SEARCH

SEARCH BY CITATION

Keywords:

  • Neonatal seizures;
  • Electroencephalography;
  • Antiquitin;
  • ALDH7A1 gene mutations;
  • Pyridoxine

Summary

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References

Purpose:  Pyridoxine-dependent epilepsy (PDE) is characterized by therapy-resistant seizures (TRS) responding to intravenous (IV) pyridoxine. PDE can be identified by increased urinary alpha-aminoadipic semialdehyde (α-AASA) concentrations and mutations in the ALDH7A1 (antiquitin) gene. Prompt recognition of PDE is important for treatment and prognosis of seizures. We aimed to determine whether immediate electroencephalography (EEG) alterations by pyridoxine-IV can identify PDE in neonates with TRS.

Methods:  In 10 neonates with TRS, we compared online EEG alterations by pyridoxine-IV between PDE (n = 6) and non-PDE (n = 4). EEG segments were visually and digitally analyzed for average background amplitude and total power and relative power (background activity magnitude per frequency band and contribution of the frequency band to the spectrum).

Results:  In 3 of 10 neonates with TRS (2 of 6 PDE and 1 of 4 non-PDE neonates), pyridoxine-IV caused flattening of the EEG amplitude and attenuation of epileptic activity. Quantitative EEG alterations by pyridoxine-IV consisted of (1) decreased central amplitude, p < 0.05 [PDE: median −30% (range −78% to −3%); non-PDE: −20% (range −45% to −12%)]; (2) unaltered relative power; (3) decreased total power, p < 0.05 [PDE: −31% (−77% to −1%); −27% (−73% to −13%); −35% (−56% to −8%) and non-PDE: −16% (−43% to −5%); −28% (−29% to −17%); −26% (−54% to −8%), in delta-, theta- and beta-frequency bands, respectively]; and (4) similar EEG responses in PDE and non-PDE.

Discussion:  In neonates with TRS, pyridoxine-IV induces nonspecific EEG responses that neither identify nor exclude PDE. These data suggest that neonates with TRS should receive pyridoxine until PDE is fully excluded by metabolic and/or DNA analysis.

For accurate treatment of neonatal seizures, it is important to first ascertain underlying etiology (Scher, 2003). Under some conditions [such as pyridoxine-dependent epilepsy (PDE), pyridoxine 5′-phosphate oxidase deficiency (PNPO), and folinic acid responsive seizures], therapy-resistant seizures (TRS) may respond to administration of pyridoxine (vitamin B6) and/or pyridoxal phosphate (Baxter, 2005; Mills et al., 2005; Wang et al., 2005; Gallagher et al., 2009). The incidence of PDE is relatively low [1:300,000 newborns in The Netherlands (Been et al., 2005)]. Classical presentations of PDE concern early, intractable neonatal seizures that cease after pyridoxine administration (Hunt et al., 1954; Baxter, 2001a). PDE is caused by an inherited metabolic disorder of lysine degradation, resulting in increased urinary alpha-aminoadipic semialdehyde (α-AASA) excretion. In addition to increased urinary α-AASA concentrations, mutations in the ALDH7A1 (antiquitin) gene can also identify PDE (Mills et al., 2006; Bok et al., 2007; Kanno et al., 2007; Plecko et al., 2007; Salomons et al., 2007; Striano et al., 2009). Although prognosis may vary even after early treatment (Rankin et al., 2007), it is conceptualized that fast and accurate PDE treatment could be beneficial (Haenggeli et al., 1991; Baxter & Aicardi, 1999; Baxter, 2001b; Bok et al., 2010).

To avoid delay by biochemical and genetic testing, diagnostic trajectories often include an empirical pyridoxine trial. Such pyridoxine trials generally involve assessment of the clinical response to pyridoxine administration and subsequent withdrawal (Pettit, 1987; Gospe, 1998; Baxter, 1999). In TRS patients, there are two common routes for pyridoxine administration: (1) The intravenous (IV) route, which is generally associated with fast clinical and electroencephalography (EEG) responses; and (2) the oral route, which is generally associated with slower clinical and EEG responses (see for review: Baxter, 2001a; Gospe, 2006). In perspective of the faster response, the IV route is often applied in neonates with a high seizure frequency.

Prior to pyridoxine-IV, neonatal EEG signals have been characterized by a discontinuous background activity (Nabbout et al., 1999; Hellstrom-Westas et al., 2002; Naasan et al., 2009), with bilateral high-voltage delta waves and with central/temporal (poly)spikes (Nabbout et al., 1999). After pyridoxine-IV, epileptic activity may disappear instantaneously (Goto et al., 2001; Baxter, 2003). This effect can be accompanied by other EEG characteristics, such as flattening of the EEG trace (Mikati et al., 1991; Nabbout et al., 1999; Hellstrom-Westas et al., 2002). However, pyridoxine-IV EEG responses can vary among PDE patients (Naasan et al., 2009), and the specificity of EEG signals can be affected by many different antiepileptic drug (AED) treatments that children with PDE often receive prior to pyridoxine-IV (Baxter, 2000; Hellstrom-Westas et al., 2002). In addition, one should also take into account that EEG signals in PDE neonates would be expected to differ from EEG signals in elder children with PDE (Baxter, 2001a). Finally, it is indicated that pyridoxine-IV may also induce a favorable antiepileptic effect in the absence of PDE (Tsuji et al., 2007). Accordingly, we have shown that pyridoxine-IV can nonspecifically reduce the EEG amplitude in non-PDE neonates with TRS (Teune et al., 2007). In neonates with TRS, we, therefore, reasoned that nonspecific EEG effects could hamper the clinical interpretation of the pyridoxine-IV trial.

Given the preceding, we aimed to investigate whether direct EEG responses to pyridoxine-IV allow prompt identification of neonatal PDE. For this purpose, we compared quantitative (digital) and qualitative (visual) EEG responses to pyridoxine-IV between neonates characterized by PDE (by ALDH7A1 gene mutation) and neonates characterized by non-PDE. To the best of our knowledge, such quantitative EEG responses to pyridoxine-IV have not been systemically investigated before in a homogeneous neonatal PDE cohort.

Patients

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References

In TRS (PDE and non-PDE) neonates, we retrospectively collected and compared EEG recordings during first pyridoxine-IV exposure. The medical ethical committee of the University Hospital Groningen approved the present study. TRS were defined as seizures that do not respond to adequate administration of at least two first-line AEDs. Prescribed antiepileptic drugs prior to the pyridoxine-IV administration involved therapeutic dosages of benzodiazepines, phenytoin, lidocaine, and/or phenobarbital (Table 1). PDE was defined according to the clinical criteria of Baxter (1999) and confirmed by ALDH7A1 gene mutation analysis (Bok et al., 2007; Salomons et al., 2007). All PDE data were derived from the Dutch study cohort [until January 2010 consisting of a total of 15 PDE children; see for further description Bok et al. (2007)]. For analysis of PDE data, we selected all six neonatal (<3 months of gestation) EEG recordings that were obtained during first pyridoxine-IV exposure. All six PDE neonates were born, treated, and recorded between 1999 and 2008, at five different university hospitals in The Netherlands (Academic Medical Center Amsterdam; Erasmus Medical Center Rotterdam; University Medical Center Groningen; University Medical Center Nijmegen; University Medical Center Utrecht). After parental informed consent (2008), we retrospectively performed EEG analysis in PDE neonates at the University Medical Center Groningen (between 2008 and 2009). Except for PDE neonate 6, all other (n = 5) included PDE neonates were homozygous for the common Dutch- c.1195G>C mutation in ALDH7A1. PDE neonate 6 appeared compound heterozygous with a c.1195G>C mutation and a c.244C>T mutation (Salomons et al., 2007). In all PDE neonates, urine α-AASA levels were increased (≥9 mmol/mol creatinine; controls <1 mmol/mol creatinine), whereas in all non-PDE neonates urine α-AASA levels were normal.

Table 1.   Clinical data of neonates with therapy-resistant seizures (TRS)
 InfantSignalAED at pyridoxine-IVSeizure classificationSeizure typeSeizure onsetAge pyrid -IVUrine conc. α-AASA
  1. PDE, pyridoxine-dependent epilepsy; non-PDE, non–pyridoxine-dependent epilepsy; AED, antiepileptic drug; IV, intravenous; conc., concentration; α-AASA, alpha-aminoadipic semialdehyde in mmol/mol creatinine (normal value <1); D, digital; aEEG, amplitude integrated electroencephalography (EEG); P, paper; B, benzodiazepines; P, phenytoin; LD, lidocaine; PB, phenobarbital; C, cryptogenic; S, symptomatic; G, generalized; MF, multifocal; d, postnatal day(s).

PDE1DB, PBGMyoclonic0 d3 d9.6
2DB, LDMFMiscellaneous0 d3 d71
3DB, PB, PGTonic0 d15 d32
4DPB, PGMyoclonic0 d2 d24
5aEEGB, PB, LDMFMiscellaneous2 d7 d32
6PB, PB, PMFTonic–clonic0 d3 d12
non-PDE1DB, PBC, GTonic–clonic0 d12 d<1
2DB, PB, LDS, GSubtle0 d2 d<1
3DB, PBS, GTonic–clonic1 d7 d<1
4DB, LD, PBS, GTonic–clonic1 d13 d<1

All non-PDE neonates were born, treated, and recorded between 2001 and 2004 at the University Medical Center Groningen [see for patient characteristics Teune et al. (2007)]. After parental informed consent (obtained in 2005), we retrospectively performed non-PDE EEG analysis at the University Medical Center Groningen [between 2005 and 2007 (Teune et al., 2007)]. One of the analyzed pyridoxine responsive TRS neonates (former case 3) was definitely diagnosed with PDE and was, therefore, assigned to the present PDE group. We consecutively obtained pyridoxine-IV EEG data for four remaining non-PDE neonates [<3 months of age; i.e., former cases 1, 2, 5, and 6 (Teune et al., 2007)]. The above-described patient selection criteria resulted in relatively homogeneous PDE and non-PDE characteristics, both regarding seizure onset (0–2 days after birth) and regarding the timing of EEG recordings (2–15 days after birth). Patient data are shown in Table 1.

Methods

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References

In all 10 included neonates with TRS, we visually analyzed EEG recordings during the first pyridoxine-IV administration. All TRS neonates received 100 mg pyridoxine-IV, except for one PDE neonate and one non-PDE neonate who received 50 and 300 mg, respectively. Both 5–15 min before and after pyridoxine-IV, we obtained two, 1-min EEG segments involving identical vigilance states. Segments were excluded for artifacts. All EEGs of TRS neonates were subsequently assigned and compared between PDE and non-PDE subgroups.

Visual EEG analysis

We visually compared the EEG response during pyridoxine-IV between neonatal PDE and non-PDE (six and four neonates, respectively). EEG recordings of PDE neonates consisted of four digital 21-channel registrations, one digital amplitude integrated EEG (aEEG) registration, and one paper 21-channel registration. EEG recordings in non-PDE neonates consisted of four digital 21-channel EEG registrations.

Quantitative analysis of digital EEG recording

We assessed the same EEG segments for both digital and visual analysis. In eight TRS neonates, we digitally analyzed eight EEG recordings for amplitude, total and relative power (in delta-, theta-, and beta-frequency-bands). Quantitative EEG data of TRS neonates were subsequently subdivided and compared between PDE (n = 4) and non-PDE (n = 4) subgroups.

Total power reflects the magnitude of background activity for an individual frequency band. Relative power indicates the relative contribution of a frequency band to the entire spectrum (Lopes da Silva, 2005). Digital EEGs were analyzed by Brainlab (version 4.00–0.00; OSG BVBA, Rumst, Belgium) and Brain Vision Analyzer (version 1.030002; Brain Products GmbH, Gilching, Germany). Bipolar recordings were evaluated at frontal (F3-F4), centrotemporal (T3-C3; C4-T4), central (C3-C4), and occipital (O1-O2) electrodes. The mean amplitude was calculated for each segment of 60 s (sampled at 256 Hz), after rectification. For power calculations, we applied a Fast Fourier Transform algorithm, as implemented in Brain Vision Analyzer, employing a 10% Hanning window and using segments of 60 s, resulting in a spectral resolution of 0.016 Hz. Before and after pyridoxine-IV, we compared relative power in each frequency band. Relative power calculation was performed according to the formula (Prelative [f1, f2] = P[f1, f2]/P[0.50] × 100%) (Teune et al., 2007), where P[f1, f2] is the total power in the frequency band between f1 and f2 Herz (Lopes da Silva, 2005).

Statistical analysis

Before and after pyridoxine-IV, we applied Wilcoxon signed-rank test to assess EEG alterations (for amplitude, total power, and relative power). We applied the Mann-Whitney U-test to compare quantitative EEG alterations (regarding amplitude, total power, and relative power) between PDE and non-PDE subgroups. We stratified results for vigilance state (i.e., sleeping or waking) by the Mann-Whitney U-test. The level of significance was set by α = 0.05.

Results

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References

Visual EEG assessment

We observed a flattening of the EEG response (i.e., isoelectrical trace) and attenuation of epileptic activity in two of six neonatal PDE trials and in one of four non-PDE trials. Both 15 min before and after pyridoxine-IV, visual assessment did not reveal diagnostic EEG features that can identify PDE or non-PDE in a sensitive or specific way. The recording of the aEEG in the PDE neonate showed a continuous normal voltage pattern with normal sleep cycles. Until 48 h after pyridoxine administration, we did not observe a clear aEEG response.

Quantitative analysis of digital EEG recordings

Amplitude

Digital EEG responses to pyridoxine-IV administration were assessed at central, centrotemporal, frontal, and occipital electrodes. In eight TRS neonates, EEG amplitudes declined at central, centrotemporal, and frontal electrodes (p < 0.05) [eight trials; C3-C4 median −20% (−82% to −3%; Fig. 1A); C3-T3: −21% (−78% to −2%); C4-T4: median −17% (−76% to −5%); F3-F4: median −6% (−64% to −2%)]. After subdivision according to PDE and non-PDE, EEG amplitude alterations similarly declined in both PDE [central decline in PDE: median −30% (−78% to −3%; Fig. 1B); and in non-PDE: median: −20% (−45% to −12%; Fig. 1C)]. At occipital electrodes, EEG amplitude alterations did not change significantly. Regarding EEG amplitude alterations, stratification according to vigilance levels did not show different responses between PDE and non-PDE.

image

Figure 1.   Decline in EEG amplitude by pyridoxine-IV in TRS infants (p < 0.05). Central amplitude responses to pyridoxine-IV are indicated for infants with TRS (A), PDE (B), and non-PDE (C). On the horizontal axis, administration of pyridoxine IV is indicated. On the vertical axis, central EEG amplitude alterations are expressed as percentile changes against individual reference values (of 100%). Orange (solid) lines indicate alterations in PDE neonates. Blue (interrupted) lines indicate alterations in non-PDE neonates. The median decline is indicated by an arrow (solid line), at the right side of AC. The interindividual variability of EEG responses to pyridoxine-IV prohibits distinction between PDE and non-PDE.

Download figure to PowerPoint

Total power

In all neonates with TRS, we separately assessed the effect by pyridoxine-IV on total power in delta-, theta-, and beta-frequency bands (Fig. 2). This revealed a decline in total power for all frequency bands at central, centrotemporal, and frontal electrodes (median declines −21%, −17%, −17%, respectively; p < 0.05). Subdivision according to PDE and non-PDE did not show different responses in total power between both groups [total power decline at C3C4 medians (ranges) in: (1) delta-band, PDE: −31% (−77% to −1%) and non-PDE: −16%, (−43% to −5%); (2) theta-band, PDE: −27% (−73% to −13%) and non-PDE: −28% (−29% to −17%); and (3) beta-band, PDE: −35% (−56% to −8%) and non-PDE: −26% (−54% to −8%)]; Fig. 2A–C. At occipital electrodes, total power did not decline in either of the groups. Stratification according to vigilance levels did not reveal different outcomes in total power between PDE and non-PDE.

image

Figure 2.   Decline in total power by pyridoxine-IV in TRS infants (p < 0.05). At central electrodes (C3C4), individual alterations are expressed for delta-, theta-, and beta-frequency bands (AC). At the horizontal axis, administration of pyridoxine IV is indicated. At the vertical axis, central total power alterations are expressed as percentile changes against individual reference values (of 100%). Orange (solid) lines indicate alterations in PDE neonates. Blue (interrupted) lines indicate alterations in non-PDE neonates. At the right side of AC, medians and ranges are separately indicated for PDE and non-PDE. In all frequency bands, total-power alterations by pyridoxine-IV overlap and do not distinguish between PDE and non-PDE.

Download figure to PowerPoint

Relative power

In neonates with TRS (both PDE and non-PDE), pyridoxine-IV did not affect relative power (percentages before versus after pyridoxine for delta-, theta-, and beta-band: in PDE: 50% versus 48%; 33% versus 33% and 16% versus 17% and in non-PDE: 57% versus 59%; 26% versus 26% and 15% versus 14%, respectively). Relative power did not differ significantly between neonates with PDE and non-PDE. Regarding relative power alterations, stratification according to vigilance levels and pyridoxine dosages did not show different responses between PDE and non-PDE.

Discussion

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References

In neonates with TRS, we aimed to determine whether the EEG response to pyridoxine-IV can identify PDE by ALDH7A1 gene mutations. To the best of our knowledge, quantitative pyridoxine-IV EEG responses have not been systematically investigated before in a homogeneous neonatal PDE cohort. Present results indicate that pyridoxine-IV induces EEG alterations that are neither highly sensitive nor specific for PDE by ALDH7A1gene mutations.

The ALDH7A1 gene encodes for the enzyme α-aminoadipic semialdehyde (α-AASA) dehydrogenase, which is involved in lysine degradation. α-AASA dehydrogenase deficiency causes increased levels of α-AASA in body fluids (Mills et al., 2006). α-AASA is in spontaneous equilibrium with its cyclic form, that is, l-delta(1)-piperideine-6-carboxylate, and the latter is able to “trap” pyridoxal-5-phosphate, resulting in pyridoxal-5-phosphate deficiency (Mills et al., 2006). Pyridoxal-5-phosphate is a cofactor involved in many metabolic pathways of the brain (Goto et al., 2001). Its function as cofactor for the enzymatic conversion of (excitatory) glutamate into (inhibitory) γ-aminobutyric acid (GABA) (Haenggeli et al., 1991; Baxter & Aicardi, 1999; Baxter, 2001b) could explain why pyridoxine supplementation may attenuate seizures in general (i.e., in non-PDE patients). However, it is indicated that cerebrospinal fluid (CSF) glutamate and/or GABA concentrations are not unanimously associated with the pyridoxine response (Goto et al., 2001). In accordance with this observation, present digital EEG data did not reflect enhancement of specific GABA-ergic characteristics after pyridoxine-IV administration [such as for instance increased beta activity (Whittington et al., 2000)]. Hopefully, future studies may further clarify the underlying mechanism for seizure reduction by pyridoxine.

In neonatal PDE, direct EEG alterations by pyridoxine-IV are characterized by attenuation of seizure activity and flattening of the trace (Nabbout et al., 1999; Hellstrom-Westas et al., 2002). However, in the present study, four of six PDE neonates lacked such a direct EEG pyridoxine-IV response. Interestingly, one of the non-PDE neonates showed an EEG response that appeared characteristic for PDE. This specific child had previously been diagnosed with PDE, until normal urinary α-AASA levels and absence of ALDH7A1 gene mutation excluded the diagnosis. All together, during pyridoxine-IV, visual EEG analysis of the EEG-response to pyridoxine-IV can neither identify nor exclude PDE. Regarding the quantitative EEG responses, we observed a trend consisting of larger median % amplitude and % total power declines in PDE than in non-PDE infants. However, due to the large interindividual overlap in outcomes, individual identification of neonatal PDE among non-PDE is prohibited. One of the strengths of the present study is that five of six PDE neonates were diagnosed with the same homogeneous genotype. Therefore, we cannot attribute the large interindividual variability in EEG responses to different phenotype–genotype relationships. With the presently available data, one may only speculate whether the same results would have been expected in other affected neonates with different genotypes. Although this question has to be carefully addressed in a later study, the large interindividual overlap between pyridoxine-IV EEG responses in neonates with ALDH7A1 gene mutations appears suggestive for a nonspecific effect, that is, regardless of the genotype. Given the preceding, we hypothesize that the EEG response to pyridoxine-IV is rather dependent upon other, nonspecific, factors such as medication, gestational age, encephalopathy, comorbidities, and factors other than the genotype itself.

We are aware that the small number of PDE patients may provide a limitation to this study. However, since PDE is a rare condition, we included all six neonates (derived from the Dutch PDE cohort) with an available pyridoxine-IV EEG trial. Despite this relatively small number of patients, large interindividual variations in EEG responses already hampered identification of PDE among non-PDE.

In conclusion, the present data show that pyridoxine-IV causes nonspecific EEG responses that neither identify nor exclude PDE caused by ALDH7A1 gene mutations. These data implicate that neonates with TRS should continue to receive pyridoxine until PDE is fully excluded by biochemical and/or genetic analysis.

Acknowledgments

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References

The authors would like to acknowledge Lucas Dijck for his help with preparing digital EEGs for analysis and Saskia Houterman for statistical advice.

Disclosure

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References

The authors confirm that they have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. None of the authors has any conflict of interest to disclosure.

References

  1. Top of page
  2. Summary
  3. Patients
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. Disclosure
  9. References