Does smoking increase chronic kidney disease risk?

By Dr. Priyom Bose, Ph.D.Reviewed by Benedette Cuffari, M.Sc.Jan 7 2025

Study reveals a correlation between smoking and chronic kidney disease (CKD) in observational analyses, but no causal relationship in genetic models, highlighting the role of confounding factors like hypertension and diabetes.

Study: Association of Smoking with Chronic Kidney Disease Stages3 to 5: A Mendelian Randomization Study. Image Credit: Wdnld / Shutterstock.com

A recent Health Data Science study determines how smoking may increase the risk of chronic kidney disease (CKD).

The prevention and management of CKD

Current estimates indicate that about 10% of the global population is affected by CKD, which is a debilitating disease that increases the risk of both cardiovascular diseases (CVDs) and end-stage kidney diseases. There remains a lack of effective therapeutics for CKD, which is currently managed through monitoring blood pressure levels and diabetes, as well as preventing the development of CVDs, anemia, and metabolic bone disease.

Smoking increases the likelihood of developing numerous morbidities; however, it remains unclear how this behavior may impact the risk ofCKD. This lack of an apparent association could be due to reverse causality, confounding bias, and limited sample sizes in traditional observational studies.

To overcome the limitations of traditional observational studies, the researchers of the current study utilized Mendelian randomization (MR), which is an instrumental variable approach that can yield causal insights between the exposure and outcome variables.

About the study

In the current study, both a traditional observational study and MR analyses using genetic instruments were performed to evaluate the association and causal relationship between smoking behavior and CKD, respectively. Individual-level data from the United Kingdom Biobank (UKB) and published summary-level statistical data were used.

Since one-sample MR can lead to data overfitting, a two-sample MR using summary-level statistics was also conducted. The combination of both results can provide a more holistic evaluation of causal associations.

The UKB cohort comprises over 500,000 individuals between the ages of 40 and 69 years who were recruited from 2006 to 2010. The primary outcome was defined as the development of CKD stages three to five. For algorithmically defined end-stage renal diseases (ESRD), an additional report was also used.

Lifetime smoking index, which provides information on the intensity, duration, initiation, and, when applicable, cessation of smoking, was used as the exposure. In the traditional observational study, two smoking indices were generated including smoking status and the lifetime smoking index.

Hazard ratios (HRs) and 95% confidence intervals (CIs) for CKD risk in relation to the lifetime smoking index and smoking status were calculated by fitting Cox proportional hazards models. A penalty spline with two degrees of freedom (df) was fitted to the lifetime smoking index to explore potential non-linearities.

Study findings

In Cox proportional hazards models, positive associations were observed for both the lifetime smoking index and smoking status with CKD. Moreover, smokers had a greater risk of developing CKD with an HR of 1.26 as compared to non-smokers.

In the adjusted model, the HR of CKD was 1.22 for every unit increase in the lifetime smoking index. There was a near-linear association between incident CKD and lifetime smoking index, as evidenced by penalty splines.

Fourteen single-nucleotide polymorphisms (SNPs) were selected and combined into a polygenic risk score (PRS) to serve as a genetic instrument to avoid weak instrument bias. A significant association was observed between lifetime smoking index and PRS in the one-sample MR analysis.

However, no significant associations or non-linearities were reported between CKD and the genetically predicted smoking index. Subgroup analyses also suggested similar insignificant associations.

The three confounding factors considered in the analysis included body mass index (BMI), hypertension, and diabetes. Three SNPs of rs2062882, rs4949465, and rs6962772 were associated with these factors, respectively. The SNP rs2062882 was directly associated with CKD outcome.

Overall, sensitivity and robustness analyses yielded results consistent with the main model. The MR analyses had 99% power to detect a statistically significant causal association at a type-one error rate of 5%.

In the two-sample MR analyses, 42 SNPs that were strongly associated with the lifetime smoking index were identified in a genome-wide association study (GWAS). Concerning the test for the causal effect of smoking on CKD, no significant heterogeneity nor significant pleiotropy were noted. The different methods considered consistently suggested the lack of a causal relationship between lifetime smoking index and CKD.

Conclusions

The study findings do not indicate a causal effect of smoking behavior on CKD. However, the traditional observational study revealed a positive correlation between the variables, thus suggesting that covariates such as hypertension and diabetes could be important confounding factors in observational analyses.

To clarify the biological mechanisms involved the interaction between CKD and smoking behavior, additional mediation analyses should be conducted. These future studies can better inform public health strategies to improve overall kidney health and alleviate CKD risk factors.