Four databases, namely PubMed, Web of Science, Embase, and Cochrane Library, were utilized for this study. The search formula was created by combining terms related to sarcopenia, resistance training, and inflammatory biomarkers. The search included publications without language or country restrictions until October 9, 2023. The search strategy combined the following medical subject headings with accessible terms and matching synonyms: ‘sarcopenia,’ ‘sarcopenias,’ ‘resistance training,’ ‘resistance exercise,’ ‘strength training,’ ‘strength exercise,’ ‘weight lifting strengthening program,’ ‘weight lifting exercise program,’ ‘weight bearing strengthening program,’ ‘weight-bearing exercise program,’ ‘inflammation,’ ‘innate inflammatory response,’ ‘c reactive protein,’ ‘CRP,’ ‘interleukin 6,’ ‘IL-6,’ ‘interleukin 10,’ ‘IL-10,’ ‘tumor necrosis factor-alpha,’ ‘TNF-α.’ The complete search strategy for each database is available in the Supplementary Material .

The inclusion criteria for this study were as per the PICOS protocol:

Furthermore, this study excluded the subsequent studies:

The literature screening was conducted using the Endnote citation management software (EndNote X9.3.3). Information extracted from the literature included the first author, year of publication, country/region, sample size, participant characteristics (gender, age, sarcopenia diagnostics), intervention elements (methods, duration, frequency, and period), control group (CG), and outcomes. The mean and standard deviation of the changes before and after intervention in both groups will be directly extracted or indirectly calculated using the sample size, median, range, or interquartile range (34–36). Two researchers (X.J. and K.L.) independently conducted the literature screening and data extraction. They consulted each other after completion and referred any disagreements to a third researcher (Z.Q.) for further discussion and decision.

In order to evaluate the risk of bias in the RCTs included in our study, we utilized the Cochrane Risk-of-Bias Tool for Randomized Interventions (RoB). This tool evaluates the risk of bias in six domains: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and selective outcome reporting. Each domain was categorized as ‘low risk of bias,’ ‘some concerns,’ or ‘high risk of bias.’ Additionally, we employed the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) approach to determine the certainty of evidence. The GRADE approach considers factors such as study design, consistency of results, directness of evidence, precision, and publication bias to assess the quality of evidence and assign a grade of strength. Utilizing the RoB tool and the GRADE approach, we thoroughly evaluated the risk of bias and certainty of evidence in the RCTs included in this study, ensuring a robust assessment of the research findings. In cases where there were disagreements between the initial reviewers (X.J. and K.L.), a third researcher (Z.Q.) was involved to resolve them.

We extracted the sample size for each study and assessed the difference in data between the experimental group and the CG before and after the intervention. This assessment included the average and standard deviation of inflammatory biomarkers to evaluate the effect of RT on the regulation of inflammatory markers in sarcopenia. We calculated the weighted mean difference (WMD) and 95% confidence interval (CI) for continuous outcomes. All outcome data were converted to a standardized unit before calculating the WMD. The units of IL-6, TNF-α, and IL-10 are pg/ml, while the unit for CRP is mg/l. We also calculated the I² statistic and 95% CI to assess heterogeneity. If the heterogeneity exceeded 50%, we used the random effects model; if the heterogeneity was below 50%, we utilized the fixed effects model. Heterogeneity was considered significant if the I² value was above 50% or the P value was below the predefined significance level (P = 0.05). Conversely, if the I² value was 50% or less and P ≥ 0.05, we did not consider the heterogeneity significant. The I² value less than 50% indicates low heterogeneity, and the I² value greater than 50% indicates high heterogeneity. P of effect size less than 0.05 was considered statistically significant. We conducted a sensitivity analysis by excluding each included study to assess the robustness of the combined results. We also used Begg’s regression asymmetry test and Egger’s regression plot test to assess the risk of publication bias. Additionally, we performed subgroup analysis to explore the source of heterogeneity and potential factors that may influence the results, such as gender, intervention methods, intervention period, intervention frequency, and intervention duration. All statistical analyses were conducted using Stata SE version 15.

After conducting a database search, 938 studies were found. Following the removal of duplicate studies, 721 studies remained. Among them, 43 studies were found to be eligible for inclusion based on their titles and abstracts. After a full-text review, only six studies met the inclusion criteria ( Figure 1 ).

Between 2016 and 2023, six RCTs of parallel design were conducted in Asia, specifically in China, Japan, Korea, Chinese Taipei, and Thailand ( Table 1 ).

The present study included a total of 278 individuals diagnosed with sarcopenia, with ages ranging from 56 to 81 years old. These individuals were part of six distinct research papers. Among the participants, 145 were randomly assigned to the RT intervention group, while the remaining 133 were allocated to the CG. Two studies focused exclusively on female participants (32, 38), while the other four included individuals of both genders in their samples (33, 37, 39, 40). Grip strength, skeletal muscle mass, and gait speed are commonly used to identify and characterize sarcopenia. Grip strength is employed to assess the likelihood of sarcopenia, skeletal muscle mass is used to confirm its presence, and gait speed is utilized to evaluate the severity of the condition (10, 41). The diagnosis of sarcopenia is determined by assessing muscle mass, which can be evaluated through Bioelectrical Impedance Analysis (BIA) or dual-energy X-ray absorptiometry (DXA). Three studies reviewed used BIA (32, 37, 40), and the other three used DXA to evaluate skeletal muscle mass (33, 38, 39) ( Table 1 ).

This study included six studies that examined various RT regimens for different intervention groups. The regimens consisted of kettlebell exercises, elastic band exercises, instruments, and mixed styles and were scheduled 2 to 3 times per week for an intervention period of 8 to 24 weeks. Three of these studies did not mention the duration of a single training session (37, 38, 40), while the remaining sessions lasted between 50 and 90 minutes. In four studies, participants in the CG received usual care and advice through a medical visit or a telephone call (32, 37–39). In two studies, advice on physical activity was provided (33, 40) ( Table 1 ).

Among the six studies included, five evaluated IL-6 levels (32, 33, 37, 38, 40), five measured TNF-α levels (32, 33, 37, 39, 40), four studies assessed CRP levels (32, 37–39), and two reported IL-10 levels (33, 37). Furthermore, we observed a significant imbalance in our results during the initial analysis, which we attributed to one study by Dong et al. (37) reporting a mean CRP value at least 1000 times lower than the mean of the other studies (32, 38, 39). Thus, based on previous studies of the same type (42), only three studies assessing CRP levels were included (32, 38, 39). Additionally, five of the six included studies reported the concentrations of inflammatory markers before and after the intervention (32, 33, 37–39). Only one study provided information on the change in inflammatory marker concentration (40) ( Table 1 ).

Among the analyzed studies, four exhibited low bias in generating the sequence (32, 33, 39, 40). However, the two studies could not have mentioned specific measures for generating the sequence, raising concerns about their methodology (37, 38). Regarding hidden allocation, none of the six studies provided enough information about hidden assignments, which casts doubt on bias. The blinding of participants and personnel involves withholding information about the treatment provided, and all six studies presented a high risk of bias due to the resistance training, introducing uncertainty and potential bias. Only one assessment of inflammatory markers, which involved recruiting qualified healthcare workers, was considered low risk in terms of blinding of outcome assessment (32). Conversely, the remaining five studies failed to provide detailed information on the blind outcome assessment method, raising concerns (33, 37–40). Finally, all six studies demonstrated low risk regarding incomplete outcome data, selective reporting, and other biases (Figure 2).

The effects of RT on TNF-α (I² = 88.2%, P < 0.001) and IL-10 (I² = 80.6%, P = 0.023) exhibited a significant level of heterogeneity. However, the effects of RT on IL-6 (I² = 21.5%, P = 0.278) and CRP (I² = 0.0%, P = 0.604) demonstrated consistent results across studies with no significant heterogeneity.

This study used a fixed effects model to analyze the IL-6 and CRP indicators, while a random effects model for the TNF-α and IL-10 indicators. When combining the effect size, the results demonstrated a statistically significant difference and low heterogeneity in the effect of RT on IL-6 (WMD = -0.73 pg/ml, 95% CI = -1.02 to -0.44, P < 0.001, I² = 21.5%); no statistical difference and high heterogeneity in the effect of RT on TNF-α (WMD = -1.00 pg/ml, 95% CI = -2.47 to 0.46, P = 0.141, I² = 88.2%; no statistical difference and low heterogeneity in the effect of RT on CRP (WMD = -0.45 mg/l, 95% CI = -1.14 to 0.23, P = 0.194, I² = 0.0%); no statistical difference and high heterogeneity in the effect of RT on IL-10 (WMD = 0.13 pg/ml, 95% CI = -3.99 to 4.25, P = 0.951, I² = 80.6%). These results suggest that RT improves IL-6 but not TNF-α, CRP, and IL-10 in participants with sarcopenia ( Figure 3 ).

Begg’s and Egger’s tests were used to analyze the publication bias in this study. The Begg regression asymmetry test and Egger’s regression plot indicated no significant differences in TNF-α (Begg’s test, P = 0.462; Egger’s test, P = 0.186), IL-6 (Begg’s test, P = 1; Egger’s test, P = 0.851), and CRP (Begg’s test, P = 0.602; Egger’s test, P = 0.286). These findings show the trustworthiness of the results obtained from this meta-analysis ( Supplementary Material ).

We conducted a sensitivity analysis to evaluate the stability and reliability of the result. In this analysis, we systematically excluded each study to assess the potential effect on the findings. Due to the inclusion of only two studies on IL-10, conducting a sensitivity analysis was impossible. The findings reveal that removing any single study did not yield robust results for TNF-α and IL-6, while robust results were observed for CRP, as shown in Figure 4 .

Several confounders, including gender, intervention methods, period, frequency, and duration, were performed by stratified subgroup analyses to identify the sources of heterogeneity and intervention parameters. In addition, there was no statistical difference in the overall effect size between CRP and IL-10, and CRP was of low heterogeneity, with only two studies on IL-10. Hence, this subgroup analysis focused solely on IL-6 and TNF-α. The analysis revealed a significant effect on the levels of IL-6 and TNF-α under the specific factors ( Table 2 ).

The assessments of the six studies according to GRADE are provided in Supplementary Material , with the quality of evidence for IL-6 as moderate, CRP and TNF-α as low, and IL-10 as very low ( Supplementary Material ).

IL-6, a crucial cytokine involved in the immune response, tissue regeneration, and metabolism, is secreted by various cell types, including monocytes, T lymphocytes, fibroblasts, and endothelial cells (43). In cases of infection or injury, production of IL-6 increases rapidly to boost immune and repair processes. However, excessive production of IL-6 and disruptions in its receptor signaling pathways can contribute to aging-related diseases (44). The elevation in IL-6 levels, which occurs with age, is a significant factor in the decline of skeletal muscle strength, quality, function, and training-mediated adaptation (45); TNF is a cytokine linked to chronic inflammatory and metabolic disorders, which can result in decreased protein, fat, and glycogen synthesis in skeletal muscle, leading to reduced muscle mass and strength (46). TNF-α production could be affected by various factors, including physical activity, diet, smoking, and age. The formation of reactive oxygen species is associated with this marker (47). The meta-analysis of 22 studies involving post-menopausal women and 14 studies in community-dwelling older adults with type 2 diabetes demonstrated that exercise training significantly decreased IL-6 levels (48, 49). In the meta-analysis conducted on elderly individuals, the effect of exercise on inflammatory markers revealed a reduction in IL-6 levels but no effect on TNF-α (50). Another meta-analysis on elderly individuals investigated the potential of RT to decrease IL-6 levels without altering TNF-α. (51). Our results support their findings that a significant decrease in IL-6 levels with RT for sarcopenia, with no significant change in TNF-α levels. Nevertheless, TNF-α did exhibit a significant change under certain specific factors.

Subgroup analysis was conducted by considering factors. The analysis revealed a significant effect of these specific factors on the levels of IL-6 and TNF-α. This finding aligns with the recommendations by intervention factor of the meta-analysis on the rehabilitation of elderly patients with sarcopenia (52). Regarding IL-6 and TNF-α, the genders mixed group showed a more positive effect than the pure-female group. An experiment revealed a correlation between low serum myostatin levels and low skeletal muscle mass in men, but this association was not in women (53). Therefore, subgroup analysis regarding the gender of this discrepancy in the findings could be linked with gender-specific characteristics of women.

Moreover, subgroup analysis further revealed that RT exhibits higher effectiveness when compared to conventional intervention performed at least three times a week. The International Exercise Recommendations in Older Adults (ICFSR) recommends RT thrice weekly to promote muscle mass and performance in the older (54); RT should be performed as a long-term intervention lasting for a minimum of 12 weeks to achieve optimal effectiveness. A meta-analysis examining the relationship between RT dosage and muscle strength in older adults suggests that engaging in RT for a sufficiently long training cycle can enhance muscle strength (55). At the same time, only IL-6 showed a significant change with unitary intervention types and duration equal to or less than 60 minutes, while TNF-α showed no change. Therefore, the intervention period for RT should be at least 12 weeks, with a frequency of at least three times per week.

In addition to IL-6 and TNF-α, CRP plays a critical role in sarcopenia. CRP is a clinical inflammation marker (56). When inflammation or tissue damage occurs, the liver produces CRP in response to IL-6, increasing the blood’s CRP concentration (57). In 2018, Tomeleri et al. (58) found that the 12-week RT program had a beneficial effect on reducing metabolic syndrome and CRP levels in older women. According to the meta-analysis, physical exercise can significantly reduce CRP levels in older individuals, regardless of the type of physical activity (48, 49, 59). Another meta-analysis focusing on patients with type 2 diabetes mellitus demonstrated that exercise effectively reduces CRP levels (60). Our findings contradict previous studies regarding CRP, possibly due to the difference in the number of CRP studies considered. The meta-analysis above included more than 10 CRP studies, whereas our study focused on only three. Consequently, it is imperative to conduct research that incorporates a more significant number of studies to comprehensively examine the effectiveness of RT on CRP levels in Asian participants with sarcopenia.

On the other hand, IL-10 is a cytokine known for its anti-inflammatory properties. It can hinder the activity of monocytes and macrophages and reduce the production of pro-inflammatory cytokines (61). In 2022, de Sá Souza et al. (62) found no significant difference in IL-10 and TNF-α levels among older individuals with sarcopenia who underwent RT. A meta-analysis reported that RT did not affect IL-10 levels in patients with type 2 diabetes (63). Our study’s findings align with previous findings that IL-10 levels were not significantly affected by RT. IL-10 focuses on only two studies; it is essential to consider that the sample size is too small to reflect the authenticity of IL-10 results. Consequently, it is imperative to conduct research that incorporates a more significant number of studies to comprehensively examine the effectiveness of RT on IL-10 levels in Asian participants with sarcopenia.

This study presents preliminary evidence of the effectiveness of RT in reducing inflammatory factors in participants with sarcopenia. We specifically focused on sarcopenia and conducted subgroup analyses based on participant gender and the type of RT intervention. Based on these results, our study highlights in which circumstances RT is most effective, which is a significant strength of our study.

There are still some limitations to interpreting the results. Some studies needed more information for meta-analysis, and most studies have room for improvement in random assignment concealment and double-blinding. Moreover, the included studies exhibit apparent heterogeneity, and the sensitivity analysis lacks robustness, which may affect the accuracy of the pooled results. In addition, the limited number of included studies restricts the ability to draw definitive conclusions. Although subgroup analyses identify potential sources, more detailed information on training intensity and detailed age of participants with sarcopenia is needed. Therefore, the meta-analysis results should be interpreted cautiously because of the lack of robustness of the sensitivity analysis and high heterogeneity. Future high-quality studies will be necessary for verification.