Additive Effects of Arsenic and Aristolochic Acid in Chemical Carcinogenesis of Upper Urinary Tract Urothelium

Arseniasis-associated environmental diseases have been reported in Chile, West Bengal, Bangladesh, and Taiwan (1–3). The incidence of bladder urothelial cancer is related, in a dose-dependent manner, to arsenic exposure from drinking artesian well water (4–6). On the basis of the Taiwan Cancer Registry (TCR), residents of arseniasis-endemic areas have a high risk of developing upper tract urothelial carcinoma (UTUC), as well as bladder urothelial cancer (7). The incidence of UTUC decreases with increasing distance from arseniasis-endemic areas.

Aristolochic acids (AA) are carcinogenic and nephrotoxic constituents of Aristolochia plants. Exposure to these herbs, either as contaminants of wheat flour in the Balkans, or through their use in the practice of traditional Chinese medicine, particularly throughout East Asia, has emerged as a global concern (8–12). Our previous studies showed that the unusually high incidence of UTUC in Taiwan is closely related to AA exposure (10). More than 39% of Taiwanese adults were prescribed AA-containing regimens between 1997 and 2003 (13), underscoring the magnitude of this problem. The widespread use of herbal regimens in Taiwan suggests that AA may be a confounding factor in assessing the etiology of UTUC in residents of arseniasis-endemic areas. The current investigation was undertaken to establish whether the high incidence of UTUC in Taiwan should be attributed to exposure to one of these two carcinogens or is a result of the combined effects of both.

In previous studies (14, 15), Chen and colleagues classified the degree of arseniasis in Taiwanese villages and towns (Fig. 1) into four grades (from 0 to 3) based on the prevalence of arsenic-associated diseases and the concentration of arsenic in artesian well water from these areas. There was a significant positive correlation between the arseniasis grade and the age-adjusted (world population in 2000) incidence of UTUC (Fig. 2; ref. 7). Residents in the grade 3 zone had almost twice the risk of UTUC compared with those in grade 0.

However, studies relating arsenic exposure to the high prevalence of UTUC in Taiwan were conducted prior to the discovery that AA is an important upper tract urothelium carcinogen in humans (1, 10). The goal of this study was to assess the relative contributions of AA and arsenic to the development of UTUC in Taiwanese patients. To do so, we used a case–control study to investigate the impact of AA on the risk of UTUC in a population-based cohort from graded arseniasis areas. As supportive evidence, we evaluated the association of arseniasis and AA exposure in a hospital-based UTUC patient cohort. This is the first study to investigate the combined effect of AA and arsenic exposure on UTUC.

A hospital-based UTUC patient cohort (N = 89) and a population-based cohort from the TCR with UTUC cases (N = 2,921) and matched control (N = 11,684) subjects were included in the study.

A total of 89 patients with UTUC who had a radical nephroureterectomy with bladder cuff resection at National Taiwan University Hospital (NTUH, Taipei, Taiwan) or National Cheng Kung University Hospital (NCKUH, Tainan, Taiwan) between September 2007 and December 2013 were enrolled retrospectively. The study was approved by the Research Ethics Committees at NTUH (Taipei, Taiwan) and NCKUH (Tainan, Taiwan) and the Human Subjects Institutional Review Board at Stony Brook University (Stony Brook, NY). All patients provided signed informed consent prior to donating tissue samples. Patients who received systemic chemotherapy or radiotherapy before tissue donation were excluded from the study. None of the subjects in the cohort reported a history suggestive of occupational exposure to aromatic amines.

A well-established arseniasis grading system (grade 0–3) was used in this study (5, 14). The grading system is based both on arseniasis-related clinical conditions and estimated arsenic exposure in each town/village in Taiwan (Fig. 1). Three factors were used to grade arseniasis: (i) an arsenic concentration of more than 350 ng/mL in the artesian well water in the town/village; (ii) the presence of clinical cases of Blackfoot disease (BFD); and (iii) presence of children with skin lesions characteristic of chronic arseniasis. The grade 3 zone, where all three factors listed above are present, represents the highest level of arseniasis. The grade 2 zone is defined as towns/villages with factors 1 and 3, but not factor 2, that is, no BFD cases. The grade 1 zone refers to towns/villages with only factor 1, but without BFD or skin lesions. Grade 0 areas are residential areas where the arsenic concentration of well water is 350 ng/mL or less.

Tissue collection, ³²P-postlabeling DNA adduct analysis (16), mass spectrometry adduct analysis (17, 18), and TP53 gene sequencing (10) were described previously. This information also is available in Supplementary Materials and Methods and Supplementary Table S1.

Information related to smoking behavior, occupational exposure, detailed residential history, AA exposure, and history of radiotherapy or chemotherapy was collected using a standardized questionnaire. Patients who had had lived in arseniasis-endemic towns for any length of time prior to diagnosis were classified as being exposed to arsenic. Estimated glomerular filtration rate calculated with the Modification of Diet in Renal Disease equation was used for staging chronic kidney disease (CKD; ref. 19). All surgical tissue specimens were reviewed by a single pathologist at NTUH (Taipei, Taiwan; C.T. Shun). Tumors were graded according to the World Health Organization 2004 classification and staged using the tumor–node–metastasis 2002 classification. Multisite tumors were defined as urothelial cancers that were present in two or more of the following sites: renal pelvis, upper ureter, lower ureter, and urinary bladder.

The national cohort data obtained from the National Health Insurance Research Database (NHIRD) are described in detail in Supplementary Materials and Methods. Subjects with newly diagnosed UTUC (ICD-9-CM codes 189.1–189.2) between January 1, 2004 and December 31, 2011 (N = 6,445) were identified in the NHIRD Catastrophic Illness Registry. The flow diagram of inclusions and exclusions were shown in Supplementary Fig. S1. The index dates for cases and matched controls were the date of UTUC diagnosis. A total of 1,562 patients with UTUC diagnosed with other cancers (ICD-9-CM codes 140–208) prior to the index date were excluded. For each case, four matching controls without a cancer diagnosis prior to the index date were randomly selected from the 2005 Longitudinal Health Insurance Dataset (LHID2005; Supplementary Materials). Controls were matched with each UTUC case by sex, age (10 age groups, each spanning at least 5 years from <40 to ≥80), insurance premium, administrative region of residency, urbanization level of residential areas (from level I referring to the most urbanized communities, to level IV referring to the least urbanized communities), arseniasis grades, and comorbidities defined by the following diagnoses within 12 months before the index date: renal insufficiency (ICD-9-CM codes: 585), hepatitis (ICD-9-CM codes: 070), coronary artery disease (CAD; ICD-9-CM codes: 414), chronic obstructive pulmonary disease (COPD; ICD-9-CM codes: 490–496), diabetes mellitus (DM; ICD-9-CM codes: 249–250), and hyperlipidemia (ICD-9-CM codes: 272). Matching controls could not be identified in 1,045 cases of UTUC who had synchronous comorbidities and characters, including renal insufficiency (N = 492), hepatitis (N = 132), CAD (N = 259), COPD (N = 197), DM (N = 401), hyperlipidemia (N = 266), and residing in BFD areas (N = 36). The other 897 UTUC cases with insufficient matching controls were excluded because of similar reasons, including renal insufficiency (N = 272), hepatitis (N = 47), CAD (N = 188), COPD (N = 141), DM (N = 280), hyperlipidemia (N = 222), and residing in BFD areas (N = 30). The final case–control analysis included 2,921 cases and 11,684 controls (four times the number of cases).

According to the Committee on Chinese Medicine and Pharmacy, Ministry of Health and Welfare, Executive Yuan in Taiwan, the following Chinese herbal medicinal products administered prior to November 2003 may have contained AA: Guan Mu Tong (stem Aristolochia manshuriensis), Guang Fangchi (root Aristolochia fangchi), Ma Dou Ling (fruit Aristolochia contorta or Aristolochia debilis), Qing Mu Xiang (root Aristolochia debilis), Tian Xian Teng (leaf and stem Aristolochia contorta, Aristolochia debilis), and Xi Xin (Asarum sieboldii; ref. 20). In Taiwan, importation of these AA-containing herbal products was banned in November 2003 due to cumulative evidence linking these herbs with the development of CKD and UTUC. These herbal products are used either as single products or as components of mixed herbal formulas found in traditional Chinese medicines.

We estimated both the total amounts of herbs consumed for each mixture of the Chinese herbal products, and the total dose of AA-containing herbs consumed between January 1, 1997 and December 31, 2003. To allow for an adequate induction time to develop UTUC after exposure, we calculated the cumulative dose for each herb prescribed to a subject for up to 1 year prior to the UTUC diagnosis date. We also calculated the estimated cumulative dose of AA for each case and control subject using the following estimated average doses of AA per gram of Guan Mu Tong (2.59 mg), Guang Fangchi (2.04 mg), Ma Dou Ling (0.63 mg), Qing Mu Xiang (0.009 mg), Tian Xian Teng (0.026 mg), and Xi Xin (0.042; refs. 21–23).

Contingency tables were constructed for comparisons using the χ² test. The Kruskal–Wallis rank-sum test was used to compare medians between groups. We used conditional logistic regression to assess the risk of UTUC in areas of graded arseniasis stratified by the estimated cumulative dose of AA from all AA-containing herbal products. Crude and adjusted ORs and 95% confidence intervals (CI) for the risk of UTUC were generated adjusting for covariates including aspirin, nonsteroidal anti-inflammatory drugs, and acetaminophen. The combined effects of AA and arsenic exposure on the risk of developing UTUC were assessed using a logistic regression model with details shown in Supplementary Materials and Methods (24, 25). The excess risk of UTUC contributed by arsenic was calculated from the median incidence rates of UTUC from 1990 to 2016. The excess risk of UTUC contributed by AA was calculated on the basis of the population-based case–control study. The proportional contribution to UTUC development is represented by the ratio between excess risks associated with arsenic and AA. All these analyses were conducted using SAS statistical software (version 9.2; SAS Institute). All tests were two-tailed with a P < 0.05 considered as statistically significant.

On the basis of the TCR database, there was a positive correlation between UTUC incidence and arseniasis grade (Fig. 2). The average age-adjusted UTUC incidence rates over the past 10 years were 3.92, 4.71, 5.31, and 8.35 per 100,000 person-years for arseniasis grades 0, 1, 2, and 3 areas, respectively. Increased UTUC incidence in the high arseniasis grade areas was observed in both males and females (Fig. 3). There was also a significant correlation between bladder urothelial cancer incidence and arseniasis grade (Fig. 3). In contrast, there was no association between the arseniasis grade and the incidence of renal cell carcinoma (Fig. 3).

Among the 2,921 cases and 11,684 controls in the population-based national cohort, 49.5% and 57.7% of subjects, respectively, had never consumed herbal medicinal products containing AA (P < 0.001; Table 1). There was a significant dose–response relationship between the amount of AA-containing herbal products consumed and the risk of UTUC.

In arseniasis grade 1–3 areas, the ORs for developing UTUC were 1.48 (95% CI, 1.20–1.81), 1.57 (95% CI, 0.98–2.51), and 1.80 (95% CI, 1.19–2.72) for those who consumed an estimated cumulative dose of 1–100, 101–200, and >200 mg of AA, respectively, compared with individuals who had not consumed AA-containing herbal products. Subjects who lived in the arseniasis grade 0 zone had an elevated risk of UTUC if they had taken AA-containing products compared with those who did not. The increased risk of UTUC from consuming AA-containing herbs was similar among residents living in the arseniasis grade 0 and 1–3 zones. Interaction analysis revealed that the combined effect of AA and arsenic exposure on the risk of developing UTUC was neither synergistic or antagonist, but additive (the relative excess risk due to interaction = 0.12; 95% CI, −0.13–0.39; P = 0.339).

On the basis of the median incidence ratio of UTUC among arseniasis areas and odds ratio of UTUC risk among the people taking graded doses of AA, residents of arseniasis grade 3 areas who had ingested more than 200 mg of AA had the highest risk (relative risk = 3.98) compared with residents of grade 1 areas who had not consumed AA. The relative contributions of AA and arsenic to UTUC risk are shown in Supplementary Table S2 and Supplementary Fig. S2. In general, AA contributes less in regions with higher grades of arsenic exposure.

Among the 89 patients with UTUC (40 males and 49 females), 43 (48.3%), 26 (29.2%), and 20 (22.5%) subjects lived in the arseniasis grade 0, 1 and 2, and 3 zones, respectively. The median age of patients in each cohort was 70.0, 68.5, and 66.5 years, respectively. Most (76.4%) of the UTUC tumors occurred in the renal pelvis, followed by lower ureter (25.8%) and upper ureter (18.0%). A total of 19 (21.3%) patients had synchronous bladder urothelial cancer, defined as bladder cancer diagnosed within 3 months of nephroureterectomy for UTUC. One-third of the patients in the UTUC cohort had synchronous multiple tumor locations. There were no significant differences among residents based on arseniasis grade in terms of age, smoking history, CKD stage, tumor grade, urothelial cancer stage, tumor location, synchronous bladder urothelial cancer, or multisite tumor locations (Table 2).

Mass spectrometry or postlabeling procedures established the presence of aristolactam-DNA adducts in renal cortical tissue samples in 85 (96%) of the 89 patients studied. This result confirms widespread exposure to AA in Taiwan and is consistent with our previous observations (10, 26, 27). Moreover, in 98%, 96%, and 90% of patients with UTUC living in arseniasis grade 0, 1 and 2, and 3 areas, respectively, aristolactam-DNA adducts in renal cortical tissues (Table 2) were identified (P = 0.34).

TP53 mutations were observed in 50%, 60%, and 38.9% of UTUC tissues from patients living in arseniasis grade 0, 1 and 2, and 3 areas, respectively (Table 2). A>T transversion, the most prevalent mutation associated with AA-induced carcinogenesis, was identified in 27.5%, 44.0%, and 22.2% of patients with UTUC living in arseniasis grade 0, 1, and 2, and 3 areas, respectively. There were no significant differences in the frequencies of other types of TP53 mutations among patients from different arseniasis grades (Fig. 4).

We explored whether carcinogens other than AA and arsenic could contribute to the high incidence of UTUC in arseniasis-endemic areas. Previous studies revealed possible synergy between smoking-derived carcinogens and arsenic in inducing bladder urothelial cancer (6, 28–30). However, a history of smoking was reported by only 16.9% of the 89 patients in our hospital-based UTUC cohort, and the proportion of smokers were comparable in residents of arseniasis grade 0 (14.0%) and grade 1+2+3 (19.6%) zones (not significantly different by χ²). These findings suggest that cigarette smoking was not a significant factor in the high incidence of UTUC in arseniasis-endemic areas. This conclusion was further supported by population-based data showing proportions of smokers among residents of Tainan and Chayi, known arseniasis-endemic areas, and the whole population of Taiwan to be similar (Supplementary Fig. S3). Furthermore, industrial contact with urothelial carcinogens was unlikely to play a role in this study as none of the subjects in the hospital-based UTUC cohort reported a history suggestive of occupational exposure to aromatic amines.

Arseniasis-endemic areas are present in Bangladesh, Pakistan, Taiwan, Thailand, Vietnam, Cambodia, the United States, Canada, Chile, and Argentina (1). Populations living in these endemic areas may also be exposed to other carcinogens, such as AA. Taiwan is the only country where the effects of exposure to arsenic and AA are well characterized (4, 10). Accordingly, the Taiwanese population was chosen as a model to study the impact of arsenic and AA on UTUC. Our study is the first to investigate the interaction between arsenic and AA on the risk of developing UTUC, a relationship that appears to be additive. The results of this study can provide useful information to researchers investigating coexposure in other arseniasis-endemic areas in the world.

In earlier studies, we reported that AA-containing herbal medicines were responsible for the high incidence of UTUC observed throughout Taiwan (10). AA carcinogenesis was evaluated with the use of two robust biomarkers that represent AA exposure (aristolactam-DNA adducts in renal cortex) and its carcinogenic effect (A>T transversions in tumor TP53 gene; refs. 10, 31). Using these two biomarkers in this study, we found that the vast majority of patients with UTUC in Taiwan had been exposed to AA, irrespective of where they lived, as aristolactam-DNA adducts were identified in the renal cortex of 94.4% of the hospital-based cohort with UTUC; and A>T mutations in TP53 were found in almost one-third of the urothelial tumors from this cohort. The latter analysis may underestimate the carcinogenic effects of AA as mutations in other cancer driver genes were not examined (9, 27, 32).

In this study, we demonstrate that residents of arseniasis-endemic areas in Taiwan face a higher risk of UTUC compared with those living in nonendemic locations. Furthermore, our population-based data showed that consumption of AA-containing herbal products increased the risk of UTUC in arseniasis-endemic areas compared with residents in the same areas who did not consume these herbs. A>T mutations in TP53 and aristolactam-DNA adducts were present at comparable frequencies in tissues from residents of arseniasis-endemic and nonendemic areas, providing evidence of comparable exposure with AA independent of location.

There are no published data addressing the interaction between arsenic and AA on the risk of developing cancer. One study from Taiwan analyzed the risk of developing urothelial cancer and renal dysfunction by adjusting for exposure to arsenic and suggested a possible interaction between arsenic and AA (33). As predicted, because both arsenic and AA substantially increase the risk of UTUC in Taiwan, this population provides a suitable model to assess the interaction between the two carcinogens. The epidemiologic evidence obtained from our population-based cohort, along with molecular data from our hospital-based cohort, revealed that the carcinogenic effects of AA and arsenic on UTUC are independent of one another, such that their combined action was neither synergistic nor antagonistic, but additive. Currently, there are no data to support whether one chemical is more potent than the other in terms of carcinogenic effects.

The mechanisms responsible for the carcinogenic effect of arsenic in urothelial cancer have not been defined. We were unable to find any mechanistic studies (in vitro or in vivo) investigating the combined effect of AA and arsenic. Although arsenic is not a mutagenic agent, it may act through inhibition of DNA repair (34), sister chromatid exchange (35), or gene amplification (36). Also, arsenic may stimulate cell proliferation by affecting signal transduction pathways and MAPK cascades (37–39). In contrast, AA initiates carcinogenesis by the formation of DNA adducts in a broad spectrum of genes (27, 32). The additive effects on the risk of UTUC as measured by our epidemiologic and molecular analyses confirmed that arsenic and AA act independently from each other.

Our study did not address the combined effects of the two carcinogens on tumor phenotype or the clinical outcomes of UTUC. In earlier studies, we reported that bladder cancer patients from arseniasis-endemic areas had more advanced tumor stages and worse survival outcomes compared with patients from nonendemic areas (4). However, no data were available for UTUC. With the use of TP53 A>T mutations, we previously demonstrated that patients with AA-associated UTUC had more advanced tumor stage and increased UTUC recurrence rates compared with non-AA–associated UTUC (26). Nevertheless, the bladder recurrence rate and overall survival were similar between AA-induced and non-AA–induced patients with UTUC (26). This study showed no significant differences in UTUC phenotype among areas based on arseniasis grade.

In this study, we observed a large difference in the AA exposure rates between the two UTUC cohorts: 50.5% as estimated from the health insurance database compared with 94.4% as detected by aristolactam-DNA adducts. There are several possible reasons for this apparent discrepancy. For example, traditional Chinese medicine prescriptions documented in the health insurance records represent only one route of exposure to AA-containing herbs. But these and other AA-containing herbs may be purchased over the counter or via the internet, which would not be recorded in the health insurance database. In addition, there are various AA-containing herbal regimens that are not banned in Taiwan (40). Finally, the limited 7-year period used for calculating AA doses may contribute to an underestimation of AA exposure in the Taiwanese population.

To investigate the effect of arsenic and AA in UTUC development without significant bias, our population-based case–control study was constructed with the adjustment of the possible factors which contributed to the frequency of physician visit. Therefore, a significant portion of UTUC cases were excluded because of insufficient number or lack of matching controls. The included cases had fewer proportion of renal insufficiency, hepatitis, DM, CAD, COPD, hyperlipidemia than did the overall cases (Supplementary Table S4). Fortunately, the major topic, arseniasis, did not differ between the included and overall cases. Our investigation of effects of arsenic and AA in carcinogenesis of UTUC would be not biased in most of Taiwan population, but could not extend to the people with multiple comorbidities.

Our study has several limitations. First, the number of UTUC cases in the hospital-based cohort was not sufficient to assess all clinical parameters in the multivariable analysis. Nevertheless, it served to prove our hypothesis that most patients with UTUC in arseniasis-endemic areas have also been exposed to AA. Second, this is a retrospective analysis; however, selection bias may have been partially mitigated by recruiting consecutive patients with UTUC. Furthermore, molecular analyses of tissue samples obtained prior to this study may have reduced recall bias. Third, using residential areas to assign arseniasis grading may be imprecise. Because all artesian wells in arseniasis-endemic areas are now closed and prohibited for use, the current arsenic concentrations in these wells are unknown. Nevertheless, this limitation was compensated for by the association between the incidence of arsenic-related disease and arseniasis grade. Fourth, biologic evidence of arsenic exposure was not assessed because nail and/or hair samples were not available. Fifth, residential information in the case–control study is limited to the records provided at the time of registration. Sixth, many UTUC cases were excluded because of the lack or insufficiency of matched controls for patients with multiple comorbidities.

In conclusion, our study revealed that exposure to AA via consumption of Aristolochia herbs significantly increases the risk of UTUC in both non-arseniasis-endemic areas and arseniasis-endemic areas of Taiwan. The combined effects of AA and arsenic toward the risk of developing UTUC appear to be additive.

C.-H. Chen reports grants from National Science Council, Taiwan, during the conduct of the study. Y.-S. Pu reports grants from National Science Council, Taiwan, during the conduct of the study. No disclosures were reported by the other authors.

C.-H. Chen: Conceptualization, resources, data curation, formal analysis, funding acquisition, investigation, methodology, writing–original draft, project administration. A.P. Grollman: Conceptualization, supervision, funding acquisition, validation, methodology, writing–review and editing. C.-Y. Huang: Resources, data curation. C.-T. Shun: Data curation, investigation, methodology. V.S. Sidorenko: Data curation, investigation, methodology. K. Hashimoto: Data curation, investigation. M. Moriya: Data curation, investigation, methodology. R.J. Turesky: Data curation, formal analysis, investigation, methodology. B.H. Yun: Data curation, investigation, methodology. K. Tsai: Data curation, investigation. S. Wu: Data curation, investigation. P.-Y. Chuang: Data curation, software, formal analysis, methodology, writing–review and editing. C.-H. Tang: Supervision, methodology. W.-H. Yang: Resources. T.-S. Tzai: Resources. Y.-S. Tsai: Resources, formal analysis, investigation, writing–review and editing. K.G. Dickman: Conceptualization, data curation, formal analysis, validation, investigation, methodology, project administration, writing–review and editing. Y.-S. Pu: Conceptualization, supervision, funding acquisition, writing–review and editing.

This study was supported, in part, by a research grant (104–2314-B-002–132-) to C.-H. Chen and a research grant (104-2314-B-002-121-MY3) to Y.-S. Pu, both from the National Science Council, Taiwan. This study was also supported by grant (R01CA220367) to R.J. Turesky from the NCI and (R01ES019564) to R.J. Turesky from the National Institute of Environmental Health Sciences. Mass spectrometry was supported by Cancer Center Support Grant (CA077598) from the NCI. K. Tsai and S. Wu were recipients of Stony Brook Medicine International Research Fellowships. A.P. Grollman, V.S. Sidorenko, K. Hashimoto, M. Moriya, and K.G. Dickman gratefully acknowledge the financial support provided by Marsha and Henry Laufer.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.