Dementia currently affects more than 50 million people worldwide, and its prevalence is projected to reach 150 million by 2050 as the global population continues to age [1]. The societal and economic consequences are profound, with cumulative costs estimated to exceed US $14.5 trillion between 2020 and 2050 [2,3]. Despite intensive research efforts, current therapeutic options remain limited in effectiveness, reinforcing the urgent need to identify modifiable risk factors that may help delay or prevent cognitive decline [4].

Sarcopenia, a progressive skeletal muscle disorder characterized by decline in muscle mass, strength, and physical performance, is highly prevalent among older adults, affecting approximately 10%-27% of this population globally [5]. In addition to its established associations with frailty, falls, disability, and mortality [6-8], emerging evidence suggests a link between sarcopenia and neurocognitive impairment, including mild cognitive impairment and dementia [9-12]. Importantly, unlike many neurodegenerative conditions, sarcopenia is potentially modifiable through targeted interventions such as resistance exercise, adequate protein intake, and nutritional supplementation, making it a promising intervention target for preserving cognitive health in aging populations [13-15].

However, most previous studies have focused on sarcopenia assessed at a single time point and have not accounted for its dynamic nature [9-12]. Sarcopenia may progress or regress over time due to changes in lifestyle or health status, and these transitions may differentially influence cognitive aging. Understanding the longitudinal impact of sarcopenia progression or recovery on cognitive trajectories may provide critical insights into optimal timing of interventions.

To address this gap, we analyzed data from the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative cohort of middle-aged and older adults in China. We investigated whether 2-year transitions in sarcopenia status were associated with subsequent 5-year cognitive trajectories. To our knowledge, this is among the first large-scale longitudinal studies to evaluate the cognitive implications of dynamic changes in sarcopenia status, capturing both progression and regression patterns. We hypothesized that progression to possible sarcopenia or sarcopenia would be associated with accelerated cognitive decline, whereas regression from these states would be linked to a slower rate of cognitive deterioration. Additionally, we examined whether these associations differed across subgroups defined by sex, age, and residential status, and educational attainment, as these factors influence sarcopenia risk [16-18] and modify its relationship with cognitive function.

This study used longitudinal data from a nationally representative cohort to examine the association between changes in sarcopenia status and cognitive trajectories among middle-aged and older adults. We found that progression from a nonsarcopenic state was significantly associated with accelerated cognitive decline, whereas regression from possible sarcopenia—but not from established sarcopenia—was linked to better cognitive performance. Importantly, the beneficial association of regression from possible sarcopenia with global cognition was consistently observed across multiple sensitivity and subgroup analyses. These findings highlight the importance of early detection and timely intervention during the possible sarcopenia stage, when the potential to preserve cognitive function and delay neurodegeneration may be greatest.

Previous CHARLS-based studies have reported associations between baseline sarcopenia and subsequent cognitive trajectories [11,35]. Our study extends this evidence by focusing on longitudinal changes in sarcopenia status (progression and regression), rather than baseline sarcopenia alone, using nationally representative data. In our cohort, 31.1% (n=2543) of the participants experienced a change in sarcopenia status over 2 years—15.4% (n=1264) progressed and 15.6% (n=1279) regressed—consistent with prior CHARLS findings [24]. Both progression from nonsarcopenia and progression from possible sarcopenia were consistently associated with steeper declines in global cognition, executive function, and episodic memory. In contrast, regression from possible sarcopenia was associated with improved cognitive outcomes. Notably, regression from established sarcopenia was not associated with meaningful cognitive gains, suggesting a limited therapeutic window once sarcopenia becomes entrenched.

In the subgroup analyses, participants who progressed to possible sarcopenia or sarcopenia exhibited faster cognitive decline than those who maintained a stable nonsarcopenic state. Significant associations were observed in certain subgroups (eg, men, women, individuals aged <65 y, and rural residents), whereas other subgroups showed attenuated or nonsignificant associations. None of the interaction terms for sex, age, residence, or education reached statistical significance. Consistent with guidance from Sun et al [43], the nonsignificant interactions observed in our study do not imply identical effects across subgroups but may reflect small true differences that our sample lacked adequate power to detect. Although the direction of associations between sarcopenia progression and cognitive decline was consistent across demographic strata, subtle population heterogeneity cannot be excluded. These findings underscore the need for larger, adequately powered prospective studies to validate potential subgroup-specific effects.

For instance, in the age-stratified analyses, the association between sarcopenia progression and accelerated cognitive decline was evident among participants aged <65 years but attenuated in those aged ≥65 years, although the interaction effect was not statistically significant. This finding contrasts with prior cross-sectional CHARLS research [44], which reported significant associations between sarcopenia and cognitive function in older adults. Although Du et al [44] examined cross-sectional relationships in participants aged ≥60 years, our longitudinal design captured within-person changes in cognition over time—an approach that better reflects temporal dynamics but is also influenced by attrition and survivorship effects. Selective attrition in longitudinal cohorts often leads to loss of participants with poorer cognitive function or declining health, leaving a more resilient survivor subset with reduced between-person variability [45]. Consequently, those at the greatest risk for cognitive decline may be disproportionately lost to follow-up, attenuating observed effects among older adults. Moreover, unaccounted competing risks such as mortality or severe morbidity may further underestimate associations; Austin and Fine [46] noted that treating participants who experience competing events as censored can bias estimates downward by excluding high-risk individuals. Finally, the smaller sample size in the ≥65-year group likely reduced statistical power and widened confidence intervals. Collectively, these factors likely contributed to the absence of statistically significant associations in older adults despite consistent effect directions across age strata.

Importantly, regression from possible sarcopenia was consistently associated with improved global cognitive function across all subgroups, with significant results (all P<.05) except among urban residents, where the association reached marginal significance (P=.05). These findings suggest that recovery from possible sarcopenia—representing an intermediate and potentially reversible stage—may provide a critical window for intervention to preserve cognitive health.

Our sensitivity analyses demonstrated that the main findings were robust. Incorporating wave 5 data, which reflect the most recent cognitive assessments, yielded results nearly identical to the primary analyses, underscoring the stability and reproducibility of our findings. In contrast, when wave 4 memory scores were harmonized using equipercentile equating, associations in the progression from possible sarcopenia to sarcopenia group were attenuated and no longer statistically significant. This attenuation likely reflects the conservative nature of the equating method, which reduces variability across test versions and may compress score distributions, thereby lowering statistical power despite consistent effect directions [42].

Similarly, when the original 9-category classification of sarcopenia transitions was used, the association between progression from nonsarcopenia to sarcopenia and cognitive decline became weaker. This was primarily due to the very small number of participants in this transition group, which reduced statistical power and produced less stable estimates. Moreover, the strong association observed in the nonsarcopenia to possible sarcopenia group is biologically plausible. Possible sarcopenia represents an early transitional stage characterized predominantly by reduced muscle strength, a parameter that has been shown to be a more sensitive indicator of functional decline and adverse health outcomes than muscle mass alone. Consistent with our additional analyses, low muscle strength—but not low muscle mass—was independently associated with faster cognitive decline. These findings align with evidence from population-based cohorts and Mendelian randomization studies showing that handgrip strength is more strongly linked to cognitive performance than muscle mass [47,48]. Therefore, cognitive deterioration may become detectable during the possible sarcopenia stage, before the onset of advanced muscle wasting in established sarcopenia. Across all analyses, however, regression from possible sarcopenia consistently demonstrated cognitive benefits, reinforcing the notion that this transitional stage is particularly amenable to intervention [49].

Several biological mechanisms may explain these associations. The muscle-brain axis has been proposed as a key pathway linking musculoskeletal and cognitive health [50]. Skeletal muscle secretes myokines—signaling molecules that regulate systemic inflammation, metabolic homeostasis, and neuroplasticity [51,52]. Sarcopenia may reduce myokine production and disrupt neuroprotective signaling, whereas reversal through physical activity may restore these pathways [52-54]. In addition, sarcopenia progression is associated with chronic inflammation and oxidative stress—both implicated in cognitive decline and neurodegeneration [55-58]—which may be mitigated through improved muscle function [59,60]. Emerging evidence also suggests a role of gut microbiota dysbiosis in sarcopenia-related neuroinflammation, underscoring the potential importance of the gut-muscle-brain axis [61,62].

These findings have important implications for clinical practice and public health. Routine screening for possible sarcopenia should be incorporated into cognitive risk assessment protocols for aging populations. Such screening is feasible using simple measures—such as grip strength and chair-stand tests—that can be implemented in primary care or community settings without specialized equipment. Early detection may help identify individuals at elevated risk of cognitive decline who could benefit from timely, multidisciplinary interventions. In our cohort, more than half of participants with possible sarcopenia reverted to a nonsarcopenic state (Table 1), reinforcing the reversibility of early-stage functional decline and the importance of timely intervention. Growing evidence from intervention trials suggests that resistance training, alone or combined with nutritional supplementation, can improve muscle function and sarcopenia-related outcomes [13-15], supporting the potential reversibility of early-stage sarcopenia. Future studies should evaluate whether such interventions can also yield sustained cognitive benefits.

This study has several limitations. First, its observational nature precludes causal inference. While associations were consistent across sensitivity analyses, randomized controlled trials are needed to establish causality. Second, sarcopenia was assessed using a validated anthropometric prediction equation rather than dual-energy DXA, the reference standard. Although highly correlated with DXA (R²=0.90) [23], some misclassification may have occurred. However, the 15% sarcopenia prevalence in our study aligns with previous meta-analyses using DXA or BIA, supporting its validity [5]. Third, modifications to the episodic memory test in 2018 may have introduced measurement inconsistencies; therefore, we applied equipercentile equating to harmonize memory scores across waves. Although this conservative adjustment attenuated associations in the progression from possible sarcopenia to sarcopenia group, the overall direction of effects remained consistent, supporting the robustness of the main findings [42]. Fourth, although physical exercise has been shown to be significantly associated with both cognitive function and sarcopenia [63,64], information on physical activity was missing in 61.0% (n=4995/8189) of participants in our cohort. Therefore, physical exercise was not included as a covariate in our analyses, and residual confounding related to physical activity cannot be excluded. Finally, residual confounding remains possible. Relevant factors such as hearing loss or social isolation—which may influence both sarcopenia and cognitive decline—were not fully captured and should be explored in future research [65].

Progression from a nonsarcopenic state—including progression from possible sarcopenia—is associated with accelerated cognitive decline, whereas regression from possible sarcopenia is linked to cognitive benefit. These findings identify possible sarcopenia as a critical and potentially reversible stage in the trajectory toward sarcopenia and cognitive impairment. They also support incorporating routine screening for possible sarcopenia into cognitive risk assessment protocols for aging populations. Given the simplicity, low cost, and feasibility of screening measures such as grip strength and chair-stand tests, implementation in primary care and community settings is both practical and scalable. Early identification and intervention at the possible sarcopenia stage may offer a key opportunity to preserve cognitive health and delay neurodegenerative processes. Future research should prioritize developing and evaluating targeted, evidence-based strategies to prevent sarcopenia progression and promote cognitive resilience in at-risk populations.