Association between kidney function and Parkinson’s disease risk: a prospective study from the UK Biobank | BMC Public Health

Study population

Participants were sourced from the UK Biobank ( a prospective cohort study that included approximately 501,314 participants aged 39–72 years, whose baseline data was collected between 2006 and 2010 [10]. Participants provided detailed personal, lifestyle, sociodemographic, and health information, underwent comprehensive physical evaluations, and provided blood, urine, and saliva samples for analysis. Ethical approval was obtained from the Northwest Multi-Center Research Ethics Committee, and all participants provided written informed consent. This study was conducted under UK Biobank application number 104,811.

Participant selection

Initially, 501,434 participants were considered for this study. Individuals with incomplete baseline data were first excluded (n = 99,151), as well as those with a recorded history of renal replacement interventions, including kidney transplantation, hemodialysis, or peritoneal dialysis (n = 415), and cases of participants who lost during the study period (n = 1,297). And ensure that there are no PD patients at baseline. Finally, a total of 400,571 participants were included in the final analysis.

Kidney function exposures

The eGFR was computed utilizing the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation, incorporating serum creatinine (Scr) and serum cystatin C (Scys). The CKD-EPI creatinine-cystatin C equation, expressed as a single equation [11], is:

(Where Scr is serum creatinine, Scys is serum cystatin C, κ is 0.7 for women and 0.9 for men, α is -0.248 for women and − 0.207 for men, min indicates the minimum of Scr/κ or 1, and max means the maximum of Scr/κ or 1).

The calculation yielded eGFR values, which were then categorized into five distinct groups: ≥105 (reference group), 90–104, 60–89, 30–59, and < 30 ml/min/1.73 m² in this study [12, 13]. This stratification enabled a comprehensive assessment of renal function across varying degrees of impairment, facilitating the exploration of its association with PD risk.

Covariates

The study incorporated a comprehensive array of covariates, including age, gender, educational attainment, ethnic background, body mass index (BMI), polygenic risk score (PRS), smoking status, alcohol intake, Townsend deprivation index (TDI), hypertension, and diabetes.

Age was computed by juxtaposing the birth date with the date of the baseline assessment. Gender (Male/Female) is chosen by the subjects. Ethnic categorizations were streamlined into “White” and “Non-White” with the latter encompassing groups such as Asian or Asian British, Black or Black British, Chinese, Mixed Race, and other specified ethnicities. Educational qualifications were classified as “College” and “Non-College”. BMI, expressed in kg/m², was calculated as the quotient of weight (in kilograms) and square of height (in meters). The Townsend Deprivation Index serves as a reflection indicator of socioeconomic standing [14]. PRS represents an individual’s susceptibility to disease (here: PD), which has been calculated previously in UKB from genome-wide genotyping data and categorized into “High Risk”, “Medium Risk” and “Low Risk” [15]. Smoking practices were divided into never-smokers, previous smokers, and current smokers. Alcohol intake was derived from touchscreen on digital interfaces that solicited information on the volume of each alcohol variety consumed weekly and automatically converts scores in the system. Further covariate considerations included incidence metrics such as hypertension and diabetes, which were captured via standardized touchscreen questionnaires and rendered as binary choices (yes or no). The urinary creatinine and urinary phosphate levels as covariates for evaluation in laboratory indicators.

Model adjustment

A multifactorial Cox regression model was used to determine the relationship between eGFR and PD, with three adjustment models. Model 1 adjusted for age, gender, ethnicity, and education. Subsequently, Model 2 further adjusted for BMI, PRS, smoking status, and alcohol intake. Finally, all variables were included in Model 3, including age, gender, ethnicity, educational attainment, BMI, PRS, smoking status, alcohol intake, hypertension and diabetes.

Neuroimaging

Cranial imaging data were obtained using a standard Siemens 3T MRI scanner. The scanned image has undergone preliminary processing and analysis, detailed information about the MRI processing can be found in the UK Biobank Protocol ( Gray matter volumes (T1-weighted MRI) from 24 brain regions were analyzed separately for the left and right hemispheres. Linear regression analysis was performed to assess the association between eGFR values and brain volume changes, with a correlation heatmap generated using the “ComplexHeatmap” package in R software.

Statistical analysis

Missing data rows were eliminated using R software (version 4.3.0). Continuous variables in the baseline characteristics are presented as mean ± standard deviation, and categorical variables were expressed as percentages. Comparisons across eGFR categories were performed using one-way ANOVA (perform Bonferroni correction) for continuous variables and the chi-square test for categorical variables.

Cox regression analyses were used to explore the association between various predictors and the PD incidence, with results presented as hazard ratios (HR) and 95% confidence intervals (95% CI). The Kaplan-Meier method depicted cumulative PD incidence, with log-rank tests assessing differences across eGFR categories. Restricted Cubic Spline (RCS) models analyzed the nonlinear relationship between eGFR and PD risk using the “rms” package in R. A nomogram model was created to predict survival rates (the 5-, 10-, and 15-year) using “nomogramEx” package in R, with ROC analysis evaluating model accuracy using “pROC” package in R. And area Under Curve (AUC) is used to measure the performance of the ROC model. All statistical analyses were performed using R (version 4.3.0), and P < .05 was considered statistically significant.