An Updated Meta-analysis on the Risk of Urologic Cancer in Patients with Systemic Lupus Erythematosus

1Rheumatology Division, Department of Internal Medicine, Gachon University Gil Medical Center, Incheon, Republic of Korea 2Department of Family Medicine, Incheon St. Mary’s Hospital, Incheon, Republic of Korea 3Yonsei University Graduate School of Medicine, Seoul, Republic of Korea 4Department of Family Medicine, Gil Medical Center, Gachon University College of Medicine, Incheon, Republic of Korea


Introduction
Systemic lupus erythematosus (SLE) affects virtually every organ and causes diverse clinical manifestations and laboratory findings. 1 SLE is usually treated with systemic steroids and/or immune suppressants. 2 Recently, biologic agents have been used to increase the survival rate of SLE patients. 3 However, associated conditions (e.g., dyslipidemia, diabetes mellitus, and osteoporosis), and particularly malignancies, still play a crucial role in the long-term outcomes of SLE patients. 4 Several studies have indicated that SLE patients are at increased risk for malignant tumors caused by a decline of the intrinsic immune system and exposure to cytotoxic agents. 5,6 The association between SLE and malignant tumors is still not fully understood, and cancer surveillance for patients with SLE is the same as that for the general population. 7 Urologic cancers are relatively common. Prostate cancer remains the second most frequent cancer and the fifth leading cause of male cancer death, accounting for 7.1% of all new cases of cancer and 3.8% of all cancer deaths worldwide in 2018. Bladder cancer ranks as the tenth most common cancer, with approximately 549 000 new cases and 200 000 deaths worldwide in 2018. 8 Kidney cancer accounted for nearly 403 000 new cases and 175 000 deaths in 2018.
Cammarata et al reported an association between SLE and lymphoma in 1963. 9 Since then, the relationship between SLE and malignancies, including urologic cancers, has been studied. According to previous researches, some potential mechanisms in the development of urologic cancers, especially prostate cancer, are related to hormonal effects and cytokine pathways. 10,11 Until now, several meta-analyses have shown inconclusive results concerning incident urologic cancers in SLE patients. 12 13,14 while Cao et al found that SLE had no effect on the risk of kidney cancer. 12 Recently, Eligibility Criteria and Study Selection Eligible studies fulfilled the following inclusion criteria: 1) observational, case-control, or cohort study; 2) SLE as one of exposures; 3) urologic cancer (prostate, bladder, or kidney) as one of outcomes; 4) general population as control; and 5) providing standardized incidence rate, hazard ratio (HR), relative risk, or odds ratio and the corresponding 95% confidence intervals (CIs). The exclusion criteria were: 1) reviews, editorials, letters, case reports or expert opinions; or 2) duplicated publications or overlapping study populations. Two investigators independently identified articles that met the eligibility criteria through screening all titles and abstracts, followed by reviews of full-text. When there were duplicates from the same database or overlapping populations, we chose the most appropriate study for inclusion in our analysis. Any inconsistency in opinions between the two investigators was resolved by an independent investigator (I.C.H).
Data Collection and Quality Assessment Two investigators (J.Y. and M.S.S) collected the relevant data independently. The extracted data included: first author, year published, country, data source, gender, design type, follow-up periods, observed/estimated number, estimates and 95% CIs, and adjusted covariates. We used a Newcastle-Ottawa Scale (NOS) to assess the methodological quality of study. 17 Based on the median values in our analysis, high quality studies were defined as those with an NOS score of 7 or more. Any disagreements between two investigators were solved through discussion and consensus with all authors.
Statistical Analysis Urologic cancer risk was evaluated using pooled HRs with 95% CIs, which were calculated cumulatively across the selected publication. We also performed subgroup analyses by ethnicity, study design, data source, follow-up period, and study quality. In contrast to the other two cancers, prostate cancer was analyzed only for male subjects. Between-study heterogeneity was estimated by Higgins I 2 (0-100%). 18 When heterogeneity was substantial (I 2 ≥50%), a random (DerSimonian-Laird) effect model was applied. 19 Furthermore, we performed a sensitivity analysis to reflect the impact of different decisions on outcomes by excluding specific study. We checked for publication bias using Begg's rank correlation. We performed statistical analyses using Stata v12.1 (Stata Corp, College Station, TX, USA).

Search and Characteristics
A flowchart of the study identification procedure is shown in Figure 1. A total of 1,011 articles were initially screened through a systematic search. We excluded duplicate publications (n = 430) and papers that appeared irrelevant based on the title (n = 447). After screening the abstracts, we excluded 117 studies due to the following reasons: not original articles (n = 36), not focusing on cancer risk (n = 23), and not including urologic cancers (n = 58). We then conducted a full-text review of 17 articles that fulfilled our eligibility criteria. Six papers were further excluded because of overlapping populations and three papers were added after conducting a manual literature search. Thus, 14 articles were finally included in the analysis.
Detailed information about the studies are presented in Table 1. Our meta-analysis included a total of 83,860 patients with SLE. The publication period was from 1996 to 2018. Eight studies were conducted in Europe, two in Asia, and three in North America. There was also one multinational study from North America, Europe, and Korea. Eleven of the studies were retrospective, and three were prospective. Nine of the studies were hospital-based, while the other five were population-based. Supplementary Table 1 shows the methodological quality of studies. The mean NOS score among the 14 studies was 6.28. Six of the studies were classified as high quality.
Urologic Cancers in SLE There were 10 studies providing HRs for prostate cancers. The pooled HR for prostate cancer among patients with SLE was 0.92 (95% CI, 0.65-1.30; I 2 =74.7%; Figure 2), suggesting that SLE was not a significant risk factor for prostate cancer. However, subgroup analyses restricted by follow-up duration showed that patients with SLE had a reduced risk of prostate cancer in <10-year follow-up studies within the fixed-effects model (HR, 0.68; 95% CI, 0.51-0.89; Table 2).
There was no marked asymmetry in the funnel plot ( Figure 3). No publication biases were noted from the Begg's (P = 0.621) and Egger's tests (P = 0.733).

Discussion
A number of studies have examined the association between SLE and cancers and found an increased overall cancer risk in patients with SLE. [20][21][22] There is accumulating evidence that the risks of non-Hodgkin's lymphoma, leukemia, cancer of the vulva, and lung cancer are increased in patients with SLE, whereas results have been inconsistent for some other solid cancers, including urologic cancers. 23,24 Several meta-analyses have attempted to clarify the relationship between SLE and urologic cancers, but the results have been mixed. [12][13][14][15] We identified a critical error in the data selection for the most recent meta-study by Song et al, 15 who included duplicate data from three studies. 21,25-26 On the other hand, some studies used more updated data. 23,[27][28] Another crucial problem with previous meta-analyses of SLE and urological cancer risk is that some of the metaanalyses missed some principal studies. We included Liu's study instead of Bjornadal's study in our meta-analysis, because the former used more inclusive data. 20,29 We also included Rees's study. 30 Complementing some of the limitations of Song's meta-analysis, 15 our study provides more specific estimates from unduplicated and recent data. In this study, bladder cancer was more commonly developed in SLE patients compared with the general population. Previous meta-studies have shown inconsistent results. [12][13][14][15] However, some researchers suggested that the high risk for bladder cancer in SLE patients might be due to lupus cystitis, viral infection, or SLE treatment such as cyclophosphamide. [31][32][33] In addition, stratified analyses suggested several factors (i.e., data sources, study design and follow-up duration) as potential mediators for the relationships between SLE and bladder cancer. On the other hand, the association of SLE with prostate  cancer remained insignificant. An early meta-analysis reported low risk of prostate cancer in SLE patients; 13 but two subsequent meta-studies did not show a significant relationship between SLE and reduced risk of prostate cancer. 12, 14 Huang's study was hampered by missing data, however, whereas Mao's study contained duplicate data.
To clarify that inconsistency, we used the most recent available data without duplicates. There are some clinical considerations that might shed some light on our result. First, SLE is a female-dominant disease, 34 but prostate cancer only occurs in males, so there might be insufficient data to examine the incident prostate cancer among patients with SLE. Second, prostate cancer has peak incidence in males 65 years of age or older. 35 Interestingly, our subgroup analysis presented a low risk of prostate cancer among studies with <10-year follow-up period. It may be attributed to hormonal effect, which is remarkable in patients at early stage of SLE, a relatively large proportion of whom have severe symptoms. Given the high disease activity of SLE correlated to hypoandrogenism, 10,36 we assumed that patients at early stage SLE have high disease activity but low testosterone level, leading to lower risk for developing prostate cancer. For kidney cancer, we did not find an overall significant association, in contrast to two previous meta-studies. 14,15 In our subgroup analyses, however, the risk of kidney cancer was increased in Western SLE patients but not in Asian SLE patients. The difference in the risk of kidney cancer by ethnicity may be explained partly by genetic backgrounds which influence immunity and tumor genesis. 37 It is well known that the incidence of kidney cancer in the general population is higher in Europe and North America compared to Asia. 38 Considering the higher incidence of kidney cancer in Western populations, we suggest that Western SLE patients should be attentively monitored for the development of kidney cancer.
We found high risk for kidney cancer among nationwide studies but not in hospital-based studies. The nationwide data might have an advantage in that they included the actual primary diagnosis, whereas the hospital-based data were secondary to the identification of cases. Despite efforts to minimize data errors, even the nationwide data may contain coding and keying errors. The use of hospitalbased data requires careful considerations to account for imperfect ascertainment of disease status, 39 because hospital-based studies use administrative data and various algorithms to identify SLE, which may result in reduced quality of diagnoses.
We found that SLE patients had a high risk for kidney cancer among studies with <10 years of followup. That might be partly explained by the intensive use of immunosuppressive agents during the early stage of SLE. 40 Most studies, including our meta-analysis, did not consider the use of immunosuppressants. The cumulative effects of immunosuppressants are another important consideration. 41 Studies with long follow-up reflecting long-term use of immunosuppressive drugs are required to clarify our results.
Although the relationship between SLE and urologic cancers remains unclear, there are some plausible hypotheses that might explain the elevated risk of urologic cancers in patients with SLE. First, the chronic inflammation in SLE can influence the development of urologic cancers. Chronic inflammation enhances the production of inflammatory cytokines, which might increase the development and growth of cancer. 42 In urologic cancers, chronic inflammation promotes tumor evasion and progression. [43][44][45] Second, SLE per se decreases immune-cell function, which results in dysregulation of the immune system. Generally, a healthy immune system removes aberrantly replicating cells to prevent malignancies, whereas this process may be impaired in SLE patients. 46 In particular, the use of immunosuppressive agents has an influence on the risk of cancer in SLE patients. 47 Several case series suggested that immunosuppressive agents increased the risk of bladder cancer in patients with SLE. [48][49][50] Bernasky et al reported that immunosuppressive therapy in patients with SLE increased the bladder cancer risk by 25%. 51 SLE patients who receive immunosuppressive therapy may be exposed to harmful side effects and immune impairment, which may lead to cell damage and promote tumor genesis. 32 Immunosuppressants damage cells directly, and the immune-compromised state impairs the body's ability to repair destroyed or damaged cells and cellular DNA. 52 This study has several limitations. First, there was high heterogeneity in the rates of kidney cancer and bladder cancer across the studies. This might hinder accurate estimation of the risks associated with SLE; however, we conducted subgroup analyses to minimize the confounding factors that contributed to heterogeneity. Second, we could not evaluate the cumulative effects of exposure to immune suppressants and disease activity because of lack of studies that included those data. Third, we could not assess confounding variables, such as smoking, alcohol history, and chronic diseases.
Despite the limitations, our results strongly indicate that SLE increases the risk of bladder cancer, but not those of prostate cancer and kidney cancer. Our results might be helpful to guide physicians that care for patients with SLE and urologic cancers. Our results also add to the body of evidence that can be used to develop better strategies to screen for and follow-up on incident urologic cancers in SLE patients. More well-designed prospective studies and more experimental data are needed to elucidate the underlying mechanisms of the associations between SLE and urologic cancers.