Subjects
Fifteen healthy young adults (8 men, 7 women) participated in this experiment. Mean (standard deviation (SD)) age, height, weight, foot length (FL), and auditory threshold were 22.7 (4.7) years, 166.9 (8.3) cm, 60.0 (8.8) kg, 24.4 (1.4) cm, and 28.0 (4.6) dB, respectively. No subject had any history of neurological or orthopedic impairment. Informed consent was obtained from all subjects following an explanation of the experimental protocols, which were approved by the ethics committee at Kanazawa University (No. 946).
Apparatus and data recording
A force platform (FPA34; Electro Design, Nagareyama, Japan) was used to measure the center of pressure in the anteroposterior direction (CoPap). CoPap signals were sent simultaneously to one computer (PC9801BX2; NEC, Tokyo, Japan) to determine CoPap position online and to another computer for analysis offline. The former received CoPap data via an A/D converter (PIO9045; I/O-Data, Kanazawa, Japan) at 20 Hz with 12-bit resolution and could generate a buzzer sound when CoPap was within ±1 cm of the position for the quiet standing (QS) posture. During QS, the frequency of body sway is below 5 Hz; especially, the main frequency is below 1.5 Hz [29, 30]. Therefore, the 20-Hz sampling frequency is commonly used in the studies on the postural sway [31]. CoPap position was calculated as the percentage distance from the heel in relation to FL (%FL). The platform was fixed to a handmade table that was movable horizontally in an anteroposterior direction by a linear motion guide actuator (SKR4610A-0290-1-1001; THK, Tokyo, Japan) with a computer-controlled electric motor (SANMOTION model No. PB PBBR604; Sanyo Denki, Tokyo, Japan). The direction, velocity, and amplitude of platform movement were adjusted by the motor. S1 was an auditory stimulus delivered via earphones at a frequency of 2000 Hz, 35 dB above the threshold and lasting 50 ms. S2 was a forward floor translation. Floor translation was detected by an accelerometer (AS-2GB; Kyowa, Tokyo, Japan) fixed to the platform.
The position of the body in the sagittal plane was recorded using the Position Sensor System (C5949; Hamamatsu Photonics, Hamamatsu, Japan). This system comprises a sensor head and light-emitting diode (LED) targets and emits analog outputs of the coordinates of the LED targets in two dimensions. The sensor head was placed 4 m from the left side of the subject. LED targets were placed over the platform and the following landmarks on the left side: vertebra prominens (C7); midpoint of the greater trochanter (hip); lateral condyle of the femur (knee); and lateral malleolus (ankle). The x- and y-coordinates of LED targets were recorded at a 0.3-mm resolution.
Ag-AgCl cup electrodes (8-mm diameter) for recording EEG were placed on the scalp at Fz, Cz, and Pz in accordance with the international 10-20 system, and referenced to linked ear lobes. A ground electrode was placed at Fpz. Electrooculography (EOG) was recorded from a pair of electrodes placed above and below the right eye. To fix the eye position, subjects were instructed to gaze at a fixation point presented on an eye-trek face-mounted display (FMD011F; Olympus, Tokyo, Japan). Surface electrodes (P-00-S; Ambu, Ballerup, Denmark) were used in bipolar derivation to record surface electromyography (EMG) of the following muscles on the right side: rectus abdominis (RA), erector spinae (ES), RF, biceps femoris (BF), TA, medial head of gastrocnemius (GcM), and soleus (Sol). For each muscle, electrodes were fixed after shaving and cleaning the skin with alcohol. The electrodes were aligned along the long axis of the muscle with an inter-electrode distance of about 3 cm. Electrode input impedance was <5 kΩ. Signals from electrodes were amplified (EEG, ×40000; EOG, ×4000; EMG, ×4000) and band-pass filtered (EEG, 0.05–100 Hz; EOG, 0.05–30 Hz; EMG, 5–500 Hz) using an amplifier (Biotop 6R12; NEC-Sanei, Tokyo, Japan). In many CNV studies with 2-s inter stimulus interval, the high-pass filter around 0.05 Hz has been used [32–35].
For subsequent analyses, all electrical signals including CoPap, EEG, EOG, and EMG were sent to the computer for analysis (Dimension E521; Dell, Kawasaki, Japan) via an A/D converter (ADA16-32/2(CB)F; Contec, Osaka, Japan) at 1000 Hz with 16-bit resolution.
Joint fixation of the ankle
The outline of the method for ankle joint fixation is shown in Fig. 1. The lower legs and feet were fixed using aluminum frames and a wooden block mounted on the platform. While the subject maintained QS posture with the heels slightly touching the wooden block, the legs were secured with three horizontal bars of the frames, two from the front of the legs and another from the back, and two belts each for the left and right legs. The front bars had two wooden boards along with the long axis of the legs, for which the angle of inclination could be adjusted to the legs. Fixation of foot position was ensured by wrapping the ankle joint and wooden block together with another two belts for each foot. To avoid overly tight wrapping, a towel or sponge was placed between the legs and fixation tool. During these fixation processes, the buzzer sound was continually generated to indicate QS posture.
Procedure
All measurements were performed while the subject stood barefoot, with feet 10 cm apart and parallel on the force platform and the upper limbs crossed in front of the chest (Fig. 1). Mean CoPap position was initially measured for 10 s with the subject maintaining QS posture. The mean value of five trials was adopted as the QS position. Next, mean CoPap position during extreme backward leaning (EBL) was measured twice. Subjects gradually leaned backward from QS posture for approximately 5 s, pivoting at the ankles, and then maintained this EBL posture for 3 s. The more posterior mean CoPap position of two trials was adopted as the EBL mean position, and the posterior peak position of CoPap in the adopted trial was defined as the EBL peak position.
Velocity and amplitude of floor translation was set for each subject based on EBL mean and peak positions [16, 36]. To begin, a 5- or 10-cm floor translation was applied at a velocity of 10 cm/s. If the posterior peak of CoPap after translation at either amplitude was located between EBL mean position and EBL peak position, 10 cm/s was adopted as the translation velocity. If not, velocity was reduced or increased until the posterior peak at either amplitude was located between these positions (change in 5-cm/s increments). Second, a linear regression line was drawn through the two coordinates of the floor translation amplitude (5 and 10 cm) and the posterior peak of CoPap at each of the two floor translation amplitudes at the determined velocity. Based on this line, the translation amplitude at which the posterior peak would be located between the EBL mean position and EBL peak position was determined. Mean (SD) values for adopted translation velocity and amplitude were 20.0 (5.0) cm/s and 7.3 (2.3) cm, respectively.
The experimental session was carried out as follows. In both setting the translation intensity and in the experimental session, subjects maintained CoPap position within the QS position ±1 cm, as presented by a buzzer sound for at least 3 s, until S2 onset. S1 was randomly presented 1–2 s after the experimenter stopped the buzzer sound, then S2 started 2 s after S1. Subjects were instructed to avoid changing the initial foot position in response to S2. A set of 20 trials was repeatedly performed twice for each condition (early- and latter-half sets). Trials were excluded if a change in foot position was observed or if CoPap deviated more than ±1 cm from QS position before S2. Subjects were given a standing rest period of 30 s between trials and a seated rest period of 3 and 10 min between every ten trials and between fixation conditions, respectively.
Data analyses
All data analyses were performed using BIMUTAS II software (Kissei Comtec, Nagano, Japan). To evaluate the magnitude of backward disturbance in response to floor translation, the posterior peak of CoPap after S2 was identified in each trial and the distance from EBL mean position to this peak position was calculated. The mean value from all trials in each set was defined as the CoPap displaced position.
EEG, EMG, CoPap, and position sensor data from 500 ms before S1 to 3000 ms after S2 were averaged for each set. Trials with eye blinks or movement artifacts (voltage at EOG or any EEG electrode exceeding ±100 μV) between 500 ms before S1 and S2 were excluded from averaging. At least 12 trials were included in the average for each set, as the minimum number of averagings in previous CNV studies [37]. For EEG, mean amplitude for the 500-ms period before S1 was used as a baseline of averaging. Prior to averaging, EMG data were high-pass filtered at 40 Hz using a seventh-order Butterworth filter to exclude electrocardiographic and movement artifacts, then full-wave rectified. For position sensor data, waveforms of the x- and y-coordinates were first smoothed by an 89-point moving average, corresponding to a 5-Hz low-pass filter. Hip angle (knee − hip − C7), knee angle (ankle − knee − hip), and ankle angle (inclination of ankle − knee from vertical line) were then calculated from the coordinates, using Excel 2010 software (Microsoft, Tokyo, Japan). Joint angles were calculated for every data point, and the waveform was then averaged.
In the averaged waveforms after S2, joint angle and EMG were analyzed. Movement angle of each joint was defined as the difference between maximal and minimal values after S2. EMG was analyzed only in the frontal postural muscles (RA, RF, and TA), since the direction of postural disturbance was backward only and burst activity just after S2 was observed mainly in these muscles. In order to smooth this averaged EMG waveforms, an 11-point moving average corresponding to a 40-Hz low-pass filter was used with reference to the previous studies [3, 38]. The maximum peak after S2 for these muscles was identified, and peak amplitude and latency were measured relative to the baseline and S2, respectively.
For analysis in the period before S2, averaged EEG, EMG, and CoPap waveforms were smoothed by a 111-point moving average, corresponding to a 4-Hz low-pass filter. CNV peaked just before S2 and then changed positively in 12 of 15 subjects, and in the other 3 subjects, the negative potential was not clearly increased. Thus, the following analyses between S1 and S2 were performed for the 12 subjects. Mean amplitudes for every 100-ms period from 100 ms before S1 to S2 onset were then calculated. CNV analyses used averaged EEG waveforms recorded from Cz, in which late CNV was maximal in all sets. CNV can be classified into early and late components [39]. The early component has been reported as the potential between 300 and 700 ms after S1, and the late component as the negative potential, which gradually increases toward S2 following the early component [40–42]. The periods from 700 ms after S1 to S2 were thus used for the analysis of mean amplitudes for every 100 ms of CNV, EMG, and CoPap. For the mean amplitude of CoPap, the difference from the mean amplitude for the 500-ms period before S1 was also calculated, since forward deviation was observed early after S1.
A maximal negative potential identified from 1400 ms after S1 to S2 was defined as the CNV peak, and the latency relative to S2 was calculated as CNV peak time. Changing patterns of frontal muscle activity and CoPap movement around the CNV peak were analyzed as follows [20]. The 500-ms period from 700 to 200 ms before the CNV peak was defined as the base period. For EMG, a continuous increase in preparation for S2 was identified as activation exceeding 2 SD above the mean amplitude of the base period for >50 ms until S2 and the minimum value of this activation was defined as the onset point. When a continuous increase of EMG was not observed, presence of a CoPap forward shift exceeding 2 SD above the mean amplitude in the base period until S2 was checked. If such a CoPap forward shift was observed, the inflection point was identified based on the second-order differentiated waveform and the start of the transient increase in EMG activity just before the inflection of the CoPap shift was identified in the same manner as the start of the continuous increase. The time difference between the start of the increase in EMG and S2 was defined as the EMG start time before S2.
Statistical analyses
Shapiro-Wilk tests confirmed that all data satisfied assumptions of normal distribution. Two-way repeated-measures analysis of variance (ANOVA) was used to assess the main effects and interaction of condition (no-fixation, fixation) and set (early half, latter half) on analysis parameters after S2, mean amplitude of CNV, EMG, and CoPap for every 100-ms period between S1 and S2, and CNV peak time and peak amplitude. When a significant interaction was shown, paired t tests were used for post hoc comparison to investigate differences within each factor. For EMG peak latency after S2, to assess whether effects of condition (no-fixation, fixation) and set (early half, latter half) differed among muscles (RA, RF, TA), three-way ANOVA was performed first, then the post hoc Games-Howell test was used to further investigate differences among muscles. To investigate the effects of joint fixation, a paired t test was used to compare each parameter across the latter half under the no-fixation condition and the early half under the fixation condition. To investigate changing patterns of CNV and EMG waveforms in the period corresponding to the late CNV, with the mean amplitude for every 100-ms period, the difference between the period of 700–800 ms after S1 and other periods was evaluated using Dunnett’s test. A one-sample t test was used to assess whether the CoPap displaced position and CoPap mean amplitude for every 100-ms period between S1 and S2 differed significantly from the EBL mean position and mean amplitude for the 500-ms period before S1, respectively. Pearson’s correlation was used to assess the relationship between CNV peak time and EMG start time before S2. The alpha level was set at p < 0.05. All statistical analyses were performed using IBM SPSS Statistics 19 (IBM Japan, Tokyo, Japan).