Causal Effects of Lifestyle and Dietary Factors on Urinary Stone Risk in a Genetic and Population-Based Study

Summary

This protocol aims to use Mendelian randomization (MR) and NHANES data to investigate links between lifestyle, diet, and urinary stones.

Abstract

This study evaluated how lifestyle habits and dietary factors influence urinary stone risk by integrating Mendelian randomization (MR) analysis with data from the National Health and Nutrition Examination Survey (NHANES). Instrumental variables were obtained from genome-wide association study datasets, while urinary stone outcomes were sourced from the FinnGen database. The primary MR analysis used the inverse-variance weighted approach, and additional MR methods verified the robustness of the findings. NHANES data further assessed the relationships between lifestyle characteristics, dietary patterns, and kidney stone occurrence. The MR analysis demonstrated that higher pre-tax income was inversely associated with urinary stone risk. Greater consumption of fresh fruits and tea, along with increased intake of potassium and vitamin E, also correlated with a lower likelihood of stone formation. NHANES analysis identified sleep deprivation, sleep disorders, depression, and elevated body mass index as significant risk factors for nephrolithiasis. In addition, pro-inflammatory dietary patterns characterized by higher Dietary Inflammatory Index scores increased stone risk, whereas adherence to HEI-2020, DASH, and aMED dietary patterns showed protective associations. These findings provide further insight into modifiable dietary and lifestyle determinants of urinary stone disease and may support the development of preventive and therapeutic strategies.

Introduction

The global incidence of urinary stones has escalated, affecting approximately 10% of the world's population at some point in their lives, with higher prevalence rates observed in industrialized countries^1,2,3. This condition imposes a significant burden on healthcare systems because of the costs associated with diagnosis, treatment, and management of recurrent stone episodes. Additionally, urinary stones can lead to severe complications, including renal colic, hematuria, urinary tract infections, and chronic kidney disease, thereby affecting the patients' quality of life and increasing morbidity^4,5.

Despite its high prevalence and substantial healthcare burden, the pathogenesis of urinary stones remains unclear. Current research has identified various risk factors associated with the formation of urinary stones, including dietary habits, fluid intake, obesity, metabolic disorders, and genetic predispositions^6,7,8,9. While previous observational studies have individually linked various lifestyle behaviors or dietary factors to urinary stone risk, most have examined these exposures in isolation or without accounting for genetic confounding. Consequently, there is a critical need for robust methodologies to uncover the causal relationships between these risk factors and the development of urinary stones. Mendelian randomization (MR) is an analytical approach that utilizes genetic variants as instrumental variables to assess the causal effect of exposure on an outcome, thereby mitigating confounding and reverse causation¹⁰. By applying MR, researchers can strengthen causal inferences between risk factors and disease outcomes, thereby offering clearer insights into disease etiology.

In this study, we employed an MR approach to investigate the causal relationships between various risk factors and the risk of developing urinary stones. Moreover, we sought to validate our findings using data from the National Health and Nutrition Examination Survey (NHANES), a comprehensive and nationally representative dataset. By integrating genetic epidemiology with traditional epidemiological data, this study aimed to provide a more definitive understanding of the factors contributing to urinary stone formation and identify potential targets for prevention and intervention. We hypothesized that lifestyle and dietary factors exert causal effects on the risk of urinary stone formation, such that unhealthy behaviors increase risk while specific beneficial dietary components reduce risk.

Protocol

The research protocols for the National Health and Nutrition Examination Survey (NHANES) were approved by the Research Ethics Review Board (ERB) of the National Center for Health Statistics (NCHS). As NHANES datasets are publicly available, researchers do not need to seek separate approval from their own institutional review boards (IRBs). The summary statistics from this genome-wide association study (GWAS) are also publicly accessible. Consistent with the data repository’s terms of use and the approvals granted to the original study investigators, this secondary analysis did not require new IRB approval or additional individual informed consent. Each contributing GWAS study included details of ethical oversight and consent procedures in its original publication. All analyses were performed in accordance with institutional policies and the principles of the Declaration of Helsinki. The full protocol is detailed in Figure 1.

MR design

Two-sample MR is considered a method of identifying the causal relationship between the phenotype of exposure and the outcome with the use of genetic instruments (single-nucleotide polymorphisms [SNPs] ) as instrumental variables (IVs) from an accessible public dataset from large-sample genome-wide association studies (GWAS). By compensating for the typical drawbacks of residual confounding and reverse causality in observational studies, two-sample MR could reinforce the ability to infer the causality of an exposure-outcome association. A well-designed MR study was based on the following three assumptions: (i) relevance assumption: genetic variants are associated with risk factors; (ii) independence assumption: genetic variants are independent of confounding factors; and (iii) exclusion restriction assumption: genetic variants affect outcomes only through risk factors^11,12.

Data source and selection of IVs

For this MR analysis, publicly available GWAS databases, including IEU openGWAS, the GWAS catalog, and published GWAS summary-level data, were searched to obtain eligible datasets for exposure. Therefore, additional ethical approval was not needed. The exposures were categorized into two groups: (i) lifestyle and (ii) dietary factors. For each exposure, the selection criteria were as follows: (1) the GWAS was conducted in individuals of European ancestry to minimize population stratification; (2) the sample size was sufficiently large (generally >100,000 participants) to ensure adequate statistical power for instrument variable selection; (3) the GWAS provided summary-level data including SNP-level effect estimates, standard errors, and p-values; and (4) the phenotype definition was consistent with our exposure of interest. When multiple datasets were available for the same exposure, we preferentially retained the dataset with the largest sample size, as larger samples provide more precise effect estimates and a greater number of genome-wide significant SNPs for use as instrumental variables. In cases where sample sizes were comparable, we selected the dataset with the most rigorous phenotype definition or the one that excluded overlapping samples with the outcome GWAS to avoid potential sample overlap bias.

Initially, we established a genome-wide significance threshold of p < 5E-8 to identify highly correlated SNPs with each exposure. However, due to the limited number of SNPs identified for certain potential risk factors when considered for exposure, we opted for a slightly higher cutoff ranging from p < 1E-7 to p < 1E-5. Detailed information on the exposures used in this study is provided in Supplementary Table 1. All SNPs were clumped to avoid the linkage disequilibrium under a strict clump window (r² = 0.01 and kb = 5,000). Harmonization was performed in strict mode (action=3) to eliminate palindromic SNPs with intermediate allele frequencies. The strengths of the correlations between IVs and exposure factors were assessed using the F statistic, and IVs with F < 10 were eliminated to ensure the association strengths of genetic instruments for each putative risk factor and avoid weak instrument bias. Data for urinary stones (ncase = 9713, ncontrol = 366,693) were extracted from the FinnGen database (finngen_R9_N14_CALCUKIDUR), a large public project covering over 370,000 Finnish biobank participants¹³. Urinary stones were diagnosed according to the ICD-10.

MR statistical analysis

Instrumental variant selection, Mendelian randomization analyses, and data visualization were conducted in R using the packages that primarily support TwoSampleMR and MRPRESSO. Multiple MR approaches were applied, including the inverse-variance weighted random-effects model (IVW)^14,15, MR-Egger¹⁶, weighted median¹⁷, weighted mode¹⁸, and simple mode¹⁹. The random-effects IVW model was used as the principal analytical approach because it provides reliable causal estimates through meta-analytic integration of Wald ratios for each instrumental variable when directional pleiotropy is not present. MR-Egger, weighted median, weighted mode, and simple mode analyses were additionally performed to evaluate the consistency and robustness of the findings.

Cochran’s Q statistic assessed heterogeneity among individual SNPs. Stability of the MR estimates was further examined through leave-one-out sensitivity analysis by sequentially removing each instrumental variable. The MR-Egger intercept test and the Mendelian Randomization Pleiotropy RESidual Sum and Outlier method evaluated the influence of pleiotropic and outlier SNPs on causal inference. Multiple-testing correction was performed using the Benjamini–Hochberg procedure with false discovery rate adjustment, and an adjusted p-value threshold of 0.0520 indicated statistical significance. Statistical power calculations for the MR analyses were conducted using an online tool available at https://sb452.shinyapps.io/power/. Causal associations were reported as odds ratios with corresponding 95% confidence intervals. All statistical analyses were carried out using R software version 4.2.2 developed by the R Foundation for Statistical Computing.

Observational study population and design

The 20,797 participants selected for the cross-sectional study were from the 2009–2020 cycle of the NHANES, a cross-sectional survey of a nationally representative sample of the U.S. civilian non-institutionalized population based on a stratified, multistage probability sampling design. The whole inclusion and exclusion criterion is shown in Supplementary Figure 1. Among the 24,593 participants who completed the Dietary Interview and the Kidney Conditions Questionnaire, those with missing data such as lifestyle and demographic information were excluded. All participants answered the Kidney Conditions Questionnaire and were asked “Ever had a kidney stones ? ” They were asked, A “ Yes ” answer was defined as a kidney stone.

Dietary pattern scores

Comprehensive dietary intake data were collected from NHANES participants to estimate energy consumption, nutrient intake, and additional food components derived from foods consumed during the 24-hour period preceding the interview. Because face-to-face dietary interviews provide greater accuracy and consistency, dietary data from the initial interview were used to evaluate diet quality and generate dietary quality indices. The Dietary Inflammatory Index (DII) was developed to assess the inflammatory characteristics of individual dietary patterns by incorporating 45 dietary components with either pro-inflammatory or anti-inflammatory properties²¹. Twenty-eight foods from the NHANES were used to calculate the DII. Higher positive DII scores were associated with greater pro-inflammatory capacity, whereas higher negative DII values were associated with greater anti-inflammatory capacity. The Healthy Eating Index-2020 (HEI-2020) is an updated version that assesses the fit between dietary intake and the new 2020–2025 U.S. Dietary Guidelines (DGA) and includes 13 groups: total fruits, whole fruits, vegetables and legumes, whole grains, dairy products, total protein foods, seafood and plant proteins, fatty acids, refined grains, sodium, added sugars, and saturated fat consumption^22,23. The HEI-2020 scores range from 0 to 100, with higher scores reflecting healthier diets. The Dietary Approaches to Stop Hypertension (DASH) score consists of nine components (total fat, saturated fat, protein, fiber, cholesterol, calcium, magnesium, potassium, and sodium), with scores ranging from 0 to 9, with higher scores indicating greater adherence to the DASH dietary pattern²⁴. The alternative Mediterranean Diet model (aMED) consists of nine components: vegetables, legumes, fruits, nuts, whole grains, red and processed meats, fish, alcohol, and the ratio of monounsaturated to saturated fats²⁵. The aMed scores ranged from 0 to 9, with higher scores indicating greater adherence to the Mediterranean Diet model. All the dietary pattern scores described above were computed using the “dietaryindex” R package.

Lifestyle and covariates

Questionnaire data provided information on participant demographics, health conditions, and lifestyle characteristics, including age, sex, race or ethnicity, educational attainment, household income, physical activity, smoking habits, and alcohol consumption. Race and ethnicity were classified into non-Hispanic White, non-Hispanic Black, Mexican American, other Hispanic groups, and other racial or ethnic categories. Educational attainment was grouped into less than high school education and education beyond high school. Household income level was defined by the poverty income ratio (PIR): low (PIR < 1.30), moderate (1.30 ≤ PIR < 3.50), and high (≥ 3.50). A history of diabetic disease was obtained based on self-report (yes or no). Leisure time physical activity (LTPA) was categorized into three categories based on the intensity and duration of the exercise: moderate-intensity exercise < 75 min; 75–150 min; and ≥ 150 min, or vigorous activity ≥ 75 min. Sedentary behavior was classified as yes or no (≥ 6 and < 6 h) based on sedentary time. A depression score of ≥ 10 was categorized as a clinical depression threshold according to the PHQ-9 questionnaire score and depression. Smoking was categorized as having smoked (more than 100 cigarettes in total in the past). Alcohol consumption was categorized as never drinking, moderate drinking (≤ 2 drinks per day for men and ≤ 1 drink per day for women), and heavy drinking (> 2 drinks per day for men and > 1 drink per day for women). Sleep duration was classified into three categories (< 7, 7–9, and ≥ 9 h), and sleep problems were categorized as yes or no based on self-reporting. Body Mass Index (BMI) was categorized as thin or normal (< 25 kg/m²), overweight (25–30 kg/m²), or obese (> 30 kg/m²).

Statistical analysis

Participant characteristics were summarized for the overall study population and further categorized according to kidney stone status. Continuous variables were presented as mean values with standard errors, whereas categorical and ordinal variables were expressed as frequencies and percentages. Because NHANES uses a complex, nationally representative sampling design, 12-year sample weights were applied during the analyses. Logistic regression was used to assess the associations between dietary pattern scores, lifestyle, and kidney stones. In this case, the dietary pattern scores were divided into tertiles. In the regression models, Model 1 was adjusted for age (consecutive years), sex (male and female), and race (non-Hispanic white, non-Hispanic black, Hispanic, etc.), and additionally adjusted for education level, PIR, and history of diabetes in Model 2 based on Model 1. All analyses were two-sided tests, and a p-value < 0.05 was considered statistically significant. NHANES data were statistically analyzed using SAS, version 9.4.

Results

Mendelian randomization analysis of lifestyles with urinary stones

Figure 2 shows the random-effects IVW results between lifestyle factors and risk of urinary stones. We found that alcohol intake frequency (OR = 1.37, 95% CI = 1.14–1.66, p = 0.001) was causally associated with an elevated risk of urinary stones for genetically predicted 1-SD increases, whereas income before tax (OR = 0.31, 95% CI = 0.16–0.63, p = 0.001) showed a ...

Discussion

Based on Mendelian randomization findings, higher income, fresh fruit intake, tea consumption, and higher levels of potassium and vitamin E significantly reduce the risk of urinary stones. Based on NHANES data, insufficient sleep, sleep disorders, depression, high BMI, and a pro-inflammatory diet (high DII) increase the risk of kidney stones, whereas the HEI-2020, DASH, and aMED healthy dietary patterns reduce the risk.

The validity of this integrated Mendelian randomization (MR) and NHANES fr...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
GWAS exposure databases	IEU OpenGWAS project	https://gwas.mrcieu.ac.uk/	Publicly available genetic data related to exposure factors
GWAS outcome databases	FinnGen	https://www.finngen.fi/en	Publicly available genetic data related to kidney stones
NHANES 2009–2020 Demographics & Socioeconomic dataset	CDC NHANES	DEMO	Publicly available dataset with demographic, socioeconomic, and lifestyle variables
NHANES 2009–2020 Dietary intake	CDC NHANES	24h recall	Dietary Intake Assessment
NHANES 2009–2020 Questionnaire	CDC NHANES	Questionnaire	Describe whether you have kidney stones
R	R Foundation	https://www.rproject.org(version 4.2.2 )	Software used to analyze NHANES and mendelian data
SAS	SAS company	http://www.sas.com/(version 9.4)	Software used to analyze NHANES data