Chronic kidney disease (CKD) affects 20–50% of people with Type 2 diabetes (T2D) and is the leading cause of kidney failure worldwide. The mechanisms linking T2D to kidney dysfunction are incompletely understood, and few interventions specifically prevent CKD in people with diabetes. The authors hypothesized that T2D-driven changes in circulating proteins contribute to kidney function decline and sought to identify protein mediators using genetic approaches.
The study applied a two-sample Mendelian randomization (MR) approach leveraging large-scale proteomic genome-wide association studies (GWAS) and disease trait GWAS. Key data sources included a circulating proteomic GWAS from deCODE with 35,559 individuals and a T2D GWAS with 80,154 cases. Kidney outcome GWAS used for downstream analyses included datasets with up to 1,004,040 participants. Additional proteomic GWAS used for replication were from UKB-PPP, Fenland, ARIC, and EPIC-Norfolk. Observational analyses used 37,854 UK Biobank participants.
In the first stage, the authors performed a proteome-wide MR analysis to identify circulating proteins whose levels were altered by genetic liability to T2D. This screen identified 71 proteins that showed evidence of being influenced by T2D genetic susceptibility based on the deCODE proteomic GWAS and T2D GWAS.
From the T2D-influenced proteins, the investigators then used cis-genetic variants (pQTLs located near the encoding gene) to proxy the causal effect of protein levels on kidney-related traits. Using this cis-MR strategy, five proteins were found to affect kidney phenotypes: INHBC (inhibin beta C-chain), GNPTG, LPO, AGRN, and CTSD. These proteins were associated with three kidney outcomes in GWAS analyses: blood urea nitrogen (BUN), estimated glomerular filtration rate (eGFR), and CKD risk.
Notably, MR estimates indicated that higher circulating INHBC levels led to lower eGFR and higher BUN, consistent with impaired kidney function. The study highlights INHBC in particular as a candidate mediator because it is a member of the TGF signaling pathway with known roles in tissue fibrosis.
To test robustness against platform and cohort variation, the authors repeated the MR analyses using proteomic GWAS from four independent cohorts: UKB-PPP, Fenland, ARIC, and EPIC-Norfolk. The direction of effect for the implicated proteins, including INHBC, was consistent across all four replication datasets, supporting that findings are not specific to a single proteomic platform or cohort.
In observational analyses of 37,854 UK Biobank participants, increased circulating INHBC levels were associated with an increased hazard of a recorded kidney disease diagnosis. Using mediation analysis, the authors estimated that circulating INHBC levels mediate 1.3% of the association between T2D and kidney disease diagnosis (95% confidence interval 0.85%–1.9%). This quantifies a modest proportion of the total T2D–CKD association attributed to INHBC in the datasets studied.
The authors acknowledge several important limitations:
Some MR assumptions (for example, absence of horizontal pleiotropy or validity of instruments) are difficult or impossible to fully test; although limited evidence of violations was observed, untestable assumptions remain.
Analyses were performed using population-based GWAS for T2D and kidney function traits rather than cohorts of individuals with established diabetic kidney disease. Therefore, the findings infer mediators between genetic susceptibility to T2D and general kidney traits, not necessarily diabetic kidney disease specifically.
Functional validation and analyses within disease-specific cohorts (individuals with diabetic kidney disease) were not performed and are needed to confirm biological mechanisms.
Other study-specific caveats such as cohort composition, assay differences, and the genetic architecture of pQTLs are implicit constraints on generalizability.
Using a multi-cohort MR framework, the study identifies five circulating proteins—most prominently INHBC—as candidate mediators through which genetic susceptibility to Type 2 diabetes may influence kidney traits (eGFR, BUN, and CKD risk). Replication across four independent proteomic datasets and supportive observational associations in UK Biobank increase confidence in these signals, though the mediated proportion attributed to INHBC is modest (~1.3%).
The results nominate INHBC and the other identified proteins for further experimental and disease-specific investigation to determine mechanistic roles and potential as therapeutic targets or biomarkers in diabetic kidney disease.
Data sources and summary statistics used in the study are listed in supplementary materials and are available as described by the authors. Code repositories were provided by the authors. The study received multiple sources of funding supporting the research group and individual investigators; the funders had no role in study design, data collection and analysis, decision to publish, or manuscript preparation. Specific funding details, competing interests, and acknowledgments are reported by the authors in the original publication.