Serum Albumin Levels as a Predictive Biomarker for Low-Load Resistance Training Programs’ Effects on Muscle Thickness in the Community-Dwelling Elderly Japanese Population: Results From the Interventional Study

Shuji Sawada Juntendo University Hayao Ozaki Tokai Gakuen University Toshiharu Natsume Tokai University Daiki Nakano Ritsumeikan University Pengyu Deng Juntendo University Toshinori Yoshihara Juntendo University Takuya Osawa Japan Women's College of Physical Education Hiroyuki Kobayashi Mito Medical Center, Tsukuba University Hospital Shuichi Machida (  machidas@juntendo.ac.jp ) Juntendo University Hisashi Naito Juntendo University

Results A low-load resistance training program signi cantly increased muscle thickness at the AT. The interaction between time and groups was examined at low (<4.1 g/dL) versus normal (≥4.1 g/dL) serum albumin (Alb) levels.
Although there was no difference in muscle thickness at the AT before the training intervention, the hypertrophic effects were higher in the normal serum Alb level group than in the low serum Alb level group. The binomial logistic regression analysis showed that participants in the low serum Alb group had an odds ratio of 7.08 for decreased muscle thickness at the AT. The effect of a low-load resistance training program on lower limb muscle thickness appears to be limited in participants with low serum Alb levels before training interventions.
Conclusions Serum Alb level may act as a biomarker to predict the effects of low-load resistance training programs on muscle hypertrophy in elderly individuals.
Trial registration This study was retrospectively registered in UMIN-Clinical Trial Registry (CTR), ID: UMIN000042759. Date of registration: 14/12/2020. Background Skeletal muscle plays an important role in physical functions related to the activities of daily living [1] and in systemic energy regulation controlled by glucose and lipid metabolism [2]. Despite playing a major role in systemic homeostasis, skeletal muscle size and strength both decrease with age [3,4]. This age-related loss of muscle mass and muscle strength, called sarcopenia, is related to a wide range of chronic disorders [5][6][7][8], motor disorders [9,10], and mortality [11]. Because of this, sarcopenia is a primary therapeutic target for improving the effects of aging in the elderly, and it has been reported that exercise intervention, especially involving resistance training, is a key approach for preventing and improving sarcopenia [12,13].
Resistance training has been recommended as an effective method against the problems indicated above, even in older adults [14,15]. Although high-load resistance training is the primary method to reduce the effects of aging, this type of training generally requires high external loads, which may be di cult for older adults to conduct safely. To identify an appropriate approach for older adults, we found that a low-load resistance training program could induce muscle hypertrophy in middle-aged and older adults [16]. However, despite subjecting each participant to the same training program, the training effect solely depended on the individual. It is well known that blood parameters are effective screening tools for identifying the risk of metabolic syndromes, locomotive syndromes, and other multi-factorial diseases [17][18][19]. To diagnose diseases early, annual medical checkups that include complete blood counts and blood biochemistry panels have been conducted in Japan. The consultation rate of annual medical checkups was reported to be 63.7% [20], which indicates the percentage of Japanese citizens managing their health through annual checkups. From this viewpoint, we focused on blood parameters, including complete blood counts and blood biochemistry panels, and hypothesized that these blood parameters might affect skeletal muscle adaptation in response to a low-load resistance training program. To test this, we rst comprehensively veri ed the blood parameters, including complete blood count and blood biochemistry, which entailed annual medical checkups conducted in Japan, before and after a low-load resistance training program.
This study aimed to evaluate whether the levels of some blood parameters in uenced the effect of a low-load resistance training program on lower limb muscle thickness in community-dwelling middle-aged and elderly individuals.

Participants
Healthy, community-dwelling Japanese middle-aged and elderly individuals were recruited to participate in this study through printed advertisements in the public information magazine. They were informed about the methods, procedures, and risks and provided written informed consent before participating. We excluded individuals who did not follow our instructions or those with medical conditions that limited their ability to participate in the resistance training program. Sixty-nine participants aged 69.4 ± 6.5 years (49 women and 20 men) volunteered for this study. Their heights, weights, muscle thicknesses, and blood parameters were evaluated before (pre) and after

Training program
Participants were instructed to engage in a low-load resistance training program using their own body weight and an elastic band. They were also instructed to avoid changing their dietary patterns throughout the training period.
The program was conducted twice a week for 12 weeks, and the total number of classes was 22 or 23. The program was composed of nine exercises: squats, split squats, push-ups, heel raises, crunches, hip lifts, seated rows, shoulder presses, and arm curls. The rst six exercises involved the participant's own body weight, and the last three exercises required an elastic band (Thera-Band®; The Hygenic Corporation, Akron, OH, USA). In the rst 2 weeks of the training period, the program was composed of only 4 exercises: squats, push-ups, crunches, and hip lifts, and the participants performed 3 sets of 8 repetitions with a 60-second rest between each set. In each repetition, they were instructed to spend 3 seconds in both the concentric and eccentric phases. After the rst 2 weeks, the number of exercises per session, repetitions, sets per exercise, and exercise times were gradually increased, and the rest interval was gradually decreased every 2 weeks throughout the 12-week training period. This training program was conducted according to the protocol followed in a previous study, and the details are

Muscle thickness
Each participant's muscle thickness was measured with a B-mode ultrasound device using a 5 to 18-MHz scanning head (Noblus; Aloka, Tokyo, Japan). We evaluated the anterior aspects of the thigh (AT) at the midpoint between the greater trochanter and lateral condyle of the femur. The participants were required to rest in the sitting position for at least 30 minutes before the measurement and to be in a supine position during the measurement.
This measurement was also conducted according to the method used in a previous study, and the details are 102 mg/dL, HbA1c = 5.3%, and FBS = 95 mg/dL. Using criteria for each blood parameter, participants were divided into the rst quartile group, which is the lowest quartile group, as participants in this group have relatively low blood parameter values, or into the combined second, third, and fourth quartile groups with participants having normal blood parameter values.

Statistical analyses
To verify the training effect on muscle thickness at the AT, we analyzed the parameters before (pre) and after (post) the training program. Data were analyzed using the paired t-tests, two-way analysis of variance, analysis of covariance, and binomial logistic regression analysis. If there was a signi cant interaction in the two-way analysis of variance between time and group, the simple main effects analysis was conducted. In the binomial logistic regression analysis and the multiple regression analysis, the variable selection was conducted using a stepwise method with p = 0.20. A p value less than 0.05 was considered statistically signi cant. Results were expressed as means and standard deviations, and the odds ratio was represented by means and 95% con dence intervals.
Statistical analyses were performed using BellCurve for Excel (Social Survey Research Information Co., Ltd., Japan).

Results
The participants' average height was 156.8±7.6 cm, and the average weight was 59.1±10.3 kg. The average participation rate of the classes was 87.4% (range: 63.6%-100.0%).
The average muscle thickness at the AT of all participants was 26.65±5.12 mm before the training intervention, which signi cantly increased to 29.48±5.38 mm after undergoing the biweekly low-load resistance training program for 12 weeks (p<0.001, Figure 1A). The average serum Alb concentration at the AT of all participants was 4.25±0.25 g/dL before the training intervention, which signi cantly increased to 4.32±0.26 g/dL after the training program (p<0.001, Figure 1B). Using the blood parameter criteria, we checked the rst quartile value of each blood parameter to establish the cutoff criteria for reduced levels (  Figure 2A). To verify the effect of participation rate of the classes, an analysis of covariance was conducted. Although regression parallelism was con rmed, the regression was not signi cant. Time × group interactions were also detected in the MCHC and serum LDL-C levels (p<0.001; Table 2), but no differences in muscle thickness at the AT either before or after the training intervention were observed between the low and normal groups, using the simple main effects analysis ( Figure 2B, 2C). To verify the in uence that serum Alb concentration and serum Alb levels (<4.1 g/dL or not) had on the change in muscle thickness at the AT, binomial logistic regression analysis was performed with muscle thickness at the AT (increased as 0, decreased as 1) as a target variable and with the serum Alb concentration and serum Alb level (>=4.1 g/dL as 0, <4.1 g/dL as 1) as explanatory variables. The serum Alb level was selected as a signi cant variable (p=0.0102), and the odds ratio for decreasing muscle thickness at the AT was 7.08, while the 95% con dence interval was 1.59-31.54.

Discussion
This study revealed that a low-load resistance training program using a participant's own body weight and elastic bands, even when performed only twice a week, induced muscle hypertrophy after 12 weeks of training intervention in community-dwelling elderly Japanese participants. The average muscle thickness at the AT increased about 10% over the 12 weeks of the low-load resistance training program. However, there were individual differences in the training effect on muscle thickness at the AT. For example, the increase was restricted in individuals with a relatively lower serum Alb before the training intervention. Our data suggested that low serum albumin levels may predict decreased e cacy of a resistance training intervention on muscle thickness in community-dwelling elderly Japanese participants.
It is common to divide participants using quartile values of blood parameters or based on physical performance tests [21,22]; here, we compared the lowest quartile with the three higher quartiles combined as reported in a previous study [23]. When we divided the participants into two groups using each blood parameter's quartile criteria (Table 1), an interaction between time and group was detected for MCHC levels, serum Alb levels, and serum LDL-C levels ( Table 2). The training effect on muscle thickness at the AT was limited in the low serum Alb group compared with that in the normal serum Alb group ( Fig. 2A). From the results of the analysis of covariance, this was not affected by the participation rate of the classes. The results of the binomial logistic regression analysis showed that classi cation in the low serum Alb group had an odds ratio of 7.08 for decreasing muscle thickness at the AT, which supports the above-mentioned limitation. Although this criterion was much higher than that of clinical malnutrition (< 3.5 g/dL), the patients with relatively low serum Alb had inhibited training effects.
This in uence was not seen in participants with lower levels of the other blood parameters, including MCHC and LDL-C ( Table 2, Fig. 2B, Fig. 2C). Our data demonstrated that the effect of a low-load resistance training program on lower limb muscle thickness appears to be limited in participants, depending on their serum Alb levels before the training intervention. Serum Alb is a clinical indicator of energy and protein de ciency; therefore, our data suggest that participants with lower serum Alb before a training intervention ought to improve their nutritional status to obtain the most optimal training effects on their muscle mass. The relatively lower serum Alb level (3.9-4.2 g/dL) was also reported to be related to cognitive decline or dementia [24][25][26]. Some other previous longitudinal observational studies reported that relatively lower serum Alb levels were associated with loss of muscle mass and muscle strength [22,27,28]. It was also reported that a relatively lower serum Alb level combined with sarcopenia would increase disability risk in older adults [29]. As these studies and our data showed, even though the serum Alb level is much higher than that of clinical malnutrition (< 3.5 g/dL), relatively lower serum Alb level might be an indicator for our healthy aging.
Measuring muscle thickness via a B-mode ultrasound device is a non-invasive method to assess the muscle mass, which can also be assessed in individual parts. The training effects are known to appear in a site-speci c manner, making this a superior method to assess the training effects on muscle mass. Some RCT studies have revealed that exercise intervention improved muscle strength, physical function, or muscle mass in older adults.
Binder et al. provided information on an exercise intervention program implemented 3 times a week, including lowto moderate-intensity resistance training for community-dwelling elderly and frail men and women for 9 months; they revealed that such a training intervention induced greater increments in whole-body fat-free mass assessed by dual energy X-ray absorptiometry [30]. Kim et al. discussed the bene ts of an intervention program provided for 60 minutes twice a week, including low-to moderate-intensity resistance training using the participants' own body weight, ankle weight, and resistance bands for community-dwelling sarcopenic elderly women for 3 months; they revealed that the training intervention induced increased leg muscle mass assessed by segmental multifrequency bioelectrical impedance analysis [31]. However, these studies targeted frail or sarcopenic older adults and did not adequately verify if exercise intervention could improve muscle mass in community-dwelling healthy older adults. Our data is therefore important, as it reveals that a low-load resistance training program is effective at increasing the site-speci c muscle mass of community-dwelling middle-aged and older adults.
Our study had some limitations. First, there was no control group when we veri ed the effects of the 12-week training intervention. Second, even though it has been reported that a combination of exercise and a nutritional approach is the most effective method to improve sarcopenia, we did not control the participants' nutrient intake during the training period. Third, we cannot verify the sex difference when examining the association between serum Alb levels and the low-load resistance training effects. To minimize these limitations, further studies must be conducted in the future.

Conclusions
In conclusion, serum Alb levels may be a predictive biomarker for the effect of a low-load resistance training program on muscle hypertrophy in community-dwelling elderly individuals.

List Of Abbreviations
AT the anterior aspects of the thigh; WBC:white blood cell count; RBC:red blood cell count Availability of data and materials The datasets during and/or analyzed during the current study available from the corresponding author on reasonable request.  Comparison of changes in muscle thickness for (A) low serum Alb and normal serum Alb levels, (B) low MCHC and normal MCHC levels, and (C) low serum LDL-C and normal LDL-C levels. Data were analyzed using two-way analysis of variance; the main effects on time, but not on group, and the time × group interactions were signi cant for all three measurements. Although the time × group interactions were detected in all three measurements, a signi cant difference after training intervention was only detected in serum Alb levels compared with the low and normal groups using Tukey's multiple comparison test as a post-hoc analysis. Data are presented as mean ± SD. N=69.