Cortical Proprioceptive Processing Is Altered by Aging

This is a self-archived version of an original article. This version may differ from the original in pagination and typographic details. Author(s): Title: Year: Version: Copyright: Rights: Rights url: Please cite the original version: CC BY 4.0 https://creativecommons.org/licenses/by/4.0/ Cortical Proprioceptive Processing Is Altered by Aging © 2018 Piitulainen, Seipäjärvi, Avela, Parviainen and Walker. Published version Piitulainen, Harri; Seipäjärvi, Santtu; Avela, Janne; Parviainen, Tiina; Walker, Simon Piitulainen, H., Seipäjärvi, S., Avela, J., Parviainen, T., & Walker, S. (2018). Cortical Proprioceptive Processing Is Altered by Aging. Frontiers in Aging Neuroscience, 10(June), Article 147. https://doi.org/10.3389/fnagi.2018.00147 2018 ORIGINAL RESEARCH published: 14 June 2018 doi: 10.3389/fnagi.2018.00147 Cortical Proprioceptive Processing Is Altered by Aging Harri Piitulainen1,2*, Santtu Seipäjärvi2,Janne Avela2,Tiina Parviainen3 and Simon Walker2 1Sensorimotor Systems Group, Department of Neuroscience and Biomedical Engineering, Aalto University School of Science, Espoo, Finland, 2Biology of Physical Activity and Neuromuscular Research Center, Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland, 3Centre for Interdisciplinary Brain Research, Department of Psychology, University of Jyväskylä, Jyväskylä, Finland Edited by: Lutz Jäncke, Universität Zürich, Switzerland Reviewed by: Rahul Goel, Baylor College of Medicine, United States Matthieu P. Boisgontier, University of British Columbia, Canada *Correspondence: Harri Piitulainen [email protected] Received: 06 March 2018 Accepted: 01 May 2018 Published: 14 June 2018 Citation: Piitulainen H, Seipäjärvi S, Avela J, Parviainen T and Walker S (2018) Cortical Proprioceptive Processing Is Altered by Aging. Front. Aging Neurosci. 10:147. doi: 10.3389/fnagi.2018.00147 Proprioceptive perception is impaired with aging, but little is known about aging-related deterioration of proprioception at the cortical level. Corticokinematic coherence (CKC) between limb kinematic and magnetoencephalographic (MEG) signals reflects cortical processing of proprioceptive afference. We, thus, compared CKC strength to ankle movements between younger and older subjects, and examined whether CKC predicts postural stability. Fifteen younger (range 18–31 years) and eight older (66–73 years) sedentary volunteers were seated in MEG, while their right and left ankle joints were moved separately at 2 Hz (for 4 min each) using a novel MEG-compatible anklemovement actuator. Coherence was computed between foot acceleration and MEG signals. CKC strength at the movement frequency (F0) and its first harmonic (F1) was quantified. In addition, postural sway was quantified during standing eyes-open and eyes-closed tasks to estimate motor performance. CKC peaked in the gradiometers over the vertex, and was significantly stronger (∼76%) at F0 for the older than younger subjects. At F1, only the dominant-leg CKC was significantly stronger (∼15%) for the older than younger subjects. In addition, CKC (at F1) was significantly stronger in the non-dominant than dominant leg, but only in the younger subjects. Postural sway was significantly (∼64%) higher in the older than younger subjects when standing with eyes closed. Regression models indicated that CKC strength at F1 in the dominant leg and age were the only significant predictors for postural sway. Our results indicated that aging-related cortical-proprioceptive processing is altered by aging. Stronger CKC may reflect poorer cortical proprioceptive processing, and not solely the amount of proprioceptive afference as suggested earlier. In combination with anklemovement actuator, CKC can be efficiently used to unravel proprioception-relatedneuronal mechanisms and the related plastic changes in aging, rehabilitation, motor-skill acquisition, motor disorders etc. Keywords: proprioception, sensorimotor cortex, aging, somatosensory, passive movement, balance, coherence, sensorimotor integration Abbreviations: CKC, corticokinematic coherence; COP, center of pressure; EEG, electroencephalography; F0, movement frequency; F1, first harmonic of the movement frequency; fMRI, functional magnetic resonance imaging; MEG, magnetoencephalography; SM1, primary sensorimotor. Frontiers in Aging Neuroscience | www.frontiersin.org 1June 2018 | Volume 10 | Article 147 Piitulainen et al. Cortical Proprioceptive Processing Is Altered by Aging INTRODUCTION The human neuromuscular system is strongly affected by aging (Larsson et al., 1979; Shaffer and Harrison, 2007; Mitchell et al., 2012; Rowan et al., 2012; Kulmala et al., 2014; Frontera, 2017). Aging-related deterioration of the motor system is demonstrated as a decline in motor performance such as loss of maximum muscle force (Frontera et al., 1991; Lindle et al., 1997), less accurate bimanual coordination (Maes et al., 2017), increased postural sway during static standing (Hasselkus and Shambes, 1975; Era and Heikkinen, 1985; Baloh et al., 1994), worsening dynamic balance control (Piirainen et al., 2013), and alterations in cortical control of upright stance (Ozdemir et al., 2018). Postural instability is related to the risk of falling (Maki et al., 1994; Pajala et al., 2008) and the number of balance loss events during perturbation of the standing posture (Boisgontier et al., 2017). Motor control relies on vast sensory input from the environment (vision, audition, touch) and especially on the internal state of the motor system through proprioception. This internal model of motor system state and thus proprioceptive control of movement are somewhat altered by aging (Boisgontier and Nougier, 2013). The proprioceptors are located in muscles and joints, and sense limb positions, movements and forces (Proske and Gandevia, 2012). Impaired proprioception has been considered as a main contributor to balance loss in older adults (Lord and Ward, 1994; McChesney and Woollacott, 2000). Proprioceptive perception (joint position sense) is impaired with aging (Skinner et al., 1984; Kaplan et al., 1985; Goble et al., 2009), and is termed as ‘‘presbypropria’’ (Boisgontier et al., 2012). Little is known about aging-related deterioration of proprioception at the cortical level, but brain responses to proprioceptive stimulation have shown significant associations to postural control (Goble et al., 2011) and proprioceptive perception (Goble et al., 2012). Corticokinematic coherence (CKC) quantifies the coupling between cortical activity, measured with magnetoencephalography (MEG) or electroencephalography (EEG) and limb kinematics (e.g., acceleration) during repetitive rhythmic active (Jerbi et al., 2007; Bourguignon et al., 2011) and passive (Piitulainen et al., 2013a, 2015; Smeds et al., 2017) movements. CKC peaks at the movement frequency and its harmonics, and it can be measured using various peripheral movement-related signals and motor tasks (Piitulainen et al., 2013b). CKC can be used for non-invasive functional cortex mapping to identify the human primary sensorimotor (SM1) cortex (Bourguignon et al., 2011, 2013). CKC primarily reflects proprioceptive processing in the SM1 cortex with negligible effect of cutaneous afference (Piitulainen et al., 2013a; Bourguignon et al., 2015). CKC corresponds to the timing of the strongest deflection of the cortical passive-movement-evoked field with an apparent latency of 50–100 ms (Piitulainen et al., 2015). However, it is not known how the strength of CKC is related with the efficiency of the proprioceptive processing, and whether the level of CKC is associated with motor performance. This relationship could be presumed if the motor tasks rely primarily on proprioception, e.g., when performing a balance task with eyes closed. As CKC reflects cortical proprioceptive processing it could potentially be used to unravel proprioception-related-neuronal mechanisms both in healthy and clinical populations. Previous MEG studies, where passive movements have been generated by pneumatic cylinders (Alary et al., 2002) or pneumatic artificial muscles (Piitulainen et al., 2015), have focused on small-scale finger or toe movements only. In order to test for the associations between CKC and functional performance, such as locomotion or balance, CKC should be applied on lower limbs at the functional range of joint rotations. Here, we introduce a novel MEG-compatible ankle-movement actuator, and use it to quantify CKC during continuous dorsi-plantarflexion movement in a similar manner as done previously for hand, finger and hallux movements (Bourguignon et al., 2011; Piitulainen et al., 2013a, 2015). The movement actuator is nonmagnetic, and is therefore both MEGand functional magnetic resonance imaging (fMRI) compatible. DCmotor-based movement actuators have been applied with EEG recordings (Desmedt and Ozaki, 1991; Mima et al., 1996; RamosMurguialday et al., 2012), but unfortunately DC-motors are not MEG-compatible. Our primary aim was to study the effect of aging on CKC strength evoked by passive movements of dominant and non-dominant leg. We also aimed to study whether CKC predicts postural stability in a task relying on both vision and proprioception (standing eyes-open) or on proprioception alone (standing eyes-closed). We hypothesized that older subjects will show weaker CKC than the younger subjects, indicating impaired cortical proprioceptive processing. We also hypothesized that weaker CKC strength (i.e., worse cortical proprioceptive processing) would be reflected by higher postural sway during eyes-closed standing (Baloh et al., 1994), and thus CKC strength would be a significant predictor of the degree of postural sway. It is important to note that habitual physical activity levels, body mass and height were matched between the groups, and thus our results emphasize age-related effects. MATERIALS AND METHODS Subjects Fifteen younger (mean age 24.8 years, range 18–31 years, height 171 ±11 cm, body mass 69 ±12 kg; four males) and eight older sedentary volunteers (mean age 69.3 years, range 66–73 years, height 167 ±9 cm, body mass 73 ±15 kg; five males) were studied. They did not report any history of movement disorders or neurological disease, and were not taking medication affecting the central nervous system. The subjects were pre-screened to confirm that there were no differences in the level of physical activity performed between groups (i.e., all subjects habitually performed less than the recommended 150 min aerobic exercise per week and none were engaged in strength training). The subjects were specifically asked to declare the type of aerobic exercise e.g., walking or cycling et cetera (performed for at least 20 min of duration) and the number of times per week that they engaged in these aerobic exercises. The same questions were used for strength training. Subjects were blind to the eligibility Frontiers in Aging Neuroscience | www.frontiersin.org 2June 2018 | Volume 10 | Article 147 Piitulainen et al. Cortical Proprioceptive Processing Is Altered by Aging criteria of the study so as not to influence their pre-study reporting. Average weekly aerobic exercise duration in younger (128 ±76 min) was similar to those of older (124 ±57 min). Their dominant leg was defined as the preferred leg to kick a ball (right leg in all subjects). This study was carried out in accordance with the recommendations of the Declaration of Helsinki, and the ethics committee of the University of Jyväskylä. The protocol was approved by the ethics committee of the University of Jyväskylä. All subjects gave written informed consent in accordance with the Declaration of Helsinki. Movement Actuator Figure 1 shows the custom-made non-magnetic anklemovement actuator. The principle of this stimulator is similar as in our previously built device to generate finger and toe movements (Piitulainen et al., 2015). A pneumatic system is embedded into a PVC-plastic frame designed to support the individual’s foot. Three pneumatic artificial muscles (DMSP10-100 AM-CM, diameter 10 mm, length of the contracting part 100 mm; Festo AG & Co., Esslingen, Germany) were attached vertically to the lower plate of the frame and the upper plate where the individual’s foot was resting. The pneumatic artificial muscles moved in a vertical direction when its internal air pressure (1–7 bar) changed, and thus rotated the upper plate. The pressure was regulated by a solenoid valve (SY52206LOU-01F-Q, SMC Corporation, Tokyo, Japan) that was controlled by computer-generated trigger pulses. The solenoid valve was placed outside the magnetically shielded room and three 3.5-m non-elastic tubes (internal diameter 2.5 mm) conveyed the airflow to the pneumatic artificial muscles. The pneumatic artificial muscles were first shortened simultaneously by increasing the air pressure (opening of the valve), thereby dorsiflexing the ankle joint, and then returned back to the initial position (plantarflexion of the ankle joint) when the air pressure was released (closing of the valve). Experimental Protocol During MEG recordings, the subjects were sitting their eyes open with the stimulated leg on the movement actuator placed on the floor in front of them. The other leg was resting on the floor. The tested foot was attached to the actuator with an elastic strap to ensure its stability. Earplugs were used and Brownian noise was played in the background via a flat-panel speaker (Panphonics 60 ×60 SSHP, Tampere, Finland) to minimize concomitant auditory noise that arose from the movement actuator. A white A3-sized paper sheet, taped vertically to the MEG gantry, prevented the subjects from seeing the moving leg. Subjects were instructed to fixate on a black dot on the wall of the magnetically shielded room, 3 m in front of them. Brisk continuous dorsiplantarflexion movement (movement range 6.6◦, peak angular velocity 80◦/s) was generated for the ankle joint at 2 Hz for 4 min in each leg separately. The stimulation order of the legs was randomized for each individual. Four minutes resting state data were collected similarly while no movements were generated. On a separate day (1–4 days after the MEG session), postural stability was measured while subjects were standing quietly, unshod on a force plate (AMTI, OR6-6 model, Watertown, MA, USA) with either eyes open or eyes closed for 2 min (continuously) in each task. Typically a minimum of 90 s is required for reliable bipedal postural sway recordings (Ruhe et al., 2010). The subjects held their hands together in front at hip level and were asked to stand as still as possible. Feet were placed hip-width apart with toes facing directly forward. During the eyes-open task, the subjects gaze was fixated on a black dot at eye level 3 m in front. After a short break (<1 min), during which time the subject’s feet remained in the same position, the second task was performed. Measurements MEG The measurements were carried out at the Jyväskylä Centre for Interdisciplinary Brain Research. MEG signals were recorded in a magnetically shielded room (Magnetical Shielding Cabin, VACOSHIELD, Vacuumschmelze GmbH & Co. KG, Hanau, Germany) with a 306-channel whole-scalp neuromagnetometer (Elekta NeuromagrTRIUXTM, Elekta Oy, Helsinki, Finland). The recording passband was 0.1–330 Hz and the signals were sampled at 1 kHz. The individual’s head position inside the MEG helmet was continuously monitored by feeding current to five head-tracking coils. The coils were attached on the scalp prior to measurement and their locations were determined with respect to anatomical fiducials with an electromagnetic tracker (Fastrak, Polhemus, Colchester, VT, USA). Acceleration and Trigger Signals Hallux acceleration was recorded with a 3-axis accelerometer (ADXL335 iMEMS Accelerometer, Analog Devices Inc., Norwood, MA, USA) attached on the skin over the metatarsal bone. Acceleration was low-pass filtered at 330 Hz and sampled at 1 kHz, time-locked to MEG signals. Stability of the peak acceleration magnitude was quantified as the coefficient of variation for the peak value of the Euclidian norm of the three orthogonal accelerometer signals across all dorsiflexions separately for each individual and for left and right legs. Data Processing Preprocessing Continuous MEG data were first preprocessed off-line using temporal signal-space-separation with head movement compensation to suppress external interferences and to correct for head movements (Taulu and Simola, 2006). The MEG and acceleration signals were band-pass filtered offline at 0.4–195 Hz. Coherence and Power Analysis For coherence analyses, the continuous data were split into 2-s epochs with 1.6-s epoch overlap, leading to a frequency resolution of 0.5 Hz (Bortel and Sovka, 2007). MEG epochs with magnetometer signals >3 pT or gradiometer signals >0.7 pT/cm were excluded to avoid contamination by eye movements and blinks, muscle activity, or external MEG artifacts. We then performed coherence analysis (Halliday et al., 1995)—yielding cross-, powerand coherence-spectra, as well as crosscorrelograms—between MEG signals and the Euclidian norm of Frontiers in Aging Neuroscience | www.frontiersin.org 3June 2018 | Volume 10 | Article 147 Piitulainen et al. Cortical Proprioceptive Processing Is Altered by Aging FIGURE 1 | Ankle-movement actuator and representative signals. (A) Technical drawing of the actuator. (B) Individual’s right foot rested on the foot rest while repetitive ankle flexions were generated. (C) Ankle-angle signal while the ankle-movement actuator operated at 2 Hz. (D) Representative time-locked magnetoencephalographic (MEG) and accelerometer signals as a function of time when the right ankle of Subject 1 was moved at 2 Hz. Rows from top to bottom 1–120-Hz and 1–10-Hz MEG (from the most responsive channel), and Euclidian norm of the three orthogonal acceleration signals. The gray vertical lines indicate the onsets of dorsiflexion. the three orthogonal accelerometer signals. Before the coherence analysis, each epoch of acceleration was normalized by its Euclidian norm (Bourguignon et al., 2011). Peak CKC strength was quantified as the strongest coherence value across all MEG gradiometers at the movement frequency (F0; 2 Hz) and its first harmonic (F1; 4 Hz) separately from the respective coherencespectra. To study the possible effect of background-MEG power on CKC strength, the power spectra were computed for restingstate-MEG recording for the same gradiometer that showed the peak CKC value at F0 and F1. This power analysis was otherwise identical to the coherence analysis, but flat-topweighted windows were used to allow power comparisons across the subjects. From the power spectra, power values for F0 and F1 were extracted and were further analyzed. Cortical Movement-Evoked Fields MEG signals were averaged with respect to the movement onsets of the ankle dorsiflexions to compute peak-to-peak-amplitude of the sustained-movement-evoked field for both legs, i.e., the strength of cortical activity related to the continuous 2-Hz movement. Cortical Source of CKC Cortical source for CKC was approximated from spatial distribution of cross-correlograms and the corresponding magnetic field patterns. No further detailed source localization was performed, since subjects’ anatomical MRIs were not available. The cross-correlograms were first band-pass filtered at 1–40 Hz and the source analysis was performed in the time domain, on the spatial distribution of the filtered Frontiers in Aging Neuroscience | www.frontiersin.org 4June 2018 | Volume 10 | Article 147 Piitulainen et al. Cortical Proprioceptive Processing Is Altered by Aging cross-correlogram, as previously done in CKC studies (Bourguignon et al., 2011, 2013; Piitulainen et al., 2013b). Equivalent current dipole was estimated within the spherical head model at the most prominent peak of the filtered cross-correlogram, using a fixed selection of 84 sensors (52 gradiometers and 26 magnetometers) over the vertex. The most prominent peak of the cross-correlogram depicts the time-point of strongest coherence between two signals. Postural Stability Anterior-posterior (x) and medio-lateral (y) force signals of the force plate were sampled at 1000 Hz and low-pass filtered offline at 20 Hz. Postural sway was quantified by first computing xy-magnitude of center of pressure (COP) distance (mm) from sample-to-sample, and then multiplying it with the sampling frequency (1000 Hz) to obtain the mean velocity (mm/s) of the COP for both tasks separately. This measure combines both mediolateral and anteroposterior postural sway, and thus provides a measure to estimate overall postural stability performance. The effect of vision for the postural sway was quantified by subtracting the postural sway of the eyes-open task from the eyes-closed task, i.e., indicating the increase in the postural sway due to closing the eyes. Statistical Analysis All statistical analyses were performed in IBM SPSS Statistics software (ver. 24). Data was first ensured to have normal distribution using the Shapiro-Wilk test. Independent-samples t-test was used in between-group (younger vs. older) comparisons for CKC strength, number of averages, sustained field peak-to-peak amplitude, acceleration magnitude and postural sway. Paired-samples t-test was used to compare CKC strength, number of averages, sustained field peak-topeak amplitude, MEG power and acceleration magnitude for dominant and non-dominant legs, and to compare postural sway between eyes-open and eyes-closed tasks. Pearson correlation coefficient was computed across all subjects between CKC strength, and MEG response amplitude and background-MEG power to estimate the effect of the MEG signal strength to the CKC strength. Linear multiple regression analysis was performed to estimate the importance of CKC strength (for nonand dominant legs and F0 and F1 separately), body mass, height, gender, age, group (younger, older) on predicting postural stability (during eyes open and closed conditions, and their difference) using the enter method in SPSS. Results are indicated as mean ±standard deviation. RESULTS Figure 1 illustrates the ankle-movement actuator, and MEG and acceleration signals of a representative individual (Subject 1) during the continuous 2-Hz movement. The actuator did not produce notable artifacts in the MEG signals, and thus the strong fluctuations at the movement frequency reflect the cortical processing of proprioceptive afference. The acceleration signals contained two clear peaks for each movement cycle, reflecting the initial phases of the dorsiand plantarflexions. Peak acceleration magnitude (for dorsiflexion) was 3.0 ±0.36 m/s2 for the dominant foot (coefficient of variation across movements 6.4 ±4.7%) and 3.1 ±0.46 m/s2for the non-dominant foot (coefficient of variation 5.0 ±2.4%). There were also no differences in the peak acceleration magnitude between feet (p= 0.548) or groups (younger 3.12 ±0.16 m/s2vs. older 3.08 ±0.35 m/s2,p= 0.806). No significant differences were observed in body mass (p= 0.579), height (p= 0.422) or habitational level of physical activity (p= 0.874) between the groups. Corticokinematic Coherence Younger vs. Older Figure 2A shows the MEG–acceleration coherence spectra superimposed for all subjects. Coherence showed a clear peak in all subjects and both legs at the movement frequency (F0) and its first harmonic (F1). Due to the regular movement, several harmonics are visible. Figure 2B shows peak CKC strength for dominant and non-dominant legs at F0 and F1. CKC ranged from 0.06 to 0.81 (mean ±SD, F0: 0.26 ±0.13; F1: 0.32 ±0.15), peaking at the gradiometers over the vertex. At F0, CKC was ∼76% stronger for the older than younger subjects when the legs were pooled together (older 0.36 ±0.14 vs. younger 0.20 ±0.09, p= 0.001). CKC was stronger in older than in younger subjects for the dominant (∼69%, older 0.33 ±0.13 vs. younger 0.20 ±0.09, p= 0.007) and non-dominant (∼83%, older 0.38 ±0.15 vs. younger 0.21 ±0.09, p= 0.003) legs separately as well. At F1, CKC was not significantly different between the groups when the legs were pooled together (older 0.38 ±0.15 vs. younger 0.29 ±0.10, p= 0.077) or for the non-dominant leg (older 0.37 ±0.12 vs. younger 0.33 ±0.13, p= 0.433). However, CKC was ∼60% stronger for the older than younger subjects for the dominant leg (older 0.39 ±0.21 vs. younger 0.25 ±0.11, p= 0.035). Dominant vs. Non-dominant Leg At F0, CKC strength did not differ between the legs in younger (dominant 0.20 ±0.09 vs. non-dominant 0.21 ±0.09, p= 0.650) or older (dominant 0.33 ±0.13 vs. non-dominant 0.38 ±0.15, p= 0.310) subjects, or when the groups were pooled together (dominant 0.24 ±0.12 vs. non-dominant 0.27 ±0.14, n= 23, p= 0.263). However, at F1, in the younger subjects, CKC was stronger (p= 0.037) in the non-dominant (0.33 ±0.13) than dominant leg (0.25 ±0.11). Number of Epochs, Response Amplitude and Background-MEG Power The number of epochs in the coherence analysis did not differ between the legs (dominant: 560 ±74 vs. non-dominant: 572 ±40, p= 0.411) or groups (younger: 574 ±46 vs. older: 551 ±51, p= 0.275). Sustained field peak-to-peak amplitude generated by the 2-Hz movement did not differ between the legs (dominant 24.4 ±8.0 fT/cm vs. non-dominant 23.9 ±6.8 fT/cm, p= 0.784) or groups (younger 24.1 ±5.6 fT/cm vs. older 24.2 ±6.1 fT/cm, p= 0.958). Similarly, no differences were observed in the Frontiers in Aging Neuroscience | www.frontiersin.org 5June 2018 | Volume 10 | Article 147 Piitulainen et al. Cortical Proprioceptive Processing Is Altered by Aging FIGURE 2 | Individual corticokinematic coherence (CKC) spectra and group averages. (A) CKC spectra showed clear peaks at movement frequency (F0) and its first harmonic (F1) for dominant and non-dominant ankle movements. Maximum coherence between the acceleration signal and MEG signal of the sensor showing the peak CKC at F0 is shown. (B) CKC at F0 and F1 for dominant and non-dominant legs in younger and older groups. Error bars indicate range. Vertical line indicates median. Horizontal boundaries of the boxes indicate quartiles. background-MEG power between the legs at F0 (dominant 131 ±77 fT/cm2/Hz vs. non-dominant 131 ±88 fT/cm2/Hz, p= 0.983) and at F1 (dominant 71 ±36 fT/cm2/Hz vs. non-dominant 70 ±36 fT/cm2/Hz,p= 0.849), or the groups at F0 (younger 146 ±85 fT/cm2/Hz vs. older 103 ±43 fT/cm2/Hz, p= 0.361) and at F1 (younger 73 ±40 fT/cm2/Hz vs. older 65 ±28 fT/cm2/Hz,p= 0.980). The sustained field peak-to-peak amplitude or background-MEG power did not correlate with CKC strength or postural stability. Cortical Source of CKC Figure 3 shows the cross-correlograms for all gradiometers of the MEG sensor array and the respective magnetic field patterns for Subject 1. Equivalent current dipole was estimated at the time of the most prominent peak of the cross-correlogram, and showed clear dipolar field patterns centered on interhemispheric sensors near the central sulcus. Similar patterns were observed in all subjects. The applied fixed selections of MEG sensors are outlined. Postural Stability Figure 4 shows superimposed COP distributions during standing with eyes open and closed for two older subjects with clearly different balance performance. COP distributions increased especially in the older subjects when they closed their eyes. On average, postural sway was higher during the eyes-closed task for younger (eyes open 11.14 ±1.74 mm/s vs. eyes closed 14.32 ±2.43 mm/s, p<0.001) and older (eyes open 13.82 ±4.53 mm/s vs. eyes closed 23.48 ±11.56 mm/s, p= 0.016) subjects. Between-group comparisons showed that postural sway was ∼64% higher for the older than younger subjects while standing with eyes closed (p= 0.007), and the sway increased more from the eyes-open to eyes-closed task in older than younger subjects (∼29% in younger 3.18 ±2.09 mm/s vs. ∼70% in older 9.66 ±7.71 mm/s, p= 0.006). During eyes-open standing, postural sway was ∼24% higher for the older than younger subjects, but did not reach the level of statistical significance (p= 0.055). Predicting Postural Stability Using the enter method, a significant regression models emerged for postural stability explaining 68% (adjusted R square = 0.68) of postural sway during eyes-open (F(9,12)= 2.89, p= 0.045) and 72% during eyes-closed (F(9,12)= 6.97, p= 0.001) condition, and 71% of their difference (F(9,12)= 6.66, p= 0.002). Table 1 presents standardized Beta coefficients and p-values for each predictor variable of the linear regression models. Significant Frontiers in Aging Neuroscience | www.frontiersin.org 6June 2018 | Volume 10 | Article 147 Piitulainen et al. Cortical Proprioceptive Processing Is Altered by Aging FIGURE 3 | The cross-correlograms for all gradiometers and the corresponding magnetic field patterns superimposed on the MEG sensor array for Subject 1 during non-dominant (left panels) and dominant (right panels) ankle movements at 2 Hz. The pre-selected subsets of sensors are outlined. Magnetic field patterns were obtained from Equivalent current dipole estimation at the main peak of the cross-correlogram. The red isocontour lines indicate the flux out of the skull and the blue lines flux into the skull. The arrow depicts the surface projection of the equivalent current dipoles orientation. variables were CKC at F1 for dominant leg (eyes open: beta 0.722, p= 0.011; eyes closed: beta 0.911, p<0.001; difference: beta 0.903, p<0.001) and age (only for difference: beta 2.371, p= 0.036). Body mass, height, gender, group and CKC at F0 or at F1 in non-dominant leg were not significant predictors in these models. DISCUSSION Our results indicated that cortical proprioceptive processing is affected by aging. Older subjects showed stronger CKC for ankle movements (contrary to our hypothesis) and higher postural sway in line with previous evidence (Hasselkus and Shambes, 1975; Era and Heikkinen, 1985; Baloh et al., 1994). The degree of postural sway was predicted by CKC strength, and also with subjects’ age when the postural task shifted from eyes-open to eyes-closed condition, i.e., when relying primarily on proprioception. The differences in CKC and postural sway may be explained by age-related motor deterioration, especially since the groups did not differ in habitual physical activity levels. Little is known about age-related proprioceptive deterioration, especially at the cortical level, however cortical control of upright stance seems to be altered due to impaired somatosensory processing in older subjects (Ozdemir et al., 2018). CKC could potentially be used as a tool to unravel this topic, and even as a ‘‘biomarker’’ for motor and somatosensory impairments. Indeed, stronger CKC seems to suggest lower motor performance, skill or level of usage, as the non-dominant Frontiers in Aging Neuroscience | www.frontiersin.org 7June 2018 | Volume 10 | Article 147 Piitulainen et al. Cortical Proprioceptive Processing Is Altered by Aging FIGURE 4 | Postural stability. (A,B) Superimposed center of pressure (COP) distributions during standing eyes open (black dots) and closed (gray dots) for one stable (A) and unstable (B) older subject. (C) Magnitude of postural sway for eyes open and closed tasks. (D) Difference in mean magnitudes of postural sway between eyes closed and open tasks. Error bars indicate range. Vertical line indicates median. Horizontal boundaries of the boxes indicate quartiles. TABLE 1 | Beta coefficients from linear regression model to predict postural stability. COP eyes open COP eyes closed COP difference Predictor variable Beta p-value Beta p-value Beta p-value Weight −0.240 0.401 0.064 0.751 0.229 0.276 Height 0.480 0.176 0.253 0.307 0.093 0.709 Age −0.544 0.701 1.429 0.173 2.371∗0.036∗ Group 0.636 0.645 1.092 0.278 1.937 0.071 Gender 0.484 0.247 0.523 0.090 0.479 0.123 CKC F0 dominant −0.186 0.525 −0.098 0.637 −0.036 0.866 CKC F0 non-dominant 0.660 0.052 0.281 0.222 0.030 0.895 CKC F1 dominant 0.722∗0.011∗0.911∗∗∗ 0.000∗∗∗ 0.903∗∗∗ 0.000∗∗∗ CKC F1 non-dominant 0.244 0.353 −0.155 0.404 −0.363 0.071 ∗p<0.05, ∗∗∗ p<0.001. leg showed stronger CKC among the younger subjects. Finally, the novel ankle-movement actuator led to a strong coherence between kinematics and cortical MEG signals, in accordance with earlier findings of finger and toe movements (Piitulainen et al., 2015). CKC in Younger and Older We have shown previously that CKC primarily reflects proprioceptive processing in the SM1 cortex with negligible effect of cutaneous input when measured with MEG (Piitulainen et al., 2013a; Bourguignon et al., 2015), and thus CKC has Frontiers in Aging Neuroscience | www.frontiersin.org 8June 2018 | Volume 10 | Article 147