Elucidating the causal nexus and immune mediation between frailty and chronic kidney disease: integrative multi-omics analysis

Abstract

Background

Empirical research has consistently documented the concurrent manifestation of frailty and chronic kidney disease (CKD). However, the existence of a reverse causal association or the influence of confounding variables on these correlations remains ambiguous.

Methods

Our analysis of 7,078 participants from National Health and Nutrition Examination Survey(NHANES) (1999–2018) applied weighted logistic regression and Mendelian Randomization (MR) to investigate the correlation between the frailty index (FI) and renal function. The multivariate MR analysis was specifically adjusted for type 2 diabetes and hypertension. Further analysis explored 3282 plasma proteins to link FI to CKD. A two-step network MR highlighted immune cells’ mediating roles in the FI-CKD relationship

Result

Genetically inferred FI and various renal function markers are significantly correlated, as supported by NHANES analyses. Multivariate MR analysis revealed a direct causal association between the FI and CKD. Additionally, our investigation into plasma proteins identified Tmprss11D and MICB correlated with FI and CKD, respectively. A two-step network MR to reveal 15 immune cell types, notably Central Memory CD4+ T cells and Lymphocytes, as crucial mediators between FI and CKD.

Conclusion

Our work establishes a causal connection between frailty and CKD, mediated by specific immune cell profiles. These findings highlight the importance of immune mechanisms in the frailty-CKD interplay and suggest that targeting shared risk factors and immune pathways could improve management strategies for these conditions. Our research contributes to a more nuanced understanding of frailty and CKD, offering new avenues for intervention and patient care in an aging population.

Introduction

Chronic kidney disease (CKD) has emerged as a critical concern in global public health, with the Global Kidney Disease Distribution report from 2017 revealing a prevalence of approximately 850 million individuals worldwide [1]. Forecasts position CKD as potentially the fifth leading cause of death by the year 2040 [2]. International clinical guidelines delineate CKD based on a decline in kidney function (eGFR below 60 mL/min/1.73 m²), the presence of markers of kidney damage, or both for a duration exceeding three months, with End-Stage Kidney Disease (ESKD) categorized by an eGFR below 15 mL/min/1.73 m² [3]. The management for ESKD patients is either kidney replacement therapy or conservative treatment, both of which contribute substantially to healthcare costs at the societal and familial levels.

Frailty, characterized by increased susceptibility to adverse health outcomes following stressors, has gained recognition as a critical issue in healthcare, influencing clinical practices and public health initiatives [4]. The frailty index (FI) is widely recognized for its utility in frailty assessment, highlighting a correlation between frailty prevalence and advanced age, though a significant incidence is also observed among younger populations [5, 6]. Disparities exist in the reported prevalence of frailty within CKD cohorts, with estimates ranging widely from 7% to 73% [7, 8]. Frailty’s pathophysiology involves inflammatory dysregulation, oxidative stress, and cellular senescence [9], with studies linking CD4 and CD8 T cell subsets [10, 11], the neutrophil-to-lymphocyte ratio, and systemic immune-inflammatory indices to frailty risk [11], paralleling associations with CKD [12, 13]. However, research into the proteomic and immunophenotypic pathways bridging frailty and CKD risk remains limited, which can elucidate the diseases’ causal relationships and underlying mechanisms.

Mendelian randomization, leveraging genome-wide association study (GWAS) summary data, aims to address the confounding and reverse causality issues inherent in observational studies, providing stronger causal inference evidence [14, 15]. This study advocates for a multi-omics approach, utilizing National Health and Nutrition Examination Survey (NHANES) and GWAS data, to explore the hypothesized causal effects between frailty and renal function. By incorporating 3282 plasma proteins and 731 immune cell signatures, this research aims to unveil the mechanisms underpinning the FI-CKD association, offering valuable insights for developing preventative, therapeutic, and management strategies for CKD populations.

Materials and methods

Study design and data sources

Cross-sectional study design

Participants

This investigation leveraged data from the NHANES, encompassing ten biennial cycles between 1999 and 2018. NHANES is a complex, nationally representative cross-sectional survey designed to assess the health and nutritional status of the U.S. populace. The initial cohort comprised 55,081 individuals aged 20 years and above. Following the exclusion of 48,003 participants due to incomplete data on CKD, frailty, and essential covariates, the final analytical sample consisted of 7,078 participants. The National Center for Health Statistics (NCHS) Institutional Review Board granted approval for NHANES, ensuring the study adhered to ethical standards with regularly updated documentation and materials available at the CDC website.

Frailty assessment

The frailty index (FI) is a comprehensive assessment tool that encompasses seven domains and 49 items, including cognition, independence, mood states, health conditions, healthcare utilization and access, physical capabilities and body measurements, along with clinical laboratory indicators [16]. To integrate continuous and categorical variables, each deficit is scored on a scale from 0 to 1, reflecting its severity. The FI is calculated by dividing the sum of an individual’s deficits by the total possible deficits. A threshold of 0.21 is utilized to distinguish frailty status, with values greater than or equal to 0.21 indicating frailty and values below this threshold denoting absence of frailty [17].

Covariates

Various covariates were considered based on previous literature, including gender, age, race, education level, poverty income ratio (PIR), body mass index (BMI), smoking status, diabetes, hypertension, urine creatinine, blood creatinine, glomerular filtration rate, and C-reactive protein.

Mendelian randomization study design and data resources

An extensive analysis employing diverse genetic methodologies is schematically represented in Figure 1. A bidirectional Mendelian randomization (MR) study was conducted, drawing upon extensive summary data from genome-wide association studies (GWAS) to explore the causal dynamics among CRP, eGFRcys, eGFRcr, BUN, renal failure, CKD, and the frailty index (FI). Additionally, multivariable MR analysis was utilized, with adjustments for the causal influences of type 2 diabetes and hypertension, to further delineate the direct causal link between FI and CKD. Genetic instruments corresponding to 3282 plasma proteins and 731 immune cell signatures were identified to assess their associations with CKD and FI risks via MR approaches. Subsequent analyses, including Protein-Protein Interaction Networks (PPI), gene ontology (GO), and Kyoto Encyclopedia of Genes and Genomes (KEGG), were applied to proteins implicated in CKD and FI risk to elucidate potential mechanistic pathways connecting FI and CKD. A two-step network MR analysis was subsequently performed to probe the intermediary role of immune cells in the FI-CKD. Ethical clearance for the GWAS was obtained from the appropriate institutional review boards, with participants providing informed consent. The datasets utilized in this study are publicly available and have been stripped of any personally identifiable information, in compliance with the STROBE-MR guidelines. The sources of the Mendelian randomization data are meticulously cataloged in Table 1.

Figure 1.

Study design.

NHANES: National Health and Nutrition Examination Survey; MR: Mendelian randomization; KEGG: Kyoto Encyclopedia of Genes and Genomes; GO: gene ontology; PPI: protein-protein interaction.

Table 1.

Detailed information about these datasets.

Traits	Sample size (total or cases/controls)	Ancestry	PubMed/GWAS ID
C-reactive protein [62]	575,531	European	PMID: 35459240
Estimated glomerular filtration rate (cystatin c) [63]	1,004,040	European	PMID:34272381
Estimated glomerular filtration rate (creatinine) [63]	460,826	European	ebi-a-GCST90026654
Blood urea nitrogen levels [64]	344,052	European	PMID: 34594039
Frailty index [65]	175,226	European	ebi-a-GCST90020053
Chronic kidney disease [66]	10,039/ 396,706	European	FinnGen
CKD [67]	12,385/104,780	European	PMID: 26831199
Renal failure [66]	15,475/ 396,706	European	FinnGen
Type 2 diabetes [68]	61,714/1,178	European	ukb-b-13806
Hypertension [68]	119,731/ 343,202	European	ukb-b-14057
Immune cell signatures [69]	3,757	European	PMID: 32929287
Plasma proteome [70]	3,301	European	PMID: 29875488

Genetic instrument selection criteria

To ensure the reliability of selecting single-nucleotide polymorphisms (SNP)s from relevant GWAS databases, the requirements for the selected SNPs are as follows: (i) genome-wide significance p < 5 × 10^-8 (adjustable to p < 5 × 10^-6 if SNP quantity is low) [18]; (ii) linkage disequilibrium (LD), with an r2 < 0.001 threshold in a 10000 kb window [19, 20]; (iii) F = R^2/(1 - R^2) × (N − 1 - 1)/1, where N represents the sample size, and a total F statistic less than 10 were excluded to avoid weak instrument bias [21]. Calculating R2 aimed to determine the extent to which IVs explain differences in the exposure factor. Specifically, R2 = 2 × β2 × maf × (1 - maf) [13], where EAF is the minor allele frequency, and β is the estimated genetic effect of each instrument variable on the exposure.

Statistical analysis

Statistical analysis of the observational study

We conducted an observational epidemiological analysis of NHANES data following the analysis guidelines provided on the NHANES website. To discern the causal relationships between CRP, eGFRcys, eGFRcr, BUN, renal failure, CKD, and the frailty index (FI), as well as the influence of CKD and renal failure on FI, we employed univariable Mendelian randomization (UVMR). Additionally, multivariable Mendelian randomization (MVMR) was utilized to ascertain the independent causal impact of FI on CKD, with adjustments for the causal contributions of type 2 diabetes and hypertension. All MR analyses adhered to three fundamental assumptions: (1) The genetic instruments must be strongly linked to the exposure in UVMR and to at least one exposure in MVMR; (2) These instruments should be free from associations with confounders; (3) The influence of the instruments on the outcome is mediated solely through the exposure.

Univariable and multivariable mendelian randomization analysis

We applied UVMR to evaluate the causal effects of CRP, eGFRcys, eGFRcr, BUN, renal failure, CKD on FI and the causal effects of CKD and renal failure on FI. Additionally, MVMR was utilized to ascertain the independent causal impact of FI on CKD, with adjustments for the causal contributions of type 2 diabetes and hypertension [22, 23]. All MR analyses adhered to three fundamental assumptions: (1) The genetic instruments must be strongly linked to the exposure in UVMR and to at least one exposure in MVMR; (2) These instruments should be free from associations with confounders; (3) The influence of the instruments on the outcome is mediated solely through the exposure [24]. To mitigate potential confounding effects, we examined the correlations between each instrumental variable (IV) single nucleotide polymorphism (SNP) and potential confounding factors using the PhenoScanner GWAS database [25]. If IVs of exposure is unavailable at the summary statistics of outcomes, we identified alternative SNPs with high LD (r2 > 0.8) on the LDlink platform [18]. We calculated the Wald ratio for each SNP, defined as the SNP-outcome association divided by the SNP-exposure association in UVMR [26]. When multiple SNPs were available for genetic variations, we summarized the effect calculated by the Wald ratio using an inverse-variance weighted (IVW) analysis [27]. In the MR analysis addressing multiple outcomes, the Bonferroni method was applied to control the Bonferroni at a 5% threshold, ensuring stringent correction for multiple testing. Analyses beyond this scope did not adhere to predefined statistical significance thresholds, serving primarily as exploratory investigations. Multivariable IVW (MV-IVW) was adopted as the primary analysis method in MVMR.

Two-step network MR analysis

We conducted two-step MR to evaluate potential mediators of the relative association between immune cells and FI and CKD [28]. The first step involved estimating the causal impact of FI on CKD using UVMR (β1). The second step involved assessing the causal effects of each mediator on FI using UVMR (β2), and the third step involved assessing the causal effects of each mediator on CKD using UVMR (β3), provided that each mediator is causally related to FI in UVMR. The total causal effect equals β1 + β2 * β3. The proportion of the causal relationship between each mediator mediating FI and CKD was calculated as the product of β2 and β3 divided by the total causal effect. The 95% confidence intervals (CIs) for mediation proportions were computed using the delta method [29].

MR sensitivity analysis

We applied weighted median, weighted mode, simple mode, and MR-Egger methods as sensitivity analyses based on different assumptions in UVMR. Reliable causal estimates were provided by the weighted median method, allowing for the violation of MR assumptions in the presence of pleiotropy in up to 50% of genetic variation [30]. If SNPs contributing to the maximum cluster were effective, the weighted mode method indicates reliable causal results [31]. The MR-Egger intercept test was used to detect potential horizontal pleiotropy, and the slope coefficient of MR-Egger regression provides a consistent estimate of the causal effect in the presence of horizontal pleiotropy [32]. Cochran’s Q statistic was employed to assess heterogeneity among IVs. The MR-PRESSO was used to detect potential horizontal pleiotropic SNPs and generate causal estimates after excluding identified outlier SNPs if heterogeneity existed [33]. Statistical power was calculated using the website (https://sb452.shinyapps.io/power/). All analyses were conducted in R software, employing packages such as ‘TwoSampleMR’, ‘tidyverse’, ‘ggplot2’, ‘MendelianRandomization’, ‘data.table’, and ‘LDlinkR’ [30]. The ‘p.adjust’ R package estimated Bonferroni q-values. All results are presented as odds ratios (ORs), β coefficients, or proportions, with corresponding 95% confidence intervals.

Result

Baseline characteristics of NHANES study participants

Overall, the age of the 7,078 participants ranged from 56 to 79 years, representing approximately 179.2 million non-institutionalized US residents. Weighted baseline demographics and clinical characteristics for the entire study population are delineated in eTable 1. The outcomes of the weighted multivariable logistic regression, after accounting for various covariates, are summarized in eTable 2. The analyses across models 1–4 consistently indicated that frailty is potentially associated with elevated levels of serum creatinine, C-reactive protein (CRP), blood urea nitrogen (BUN), and an increased risk of chronic kidney disease (CKD), underscoring frailty’s role as a possible risk factor.

MR analysis results

Characteristics of included studies and selected SNPs

A total of 15 SNPs are identified as IVs for FI, all of which have an F statistic > 10, indicating low risk of weak instrument bias (eTable 3).

The IVW identifies genetic variations of FI inversely associated with eGFRcys (OR = 0.9583, CI 0.9420–0.9748; p = 1.02E-06), eGFRcr (OR = 0.9873, CI 0.9778–0.9969; p = 0.0095), and BUN (OR = 0.9700, CI 0.9456–0.9951; p = 0.0192), and positively associated with CRP levels (OR = 1.2065, CI 1.1276–1.2909; P = 5.31E-08), CKD (OR = 2.1572, CI 1.4924–3.1181; P = 4.32E-05), and renal failure (OR = 1.9411, CI 1.4385–2.6193; P = 1.44E-05) for forward MR analysis. Inconsistency between the IVW method and MR Egger method is only detected concerning the impact of eGFRcr on FI (Figure 2). The application of Cochran’s Q and MR-Egger intercept tests in the MR analysis affirmed the absence of heterogeneity and horizontal pleiotropy (eTable 21).

Figure 2.

Forest plot for forward MR estimation of the causal relationship between FI and kidney disease.

MR: Mendelian randomization;eGFR: estimated glomerular filtration rates;BUN: blood urea nitrogen;CKD:chronic kidney disease; IVW: inverse-variance weighted;SNP: single-nucleotide polymorphisms;OR: odds ratio.

The reverse MR analysis reveals no statistically significant association between genetic variations of CKD and renal failure with FI (Figure 3). To verify reverse causality, the GWAS summary data from Pattaro C et al. on CKD were included, and similarly, no reverse causality was detected between CKD and FI. (OR= 1.0326, CI 0.9970–1.0695; P_IVW = 0.0734), When adjusting for type 2 diabetes and hypertension, MVMR demonstrates a direct causal relationship between FI and CKD risk (OR = 1.9943, CI 1.0722–3.7097; p = 0.0292) (eTable 20). Cochran’s Q and MR-Egger intercept tests confirm the absence of heterogeneity and horizontal pleiotropy in MR analysis (eTable 21).

Figure 3.

Forest plot for reverse MR estimation of the causal relationship between FI and kidney disease.

CKD:chronic kidney disease; SNP: single-nucleotide polymorphisms;OR: odds ratio.

Among the 3,282 plasma proteins examined, a higher genetic prediction of Transmembrane protease serine 11D was causally associated with an increased risk of FI (OR = 1.0280, CI 1.0166–1.0394; P = 1.11E-06; P _Bonferroni = 0.0036), while a lower genetic prediction of MHC class I polypeptide-related sequence B (OR = 0.8833, CI 0.8452–0.9231; P = 3.41E-08; P _Bonferroni = 0.0001) was causally associated with an increased risk of CKD.

In this study, a total of 178 plasma proteins were potentially causally related to FI, and 153 proteins to CKD, respectively. These potential proteins were annotated to genes. Among these genes, we obtained a gene interaction network from STRING and imported 331 genes for analysis at https://www.string-db.org/, further processed in Cytoscape to produce a PPI network (eFigure 2). Additionally, GO and KEGG enrichment analysis showed that genes associated with frailty and CKD primarily regulate functions such as negative regulation of cytokine receptor activity, immune receptor activity, lymphocyte mediated immunity, leukocyte mediated immunity, and regulation of immune effector process. Moreover, these genes are involved in pathways such as natural killer cell mediated cytotoxicity, cytokine-cytokine receptor interaction, and JAK-STAT signaling pathway (eFigure S1). Through MR analysis of 731 immune cells with FI and CKD, we found 15 immune cells with potential statistically significant common effects (eTable 14 and eTable 17), which were included in the two-step network MR analysis.

We then performed mediation MR analysis in which we calculated the proportion of the effect that was mediated by 731 immune cells signatures by dividing the indirect effect by the total effect for each of the CKD and FI phenotypes. Ordered by mediation proportion, the impact of immune cells on CKD risk was mediated by five modifiable risk factors, including Central Memory CD4+ T cell Absolute Count (mediation proportion: 11.03%; 95% CI: 1.69%, 17.37%), T cell Absolute Count (mediation proportion: 9.67%; 95% CI: 3.27%, 13.19%), CD4+ T cell Absolute Count (mediation proportion: 8.67%; 95% CI: 1.29%, 13.31%), Lymphocyte Absolute Count (mediation proportion: 3.26%; 95% CI: 1.36%, 3.82%), and HLA DR++ monocyte Absolute Count (mediation proportion: 2.33%; 95% CI: 1.30%, 2.36%) (Figure 4).

Figure 4.

Bar chart of mediation analysis results.

Discussion

In this study, we aimed to reveal the intricate interplay between frailty and CKD, with a focus on understanding causal relationships and potential mechanisms. We uncovered substantial evidence supporting a positive causal link between genetically inferred frailty and various indicators of kidney function, including CRP, BUN, eGFR, CKD, and renal failure, utilizing data from the NHANES and MR analysis. Conversely, the reverse MR analysis did not establish a causal relationship between CKD or renal failure and FI. After accounting for type 2 diabetes and hypertension through MVMR, a direct causal association between FI and CKD was evident. Subsequently, Through MR analysis, 178 plasma proteins were linked to FI, and 153 were identified as potential targets for CKD. The subsequent enrichment analysis of these proteins enriched our understanding of their roles, with the PPI network shedding light on their interactions. A two-step MR analysis investigating 731 immune cell signatures in relation to FI and CKD highlighted Central Memory CD4+ T cells, T cells, CD4+ T cells, and Lymphocytes as key mediators within this network, underscoring their possible role in the pathogenesis and as potential intervention points for CKD management.

The relationship between frailty and CKD has been the subject of ongoing debate, predominantly informed by observational studies. Clinical evidence suggests a higher incidence of frailty among CKD patients, particularly in older individuals, implying an observational link between these conditions. Notably, a cohort study involving American patients aged 57 to 71 with CKD found a 7%-43% prevalence of frailty, which was inversely related to eGFR levels [7]. Notably, a cohort study involving American patients aged 57 to 71 with CKD found a 7%-43% prevalence of frailty, which was inversely related to eGFR levels [34]. Another study reported that individuals with moderately reduced eGFR had a significantly higher risk of frailty compared to those with more severely reduced eGFR, with the risk of frailty notably increasing as renal function declined among participants aged over 65 [35]. Our MR findings indicate a stronger correlation between frailty and eGFRcys than eGFRcr, suggesting eGFRcys’s superior sensitivity in reflecting renal function, corroborating similar observations in the literature [35, 36]. These results are consistent with the hypothesis that frailty contributes to the deterioration of renal function, offering insights into potential therapeutic and preventive strategies for CKD.

Data from the NHANES III, including 33,994 participants, revealed that 20.9% of patients with moderate to severe CKD were frail. Even individuals with mild or early-stage CKD had about twice the likelihood of frailty compared to those without CKD, with a higher incidence of frailty observed in those with more severe CKD stages [37]. A more recent retrospective cohort study of 1,707 CKD patients revealed that half of the patients ranged from frail to severely frail [38]. Meta-analyses have further highlighted the prevalence of frailty in CKD patients, with estimates around 38.1% and 34.5% across various studies [39, 40], though some reports have indicated a prevalence as high as 73% [8]. These variations underscore the critical need for healthcare professionals to prioritize frailty screening in CKD patients to enable early detection and intervention.

Explorations into the mediating pathways linking frailty to CKD risk remain scarce. Our study posited that higher genetic predispositions for TMPRSS11D were linked to an elevated frailty risk, while lower genetic predispositions for MICB were associated with a heightened CKD risk. While direct studies on these protein associations with frailty and CKD were lacking, literature suggests TMPRSS11D might upregulate IL-8 expression [41, 42] and MICB played a role in immune function, indicating their involvement in immune-related processes. Subsequently, the KEGG and GO analyses of potential gene-protein interactions suggested that they were primarily mediated by pathways such as Natural killer cell-mediated cytotoxicity and Cytokine-cytokine receptor interaction. Further, our Mendelian randomization analysis of 731 immune cells in relation to frailty and CKD highlighted significant associations with Central Memory CD4+ T cells, T cells, CD4+ T cells, and Lymphocytes. T lymphocytes, crucial in cellular immunity, can mediate inflammation and orchestrate other immune cells through cytokine production. Jie Li and colleagues reviewed the utilization of oxidative phosphorylation, glycolytic metabolism, and fatty acid synthesis in T cell activation to promote kidney fibrosis [43]. A controlled study involving 100 subjects aged 1–21 years found increased central memory T cells and effector memory T cells in children with CKD [44]. Similarly, T cells also had intricate relationships with frailty [45–48]. A study involving 811 participants found a negative correlation between lymphocytes and frailty, especially in the memory/native B cell ratio [49]. The results of Tze Pin Ng et al. support the role of immune senescence in physical frailty, particularly concerning the observed loss of CD28 expression in CD8 and CD4 cells [50]. This is consistent with our findings. Ilaria Buondonno et al. also reported decreased numbers of CD4+ and CD8+ T cells in frail subjects, while CD4+/CD8+, memory, and naive T cells were unaffected by frailty [51]. Interestingly, Gilda Varricchi et al. reported on a subset of CD4+ T cells-circulating TFH (cTFH) cells, emphasizing the importance and complexity of cTFH subset phenotypes in frail patients [52]. Thus, there was reason to believe that the immune response between frailty and CKD was primarily mediated by the activation of T cells as a bridge between the two factors. On the other hand, several mechanisms had been proposed to support the association between frailty and CKD, including endocrine dysregulation, oxidative stress, and mitochondrial dysfunction [4, 53].

The associations of frailty and CKD with a spectrum of adverse outcomes, including diminished quality of life and elevated healthcare utilization, were well-documented [54, 55]. The coexistence of frailty and CKD had been shown to further escalate the risks of disability and mortality [56]. Notably, literature suggests that the progression of frailty can be decelerated or even reversed through the implementation of suitable interventions [57–59]. It is imperative for healthcare professionals to prioritize multidimensional health education, focusing on the early identification of frail individuals to enhance their self-care capabilities and management of health conditions. The European Renal Best Practice Group advocates for the screening of frailty among elderly patients with CKD stages 3b-5d, with the objective of discerning patients who could benefit from geriatric assessments and targeted interventions [60]. This study’s findings are predicated on the influence of a collection of SNPs, identified as genetic predispositions to frailty [61]. Our analysis underscores that frailty elevates the risk of CKD across various age groups, emphasizing the critical need for systematic frailty screening within the CKD patient population to facilitate early detection and intervention.

Our findings should be considered in light of several limitations. First, We acknowledge that our data are confined to European populations due to the constraints of GWAS summary data, potentially limiting the generalizability to other populations. Second, the MR analyses might not be adequately powered to detect subtle effects, attributable to the limited variance elucidated by the SNPs or the sample sizes in the genome-wide association studies employed [59]. Future studies, leveraging larger GWAS datasets, are anticipated to bolster statistical power. Third, standardized tools for frailty without consensus recommendations may lead to variability in frailty assessment methods and may introduce confounding effects. Fourth, although the study adjusted for type 2 diabetes and hypertension in the multivariable MR analysis, other potential confounding factors may not have been accounted for, potentially impacting the results. Therefore, future studies should consider a more comprehensive set of confounders. Fifth, The study identified proteins and immune cells potentially involved in the frailty-CKD association, but the exact mechanisms are unclear. Further experimental studies are needed to elucidate these mechanisms and identify potential therapeutic targets. Sixth, although mediation by certain immune cell characteristics in the frailty-CKD nexus was observed, these associations are only potentially significant, rendering the involvement of these immune cells in the frailty-CKD interplay as merely indicative. Thus, it remains essential to determine the involvement of these immune cells in the interaction between frailty and CKD through additional prospective and experimental research in future studies. Lastly, despite establishing causal links between frailty and CKD and identifying possible intermediary pathways, further prospective and experimental researches are essential to elucidate the underlying biological mechanisms.

In conclusion, this study established a causal link between frailty and CKD, highlighting one protein implicated in both conditions and identifying 15 immune cell signatures as mediators in the frailty-CKD association. These immune cells, particularly T cells and CD4+ T cells, frequently act as intermediaries in the networks connecting frailty and CKD, underscoring their potential role in the mechanisms underlying these conditions. This insight prompts consideration of whether targeted interventions addressing the shared risk factors of frailty and CKD could enhance patient management and outcomes.

Funding Statement

This study was funded by the Jiangxi Provincial Health Commission project (2018A385, 202310034), Jiangxi Provincial academic and technical leaders training program (20225BCJ22009), and the National Nature Science Foundation of China (82260598).

Authors’ contributions

Study conception and design: Guanghao Zheng; data analyses: Guanghao Zheng and Yu Cheng; draft preparation: Guanghao Zheng, Yu Cheng and Chenlong Wang; Statistical analysis:Bin Wang, Xinchang Zou and Jie Zhou; supervision of the study: Tao Zeng and Lifen Peng. All authors reviewed the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Ethical approval and consent to participate

Ethical approval was not necessary for this study, as it utilized data from databases open to the public.

Data availability statement and sources

Data for Mendelian randomization can be accessed through public GWAS repositories and the GWAS catalog, whereas additional data underpinning this study’s findings are available upon request from the Finngen Consortium.