Determining Causal Effect of Kidney Function on Cancer Incidence

Patients receiving kidney replacement therapy (KRT) have a 1.5 to two times greater risk of cancer and cancer-related death. The extent of this heightened risk depends on the type of cancer and KRT. Research shows that even patients with lower-stage chronic kidney disease (CKD) may experience a higher cancer risk. However, the reasons for this remain unclear. There may be a causal relationship, the association could be due to factors such as immune suppression and inflammation, or it could result from distortion of the analyses from unidentified variables (confounding). 

To shed light on the relationship between kidney function and cancer risk, a group of researchers led by Ellen Dobrijevic, MD, performed a two-sample Mendelian randomization (MR) analysis to examine a potential causal relationship between kidney function and cancer risk. MR randomly assigns genetic variants at birth to investigate causal relationships as they reflect natural randomization, are fixed from birth, and are unaffected by clinical confounders or reverse causation. The study results were published in the American Journal of Kidney Diseases. 

Study outcomes included overall cancer incidence, cancer-related mortality, and site-specific colorectal, lung, and urinary tract cancer incidence. The exposures were estimated glomerular filtration rate (eGFR) and urine albumin-to-creatinine ratio (UACR). The researchers obtained genome-wide association studies (GWAS) summary statistics for eGFR (n=567,460; mean age 50.1 years; 48.2% male) and UACR (n=127,865; mean age 56.0 years; 45.1% male) for participants of European ancestry from the CKDGen Consortium. Estimates of the association between the genetic variants and cancer outcomes (cancer incidence and cancer-related death) came from individual-level, deidentified data from the UK Biobank database (n=407,329; mean age 56.9 years; 45.9% male) for participants recruited between 2006 and 2010.  

Univariate and multivariable MR were conducted for all outcomes using inverse-variance weighted methods. MR was conducted for each of the two exposures (eGFR and UACR) and the outcomes of overall cancer incidence, colorectal cancer incidence, urinary tract cancer incidence, lung cancer incidence, and cancer-related death. These types of cancer were selected for the analysis because they are the most common cancer types among patients with CKD.  

The associations between the genetic variants and cancer outcomes came from generalized linear mixed-effects regression models adjusted for age, sex, and the first 10 genetic principal components. Sensitivity analyses were also performed. 

Among CKDGen participants, mean eGFR and median UACR were 91.4 mL/min/1.73m² and 9.32 mg/g, respectively. For the UK Biobank participants, mean eGFR and median UACR were 90.4 mL/min/1.73m² and 9.29 mg/g, respectively. There were 98,093 cases of cancer observed. Of them, 6,664 were colorectal, 3,584 were lung, and 3,271 were urinary tract. There were 15,850 cases of cancer-related death. 

The genetic instruments for eGFR and UACR comprised 34 and 38 variants, respectively. Neither genetically predicted eGFR, UACR, nor a combination of eGFR and UACR were found to have a causal association with overall cancer incidence, cancer-related death, or site-specific lung or colorectal cancer incidence.  

The odds ratios (95% CI; P value) of genetically predicted eGFR for overall cancer incidence, overall cancer-related death, lung, colorectal, and urinary tract cancer incidences were 0.88 (0.40-1.97; P=.76), 0.92 (0.24-3.53; P=.91), 0.48 (0.03-8.49; P=.62), 0.93 (0.11-7.99; P=.95), and 0.22 (0.02-2.72; P=.24), respectively. The odds ratios (95% CI; P value) of genetically predicted UACR on overall cancer incidence, cancer-related death, lung cancer, colorectal cancer, and urinary tract cancer were 0.90 (0.78-1.04; P=.16), 0.96 (0.76-1.20; P=.70), 1.07 (0.65-1.75; P=.79), 0.74 (0.50-1.07; P=.11), and 0.93 (0.59-1.46; P=.76), respectively. 

There was no evidence of a causal association between eGFR and cancer in a stratified MR. In sum, no causal association between genetically predicted kidney function and cancer outcomes (incidence and cancer-related death) was found during MR analyses. 

Notably, the GWAS and UK Biobank data comprised participants who were generally healthy and included only a small proportion of CKD or KRT patients. Slight variations in kidney function due to the genetic variants analyzed in this study may be insufficient to cause cancer. The effects of reduced kidney function on cancer risk may be limited to those with more advanced kidney disease.  

Although a causal relationship was not identified, there was evidence of a nonlinear relationship between eGFR and cancer with an inflection point at an eGFR of 75. However, stratified MR did not show an association between genetically predicted eGFR and cancer, suggesting that increased cancer risk observed in kidney transplant recipients or patients with chronic glomerulonephritis may be related to factors such as prolonged use of maintenance immunosuppression rather than reduced kidney function.  

Limitations acknowledged by the authors include the use of a smaller GWAS for UACR, which reduced the strength of the instrument and potentially introduced population stratification. The authors did this to prevent overlapping samples.  

“In conclusion,” the authors wrote, “we did not find evidence of a causal effect of kidney function on overall cancer incidence, cancer-related death, or site-specific lung, colorectal cancer, or urinary cancer incidence. Associations with cancer observed in patients with mild to moderate stage CKD may reflect residual confounding in the observational evidence. Strategies to reduce the excess risk of cancer observed in patients with more advanced stage CKD or those with kidney transplants should focus on other mediating factors such as chronic immunosuppression use.” 

Source: American Journal of Kidney Diseases