Original Paper
HORMONE RESEARCH
Horm Res 2006;66:240–248 DOI: 10.1159/000095069
Received: September 9, 2005 Accepted: June 7, 2006 Published online: August 15, 2006
Effects of Early High-Dose Levothyroxine Treatment on Auditory Brain Event-Related Potentials at School Entry in Children with Congenital Hypothyroidism S. Marti a M. Alvarez c J. Simoneau-Roy b S. Leroux a G. Van Vliet b P. Robaey a Departments of a Psychiatry and b Pediatrics, Ste Justine Hospital and Research Center, Université de Montréal, Montréal, Canada; c Instituto de Neurología y Neurocirugía, La Habana, Cuba
Key Words Congenital hypothyroidism Event-related potential Brain development
Abstract Aims: We tested whether brain event-related potentials (ERPs) are normal in children with congenital hypothyroidism (CH) after early high-dose levothyroxine treatment. Methods: Auditory ERPs were recorded in 33 normal controls and in 15 children with CH at 5 years 9/12. Based on bone maturation at diagnosis, the CH group was divided into severe (n = 8) and moderate (n = 7) subgroups. CH patients were treated at a median age of 14 days with a mean initial dose of levothyroxine of 11.6 g/kg·day. Two ERP components (N100 and N200) were measured and clinical follow-up variables collected. Results: The functional anatomical and cognitive organisation of the auditory system, as revealed by the analyses of ERP measures, did not differ between CH and controls, or between severe and moderate CH subjects. However, N200 latency was globally longer in the CH than in the control group (p = 0.01) and was positively correlated with the over-treatment index (r = 0.61; p ! 0.05) and verbal IQ. N200 amplitude was negatively correlated with initial dose (r = –0.74; p ! 0.005). Conclusion: These data suggest
© 2006 S. Karger AG, Basel 0301–0163/06/0665–0240$23.50/0 Fax +41 61 306 12 34 E-Mail
[email protected] www.karger.com
Accessible online at: www.karger.com/hre
that sensitive tools such as ERPs can reveal differences between CH and controls and relate these differences to the adequacy of treatment of CH. Copyright © 2006 S. Karger AG, Basel
Introduction
High-dose (10–15 g/kg·day) levothyroxine (LT4) given very early after birth has effectively normalised global psychomotor development in our longitudinal cohort of children with congenital hypothyroidism (CH) at 18 months [1] and 5 years 9 months [2]. However, differences on subscales remain and are attributed both to the severity of the disease and to treatment-related factors [2, 3]. This paper aimed at testing the effects of early highdose LT4 treatment on auditory brain event-related potentials (ERPs) at 5 years 9 months, before entry in the elementary school. In CH evolution and treatment, four different phases may be identified, each with specific markers [4]: (a) Foetal hypothyroidism reflects both insufficient maternal supply and insufficient foetal secretion of thyroxine (T4) [5]. It results in delayed bone maturation, assessed by the area of the knee epiphyses at diagnosis. (b) Shortly after
Dr. Philippe Robaey Department of Psychiatry, Ste Justine Hospital, 3100 Ellendale Montreal, Quebec, H3S 1W3 (Canada) Tel. +1 514 345 4997, Fax +1 514 345 4635 E-Mail
[email protected]
birth, the baby no longer benefits from maternal T4 and plasma T4 levels at diagnosis mostly reflect the residual thyroid function of the newborn. The earliest possible treatment (before 14 days) is crucial [1, 2, 6, 7]. (c) Treatment with high-dose LT4 then quickly normalises plasma thyrotropin or thyroid-stimulating hormone (TSH) but induces high plasma T4 levels [8]. Negative effects of high T4 levels at 1–3 months on later behaviour have been reported by some [9] but not by others [10]. (d) Lastly, the treatment follow-up is characterised by constant monitoring and dose adjustments with transient phases of over- and under-treatment. Auditory pathways begin to develop during intra-uterine life and continue to mature after birth over a prolonged period, until the age of about 5 years [11, 12]. As T4 has a key role in the timing of neuronal development [13], this protracted period of maturation makes the auditory pathways durably vulnerable to abnormal T4 levels, related primarily to the severity of pre-natal hypothyroidism but also to treatment problems from initiation to 5 years of age (i.e., high initial dose of T4, number of over- and under-treatment episodes during long-term follow-up). Consistent with this view, adolescents with CH treated at a mean age of 16.4 8 22 days with a mean starting dose of 9.3 8 5 g/kg·day have poorer language and reading abilities, associated with more severe hypothyroidism initially [14]. In order to explore the neural basis of these verbal deficits, auditory ERPs (P300) were previously recorded by using an oddball paradigm [15]; as ERPs were similar in controls and CH children, the authors suggested that the maturation of the brain structures involved in the generation of the auditory P300 were relatively unaffected by CH. However, ERP studies in normally developing pre-school children showed that it is an ERP negativity preceding the P300 that indexes the brain-based auditory perceptual skills and predict language-related abilities at school age [16]. We therefore used two auditory ERP negativities preceding the P300 (N100 and N200) [17] in order to compare children with CH to a large number of age- and sex-matched normal control children. In addition, given rapid developmental changes in young children, age at evaluation in both groups was strictly restricted to the 5 years 8/12–5 years 10/12 range. The instruction to respond differently to two tones is difficult in 5-year-olds and is likely to affect the validity of ERP data, especially if we expect large inter-individual developmental differences. We thus designed a passive auditory oddball task using two series of the high-low tones presented at short interval and separately in the left and right ear. The child was only instructed to listen to
the sounds, while staring at a fixation point. During electroencephalography (EEG) recording, vigilance may fluctuate and affect ERPs; as brain activity is averaged over time, these covert changes are compared between subjects and are expected to be more consistent than those based on irregular overt responses that could not be reliably used in planned comparisons. Stimuli delivered separately in each ear are expected to produce ERP components of larger amplitude on the hemisphere contralateral to the ear of delivery, especially for an early negative N100 peak recorded over the temporal region [18]. This lateralisation with regard to the ear is the consequence of the predominantly crossed functional organisation of the early activation of various primary and secondary auditory cortical areas (Heschl’s gyrus, the posterior parietal cortex, the superior temporal gyrus and the planum temporale). Specifically, we tested the integrity of this functional crossed organisation by testing that tones presented in the right ear elicited a stronger activation in the contralateral left hemisphere in which the critical auditory areas are more developed [19]. Beyond the functional crossed connections from ear to brain, we also tested auditory pathway dominance effect by using a reversed stimulus pattern for the left and the right ear. In the left ear the low tones were rare, but in the right ear, it was the high tones that were rare. This pattern creates a potential conflict, as the high and low tones had the same frequency across both ears. In school age children as in adults, tones produce larger early negative ERPs when they are delivered in the left ear, as compared to the right, irrespective of the direction of voluntary attention to a specific ear [20]. This automatic attention effect is a marker of the well-established left ear-right hemisphere advantage (in right-handed subjects) to process tonal stimuli [21, 22]. This left ear advantage for tones mirrors the right ear (or left hemisphere) advantage (in right hand dominants) to verbal stimuli [23, 24]. We predicted that if a specific tone were rare in the dominant pathway, it would elicit larger ERP negativities, whether delivered in the left (dominant) or in the right (non-dominant) ear. This tone effect irrespective of ear (i.e. dominance) could possibly be in interaction with a topographical factor, along the left-right and/or the posterior-anterior axis. However, if brain attention processes are sufficiently developed to take advantage of the inter-stimulus interval to cancel the competition between ears, the reverse pattern between ears will be detected (low tones rare in the left ear, but high tones rare in the right ear) and the prediction is then a tone by ear interaction (i.e. attention effect), again alone or combined with a topographical fac-
Auditory ERP in CH Children
Horm Res 2006;66:240–248
241
tor. The analysis of the early brain ERPs according to ear and tone should thus reveal whether the children with CH differ from the controls as a function of the functional crossed connection (ear by hemisphere interaction), ear dominance (tone effect only) or attention (ear by tone interaction). The first aim of this study was thus to verify that early high-dose treatment makes the children with CH comparable to normal controls with regard to auditory brain processes, and the children with severe CH comparable to those with moderate CH. In addition, within children with CH, we tested whether inter-individual differences in the auditory ERPs could be accounted for by individual characteristics of the disease and its treatment. Moreover, we also tested if ERP inter-individual differences in the CH group could account for differences on an intelligence test or on a behavioural questionnaire.
Subjects and Methods Subjects For the present study, 15 full-term otherwise healthy infants with CH were enrolled over a period of 5 years. All subjects were born between December 1989 and December 1994, and were 5.8 8 0.1 years old at the time of evaluation. The intellectual and behavioural outcomes of these children have been described previously [2]. The criterion for severity of CH was the area of the knee epiphyses at diagnosis: severe CH: ! 0.05 cm2 (n = 2 boys, 6 girls); moderate CH: 1 0.05 cm2 (n = 2 boys, 5 girls). The mean (8 SD) age at starting treatment was 14.5 8 3.6 days of life with a mean initial LT4 dose of 11.6 g/kg·day (range: 6.5–16.8 g/kg·day). Thirty-three normal children (22 girls, same sex ratio as in the CH group) within the same age range were recruited in the same period and used for comparison. The three groups had similarly favourable family environments, as shown by their mean (8 SD) adversity scores [25]: severe 0.08 8 0.10, moderate 0.17 8 0.21, controls 0.09 8 0.16 (NS). Treatment and Follow-Up Plasma TSH and free T4 were obtained at 1.5, 3, 6, 9, 12, 18 months, and 2, 2.5, 3, 4 and 5.75 years. TSH concentrations between 0.8 and 6 mU/l can be considered as within established norms [26]. Between 3 months1 and 5 years 9/12, the number of times that TSH was below 0.8 mU/l was used as an index of overtreatment. During the same period, the number of times that TSH was above 6 mU/l was used as an index of under-treatment. Evaluation of Cognition and Development This was performed using the McCarthy Scales of Children’s Abilities [27] with the correction of Kaufman and Kaufman [28], and the Questionnaire for the Evaluation of Social Behaviour
1
TSH at 1 month was not used in the computation of these indices as, in the severe cases, TSH did not return to normal before the third month.
242
Horm Res 2006;66:240–248
(QESB) [29–31]. These results have been reported previously [2] and are used here only to test their association with ERP variables. Electrophysiology Evaluation EEG was recorded with 13 electrodes (Electro-cap International, Inc.) placed according to the 10–20 system [32] at Fz, Cz, Pz, F7/8, C3/4, P3/4, T7/8, P7/8) with linked earlobes as references, from 72 ms before stimulus to 440 ms after. Electrode impedance was kept below 5 k. Electro-oculogram (EOG) was recorded with 4 electrodes (two placed at the outer canthus of each eye and two infra- and supraorbital to the left eye). EEG signals were amplified with a band-pass between 0.1 and 30 Hz (!20,000 for EEG, !5,000 for EOG). The EEG was averaged from the time of the stimulus and was corrected for EOG artefacts in the frequency domain [33]. Before EOG correction, all EOG epochs with a clipping duration longer than 100 ms were rejected. After visual inspection, stimulus-locked averaged ERPs were digitally filtered offline with a band-pass of 0.15 and 37 Hz. We used a passive auditory condition in which 300 pure tones (150 of 1,000 Hz and 150 of 1,200 Hz) were presented separately in each ear (80 db SPL), without any instruction to the subject, except to fixate a cross on the centre of a screen. Hearing loss in CH subjects is mainly neurosensory loss in higher frequencies (around 4,000 Hz); it could not affect differentially low-frequency tones separated by 200 Hz only and is almost prevented by highdose LT4 treatment before the third week of life [34, 35]. For the left ear, the frequent stimulus (n = 120) was of 1,200 Hz while the rare stimulus (n = 30) was of 1,000 Hz. Conversely, for the right ear, the 1,200 Hz tone was rare (n = 30) and the 1,000 Hz tone was frequent (n = 120). Inter-stimulus interval varied randomly between 550 and 950 ms with 20 ms duration (rise/fall time: 2 ms). The total duration of the task was 6 min. Thirty frequent stimuli from a set of sequences (e.g. one, two or three frequents after a rare one) representative of the whole series were selected for ERP averaging in order to equate the signal-to-noise ratio. Measurements and Statistical Analyses Peak amplitude and latency measures were obtained for each electrode and each condition separately. The parieto-temporal N100 was defined as the maximum negative peak amplitude between 83 and 273 ms on parietal (P7/8) and temporal (T7/8) leads. The auditory fronto-central N200 was defined as the maximum negative peak amplitude between 190 and 360 ms on frontal (F7/8, Fz), central (C3/4, Cz) and parietal (P3/4, Pz) leads (fig. 1). All statistical analyses were conducted using the MANOVA procedure in SPSS v.10.0.7. The N100 measures were analysed using 2 separate groups ! [2 tones (1,200 vs. 1,000 Hz) ! 2 ears (left vs. right) ! 2 hemispheres (left vs. right) ! 2 electrodes (temporal vs. parietal)] mixed-design analysis of variance. The N200 measures were analysed using 2 separate groups [2 tones (1,200 vs. 1,000 Hz) ! 2 ears (left vs. right) ! 2 hemispheres (left vs. right) ! 2 electrodes (frontal vs. parietal)] mixed-design analysis of variance. We selected only the parietal and the frontal leads in order to differentiate the N200 topography. First, we tested the hypothesis that CH subjects were similar to age-matched normal controls. Given the small number of children, significant differences were confirmed by using a more conservative non-parametric strategy (Mann-Whitney U test). Second, based on the results of Dubuis et al. [1] at 18 months and of Simoneau-Roy et al. [2] at
Marti /Alvarez /Simoneau-Roy /Leroux / Van Vliet /Robaey
F7/8
C3/4
T7/8
–5 V P3/4
P7/8
Fig. 1. Grand average auditory ERPs
(across hemispheres, stimuli) showing the group differences between CH (thick line) and control (thin line) over frontal, central, temporal and parietal electrodes. Intervals for peak measurement of N100 (T7/8 and P7/8) and N200 (F7/8; C3/4 and P3/4) are indicated by boxes. Negativity is up.
–50
0
100
200
300
400 × 1 ms
Results
5 years 9 months on the same cohort, we tested the hypothesis of a lack of difference between the severe and moderate CH subjects. The third hypothesis (i.e. differences in disease severity and treatment variables accounted for inter-individual differences in ERP measures) was tested through correlations. T4 at diagnosis, initial LT4 dose, number of over- and under-treatment episodes were correlated using a non-parametric strategy (Spearman’s rank test) with ERP measures in all CH subjects and in each CH group separately (although these correlations based on a very small group should be seen as indicative only). In order to improve the reliability of the ERP mean amplitude and latency values, we selected the measures obtained for the frequent stimuli on the electrodes where each component was peaking: C3, Cz, C4 for N200, T7/8 and P7/8 for N100. N100 and N200 amplitude was further restricted for the tones presented in the left ear (auditory dominant pathway for tones). We also computed the correlations between ERP measures and scores from intelligence test and behavioural questionnaire (table 1).
Comparison between CH and Control Groups N100: N100 amplitude analysis showed an interaction between ear, hemisphere and electrode: F(1,46) = 7.04; p = 0.01). For the sounds presented in the right ear only, N100 amplitude was larger on the left hemisphere, especially over temporal leads [hemisphere ! electrode; F(1,46) = 6.51; p = 0.01]. N100 amplitude was 8.15 V on T7 and 5.61 V on T8: F(1,46) = 8.45; p = 0.01. The same pattern was found for N100 latency, with larger N100 peaking earlier. N100 latency was always shorter in the hemisphere contralateral to the stimulated ear [ear ! hemisphere ! electrode; F(1,46) = 6.72; p = 0.01]. The tones presented in the right ear evoked an earlier N100 on the left hemisphere, as compared to the right (172 vs.
Auditory ERP in CH Children
Horm Res 2006;66:240–248
243
1.0
Group Moderate Severe Both groups 95% CI
N200 amplitude (V)
0.5
rho = –0.74 p~0.005
0.0
–0.5
–1.0
–1.5 8
a
10
12
14
16
18
Initial LT4 dose (g/kg • day) 320
rho = 0.61 p~0.05
N200 latency (ms)
300
280
260 Group Moderate Severe Both groups 95% CI
240
184 ms): F(1,46) = 5.92; p = 0.02. This pattern is consistent with the functional anatomy of the auditory pathways, but did not differ between CH and control participants or between CH subgroups. N100 was larger in response to low pitch tones, and this effect differentiated normal controls from children with CH [F(1,46) = 10.40; p = 0.002]. In normal controls, but not in children with CH, the low pitch tones elicited larger N100 (6.79 vs. 4.98 V): F(1,32) = 10.16; p = 0.003. N200: As expected, we found that the N200 was sensitive to the probability of the tones in each ear, as it was larger frontally for the locally rare tones, while it was larger on parietal leads for frequent tones [ear ! tone ! electrode; F(1,46) = 5.84; p = 0.02]. Although the N200 latency was not different on the frontal and parietal leads (274 ms), the frontal N200 peaked earlier on the left hemisphere (270 vs. 276 ms), and the parietal N200 on the right hemisphere (271 vs. 278 ms): electrode ! hemisphere; F(1,46) = 9.81; p ! 0.005. These interactions reflect topographical differences in the processing of the reversed auditory pattern of stimuli, but did not differentiate the children with CH from the normal controls, or between children with severe or moderate CH (fig. 2).
220 0
b
2
4
6
8
Over-treatment index
10
Fig. 2. 2D scatter-plot of a N200 amplitude (V) vs. the initial dose of LT4 (g/kg·day), b N200 latency (ms) vs. the number of over-treatment phases between 3 months and 5 years 9/12.
Table 1. Spearman correlations between starting dose of LT4 and over- and under-treatment index (number of times TSH was 6.0 mU/l between 3 months and 5 years 9 months), the amplitude of the auditory N200 and the verbal IQ in the whole group of CH children (with results restricted to the severe CH subgroup in parentheses)
N200 Starting LT4 dose Over-treatment index Under-treatment index Verbal IQ
Amplitude Spearman’s rho –0.74 (–0.93)a 0.16 (0.59) –0.51 (–0.75)b –0.48 (–0.62)
p value