Subjects

A total of 111 subjects including 88 elderlies (age: 61~ 83 years) and 23 young (age: 19~ 26 years) participated in this study. The subjects living in a nursing home were included in the elderly group. The young subjects were male university students and not participated in any competitive sports events. Twenty-seven females were included in elderly group.

The subjects gave written informed consent for the study after receiving a detailed explanation of the purposes, potential benefits, and risks associated with participation. All procedures used in this study were approved by the Committee for Human Experimentation at the Graduate School of Human and Environmental Studies, Kyoto University and the Research Ethics Committee of Chukyo University, and were in accordance with the Declaration of Helsinki.

Experimental design

To familiarize themselves with the motor tasks used in the present study, all subjects came to the laboratory > 1 week before the experimental day. The subjects performed maximal voluntary contraction (MVC) and submaximal incremental ramp contraction during unilateral isometric knee extension. External force at the distal portion of the shank that fixed in a custom-made dynamometer (Takei Scientific Instruments Co., Ltd., Niigata, Japan) with a force transducer (LU-100KSE; Kyowa Electronic Instruments, Tokyo, Japan) was measured during isometric knee extension. During knee extension, both hip and knee joint angles were flexed at 90° (180° corresponds to full extension). Since European consensus on definition and diagnosis used this knee joint angle [6], the present study selected this knee joint angle.

The MVC trial consists of a gradual increase in knee extension force from baseline to maximum in 2–3 s and a plateau phase at maximal contraction for 2–3 s. The timing of the task was based on a verbal count given at 1-s intervals with vigorous encouragement from the investigators. After three submaximal trials at approximately 50, 70, and 90% of MVC as warming up, the subjects performed two MVC trials with ≥ 2 min rest in between. MVC force was mean value of 1 s, when the highest force was produced, during a plateau phase. MVC torque was calculated as the product of MVC force and the distance between the estimated knee joint center and center of force transducer at the distal portion of the shank. To normalize difference in body type among the subjects, MVC torque relative to body mass (MVC/BM) was also calculated and used for further analysis.

After MVC trials, the subjects performed submaximal ramp contractions from 0 to 70% MVC in 35 s (rate of force increase: 5% MVC/s) (Fig. 1). Target and performed forces were shown to the subjects on a monitor. Two submaximal contractions were performed with ≥ 2 min rest in between. Out of two trials, the one trial with the smaller error between the targeted and performed forces was selected with visual inspection and used for analysis.

Multi-channel surface EMG

Surface EMG signal was recorded from the VL muscle with a semi-disposable adhesive grid of 64 electrodes (ELSCH064R3S, OT Bioelectronica, Torino, Italy). The two-dimensional electrode grid is comprised of 13 rows and 5 columns of electrodes with 1 missing electrode in the upper left corner. Each electrode is 1-mm diameter and inter electrode distance was 8-mm inter-electrode distance in both directions (Fig. 1). Before attaching the electrode, the skin was cleaned with alcohol. The line between the head of the great trochanter and inferior lateral edge of the patella which was defined and used as reference line for positioning the electrode grid. The center of electrode grid was positioned at middle point of the reference line and columns of electrodes aligned with the reference line. The position of the missing electrode was located proximally. To assure proper electrode contact with the skin, conductive gel was inserted into the cavities of the grid electrode. The grid electrode was connected to the amplifier through 4 connectors which were fixed to the skin by elastic tape. A reference electrode (C-150, Nihon Kohden, Tokyo, Japan) was placed at the iliac crest.

Fifty-nine bipolar surface EMG signals along the columns were made from 64 electrodes. Root mean square (RMS) values were calculated from rom bipolar signal sampled over 1 s at 5% increment from 20 to 65% of MVC ramp contraction. Sampled signals were overlapped by 0.5 s between neighboring contraction levels, since performed ramp rate was 10% of the MVC force / 1 s.

At the center of electrode location, longitudinal ultrasonographic images (SSD-900, ALOKA, Tokyo, Japan, or Fazone CB, Fuji-film, Tokyo, Japan) were taken to measure the thickness of the subcutaneous tissue and VL muscle before attaching the electrode grid. On NIH ImageJ software, the vertical line of the image was defined on the center of horizontal axis for measurements. Distance between the intersections of that vertical line and deeper and superficial aponeurosis of the VL muscle were measured as muscle thickness of VL. Thickness of subcutaneous tissues was distance between the intersections of that vertical line and superficial aponeurosis of the VL muscle and skin surface.

Statistics

All data are provided as the mean and standard deviation. To test relationship between muscle strength and neuromuscular function, spearman’s rank correlation coefficient was calculated for MVC with multi-channel surface EMG variables and others for the elderly and young groups, respectively. We also performed a stepwise regression analysis for MVC in the elderly group. Nine independent variables, i.e., age, muscle thickness of VL, thickness of subcutaneous tissues, and six surface EMG variables, were entered the stepwise regression if they represented a significant contribution to the explained variance corresponding to an alpha level of p < 0.05. Six EMG variables were modified entropy, correlation coefficient value, mean RMS across all 59 channels, standard deviation of RMS across all 59 channels, mean RMS across all 59 channels normalized by those at 20% of MVC (normalized RMS), and standard deviation of mean RMS across all 59 channels normalized by those at 20% of MVC (normalized standard deviation) and these variables calculated at 65% of MVC were used for correlation analysis.

To compare neuromuscular function among the elderly with different muscle strength, 88 elderly subjects were divided into three groups based on MVC using cumulative frequency distribution, i.e., (Weak) < 33.3%, (Mid) 33.4~ 66.6%, and (Strong) > 66.7%, and the weak and strong groups were used for further analysis. Age, height, body mass, MVC, MVC/BM, muscle thickness of VL, and thickness of cutaneous tissues, and modified entropy, correlation coefficient value, and mean RMS across all 59 channels at from 20 to 65% of MVC were compared between the weak and strong strength groups by using Mann-Whitney test. EMG variables for the elderly groups with weak and strong strength were compared with young subjects by using Mann-Whitney test.

We also compared modified entropy, correlation coefficient value, and mean RMS across all 59 channels at from 20 to 65% of MVC between the elderly and young groups which are matched for MVC. MVC/BM, body mass, muscle thickness of VL, and thickness of cutaneous tissues were also matched between these groups.

The level of significance was set at p < 0.05. Statistical analysis was performed using SPSS (version 15.0, SPSS, Tokyo, Japan) and MATLAB (R2009b, MathWorks GK, Tokyo, Japan).

Correlation analysis

There were significant correlations with MVC in muscle thickness of VL (p < 0.001) (Fig. 2 left), RMS (p = 0.001) (Fig. 2 right), and normalized RMS (p = 0.003) in the elderly group (Table 1).

		R	P
MVC	Age	− 0.111	0.324
	MT	0.524	< 0.001	*
	ST	−0.150	0.182
	ME	0.056	0.622
	RMS	0.361	0.001	*
	NRMS	0.327	0.003	*
	CRR	−0.180	0.108

MVC, maximal voluntary isometric knee extension joint torque; MT, Muscle thickness of vastus lateralis; ST, subcutaneous tissue thickness; ME, modified entropy of surface electromyography at 65% of the maximal voluntary contraction; RMS, root mean square value of surface electromyography for all channels in electrode grid at 65% of the maximal voluntary contraction; NRMS, root mean square of surface electromyography for all channels in electrode grid at 65% of the maximal voluntary contraction normalized by that at 20% of the maximal voluntary contraction; CRR, correlation coefficient value in root mean square of surface electromyography between 20 and 65% of the maximal voluntary contraction

* indicates significant correlation

MVC = 34.619 x muscle thickness of VL + 349.554 x RMS + 8.921 x normalized RMS.

The correlation coefficient R, R² and adjusted R² for this model was 0.664, 0.441, and 0.419, respectively. Also, there were no multicollinearity (Variance Inflation factor < 10.0, for all) among the selected variables, meaning that the selected variables are independent.

Comparisons among different muscle strength in elderly

Elderly with < 91.5 Nm (n = 27) and > 113.9 Nm (n = 28) of MVC were selected as weak and strong strength groups. There were significant differences in height, body mass, MVC, MVC/BM, and muscle thickness of VL (p < 0.05). No significant differences were found in RMS, correlation coefficient value, and modified entropy across all 59 channels at from 20 to 65% of MVC between the elderly groups with weak and strong strength (p > 0.05) (Figs. 3, 4 and 5).

Comparisons between elderly and young groups

In comparisons to young group for multi-channel surface EMG parameters, significant differences in RMS were observed for the weak and strong groups at 45–65% of MVC (p < 0.05) (Fig. 3). In correlation coefficient with 20% of MVC, there were significant differences with young group for the weak strength group at 35%, 40%, and 50–65% of MVC and for the strong strength group at 30% and 40% of MVC (p < 0.05) (Fig. 4).

Ten elderly men (67.6 ± 3.9 ys) and nine young men (21.1 ± 1.1 ys) were selected as the MVC -matched elderly and young groups. Between the MVC-matched elderly and young groups, significant differences were seen in RMS at 20% and 25% of MVC (p < 0.05). There were no significant differences in modified entropy and correlation coefficient value of surface EMG (p > 0.05).

Positive correlation between muscle strength and surface EMG amplitude in the elderly

In the elderly group, muscle thickness, RMS, and normalized RMS were significantly correlated with MVC (p < 0.05) (Table 1) and in stepwise regression analysis muscle thickness, RMS, and normalized RMS were selected as determinants of MVC (Table 3). Also, stepwise regression analysis shows high standardized regression coefficient in RMS (0.288, p = 0.001) in addition to MT (0.519, p < 0.001) (Table 3). Although significant correlation was not observed, tendency to correlate between MVC and RMS was also shown in young (p = 0.054) (Table 2). These results suggest that the variables related to neuromuscular function are also major determinants of muscle strength in addition to indicators of muscle volume such as muscle thickness [16, 31, 42, 43].

Since voluntary muscle contraction is regulated by central nervous system, it would be reasonable to conclude that neuromuscular function contributes to individuals’ muscle strength. EMG amplitude variables, i.e., RMS and normalized RMS, mainly reflect number of recruited motor units and its firing rate during voluntary contraction. Decreases in number of motor unit [24, 33] and in motor unit firing rate [34] with ageing are well known. These changes would induce decrease in EMG amplitude in the elderly. Merletti et al. [26] reported a significantly lower averaged rectified value of surface EMG of biceps brachii muscle during submaximal isometric contractions in elderly compared with young [26]. On the other hand, age-related decrease in EMG amplitude may not be uniformly manifested among elderly individuals. Power et al. [33] demonstrated that the estimated number of motor units in tibialis anterior muscle were significantly lower in elderly than young, but not in master runners [33]. In our recent study that used a decomposition of motor units action potential from multi-channel surface EMG, positive correlation between MVC and motor unit firing rate was demonstrated in the VL muscle for elderly [38]. Variation in EMG amplitude among individuals and relationship between EMG amplitude and muscle strength in the elderly may be brought about by these inter-individual differences in age-related alterations for anatomy and/or regulation of motor units [24, 33, 34, 38]. However, we should note that surface EMG is strongly influenced by geometric or non-physiological parameters [12, 27]. When amplitude variables are compared among the individuals, differences in the thickness of the subcutaneous tissue should be considered [28]. To quantify the effect of inter-individual differences in the subcutaneous tissue on relationship between EMG amplitude and muscle strength, we additionally calculated partial correlation coefficient between RMS and MVC adjusted for the thickness of the subcutaneous tissue in the elderly. Adjusted correlation coefficient (r = 0.291, p = 0.009) was slightly lower than the value before adjustment (r = 0.361, p = 0.001). While this means that relationship between MVC and RMS was influenced by the thickness of the subcutaneous tissue, significance in correlation coefficient value is unchanged. We thus assumed that the thickness of the subcutaneous tissue was not critical effect on our results.

Relationship between muscle strength and estimated motor unit recruitment pattern

In the present study, different manners in multi-channel surface EMG parameters were observed between the elerly with weak and strong strength when compared to the young subjects (Fig. 4). Significant differences in correlation coefficient values with the young subjects were noted at high force levels in the elderly with weak strength (< 91.5 Nm) (p < 0.05), but not in the elderly with strong strength (> 113.9 Nm) (p > 0.05) (Fig. 4). Also, as the results of comparison between the young and elderly under the condition that MVC were matched, no significant difference was observed in correlation coefficient values at all contraction levels (p > 0.05). Correlation coefficient value between RMS from same channels between 20% of MVC and other force levels was used to assess changes in spatial distribution of surface EMG within a muscle in the present study. Holtermann et al. [18] showed that alteration pattern of correlation coefficients during ascending and descending phases during ramp contraction was not the same [18]. Since a deviation between recruitment and derecruitment of motor units was reported [18], the correlation coefficient of multi-channel SEMG could be reasonably used to estimate recruitment of global motor units during force production. From the results in the present study, it can be interpreted that changes in spatial distribution patterns of RMS from 20% to 65% of MVC were lesser in the elderly with weak muscle strength and were similar in the elderly with strong muscle strength when compared to the young. Based on our assumption, motor unit recruitment pattern of the young may be different from the elderly, which is corresponds to the conclusion in our previous study [39], but not from the elderly with high muscle strength. The results of some previous studies had pointed out that importance of neural factors in muscle strength for elderly. Previous studies reported that neural activation is more adaptable for the resistance training in elderly than young [5, 19, 30]. The results of the present study could support this notion and suggests that neural factor should be also evaluated for understanding age-related muscle strength loss in addition to indicator of muscle volume.

In the young subjects, significant correlation with MVC was noted only in modified entropy of multi-channel surface EMG (Table 2). Lower modified entropy means greater heterogeneity of spatial distribution of surface EMG amplitude within an electrode grid. Our previous study demonstrated lower modified entropy at 65% of MVC in the young compared with the elderly [39]. From this result, we estimated that spatial inhomogeneity in location of different types of muscle fibers/motor units and/or numbers of newly recruited muscle fibers/motor units are greater in the young. Therefore, the result of the present study suggest that motor unit recruitment pattern strongly relates with the muscle strength in young and the determinants of muscle strength are not same between the young and elderly.

Limitations

The present study used a specific knee joint angle. Although the knee joint angle of 90° has been widely used in the related research area [6], the highest knee extension torque is observed at more extended knee joint angles (115°-140°) [36, 37]. We also should note that surface EMG and muscle thickness were measured only from the VL muscle in the present study. Knee extension joint torque is produced by four muscle components of quadriceps femoris, i.e., the VL, vastus intermedius, vastus medialis, and rectus femoris muscles. Neuromuscular activation pattern are not uniform among the four muscle components and their contribution to knee extension torque could be influenced by knee joit angle [32, 37]. Therefore, the results of the present study would be specific for one knee joint angle and the VL muscle. Different results may be observed at different joint angles and in other muscles. Also, the present study calculated maximal voluntary joint torque for knee extension from the estimated moment arm length at the resting condition and measured force. Since the soft tissue deforms during loading, the estimated moment arm dinamically change between the resting and contraction conditions [1]. Therefore, our calculated joint torque would not be consistent with the resultant knee joint moment. This is also the major issues of this study.