Mendelian randomization analysis reveals a protective association between genetically predicted systemic lupus erythematosus and renal cell carcinoma

Abstract

Observational studies have suggested that there may be a connection between systemic lupus erythematosus (SLE) and a higher likelihood of developing urological cancers, although the exact cause-effect relationship is still unclear. This study therefore investigated the causal relationship between SLE and urological cancers using the Mendelian randomization (MR) approach. Our primary MR analysis involved using the inverse variance weighted method, which employed an inverse-variance-weighted approach, to examine the causal relationship between SLE and urological conditions. In addition, we performed various sensitivity analyses, such as MR-Egger regression, tests for heterogeneity, and leave-one-out sensitivity tests, to assess the reliability of our results. The findings from our analysis using Two-Sample MR showed that genetically predicted SLE was linked to a reduced likelihood of developing renal cell carcinoma (RCC) (odds ratio = 0.9996, 95% confidence interval = 0.9993–0.9999, P value = .0159). These results suggest a possible protective impact of SLE against RCC. Nevertheless, no substantial correlation was detected between SLE and the likelihood of developing bladder cancer or prostate cancer. Collectively, these findings offer significant fresh perspectives on the possible correlation between SLE and genitourinary malignancies, specifically RCC, which will provide ideas and basis for the treatment of RCC.

1. Introduction

Kidney, bladder, and prostate cancers (PCas) are the predominant types of all cancer cases in the field of urology. These tumors are especially worrisome because they present substantial dangers to human well-being.^[1–3] The incidence of these cancers is mainly influenced by factors such as advancing age, family background, intake of a fatty diet and insufficient physical activity.^[4] To reduce the burden of urological cancers, attention should be given to other modifiable risk factors, such as autoimmune diseases. By addressing these risk factors, we can work toward preventing the incidence of urological cancers and improving overall health outcomes.

Systemic lupus erythematosus (SLE) is a condition where the immune system mistakenly attacks various organs and systems throughout the body.^[5] It can trigger a range of autoimmune reactions as the immune system of SLE patients mistakenly attacks their own tissues and organs, leading to inflammation and damage.^[6] The presence of inflammation is recognized as a possible factor in the progression of cancer, and there is a potential link between specific medications utilized for treating autoimmune disorders such as SLE and an elevated likelihood of developing cancer. An increased risk of cancer has been associated with cyclophosphamide, which is an alkylating agent utilized in the treatment of SLE.^[7] Previous research has indicated that individuals with SLE commonly face an increased likelihood of developing particular forms of cancer in comparison to the overall populace, although the risk of specific malignancies might be diminished or remain unaltered.^[8] Nevertheless, additional research is required to gain a comprehensive understanding of the correlation between SLE and cancer, along with the possible influence of SLE-triggered hormonal alterations on the risk of developing cancer. In the end, gaining a deeper comprehension of these matters may enhance the administration and care of both SLE and malignancy.

The presence of confounding and reverse causation makes it difficult to draw causal inferences from conventional epidemiological studies. Determining whether there is a causal connection between SLE and the risk of urologic cancer becomes challenging. Nonetheless, Mendelian randomization (MR) is a method that can assist in surmounting these difficulties, particularly in situations where conducting randomized controlled trials (RCTs) is impractical or unethical. MR can estimate the causal impact of exposure on the outcome through the utilization of genetic factors that determine exposure. As genetic components are randomly allocated during meiosis and remain consistent throughout an individual lifetime, MR effectively reduces the bias that impacts observational epidemiological studies. The presence of pleiotropy can now be better evaluated with the help of recent developments in MR techniques, including weighted median MR and MR-Egger, which have enhanced the accuracy of assessing the reliability of causal estimates. Hence, we utilized a 2-sample MR approach to examine the possible causal association between SLE and the likelihood of urologic malignancy. By utilizing this method, we were able to derive more dependable inferences concerning the cause-and-effect association between SLE and the susceptibility to urologic malignancy, ultimately enhancing our comprehension of the ailment and fostering the creation of more efficient therapeutic approaches.

2. Methods

2.1. Overall study design

The information utilized in this research was obtained from published studies that obtained approval from an institutional review board, and the individuals who took part in the original study gave their informed consent.^[9] Hence, there was no need for additional sanctions.^[10] To examine the causal relationship between urologic cancers and SLE, a study using the 2-sample MR approach was conducted. In this research, genetic variations known as single nucleotide polymorphisms (SNPs) were identified as instrumental variables (IVs). Using SNPs as a modeling technique similar to RCTs enables the discovery of causal connections between exposure factors like SLE and specific types of cancer, including kidney, bladder, and PCa.

3. Data sources

3.1. Genetic instrument variants for exposure

The SLE information used in this research was acquired from the latest and comprehensive genome-wide association study (GWAS) carried out by Wang et al. The study comprised 4222 cases and 8431 controls. Approval from the institutional review board was obtained, and all participants involved in the original study provided informed consent.

SNPs were selected based on comprehensive criteria. To begin with, SNPs that showed a strong correlation with SLE at a significance level across the entire genome (P = 5 × 10E-8) were incorporated. Secondly, SNPs were chosen to be independent of each other, ensuring that the impact of linkage disequilibrium on the results was minimized. To prevent bias caused by linkage disequilibrium, a strict threshold of r²= 0.01 was implemented, along with a window size of 10,000 kb. Furthermore, the association between the IVs and the exposure factors was evaluated by analyzing the F-statistic of the SNPs. Typically, IVs with an F-statistic exceeding 10 are regarded as unbiased indicators.

4. Genetic instrument variants for outcome

Data on renal cell carcinoma (RCC) were available from UK biobank and includes 1114 cases and 461,896 controls of European ancestry. Data on bladder cancer (BCa) were available from Finn database and includes 1115 cases and 217,677 controls. Data on PCas were available from UK Biobank and includes 9132 cases and 173,493 controls. The ethical approval was obtained for the original studies included in the analysis.

4.1. Statistical analysis

To offer a valid explanation for MR analysis, it is crucial to take into account these hypotheses.^[11] The connection between instrumental variables (IVs) and SLE has been firmly established in scholarly works. The development of urologic cancer is solely impacted by the consequences of IVs linked to SLE deficiencies. The IVs confirmed that there were no confounding factors in the relationship between SLE and urologic cancer.

Genetic variation can potentially introduce bias in the causal estimates through a single pathway known as horizontal pleiotropy, rather than through separate exposures. The situation goes against the presuppositions of MR. To address this issue, the MR analysis utilized 3 separate analytical approaches, each relying on a distinct horizontal pleiotropy model. The credibility of the findings is enhanced by comparing the results obtained from these 3 methods, as it demonstrates consistency across approaches. The primary analysis utilized an inverse variance weighted (IVW) approach, employing inverse variance-weighting, which yielded the most precise estimates.^[12] IVW is a statistical method used to estimate the effect of genetic variation on complex diseases. The advantage is that large-scale genetic and epidemiological data can be used to estimate the combined influence of multiple genetic variants on disease. In addition, the method can also take into account the influence of various confounding factors, so as to improve the accuracy and reliability of the estimation. However, it made the assumption that all SNPs were valid IVs. In case a single SNP fails to meet the assumption of IVs, the random-defect IVW method will be employed to introduce a bias, which assigns weight to each rate based on its standard error while taking into account potential heterogeneity. In order to meet the requirement of a valid instrumental variable, the weighted median method necessitates a minimum of 50% SNPs.^[13] By arranging the included SNPs according to their weights, we calculated the median of the corresponding distribution function based on the outcomes of our experiments. Moreover, in the absence of pleiotropic effects, MR-Egger regression can be used to obtain an estimate of the effect from the genetic instrument.^[14] We evaluated the pleiotropic impact by utilizing MR-Egger intercept. Moreover, the presence of a directional multiplicative impact cannot be established unless there is a significant deviation of MR-Egger intercept from zero.^[15]

5. Sensitivity analysis

Funnel plots can accurately demonstrate the extent of pleiotropy in IVs by presenting a solitary Wald ratio for each SNP. Nevertheless, evaluating horizontal pleiotropy using funnel plots becomes difficult as a result of the restricted number of IVs incorporated in the analysis. Significantly, the funnel plot illustrated a nearly balanced pattern, indicating the causal effect. To examine the potential prejudice or impact of specific SNPs on the IVW analysis estimates, leave-one-out analyses were performed. These analyses involved running meta-analyses using the reanalyzed IVW results after excluding one SNP at a time. After removing each SNP, a subsequent MR analysis was conducted systematically for the remaining SNPs. Remarkably, the results consistently indicated a significant causal relationship across all the SNPs. In the framework of MR analysis, the second proposition suggests that SNPs influence the exposure of interest exclusively, without any participation from other confounding pathways. MR-Egger regression was used to determine the intercept and P value for horizontal pleiotropy. Importantly, the MR-Egger regression intercept did not show any indication of horizontal pleiotropy (P > .05), which strengthens the argument that pleiotropy did not introduce any bias to the estimation of the causal effect.

Furthermore, upon examining the published GWAS data, no significant correlations were found between the SNPs linked to SLE and any other traits, except for SLE itself. The discovery indicates that the presuppositions supporting the third MR analysis were maintained without any breach. As a result, there was no indication that the genetic tools of the 10 SLE-related SNPs were notably linked to any other characteristic on a comprehensive genomic level, confirming our third MR hypothesis, which is improbable to be violated in our 3 study. MR and sensitivity analysis in R (version 4.2.1) were conducted using the “Two sample MR” (version 0.5.6) software package.^[16]

6. Results

6.1. MR results

To ensure that genetic variants can be used as a dependable tool for causal inference in MR, our methodology needed to meet 3 fundamental assumptions. In the first place, the genetic mutations need to be linked to the investigated exposure. Additionally, it is crucial that the genetic variations are not linked to any extraneous variables that might impact the result. In conclusion, the genetic mutations should solely impact the likelihood of the result by means of the exposure, excluding any alternative routes (Fig. 1). Satisfying these assumptions is crucial to ensure that the MR approach is valid and the conclusions drawn from it are reliable. Through a meticulous examination of these presumptions, we can enhance our comprehension of the cause-and-effect connection between SLE and the susceptibility to urologic cancer, ultimately leading to the formulation of more efficient approaches for the prevention and management of the disease.

Figure 1.

The study design of the present investigation, which involves Mendelian randomization (MR) analysis. The MR method is based on 3 assumptions: assumption 1 states that the genetic variants should have a strong correlation with the exposure (in this case, SLE); assumption 2 states that the genetic variants should not be related to any confounding factors associated with the outcome (in this case, urological cancers); and assumption 3 states that the genetic variants should only affect the outcome through exposure factors and not through alternative pathways. SLE = systemic lupus erythematosus.

In the summary statistics for BCa, PCa, and RCC, we found a combined total of 57, 57, and 12 distinct genetic variants that were linked to SLE. In SLE, the F-statistic for these SNPs exceeded 10, suggesting a minimal likelihood of weak-instrument bias. For more details about these SNPs, including a comprehensive analysis, please refer to the supplementary materials (Supplement Tables 1–3; http://links.lww.com/MD/L927; http://links.lww.com/MD/L928; http://links.lww.com/MD/L929).

According to the IVW analysis, there was no substantial connection found between SLE and bladder or PCa. The odds ratios were 1.034 (95% CI, 0.978–1.093, P = .2350) and 0.9996 (95% CI, 0.9986–1.0006, P = .4532) for bladder and PCa, respectively. Nevertheless, our research revealed an inverse correlation between genetically predicted SLE and RCC, displaying an odds ratio of 0.9996 (95% CI, 0.9993–0.9999, P = .0159) in the IVW analysis. Furthermore, our findings demonstrated a consistent association between genetically predicted SLE and all 3 urological malignancies using different MR approaches (Table 1).

Table 1.

MR results of SLE on risk of urologic cancers.

	Sample sizes	Method	SNP	b	SE	OR	95% CI	P
Bladder cancer	1115	MR Egger	50	0.0679	0.0726	1.0703	0.9283–1.2339	.3541
		Weighted median	50	0.0752	0.0423	1.0781	0.9954–1.1675	.0756
		Inverse variance weighted	50	0.0337	0.0284	1.0343	0.9783–1.0934	.2350
		Simple mode	50	−0.0059	0.0809	0.9941	0.8467–1.1672	.9424
		Weighted mode	50	0.1080	0.0537	1.1141	0.9953–1.2468	.0500
Prostate cancer	9132	MR Egger	50	−0.0009	0.0013	0.9991	0.9965–1.0016	.4849
		Weighted median	50	0.0002	0.0007	1.0002	0.9987–1.0017	.7444
		Inverse variance weighted	50	−0.0004	0.0005	0.9996	0.9986–1.0006	.4532
		Simple mode	50	0.0010	0.0014	1.0010	0.9981–1.0038	.4694
		Weighted mode	50	0.0006	0.0011	1.0006	0.9985–1.0026	.5738
Renal cell cancer	1114	MR Egger	10	−0.0008	0.0006	0.9992	0.9979–1.0003	.2189
		Weighted median	10	−0.0004	0.0002	0.9996	0.9991–0.9999	.0197
		Inverse variance weighted	10	−0.0004	0.0002	0.9996	0.9993–0.9999	.0159
		Simple mode	10	−0.0006	0.0003	0.9994	0.9988–0.9999	.0550
		Weighted mode	10	−0.0005	0.0002	0.9995	0.9990–0.9999	.0587

CI = confidence interval, MR = Mendelian randomization, OR = odds ratios, SLE = systemic lupus erythematosus, SNP = single nucleotide polymorphism.

The outcomes of our study aligned with the findings of the MR analysis, as we did not observe any notable association between SLE and bladder or PCa based on the scatter plots. In a scatter plot, each point represents a SNP, which shows the association between each SNP and exposure and outcomes. The X-axis shows the genetic association with SLE. The Y-axis represents genetic associations with BCa, PCa, and RCC risk, respectively. Each line represents a different approach to MR. The slope represents the correlation between exposure and outcome. However, the scatter plots did reveal a negative slope for SLE and RCC across various analytical methods, indicating a potential inverse relationship between SLE and RCC. The fact that these results were consistent across different analytical methods suggests that our findings are robust (Fig. 2).

Figure 2.

Scatter plots for effect sizes of SNPs for and those for urological cancers. (A) Causal estimates for SLE on BCa. (B) Causal estimates for SLE on PCa. (C) Causal estimates for SLE on RCC. The effect size of each SNP on SLE (X-axis) and its impact on BCa, PCa, and RCC (Y-axis) are represented by black points, along with their corresponding standard error bars. The slope of each line corresponds to the causal estimate using the inverse variance weighted, MR-Egger, weighted median, and weighted mode methods. BCa = bladder cancer, MR = Mendelian randomization, PCa = prostate cancer, RCC = renal cell carcinoma, SLE = systemic lupus erythematosus, SNPs = single nucleotide polymorphisms.

We conducted further analyses to test for heterogeneity in our findings using the IVW and MR-Egger methods, and found no significant heterogeneity in any of our final results (Supplement Table 4, http://links.lww.com/MD/L930). Furthermore, there was no indication of heterogeneity based on the results of the funnel plots (Supplement Fig. 1, http://links.lww.com/MD/L926). In addition, our results were not influenced by any leverage points, as indicated by the leave-one-out analysis and forest plots (Figs. 3 and 4). These additional analyses support the robustness and reliability of our findings.

Figure 3.

There are leave-one-out plots showing the association between SLE and the likelihood of developing (A) BCa, (B) PCa, (C) and RCC. BCa = bladder cancer, PCa = prostate cancer, RCC = renal cell carcinoma, SLE = systemic lupus erythematosus.

Figure 4.

Forest plot individual and combined SNP MR-estimated effects sizes for SLE on (A) BCa, (B) PCa, and (C) RCC. BCa = bladder cancer, MR = Mendelian randomization, PCa = prostate cancer, RCC = renal cell carcinoma, SLE = systemic lupus erythematosus, SNPs = single nucleotide polymorphisms.

7. Assessment of MR assumption

Our study utilized SNPs selected based on the genome-wide significance threshold of P < 5 × 10E-8. These analyses provided evidence supporting the absence of directional pleiotropy, indicating that the second MR assumption was not violated (P = .46). Additionally, the MR heterogeneity test showed no heterogeneity in our positive outcomes (P = .21). In summary, the rigorous assessment of the 3 fundamental assumptions in MR analysis was conducted in our study. The results suggested that the selected SNPs were appropriate as genetic instruments, and the relationships between genetically predicted SLE and RCC were not confounded by potential confounders or mediators.

8. Discussion

Using a comprehensive 2-sample MR approach, we conducted a study to explore the potential causal relationship between SLE and the occurrence of 3 prevalent urological malignancies: BCa, PCa, and RCC. The results of our study suggest that people diagnosed with SLE experience a notable decrease in the likelihood of developing RCC. Nevertheless, no conclusive proof was discovered to substantiate the notion that genetic factors can predict the occurrence of SLE and its impact on the risk of bladder or PCa. In general, our research offers significant fresh perspectives on the connection between SLE and urological malignancies, which may have consequences for the prevention and management of these conditions.

SLE is a multifaceted autoimmune disorder marked by the existence of numerous autoantibodies and immune complex accumulations, which can impact different organs, such as the cardiac, pulmonary, hepatic, and renal systems.^[17–19] According to recent meta-analyses, it has been suggested that SLE might have varying effects on the occurrence of various forms of cancer.^[8,20] Multiple reasons exist for the potential rise in cancer risk linked to SLE, encompassing the fundamental immune consequences of autoimmunity on cancer susceptibility, along with possible impacts of immunosuppressive medications employed in therapy.^[21]

Moreover, there exist numerous possible connections between cancer and SLE, encompassing the engagement of significant co-stimulatory compounds that have a vital function in the development and formation of SLE.^[22] The observed connections between SLE and urological cancers, along with other forms of cancer, could be influenced by intricate interplays and mechanisms.^[23] Additional investigation is required to gain a comprehensive comprehension of the intricate correlation between SLE and cancer, as well as to formulate efficacious approaches for the prevention and management of both ailments.

Prior research has not discovered a reliable connection between SLE and BCa.^[24] However, the utilization of cyclophosphamide is commonly regarded as a conventional factor that heightens the risk of BCa in individuals with SLE.^[25] BCa in SLE patients can also be caused by viral infections. Research on the potential connection between SLE and PCa indicates that individuals with SLE might exhibit decreased levels of testosterone in comparison to those without SLE.^[26,27] Consequently, the decreased levels of circulating testosterone in SLE patients could potentially lead to a reduced risk of developing PCa. Nevertheless, our research yielded no proof to substantiate the impact of SLE on the likelihood of developing bladder or PCa.

The exact biological process through which SLE decreases the likelihood of RCC is still not understood, although recent studies have provided insights into possible reasons. SLE is a chronic systemic autoimmune disease. Chemokines and/or cytokines, such as C-X-C motif chemokine ligand 10 (CXCL10) and interleukin1β, are involved in the pathogenesis of SLE and play an important role in the formation of peri-tumor inflammation, which is a determinant of tumor microenvironment.^[28] The changes of tumor immune microenvironment caused by SLE may be involved in RCC disease process. In addition, apoptosis caused by chronic inflammation may also cause changes in cancer-related genes.^[29] In addition, the presence of atypical autoantibodies that react to the body own DNA is a characteristic feature of SLE and has been detected in the bloodstream of individuals with SLE. According to a recent study, it was found that the lupus autoantibody 3E10, which has the ability to enter cells, exhibits significant toxicity toward cancer cells and tumors that lack the ability to repair DNA double-strand breaks.^[30] The reason for this is that 3E10 has the ability to attach to the single-strand tail of DNA, which effectively hinders an important stage in the repair of both single-strand and double-strand breaks in DNA.^[31] The results indicate that SLE may offer protection against RCC due to the existence of specific autoantibodies that can penetrate the nucleus, proving fatal for cancer cells that already have DNA repair deficiencies.^[32]

Although additional investigation is required to completely comprehend these mechanisms, these discoveries offer significant perspectives on the potential impact of autoantibodies in the prevention and treatment of cancer. Indeed, RCTs are necessary to further clarify the relationship between SLE and urologic cancers. Based on the results of previous studies and this study, we believe that the molecular mechanism of the potential relationship between SLE and RCC should be further explored, especially the mode of action of SLE specific lupus antibodies involved in killing RCC tumor cells, which can be used as a potential target for the subsequent development of RCC therapy.

MR analysis has a number of significant benefits. First, we systematically assessed the causal relationship between SLE and urinary cancers. MR analysis allowed for a relatively unconfusing estimate of the inferred causal relationship without being affected by reverse causal effects or confounding. Second, this study employed the broadest and most influential GWAS to investigate the association between SLE and urological malignancy, thus enhancing the credibility and applicability of our findings in different populations. Third, in addition to the traditional IVW method, we also use the weighted median method and the MR-Egger method for analysis to ensure the consistency of the causal estimation, indicating the robustness of our research results.

In this study, we also performed multiple heterogeneity and sensitivity analyses to detect and eliminate any potential pleiotropy. In order to evaluate, intercept and corresponding P values are obtained by MR-Egger regression. Notably, the intercept of the MR-Egger regression does not show evidence of horizontal pleiotropy (P > .05), which further supports that pleiotropy does not bias the estimates of causal effects, demonstrating that our MR estimates are robust and reliable without significant bias from other sources of pleiotropy.

Despite these advantages, there are still some limitations to our study. SNPs are mutations or polymorphisms that occur in a single nucleotide at allele position in a DNA sequence. There are millions of genetic variations in populations, the most common being SNPs, which are IVs in MR analyses. Populations with specific genetic backgrounds have different SNPs, and results from a particular population may have a potential impact on the generality of the findings. Our findings may not be applicable to other ethnicities due to the fact that the study population primarily consisted of individuals from Europe. Additionally, as with all MR studies, residual pleiotropy cannot be completely ruled out. Nevertheless, we tackled this matter by performing a sequence of sensitivity examinations, which contributed to enhancing the reliability of our results. In general, our research offers significant fresh perspectives on the connection between SLE and urological cancers; however, additional investigation is necessary to validate and expand upon these discoveries in various populations.

9. Conclusion

To summarize, this research offers fresh proof of a possible connection between SLE and the emergence of urological malignancies. Particularly, it suggests a decreased likelihood of RCC in individuals with SLE, while the risk of bladder and PCa remains unchanged. However, further research is needed to fully grasp the underlying mechanisms driving this association, and relevant trials are urgently needed to investigate the way SLE specific antibodies act in RCC patients, which will provide potential targets for subsequent RCC treatment options.

Author contributions

Conceptualization: Wenzhi Zhang.

Data curation: Tian An.

Formal analysis: Tian An.

Writing – original draft: Tian An.

Writing – review & editing: Wenzhi Zhang.

Supplementary Material