoleacetic acid (the stable acidic metabolite of serotonin) and an increased concentration of 3-methoxytyrosine. These biochemical abnormalities are not present in all patients.⁶ However, if they are present, they reverse with treatment.

The birth incidence of PNPO deficiency is not known.

FOLINIC ACID-RESPONSIVE SEIZURES

Folinic acid responsive seizures is the least well recognised cause of vitamin-responsive neonatal epilepsy.^7,8 Infants present in the first few days of life with multiple seizure types which are intractable to anticonvulsant drugs. They may also have features of a neonatal encephalopathy. EEG shows abnormal background activity with multifocal spike-wave complexes but no diagnostic features. Neuroimaging is normal. There can sometimes be a transient response to pyridoxine. The seizures promptly respond to folinic acid at a dose of 2.5–5 mg twice daily. However, seizures can recur later, sometimes responding to increases in folinic acid alone (up to 8 mg/kg/day) and sometimes also requiring anticonvulsant drug therapy. If left untreated, folinic acid-responsive seizures is a fatal disorder. Even with treatment and control of the epilepsy, there is appreciable mortality and survivors have global learning difficulties.

Folinic acid-responsive seizures does have a biochemical marker in cerebrospinal fluid, but its nature is unknown. It is detected by high performance liquid chromatography with electrochemical detection under the conditions required to measure homovanillic acid and 5-hydroxyindoleacetic acid. This unknown compound is not normally present in cerebrospinal fluid and decreases once treatment with folinic acid has been initiated.

The birth incidence of folinic acid-responsive seizures is not known.

APPROACH TO DIAGNOSIS

There are two possible approaches to diagnosing these treatable neonatal epilepsies. The first is to give the relevant vitamins and observe the response. In our view, this is the correct and also the easiest approach. The second method is to investigate the biochemical markers for each condition. This will require the help of a specialised (often highly specialised) metabolic laboratory. Because this approach will take time and typical biochemical abnormalities will not be present in all cases,⁶ we advise treatment before starting such investigations.

APPROACH TO NEONATAL EPILEPSY

Once precipitating conditions such as infection, hypoglycaemia or electrolyte, calcium and magnesium disturbance have been excluded or corrected, we believe the next line of investigation and treatment in neonatal epilepsy is to initiate a vitamin trial. We suggest giving two doses of pyridoxal phosphate (10 mg/kg/dose) 2 h apart, and, if the epilepsy persists, two doses of folinic acid (5 mg) 6 h apart. EEG monitoring is helpful but not mandatory. If there is no response to the vitamins, anticonvulsant therapy should be introduced and further investigations into the cause of the epilepsy carried out.

There are three lines of reasoning behind this view as follows.

Firstly, it has recently become clear that there are physiological reasons why anticonvulsant drugs which act as GABA-agonists might not be as effective in newborns as in older children and adults. These anticonvulsant drugs include phenobarbitone and benzodiazepines. There is also increasing evidence from animal studies that phenobarbitone, phenytoin and the benzodiazepines are toxic to the newborn brain, causing increased neural apoptosis. These findings suggest that standard anticonvulsant treatment of neonatal seizures may not be as effective or as safe as previously believed.^9,10

Secondly, there is no biochemical or chemical reason to believe that pyridoxal phosphate will not be as effective as pyridoxine in the treatment of PDE. PDE can be distinguished from PNPO deficiency later by measuring urinary α-AASA excretion, and a firm diagnosis made by mutation analysis of DNA. It is also important to remember that α-aminoadipic semialdehyde dehydrogenase deficiency is not the only cause of PDE, and a small proportion of cases will be missed by biochemical analysis unless pyridoxal phosphate is tried. If pyridoxal phosphate is not easily available, pyridoxine should be given as first line therapy. It must, however, not be forgotten that unsuccessful treatment with pyridoxine does not exclude PNPO deficiency.

Thirdly, delayed diagnosis of these conditions can cause severe neurological handicap and early death.