|
nif:isString
|
-
The Medical Ethical Committee of Maastricht University Hospital/Maastricht University (azM/UM) gave ethical clearance for this study. All participants, and for minors also both parents/caregivers, gave written informed consent.
Twenty-four adolescent patients with galactosemia and twenty-one healthy controls participated in this study. Classic galactosemia was diagnosed by GALT enzyme activity assay and/or GALT-gene mutation analysis. Two participants (both patients) were excluded because of difficulties executing the ERP task. Patient characteristics can be found in Table 1. Of the remaining 22 patients, 15 were female and 7 male, mean age 14.9 years (SD 2.2 y, range 10.8–19.1 y). The control group consisted of 14 females and 7 males mean age 14.2 years (SD 1.8 y, range 11.4–17.0 y). Neither gender nor age differed significantly between the groups [F(1,41) = .01, p = .92 and F(1,41) = 1.07, p = .31, respectively]. Participants had no other relevant health conditions, all had normal or corrected to normal vision, and were native Dutch speakers.
Table data removed from full text. Table identifier and caption: 10.1371/journal.pone.0052826.t001 Galactosemia patient characteristics. GALT enzyme activities indicate that all patients have the classic galactosemia type. Urine galactose and galactitol levels indicate adequate dietary compliance.1GALT activity was measured at diagnosis;2In case the GALT activity is not reported, it was confirmed by the treating physician to be severely decreased;3ND = not detected;4Urine levels were measured within three months of testing;5At some point in life;6Q188R/L195P (n = 4) or Q188R/S135W (n = 1);7L195P/K229N (n = 3) or 400Tdel/unknown (n = 2).
The Rey Osterreith Complex Figure was used to assess visuo-motor skills (Copy subtest), short term visual memory (Immediate Recall) and long term visual memory (Delayed Recall and Recognition) [44]. The Bourdon-Vos test was used to measure sustained attention skills (mean reaction time [RT]) [45]. The Digit Span (Forward and Backward) addressed verbal working memory skills [46].
Language paradigm during EEG recording: Visually animated scenes were presented to the participants. Each scene consisted of three geometrical shapes (square, triangle, or circle) having one of three different colours (red, blue or green). In each trial, one of the three geometrical figures performed an action upon another figure (one figure moves towards or bumps into another figure; described by either ‘to fly towards’ or ‘to bump into’). Participants were asked to either passively watch the scene (control task, ‘C’) or to describe the animated scene overtly using one of two possible responses that varied in syntactic complexity: using separate words, ‘W’ (e.g., “triangle”, “red”, “square”, “green”, “to bump into”; minimal syntactic planning) or using sentences, ‘S’ (e.g., “The red triangle bumps into the green square.”; sentence-level syntactic planning) [38], [40]. Participants were asked to keep the naming format of the phrases constant over trials. In the word ‘W’ naming format, lexical access of words is required, but virtually no syntactic encoding. In the sentence ‘S’ naming format, in contrast, syntactic encoding is required on local noun phrase level (e.g., inflection of adjectives) and on sentence level (e.g., inflection of the verb, determination of the word order, constructing and filling in of the syntactic frame). The control (‘C’) condition was added in this study to receive relevant information for the required non-linguistic resources (e.g., visual processes, attention).
The study was conducted in two sessions. In the first session, the neuropsychological tests were conducted in all participants after explanation and written informed consents were given (by the participant and both parents/caregiver). In the second session, the language paradigm and EEG recordings took place. After a brief explanation, participants were prepared and seated in an electrically-shielded, sound-attenuated room in front of a computer monitor. The session started with the control task ‘C’, followed by instructions and a practice version of the language task (consisting of 18 practice trials per condition) and the main language experiment. The main language task consisted of three runs in a blocked design. Each run comprised two blocks which were randomized within the run and counter-balanced between participants to exclude order effects. Each block started with a brief instruction reflecting the expected naming format (i.e., either ‘SENTENCE’ or ‘WORD’) followed by 32 trials of different scene displays, of which the content (figures, colours, action and arrangement) was randomized. Per condition and participant, a total of 96 trials were recorded. The control task consisted of three consecutive runs, having a total of 108 trials. Figure 1 gives a schematic overview of the sequences of events within a trial. The duration of animation in the scene differed (955 or 1885 ms) depending on the action format (‘to fly towards’ or ‘to bump into’, respectively). The difference in animation durations is caused by a different amount of action frames (10 versus 18 frames, where the actual ‘bump’ event occurred at frame 14, at 1520 ms after scene onset). Note that the movements in the scenes are visually identical until they diverge at the moment the ‘to fly towards’ trials freeze while ‘to bump into’ trials continue. Participants were instructed to start the description as fast and accurate as possible. The next trials started via a self-paced button push (USB-keyboard key), except for the control trials which had a fixed 2000 ms interval between trials. Control trials had approximately the same duration as the linguistic trials.
Figure data removed from full text. Figure identifier and caption: 10.1371/journal.pone.0052826.g001 Overview of the sequences of events within trials.Timing of events within an experimental trial, separated for the two action formats (‘to fly towards’ and ‘to bump into’). Time is displayed upwards. The block started with the instruction cue (‘WORD’ or ‘SENTENCE’), a fixation cross, a ready sign, and a randomized sequence of trials. For each trial type screenshots are displayed to illustrate the actual moving time period of the objects along with the moments of expected response of the participant, and the corresponding ERP epochs of interest (time-locked to scene onset and the bump event, respectively).
The EEG recording was done using an elastic cap in which 32 tin electrodes were mounted (Electro-Cap International (ECI), Inc.), positioned according to the international 10–20 system [47]. Twenty electrodes - F3, Fz, F4, FC3, FCz, FC4, C3, Cz, C4, CP3, CPz, CP4, P3, Pz, P4, O1, Oz, O2, T3 and T4 - were measured as active leads, AFz was used as the ground electrode. The left mastoid (A1) was used as online reference. Offline the signal was re-referenced to the average signal of both mastoids. Vertical eye movements and blinks were monitored by two electrodes placed at the left upper and lower orbital ridge. Horizontal eye movements were recorded with electrodes placed on the left and right cantus. The impedance of all electrodes was kept below 5 kΩ. Data acquisition was done using Brain Vision Recorder software (Brain Vision, MedCaT B.V.) and the signal was amplified using a 0.05–50 Hz band pass and sampled at a 500 Hz interval. The scene onset as well as the voice onset triggered a TTL pulse directly into the EEG recordings. The voice onset pulse was initiated whenever the sound pressure level reached a certain threshold (individually adjusted to each subject) and was transferred via a microphone.
The number of errors and self-corrections were computed using the recorded audio data and manual (online) scores. Errors were defined as any deviation from the expected utterance (i.e., incorrect figure, colour, action, naming format or ordering). Self-corrections were defined as any overt corrective effort during the response utterance. The voice onset time (VOT) was determined as the time between the scene onset and the onset of the voice response; the total speech time (TST) was cautiously estimated as the time between the onset of the voice response and the button push indicating when participants were ready to continue. VOTs<0.5 seconds and >4.5 seconds and TSTs<2 seconds and >10 seconds were considered outliers and discarded from analysis. The neuropsychological data were standardized using norm data and classified according to the guidelines of Lezak [48]. A repeated measures General Linear Model was used to analyze the behavioural data (VOT, TST, errors and self-corrections) having Condition (‘W’ versus ‘S’) as the within-subject factor and Group (patients, controls) as between-subject factor. The standardized neuropsychological data were analyzed using frequency tables (for the classified data) and univariate GLM to examine group differences. With respect to the EEG data, trials in which the participant's response was absent were excluded from analysis. The EEG data were epoched from −200 to 2500 ms post scene onset (to include the entire interval from onset of visual scene to the end of the display/onset of articulation), band-pass filtered from 0.3–30 Hz (zero phase, 24 dB) and baseline corrected (from −200 to 0 ms). Large visual artefacts were removed. In addition, data were decomposed using the infomax Independent Component Analysis (ICA) in EEGlab [49]. This method disentangles brain- and artefact-related processes by searching for maximally independent components [50]. Stereotype artefact-related components reflecting eye movements, noise and muscle activity were subsequently removed. On average, 84.5% of all trials (SD 5.2%) were kept for analysis [no difference between groups, F(1, 41) = .000, p = .988]: mean 96 trials in ‘C’, 79 in ‘W’ and 78 in ‘S’. The remaining components (the cleaned data) were back-projected into the ERP. In the back-projected ERPs, epochs were divided in two time ranges: one interval related to the scene onset (−200 to 1000 ms after scene onset), and one related to the bump event (−200 to 800 ms after the bump event, or 1320 to 2320 ms post scene onset) (see also Figure 1). Note that in the bump epoch, only ‘to bump into’ trials were included (and no ‘to fly towards’ trials), corresponding to on average 49 trials in ‘C’, 39 in ‘W’ and 40 in ‘S’. The bump epochs were baseline corrected (−200 to 0 ms after the bump event). Based on visual inspection of the grand averages, target peak ERP components and corresponding time windows were specified on which we conducted mean amplitude analyses. ERP statistics were performed on the mean amplitude data per time window, condition, and participant using repeated measures GLM with Condition as within-subjects factor (‘C’, ‘W’, ‘S’), and two within-subject topographical factors Laterality (left, central, right) and Anterior-Posterior (F, FC, C, CP, P, O). Based on visual inspection, additional analyses were performed on subsets of electrodes. Group was added as the between-subjects factor (patients, controls). Pearson's correlations were used to examine the relationship between the ERP data and behaviour (on-line measures of reaction times and accuracy) and other cognitive functions (off-line neuropsychological tests); and with patient characteristics (e.g., mutation, rest activity of the enzyme). Where necessary, corrections were made for multiple testing (Bonferroni) and for sphericity violations (Greenhouse Geisser). Age and gender were added as covariates in all analyses but the ones performed on standardized data. An alpha of 0.05 was used as significance level.
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}