Phase II Drug-Metabolizing Polymorphisms and Smoking Predict Recurrence of Non–Muscle-Invasive Bladder Cancer: A Gene–Smoking Interaction

Bladder cancer is the 11th most common malignancy and the 14th leading cause of cancer worldwide with 382,700 new cases diagnosed in 2008 (1). Non–muscle-invasive bladder cancer (NMIBC), defined as stage Ta, T1, and Tis (TNM classification; ref. 2) represent approximately 70% to 85% of the newly diagnosed cases. More than 50% of these tumors will recur at least once (3).

Cigarette smoking is the most important known risk factor for urinary bladder cancer, accounting for approximately 50% of all incident cases in men (4), possibly resulting in a reduced overall life expectancy in regular smokers (5). Considerable evidence in the literature suggests that cigarettes smoking increase the risk (up to 2.5-fold) to develop bladder cancer (6, 7), compared with nonsmokers. In a case–control study, cigarette smoking was associated to the subtypes of bladder cancer, the odds associated with regular smokers being 2.2, 2.7, and 3.7 for low-grade NMIBC, high-grade NMIBC, and muscle-invasive tumors, respectively (8). Furthermore, smoking status significantly influences the recurrence-free survival (9, 10), and progression (10) of patients with NMIBC. This is sustained by our observation, in a retrospective study of muscle-invasive bladder cancer, that smoking was an independent prognostic factor for recurrence and cancer-specific survival in patients treated with radical cystectomy (11).

More than 60 carcinogens have been found in cigarette smoke and among them, selected arylamines (including 2-naphthylamine and 4-aminobiphenyl) found in cigarette smoke and other environmental sources have been identified as human bladder carcinogens (4). The oxygenated intermediates formed in initial phase I reactions by cytochromes P450 undergo further transformations by phase II enzymes, including glutathione S-transferases (GST) and uridine diphospho-glucuronosyltransferases (UGT; ref. 12). The GSTs, with significant activity in bladder urothelium (12, 13), catalyze the conjugation of xenobiotics with glutathione, including aromatic hydrocarbons, chlorinated compounds, and some heterocyclic amines, resulting in enhanced elimination of potential carcinogens. Of those, GSTM1 is predominantly involved in the metabolism of arylamines (14). Genetic polymorphisms have been demonstrated for GST genes with one inactivating polymorphism corresponding to GSTM1 gene deletion resulting in reduced transferase activity (12). In smokers, low-activity GSTM1-null genotype has been shown to increase individual susceptibility to bladder cancer (15, 16).

Another detoxification pathway mediated by UGTs metabolize glucuronic acid conjugation to a wide variety of substrates including bilirubin, steroid hormones, drugs, and environmental carcinogens (17). In vitro studies indicate that several UGT enzymes, including the bilirubin-conjugating enzyme UGT1A1, have been implicated in the conjugation and detoxification of the tobacco carcinogens such as benzo [α] pyrene (18), nicotine and nitrosamines such as 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNAL; ref. 19). The presence of a variable number of TA dinucleotide repeat in the TATA region of the UGT1A1 gene promoter [UGT1A1*1 reference allele = TA₆ or UGT1A1*28 = TA₇ (rs8175347)] influences the transcriptional activity of the gene and leads to a significant reduction in UGT1A1 protein expression and its enzyme activity (20).

In this study, we tested whether the patient's smoking status as well as two of the most common and functional genetic variations in GSTM1 (deletion status), and UGT1A1 rs8175347 (low-activity promoter variant) predict the risk of bladder cancer recurrence in patients with NMIBC (21, 22). No study has addressed the link between these functional germline variations with bladder cancer recurrence.

A case series of 189 patients issued from two different cohorts (Table 1) and with sufficient material to perform genetic analysis was used. Details regarding eligibility were reported previously (21, 22) and all patients had provided an informed written consent. Demographics, smoking history, presentation, and clinicopathologic data as well as time to recurrence, progression of disease, and death were available for the 189 patients.

The first group consisted of 131 patients from a bank of 382 newly diagnosed patients with NMIBC recruited between September 1990 and April 1992 from 15 participating hospitals located in the province of Quebec, Canada (22). Of the 382, we had remaining blood samples for 131 patients, obtained at time of preoperative evaluation. These patients were solely treated by transurethral resection of bladder tumor (TURBT) without any immediate postoperative or adjuvant therapy upon recurrence. The resected primary tumors were histologically confirmed as stage pTa or pT1 transitional cell carcinoma (23), and cases of carcinoma in situ (CIS) were excluded for this cohort of patients (22).

The second group was collected between 1997 and 2002. A total of 131 patients with high-risk NMIBC from eight institutions in Canada were offered participation in a prospective study, and all of them received BCG therapy. Inclusion and exclusion criteria were described in a previous publication (21). Of the 131 patients, we had blood samples from the 58 patients from our institution, obtained at time of preoperative evaluation. Six of the 58 cases had concomitant CIS. The TNM classification was used for staging (24).

Genomic DNA was extracted using the QIAmp DNA Blood Mini Kit (Qiagen Inc.) and stored at −80°C. Genotyping was performed by sequencing of PCR amplicons using previously reported PCR strategies (25). Quality control included genotyping of 5% blind duplicate samples distributed across all genotyping batches. Laboratory personnel were blinded to the bladder cancer recurrence status of samples.

The primary clinical outcome for this study was NMIBC recurrence, and was defined as a pathologically confirmed (at re-TURBT) new tumor(s) identified during cystoscopy follow-up after TURBT. Smoking exposure was collected at time of initial TURBT, while other clinical and pathologic characteristics were collected at the time of initial and recurrent TURBT. For recurrence, follow-up time was counted from the date of TURBT until the date of recurrence or the date of the last cystoscopy follow-up visit. Duration of cigarette smoking and numbers of cigarettes smoked was calculated using smoking data collected at baseline. Both factors were analyzed as continuous and categorical variables.

Deviation from Hardy–Weinberg equilibrium was tested using the Pearson χ² test. Cox proportional hazards models was used to test the associations between the sequence variants and the recurrence of NMIBC (26). Cox regression was performed on individual genetic variation as well as the associations with tobacco exposure with adjustment for clinicopathologic variables or all predictors of recurrence. The genotype effect was modeled initially using an additive allelic approach (see Supplementary Table S2). On the basis of this observed effect, we then selected the optimal genetic modeling approach (recessive for UGT1A1; dominant for GSTM1; see Table 3). For all Cox models, the proportionality assumption was examined by standardized Schoenfeld residuals and tested (27). The overall adequacy of the model was assessed by examining the deviance residuals (28). HRs and their 95% confidence intervals (CI) were estimated while taking into account prognostic factors (age, gender, smoking status, stage, grade, presence of CIS, multifocality, tumor size). We then introduced interaction terms in the regression models that assumed a log-additive mode of inheritance for the genotype effects to evaluate for possible gene–smoking interactions. Statistical testing was carried using SAS, version 9.1 (SAS Institute Inc.).

Baseline characteristics of the 189 study subjects are shown in Table 1. The mean age of patients was 62.8 years. Overall, 34.4% of patients were younger than 60 years and 29.1% were older than 70 years. At time of entry, a proportion of 22.8% of patients had pT1 and 68.8% had high-grade disease (58.2% grade 2 and 10.6% grade 3). Most patients were current or previous cigarette smokers with only 15.9% having never smoked and 48.7% being current smokers. In the year before entry into study, 39.7% reported to smoke 1 to 20 cigarettes per day and 44.4% had smoked more than 20 cigarettes per day. Recurrence was observed in 65.1% of the patients, at a median time for recurrence of 12.9 months and a mean follow-up of 5.6 years, while 6.1% had tumor progression (to ≥ pT2 or M1 disease). Only 1.8% of this population died of bladder cancer.

Smoking exposure, age >70 years, and grade 3 tumors were significantly associated with recurrence, whereas no other baseline characteristics were significantly associated with outcome (Table 2). Only 36.6% of nonsmokers had recurrence compared with 70.1% for previous smokers and 70.7% for current smokers (log-rank, P = 0.008). The highest proportion of patients having recurrence (75.0%) was observed among the subgroup of heavy smokers (more than 20 cigarettes/day) at baseline (Supplementary Table S1; Fig. 1).

Table 2 presents the results of univariate and multivariate regression analyses. Previous and current smokers approximately tripled their risk of recurrences (HR = 2.76; 95% CI, 1.03–7.40 and HR = 2.93; 95% CI, 1.08–7.94, respectively). In multivariate analysis, and compared with nonsmokers, moderate smokers (1–20 cigarettes/day) have a 2-fold risk (HR = 2.38; 95% CI, 0.88–6.43), whereas heavy smokers have a 3-fold risk of recurrence (HR = 3.44; 95% CI, 1.28–9.23; data not shown).

In multivariate analysis, and upon adjustment for age, smoking status, stage, grade, gender, and presence of CIS, carriers homozygous for UGT1A1*28 and those with GSTM1 +/− and −/− genotypes were significantly at the risk of tumor recurrence (HR = 1.91; 95% CI 1.01–3.62, and HR = 10.05; 95% CI, 1.35–75.1, respectively; Tables 3a and 3b, respectively).

The risk of bladder cancer recurrence was compared between nonsmokers, previous, and current smokers and the UGT 1A1 promoter status (Table 4a). When compared with nonsmokers and homozygous for the reference UGT1A1*1 allele and those *1/*28 allele (reference group), HR was 2.14 (95% CI, 1.02–4.47) for previous smokers with the UGT1A1*1/*1 and UGT1A1*1/*28 genotypes compared with a risk of 4.95 (95% CI, 1.02–24.0) for homozygous carriers of the UGT1A1*28 alleles. Compared with nonsmokers, a similar interaction was demonstrated for current smokers with a risk of bladder cancer recurrences of 1.91 (95% CI, 0.93–3.92) for UGT1A1 *1 and *1/*28, compared with a risk of 5.32 (95% CI, 2.07–13.7) for homozygous UGT1A1*28/*28 (Likelihood ratio test of 0.039). Similar data were observed when we compared nonsmokers and smokers based on the number of cigarettes smoked per day and UGT 1A1 promoter status (Table 4b; Likelihood ratio test of 0.006; Fig. 2)

Smoking is believed to be one of the main causes of bladder cancer, and similar to our findings, recent reports involved smoking for the risk of recurrence of NMIBC (9, 10). In one of these reports, 46.8% of the ex- or current smokers were recurrence-free compared with 62.3% of nonsmokers (log-rank test; P = 0.005) after a mean follow-up of 2.5 years (9). Here, we demonstrate that NMIBC patients smoking at least 20 cigarettes/day presented a 3-fold increased risk of recurrence compared with nonsmokers. After a mean follow-up of 25 months, 29.9% and 29.3% of the ex- or current smokers were, respectively, recurrence-free compared with 63.3% of nonsmokers (log-rank test; P = 0.008; Supplementary Table S1). Moreover, the multiple smoking exposure definitions suggest a dose–response type of association. Our observations clearly support the detrimental role of smoking on the risk of recurrence of NMIBC. Urologists should certainly recommend to current smokers with bladder cancer to quit smoking. Moreover, our study showed that the genetic status of NMIBC patients for the UGT1A1 and GSTM1 metabolic genes was an additional risk factor.

The homozygous deletion of the GSTM1 genes (i.e., null genotype) results in complete deficiency in enzyme activity (29). Studies have found that the GSTM1 null genotype is associated, alone and in smokers, with an increased risk of bladder cancer (30, 31), but equivocal or negative results have been obtained for GSTT1 or GSTP1 genotypes (30, 31). Only one study has reported, so far, the potential prognostic value of GSTM1 tissue genotype to predict recurrence of bladder cancer (32). In this work (32), the authors studied 165 bladder cancer tumors, of which 61.8% (n = 102) were NMIBC, the remaining being muscle-invasive or N+/M+ bladder cancer patients. In the subgroup of NMIBC, Kaplan–Meier estimate curves demonstrated an increased risk of recurrence for patients harboring a GSTM1-null genotype. In our study, patients carrying at least one GSTM1 deletion displayed an increased risk of NMIBC recurrence (HR = 10.05; 95% CI, 1.35–75.1), while patients harboring a GSTM1-null genotype did not (Table 3b). We are unaware of other studies evaluating the effect of inherited germline variation in GSTM1 on the risk of NMIBC recurrence. From our results and those of others (32), patients with at least one deleted copy of the GSTM1 gene or a null genotype are at an increased risk to develop recurrent bladder cancer, in addition to be associated with an increased risk of bladder cancer (15, 16). This detrimental effect may be related to the inability of those patients to metabolize the potent carcinogens in circulation and hence prolong tissue exposure.

A genome-wide association study also identified the UGT1 locus as a susceptibility gene associated with bladder cancer risk (16), an observation supported by subsequent studies (33, 34). No study has reported, so far, the potential prognostic value of the common UGT1A1 promoter variant associated with mild hyperbilirubinemia (i.e., Gilbert syndrome) on bladder cancer recurrence. In this work, we observed an increased risk of NMIBC recurrence for patients homozygous for the UGT1A1*28 alleles (HR = 1.91; 95% CI 1.01–3.62; Table 3a).

Another noteworthy finding is the interaction between tobacco exposure and UGT1A1 promoter status and the risk of NMIBC recurrence (Table 4). For instance, comparison of nonsmokers to previous smokers based on their UGT1A1 status demonstrated a progressive risk for UGT1A1*1/and *1/*28 individuals and previous smokers (HR = 2.14; 95% CI, 1.02–4.47) compared with nonsmokers. This risk increased to 4.95 (95% CI, 1.02–24.0) for previous smokers carrying the UGT1A1*28/*28 genotype. A similar interaction was demonstrated for current smokers, compared with nonsmokers, with a 1.91 risk of recurrence (95% CI, 0.93–3.92) for those harboring UGT1A1*1/*1 and *1/*28, compared with a risk of 5.32 (95% CI, 2.07–13.7) for UGT1A1*28/*28 (Table 4a; Likelihood ratio test of 0.039). A similar interaction was demonstrated when comparing nonsmokers to smokers but based on numbers of cigarettes smoked per day (Table 4b; Likelihood ratio test of 0.006). It seems that based on a patient's UGT1A1 genotypes and smoking history, we may identify those having a higher likelihood to experience bladder tumor recurrence. In our population of NMIBC, almost 100% of the patients having a recurrence during follow-up were UGT1A1*28/*28 genotype and smoked more than 20 cigarettes per day compared to 14% of recurrence for nonsmokers with the same genotype (Table 4b). The slow metabolism phenotype associated with the UGT1A1*28 low-activity allele (35), may enhance tobacco carcinogens exposure and thus, tumor recurrence. Furthermore, we observed no interaction between GSTM1 and smoking. To our knowledge, no other study evaluated carcinogens exposure such as tobacco and GSTM1 and UGT1A1 genotype abnormalities in NMIBC.

Limitations include the fact that the study is a retrospective study based on two prospective and very different cohorts of patients, recruited many years ago. Different treatment strategies, grading, and staging procedures have been used for them, over time. However, we observed similar gene and smoking effects in each subcohort, suggesting a similar effect despite clinical time-dependent disparities. We have also adjusted for all known clinicopathologic variables, which did not alter substantially the effect, suggesting that residual confounding is unlikely. Population stratification appears also unlikely, as our study cohort is extremely homogeneous (Caucasians). Another potential limitation of this study, and of all candidate gene studies, is the possibility of a false positive association. However, the candidate genes examined herein were chosen based on a biologic rationale and for both loci, prior associations with bladder cancer risk from genome-wide association study (16) support the role of this metabolic pathway in bladder cancer.

We conclude that smoking status and GSTM1 and UGT1A1*28 genotypes predict risk of tumor recurrence. To our knowledge, this is the first publication in bladder cancer patients associating carcinogen metabolic capacity and smoking to the risk of NMIBC recurrence. Studies on larger cohorts of patients will be necessary to validate our findings.

No potential conflicts of interest were disclosed.

Conception and design: L. Lacombe, V. Fradet, E. Levesque, C. Guillemette

Development of methodology: L. Lacombe, V. Fradet

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L. Lacombe, H. Hovington, M. Harvey, Y. Fradet, C. Guillemette

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Lacombe, V. Fradet, E. Levesque, F. Pouliot, A. Caron, M. Nguile-Makao, C. Guillemette

Writing, review, and/or revision of the manuscript: L. Lacombe, V. Fradet, E. Levesque, F. Pouliot, H. Larue, A. Bergeron, H. Hovington, Y. Fradet, C. Guillemette

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Lacombe, A. Bergeron, Y. Fradet

Study supervision: L. Lacombe

This work was supported by intern funds from the Laval University Uro-oncology Research group (to L. Lacombe) and the Canada Research Chairs program (C. Guillemette holds a Tier I Canada Research Chair in Pharmacogenomics).

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.