PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Int J Cancer. Author manuscript; available in PMC 2014 March 12.
Published in final edited form as:
PMCID: PMC3951299
NIHMSID: NIHMS527574

Nonsteroidal anti-inflammatory drugs and other analgesic use and bladder cancer in northern New England

Abstract

A few epidemiologic studies have found that use of nonsteroidal anti-inflammatory drugs (NSAIDs) is associated with reduced risk of bladder cancer. However, the effects of specific NSAID use and individual variability in risk have not been well studied. We examined the association between NSAIDs use and bladder cancer risk, and its modification by 39 candidate genes related to NSAID metabolism. A population-based case–control study was conducted in northern New England, enrolling 1,171 newly diagnosed cases and 1,418 controls. Regular use of nonaspirin, nonselective NSAIDs was associated with reduced bladder cancer risk, with a statistically significant inverse trend in risk with duration of use (ORs of 1.0, 0.8, 0.6 and 0.6 for <5, 5–9, 10–19 and 201 years, respectively; ptrend = 0.015). This association was driven mainly by ibuprofen; significant inverse trends in risk with increasing duration and dose of ibuprofen were observed (ptrend = 0.009 and 0.054, respectively). The reduced risk from ibuprofen use was limited to individuals carrying the T allele of a single nucleotide polymorphism (rs4646450) compared to those who did not use ibuprofen and did not carry the T allele in the CYP3A locus, providing new evidence that this association might be modified by polymorphisms in genes that metabolize ibuprofen. Significant positive trends in risk with increasing duration and cumulative dose of selective cyclooxygenase (COX-2) inhibitors were observed. Our results are consistent with those from previous studies linking use of NSAIDs, particularly ibuprofen, with reduced risk. We observed a previously unrecognized risk associated with use of COX-2 inhibitors, which merits further evaluation.

Keywords: bladder cancer, nonsteroidal anti-inflammatory drugs, gene–drug interaction, CYP3A

Nonsteroidal anti-inflammatory drugs (NSAIDs) and analgesics are among the most commonly used over-the-counter medications in the United States.1 Among adults 20 years of age or older, the lifetime prevalence of daily use for 1 month or more is approximately 10, 7 and 5% for NSAIDs, aspirin and acetaminophen, respectively.1,2 NSAIDs, including selective cyclooxygenase (COX-2) inhibitors, have been investigated as cancer chemopreventive agents in experimental and observational studies.3,4 The antitumor properties of NSAIDs are thought to be attributable to their capacity to inhibit the COX-2 enzyme, inhibit proliferation, and induce apoptotic cell death5,6; however, other COX-independent pathways may also contribute to the antitumor characteristics of these agents.4,7,8

A number of epidemiologic studies have examined the association between use of NSAIDs and bladder cancer risk. Three case–control studies have observed associations for several classes of NSAIDs with reduced risk of bladder cancer; however, the strength of the association varies by formulation.911 One cohort study reported increased risk,12 two had null findings13,14 and a recent pooled analysis of three large cohort studies suggested a significant protective effect among nonsmokers who reported daily use of nonaspirin NSAIDs.15 The effects of regular intake of selective COX-2 inhibitors on bladder cancer risk have not been studied extensively, because these drugs were introduced to the market in the 1990s.3

The primary goal of this large, population-based case–control study was to determine the reasons for the persistently elevated mortality and incidence rates of bladder cancer among men and women in northern New England.16 In this context, we evaluated use of NSAIDs and analgesics and bladder cancer risk in northern New England based on a detailed assessment of drug composition and dose, obtained through in-person interviews. We also evaluated effect modification by 39 candidate genes in the NSAID metabolism pathway for ibuprofen, a commonly used drug for which we observed the strongest main effect.

Material and Methods

Study design and population

Cases included all patients with a histologically confirmed carcinoma of the urinary bladder (including carcinoma in situ) newly diagnosed between September 1, 2001 and October 31, 2004 (Maine and Vermont) or between January 1, 2002 and July 31, 2004 (New Hampshire) among residents of these three states aged 30–79 years. A total of 1,213 bladder cancer cases (65% of 1,878 eligible cases) were interviewed. Of the 665 cases who did not participate, 50% refused, 22% were deceased, 12% were too ill, 5.5% did not speak English, 5% had a physician refusal and 5% could not be located.

Control subjects aged 30–64 years were selected randomly from Department of Motor Vehicle (DMV) records in each state, and controls aged 65–79 years were selected from beneficiary records of the Centers for Medicare and Medicaid Services (CMS). Controls were frequency matched to cases by state, gender and age at diagnosis/control selection within 5 years. We interviewed 1,418 (594 DMV and 824 CMS) controls (65% of 916 eligible DMV and 65% of 1274 eligible CMS controls). Of control subjects who did not participate, 70% and 65% of DMV and CMS control subjects, respectively, refused; we could not locate 24% of DMV and 11% of CMS control subjects; 3% of DMV and 10% of CMS control subjects did not speak English; 1% of DMV and 7% of CMS control subjects were too ill; and 1% of DMV and 7% of CMS control subjects were deceased.

Individuals who agreed to participate were interviewed at home by a trained interviewer using a computer-assisted personal interview. We obtained detailed information on demographics, use of tobacco products, use of NSAIDs and other analgesics, occupational and residential histories, fluid intake, use of hair coloring products, family history of cancer and dietary factors. The mean and median values for number of months from diagnosis to interview were 8 months and 6 months, respectively. The corresponding values for controls, time from date of selection to interview date, were 8 and 7 months.

Pathology review

We conducted a uniform pathology review of all interviewed cases. An expert pathologist reviewed the initial diagnostic slides and histologic classification. When the pathology material was not available for review (98 cases out of 1,213, 8.1%), we relied on the histology abstracted from the hospital pathology report. Based on the expert pathology review, we excluded 20 patients who were found not to have bladder cancer, and 22 nonurothelial carcinoma cases, leaving 1,171 urothelial carcinoma cases for analysis. Histologic classification was based on the 1973 World Health Organization (WHO)17 and 1998 WHO/International Society of Urologic Pathology classification.18

Drug information

Participants were asked to report lifetime use of NSAIDs and other analgesics. A show card of the 124 brand names of commonly used drugs was provided to the subjects during the interview to aid recall. Drugs were recorded, if they had been used at least 20 times. Detailed information on dose and time periods of use was collected, if the drug was used on a “regular” basis (i.e., two or more times per week for at least 1 month). If the subjects had used a drug more than 20 times but not regularly, they were considered “nonregular” users. Subjects who had never used the drug or had used it less than 20 times were defined as “never” users.

Assessment of exposure to active ingredients

Two pharmacologists developed a drug matrix to translate information on brand names into doses of specific active ingredients over time. The matrix was based on information from publications of the pharmaceutical industry and the Food and Drug Administration, expert opinions of pharmacists, the American Drug Index and Medline Plus. The following categories of analgesics based on their active ingredients were created for analysis: acetaminophen, phenacetin, aspirin, “nonaspirin nonselective NSAIDs” and “selective COX-2 inhibitors.” The compounds included in each category are given in Appendix. For each active ingredient, we calculated duration of use (years) and cumulative lifetime dose (g). To compute lifetime cumulative dose for an active ingredient, we first computed the product of each yearly dose based on the reported dose and frequency of use, as well as the grams of the active ingredient of interest from the drug matrix. We then summed the dose of the active ingredient over the duration of use and over all drugs with the active ingredient.

Genotyping and single nucleotide polymorphism selection

DNA for genotyping was extracted from exfoliated buccal cells collected from mouthwash samples using standard phenol–chloroform extraction methods or using the Autopure protocol (QIAGEN). Genotyping was performed at NCI's Core Genotyping Facility (http://cgf.nci.nih.gov/operations/multiplex-genotyping.html) as part of a genome-wide association scan of bladder cancer, which included subjects in the Maine and Vermont components of the New England Bladder Cancer Study, as previously described in detail (603 cases and 759 controls).19 New Hampshire subjects were not included in the genotyping step. From the available data, we selected 39 candidate genes in NSAID metabolism pathways, defined to include genes known to play a role in the biotransformation of these substrates. Additional genotyping for NAT1 and NAT2 acetylation was conducted at the University of Louisville, School of Medicine, which included subjects from all three states, as described elsewhere.20 For each candidate gene, single nucleotide polymorphisms (SNPs) were selected across the gene as well as 20 kb upstream of the start of transcription and 10 kb downstream of the polyA tail. SNPs were excluded (N = 4), if they showed evidence of deviation from fitness for Hardy–Weinberg proportion (p <1 × 10−5). SNPs with minor allele frequency <5% (N = 254) were excluded from analysis due to limited power. As a result, 984 SNPs in 39 candidate genes in NSAID metabolism pathways were identified (Supporting Information, Table 1).

Table 1
ORs and 95% CIs by analgesic and NSAIDs use status and duration of use1

Statistical analysis

We computed odds ratios (ORs) and 95% confidence intervals (95% CI) using unconditional logistic regression for frequency of use (never, nonregular and regular use), duration of use, and cumulative dose, adjusting for age (<55, 55–64, 65–74 and 75+ years), sex, race/ethnicity (white only, Native American/white and other races), Hispanic status (yes/no), state (Maine, New Hampshire and Vermont) and smoking status (nonsmoker, occasional, former and current smoker)21. More detailed adjustment for smoking as well as further adjustment for other drugs under study and employment in a high-risk occupation for bladder cancer had minimal impact on the ORs, and these factors were not included in the final models. Subjects with missing data on the number of years of use or dose were excluded from the duration and cumulative dose analyses. The following number of cases and controls were missing either duration or dose, or both: acetaminophen, 50 cases and 60 controls; phenacetin, 10 cases and 20 controls; aspirin, 38 cases and 53 controls; ibuprofen, 22 cases and 35 controls; naproxen, 11 cases and 13 controls; celecoxib, 4 cases and 15 controls and refecoxib, 3 cases and 11 controls.

To explore the possibility that drug use was related to symptoms prior to diagnosis, we examined the impact on the risk estimates of excluding drug use during the year prior to the reference date and found little or no impact on estimates of risk.

The reference group in the computations of ORs included subjects who reported “never” using the active ingredient or analgesic category of interest. The cutoff points for cumulative lifetime dose were based on the following: median for phenacetin (because of small numbers of exposed subjects); quartiles with an additional split of the last quartile at the 90th percentile for acetaminophen, aspirin, ibuprofen and naproxen/naproxen sodium; and quartiles for celecoxib and refocoxib among controls.

We used the Wald test to test for linear trend, treating the median value for each category among control subjects as continuous. We also evaluated the effect of exclusive regular use of each active ingredient (i.e., people were treated as exposed, if the active ingredient of interest was the only one that they had ever used regularly), with the reference group defined as those who had never used any NSAIDs or other analgesics. Because 31% of cases and 29% of controls reported regular use of more than one of these drugs, for some NSAIDs the sample size was limited to carry out analyses limited to exclusive users.

To examine whether the association between use of NSAIDs and bladder cancer was modified by smoking, we conducted both stratified analyses and formal tests of interaction. We also examined multiplicative interaction between ibuprofen use (never, nonregular and regular) and SNPs (dominant genetic model) in the NSAID metabolism pathway. Because of lower prevalence of NSAID use and small minor allele frequencies for some SNPs, we chose to use a dominant genetic model to improve on the power to detect an association and to avoid sparse cell counts.

The p value for each SNP–ibuprofen interaction was computed by comparing nested models with and without the cross-product term using a likelihood-ratio test with one degree of freedom. We applied the false discovery rate (FDR; Benjamini–Hochberg adjustment)22 method to identify findings that are not likely due to chance. Interactions were deemed noteworthy at an FDR = 0.20 level. ORs for the joint association between ibuprofen use and genotype on risk of bladder cancer were adjusted for age, race and smoking status and presented for the remaining noteworthy interactions (three SNPs in the CYP3A gene). To reduce false positives and to increase power for detecting more plausible forms of interaction, we applied two additional tests. For the most significant interaction, we applied a new interaction test (constrained test), which imposes biologically more plausible constraints to reduce false positives and to enhance power for plausible interactions.23 This method constrains parameters to allow for only “quantitative” interaction, avoiding effect-reversal or qualitative interactions, where effects for one exposure could go in the opposite direction depending on the level of another exposure. Such qualitative interactions are generally unnatural and are likely to be rare, if they exist. Additionally, we applied an additional interaction test that exploits the independence between the genetic and environmental component.24,25

Results

Table 1 shows bladder cancer risk by use of acetaminophen, phenacetin, aspirin, nonaspirin nonselective NSAIDs and selective COX-2 inhibitors, celocoxib and refocoxib. Compared to never users of acetaminophen, regular users showed a small significant increase in bladder cancer risk (OR = 1.3, 95%CI = 1.1–1.7), but no trend in risk with duration of use was apparent (ptrend = 0.966). Regular use of phenacetin was unrelated to risk of bladder cancer overall (OR = 1.0, 95% CI = 0.6–1.7); we observed an OR of 1.8 (95%CI = 0.7–4.5) for regular use of 3 years or more, but with wide CIs. Regular use of aspirin and other salicylates showed a small, nonsignificant increase in risk (OR = 1.2, 95%CI = 0.9–1.4), but no trend with duration of use was observed (ptrend = 0.460). Regular use of nonaspirin nonselective NSAIDs was associated with a 20%, nonsignificant reduced bladder cancer risk (OR = 0.8, 95%CI = 0.7–1.0), with a significant inverse trend in risk with increasing duration of use (OR of 1.0, 0.8, 0.6 and 0.6 for regular consumption of <5, 5–9, 10–19 and 20+ years, respectively, compared to those who never used these drugs (ptrend = 0.015). Specifically, for use of ibuprofen, the most commonly consumed nonaspirin nonselective NSAIDs, a significant inverse trend in risk with increasing duration was observed (ptrend = 0.009). When we excluded the use of ibuprofen from the analysis, the observed pattern of decreasing risk with increasing duration of use for nonaspirin nonselective NSAIDs remained the same, but the trend was no longer significant (ptrend = 0.146).

In contrast, regular use of selective COX-2 inhibitors was associated with slightly increased risk of bladder cancer overall (OR = 1.3, 95%CI = 1.0–1.7). The trend in risk with increasing duration of use was statistically significant, but not consistent (OR of 1.2, 1.9 and 1.5 for regular use for <2, 2– <3 and 3+ years; ptrend = 0.014).

When specific drugs were examined, use of celecoxib had a positive trend in risk with increasing duration of use (OR of 1.1, 3.1 and 2.0 for regular consumption of <2, 2–<3 and 3+ years, respectively; ptrend = 0.002). The OR for regular use of refocoxib was 1.3 (95%CI = 0.8–1.9) and lacked a clear duration–response relationship with duration (ptrend = 0.733). Numbers were insufficient to compute risk estimates for valdecoxib (three cases, five controls).

For exclusive regular use of a specific drug, the ORs for the following drugs were similar to those presented above: acetaminophen (24 cases, 27 controls, OR = 1.2, 95%CI = 0.6–2.2), aspirin (94 cases, 95 controls, OR = 1.3, 95%CI = 0.8–2.0), nonaspirin nonselective NSAIDs as a group (12 cases, 31 controls, OR = 0.5, 95%CI = 0.2–1.1), ibuprofen (two cases, nine controls, OR = 0.2, 95%CI = 0.2–1.2). Numbers were inadequate to evaluate exclusive use of phenacetin (0 cases, 0 controls), naproxen/naproxen sodium (two cases, five controls), selective COX-2 inhibitors (three cases, five controls), celecoxib (two cases, two controls) and refocoxib (0 cases, two controls).

There was no difference in risk among regular users of acetaminophen, aspirin or naproxen by smoking status. Risk among regular users of ibuprofen was slightly modified by smoking; the ORs were 0.8 (95%CI = 0.5–1.4), 0.8 (95%CI = 0.6–1.1) and 0.6 (95%CI = 0.4–1.0) for never, former and current smokers, respectively. The OR among regular users of celecoxib who never smoked was 2.2 (95%CI = 1.0–4.6), while the ORs for former and current smokers were 1.2 (95%CI = 0.8–1.8) and 4.8 (95%CI = 1.1–21.3), respectively.

Table 2 shows bladder cancer risk by cumulative lifetime dose of each drug (acetaminophen, phenacetin, aspirin, ibuprofen, naproxen/naproxen sodium, celocoxib and refocoxib). We observed a statistically significant inverse trend in risk with cumulative dose of ibuprofen (ptrend = 0.054), with an OR of 0.6 (95%CI = 0.3–1.2) in the highest category of dose. A significant positive trend in risk with increasing dose of celecoxib was found (ptrend = 0.004); the OR for the highest quartile of cumulative dose of celecoxib was 2.3 (95%CI = 1.0–5.1). Dose-response relationships were not detected among regular users of acetaminophen (ptrend = 0.682), phenacetin (ptrend = 0.695), aspirin (ptrend = 0.879), naproxen/naproxen sodium (ptrend = 0.776) or refocoxib (ptrend = 0.599).

Table 2
ORs and 95% CIs by cumulative lifetime dose of analgesics and NSAIDs (g)1

Interactions between NSAIDs candidate genes and ibuprofen use in particular were pursued because of the significantly decreased bladder cancer risk observed for increasing duration and cumulative dose of ibuprofen use. Table 3 shows the SNP-ibuprofen interactions with p-interaction ≤ 0.05, which included SNPs in 13 candidate genes. The most significant p-interaction was observed in the CYP3A locus (p-interaction = 5.969 × 10−6), and this gene also had the highest number of observed interactions with p ≤ 0.05 (N = 24). No other gene in the NSAIDs candidate gene pathway had an interaction with p ≤ 0.05. After applying the FDR correction, three correlated SNPs in CYP3A remained noteworthy (FDR = 0.20 level). Table 4 shows the joint effect of these three CYP3A SNPs and ibuprofen use on bladder cancer risk. Subjects carrying at least one variant allele in rs4646450, or rs15524 who were regular users of ibuprofen had significantly decreased risks of bladder cancer (OR = 0.5, 95%CI = 0.3–0.8; OR = 0.3, 95%CI = 0.2–0.7, respectively), compared to never users with the homozygous wild-type genotype. Given the observation that the most significant interaction in rs4646450 shows a qualitative interaction—where the effect of ibuprofen is protective in a subgroup with “CT+TT” genotypes, whereas it has the opposite effect in a subgroup with “CC” genotype (Table 4) and vice versa for the SNP effect, we applied a new constrained test (see “Material and Methods” section) to investigate if significance of the interaction is driven by the qualitative effects of the SNP and ibuprofen, and to investigate if the interaction is still detectable after ensuring a quantitative interaction where the effects of one exposure are constrained to go in the same direction within each level of the other exposure (and vice versa). The interaction remained statistically significant (constrained test p-value = 1.1 × 10−4), although significance slightly decreased compared to the unconstrained likelihood-ratio test with two degrees of freedom (p-value = 2.9 × 10−5; Fig. 1). Similarly, a modest statistical significance of the interaction remained after applying the independence assumption test (independence assumption, empirical bayes p-value = 0.004).

Figure 1
Constrained/unconstrained test estimates and p-value for rs4646450–ibuprofen interaction.
Table 3
Overview of SNP-ibuprofen interactions where p ≤ 0.05 in cases and controls from Maine and Vermont only1
Table 4
Joint association between CYP3A SNPs and ibuprofen use and risk of bladder cancer from Maine and Vermont only1

Discussion

Our findings suggest that regular use of nonaspirin nonselective NSAIDs, particularly ibuprofen, reduces bladder cancer risk, particularly among regular users for 10 years or more. We also observed a previously unrecognized elevated risk of bladder cancer associated with use of selective COX-2 inhibitors, particularly celecoxib. No consistent relationships were observed for other analgesics or NSAIDs.

Our finding of an inverse association for nonaspirin non-selective NSAIDs has been observed previously,911,15 but not in all13,14 studies. A large population-based case–control study conducted in Southern California found a reduced risk of bladder cancer among nonaspirin NSAIDs users, with estimated ORs between 0.5 and 0.7.9 Regular use of nonaspirin nonselective NSAIDs was associated with a similar reduction in risk in a large, hospital-based case–control study in Spain.10 A population-based case–control study in New Hampshire covering the period from 1998 to 2001 observed an inverse association between bladder cancer risk and regular use of any NSAIDs, including aspirin, with a significant inverse trend in risk with duration of use.11 When analyses were restricted to ibuprofen use, the ORs remained below one for heavy users, although a trend of decreasing risk with increasing duration was not statistically significant. A record linkage cohort study in Denmark showed no association between bladder cancer risk and use of prescription nonaspirin NSAIDs.14 No association was observed with overall use of nonaspirin NSAIDs in a large cohort of U.S. men in the Health Professionals Follow-Up Study.13 Neither cohort study had information on duration of use or dose. A recent pooled analysis of three large cohort studies with information on frequency of use found a significant protective effect but only among nonsmokers reporting regular use of nonaspirin NSAIDs.15 In contrast, we did not observe any significant interaction between NSAIDs and smoking; risk reduction was apparent in both smokers and nonsmokers.

Ibuprofen is a nonselective COX inhibitor, which inhibits both COX-1 and COX-2 enzymes. Overexpression of COX-2 is associated with proliferation, angiogenesis and deregulation of apoptosis in bladder cancer cells.26 COX-2 is up-regulated in urothelial carcinomas, 27with a higher likelihood of expression in invasive, high-grade tumors than in superficial and low-grade tumors.28,29 COX-2 inhibitors have been shown to reduce bladder tumor growth in mouse and canine models.6,30

Ibuprofen also induces expression of a cell surface receptor glycoprotein, p75 neurotrophin receptor (p75NTR), which is a COX-independent mechanism that might account for ibuprofen's potential to lower bladder cancer risk in addition to its ability to inhibit the COX-2 enzyme.7,31 p75NTR shares both structural and sequence homology with the tumor necrosis factor receptor superfamily of proteins, and was identified as a tumor suppressor and metastasis suppressor of bladder cancer cells.32 Interestingly, ibuprofen was ranked highest among NSAIDs for its ability to induce expression of p75NTR and to inhibit bladder cancer cell survival in in vitro studies.7 It is noteworthy that NSAIDs that lack COX-inhibitory activity (e.g., R-flurbiprofen) have been shown to produce similar anticancer effects in in vivo33 and in vitro studies.8 Differences in COX-independent mechanisms may result in variation in the antitumor effects of individual NSAIDs.

A recently introduced selective COX-2 inhibitor, celecoxib, has been proposed for chemoprevention and treatment of bladder cancer.6,34 Thus, our finding of an increased risk of bladder cancer associated with use of selective COX-2 inhibitors was unexpected. Celecoxib, in contrast to other commonly used NSAIDs, may not inhibit but activate nuclear factor-kappa B (NF-κB) and NF-κB gene transcription,8 especially at high doses.35 Increased NF-κB expression is associated with bladder cancer progression.36 Further, COX-2 inhibitors have been reported to impair bladder contraction, resulting in increased acute urinary retention37,38 a probable risk factor for bladder cancer.39 While this finding is intriguing, the lack of dose–response for the individual drugs as well as an inadequate latent period to evaluate risk demand cautious interpretation of our results. Given that selective COX-2 inhibitors became commercially available in the late 1990s, and our cases were diagnosed between 2001 and 2004, unless selective COX-2 inhibitors operate as powerful late stage bladder carcinogens, we cannot rule out chance as a possible explanation of our finding.

In our study, aspirin did not modify bladder cancer risk. The potential effect of aspirin on bladder cancer risk remains equivocal. Three population-based case–control studies found an inverse association between regular use of aspirin and bladder cancer risk,9,11,40 while a large, hospital-based case–control study observed no association.10 Additionally, four cohort studies have reported null associations for aspirin use,13,15,41,42 although one cohort study showed an increased risk among women.12 Phenacetin, an analgesic used in many drug mixtures before being banned in the United States in the early 1980s, has been associated with an elevated risk of cancer of the renal pelvis.43 In 1987, the International Agency for Research on Cancer classified phenacetin as a probable carcinogen in humans (Group 2A).44 Use of phenacetin has been associated with increased bladder cancer risk in epidemiologic studies.16 Although our results do not support an overall association with phenacetin, an increased risk was observed among long-term users. This finding, however, is based on small numbers because of the low prevalence of phenacetin use in our study population. The epidemiologic data for acetaminophen, a phenacetin metabolite and bladder cancer risk are relatively sparse and inconclusive, with two studies showing support for a small increase in risk10,45 and others indicating no effect or decreased risk.9,11,13 Acetaminophen showed a small nonsignificant increased risk among regular users, however, no dose–response relationship with duration of use or cumulative dose was apparent.

Our study provides new evidence that the protective effect of ibuprofen might be modified by polymorphisms in the CYP3A locus. We investigated whether significance of the interaction was driven merely by the qualitative effects of the SNP and ibuprofen, and whether the interaction was still detectable after ensuring a quantitative interaction where the effects of one exposure were constrained to go in the same direction on each level of the other exposure, but differing only in the magnitude of effects. Application of a new constrained test showed significance of the interaction (constrained test p-value = 1.1 × 10−4), which implies that interaction between ibuprofen and rs4646450 still exists after ensuring a quantitative interaction. Because we also observed a similar effect with two other highly correlated SNPs, rs15524 and rs4646457 and there is limited information about the functionality of all three SNPs, it is unclear, which may be driving the association.

The CYP3A locus is located on chromosome 7 and consists of genes CYP3A43, CYP3A4, CYP3A7 and CYP3A5.46 Some in vitro studies using liver microsomes suggest that at lower concentrations, ibuprofen undergoes metabolism by 2-hydroxylation and 3-hydroxylation almost exclusively through processes mediated by CYP2C9; however, as concentrations of ibuprofen increase, CYP3A4 expression is activated to assist with the metabolism of excess ibuprofen.47,48 It should be noted that the SNPs that showed the most significant interaction with ibuprofen are located around the CYP3A5 region. CYP3A5 shares ~ 90% sequence identity of its cDNA with CYP3A4.46 If the variant alleles of CYP3A are involved with decreasing the expression of active CYP3A, the increased metabolism of excess ibuprofen by CYP3A would not occur, providing more available active ibuprofen to exert anticancer effects, thus supporting the observed reduction in bladder cancer risk. Further work will be needed to attempt to replicate the finding and if positive, to identify the functional alleles responsible for the interaction.

Our study has several strengths, including its population-based design, large sample size and detailed assessment of drug composition by duration of use and dose, allowing us to compute cumulative lifetime exposures and to discriminate among the effects of specific NSAIDs. In addition, despite the large proportion of people who consumed two or more types of NSAIDs, we were able to conduct analyses among exclusive users to isolate the effects of specific NSAIDs (when numbers permitted) as well as control for other potential confounders.

Our study has several possible limitations. First, it is possible that knowledge of the withdrawal of refocoxib from the U.S. marketplace in 2004 may have influenced cases' recall of past use of selective COX-2 inhibitors. However, it is unlikely that our findings were subject to such recall bias given that safety concerns regarding selective COX-2 inhibitors were unrelated to bladder cancer. Furthermore, risk estimates for selective COX-2 inhibitors observed among subjects interviewed prior to 2004 were similar to those interviewed later. Second, we had a very short latency period to examine risk associated with selective COX-2 inhibitors. Third, selection bias may have occurred if nonparticipants differed from participants with respect to drug use and their decision as to whether to participate was related to their disease status. It is improbable, however, that study participation differed between cases and controls in a dose-dependent manner. If study participation was related to drug use in the same way for cases and controls, this would have biased the estimates of risk toward the null. Such nondifferential selection bias seems improbable, because in general, participants and non-participants were similar with respect to sociodemographic factors such as age, sex and state of residence (data not shown). Another potential limitation is that the case patients who started taking NSAIDs only in the last 2 months prior to the diagnosis of their bladder cancer were classified as “regular users.” If the case patients' use of NSAIDs was triggered by their symptoms related to their undiagnosed tumor, this would distort the ORs resulting in erroneously observed risk between bladder cancer and NSAIDs use. To address this concern, we reran our analysis after redefining regular use to exclude NSAIDs use one year prior to diagnosis/control selection. Our results were not changed.

As mentioned earlier, the main purpose of our study was to determine the reason for the excess bladder cancer incidence and mortality in northern New England. The prevalence of regular use of ibuprofen among our controls was similar to that of the U.S. population (i.e., 4.4 and 7% among men and 8.6 and 8% among women over 65 years of age, respectively),49 so it is unlikely that low prevalence of use of ibuprofen could explain the excess of bladder cancer in New England. In addition, risk estimates for ibuprofen in our study are similar to those in a study conducted in other parts of the country.50 Use of celecoxib is unlikely to play a role in the long standing New England excess of bladder cancer, because this drug was introduced into the market in 1998.

In summary, our data provide further evidence that use of nonaspirin nonselective NSAIDs, particularly ibuprofen, reduces bladder cancer risk. We provide new evidence that the protective effect of ibuprofen might be modified by polymorphisms in genes that metabolize this drug. We also found an unexpectedly increased risk of bladder cancer in relation to use of selective COX-2 inhibitors, particularly celecoxib, warranting further investigation.

Acknowledgments

The authors thank Anna McIntosh, Paul Hurwitz, Patricia Clark and Vanessa Olivo (Westat, Rockville, MD) for their support in study and data management, and Anne Taylor and Mary McAdams (IMS, Silver Spring, MD) for their programming help. They acknowledge Dr. Michael Jones (Maine Medical Center), Sue Ledoux and Dawn Nicolaides (Maine Cancer Registry), Kimberley Walsh and Christina Robinson (Dartmouth Medical School), Dr. Masatoshi Kida (University of Vermont), William Apao and Carolyn Greene (Vermont Cancer Registry) for their contributions during the fieldwork and data collection phases. They also thank all fieldwork staff, interviewers and data abstractors for their dedicated work, and our study participants for agreeing to be part of this study. None of the authors have conflicts of interest that are relevant to the subject matter or materials discussed in the manuscript. The authors had full responsibility for the design of the study, collection of the data, the analysis and interpretation of the data, the decision to submit the manuscript for publication, and the writing of the manuscript.

Grant sponsor: Intramural Research Program of the National Institutes of Health, National Cancer Institute, Division of Cancer Epidemiology and Genetics; Grant number: N02-CP-01037

Appendix

Acetaminophen

Phenacetin

Aspirin and other salicylates

salyclic acid = magnesium salicylate, magnesium salicylate tetrahydrate, salicylamide, choline salicylate, diflunisal, salsalate, sodium salicilate, other drugs that contain salicylate

Nonaspirin NSAIDs

propionic acid = ibuprofen, ketoprofen, flurbiporfen, fenoprofen, ibuprofen potassium, naproxen, naproxen sodium, oxaprozin acetic acid= indomethacin, sulindac, etodolac, tolmetin sodium, diclofenac sodium, diclofenac potasium

fenamic acid= meclophenamic acid, mefenamic acid, meclofenamate sodium oxicam=piroxicam, meloxicam pyrazolidine derivatives=phenylbutazone

Selective COX-2 inhibitors (celecoxib, rofecoxib, valdecoxib)

Footnotes

Additional Supporting Information may be found in the online version of this article.

*Published 2012. This article is a US Government work and, as such, is in the public domain of the United States of America.

References

1. Paulose-Ram R, Hirsch R, Dillon C, et al. Prescription and non-prescription analgesic use among the US adult population: results from the third National Health and Nutrition Examination Survey (NHANES III) Pharmacoepidemiol Drug Saf. 2003;12:315–26. [PubMed]
2. Paulose-Ram R, Hirsch R, Dillon C, et al. Frequent monthly use of selected non-prescription and prescription non-narcotic analgesics among U.S. adults. Pharmacoepidemiol Drug Saf. 2005;14:257–66. [PubMed]
3. Harris RE, Beebe-Donk J, Doss H, et al. Aspirin, ibuprofen, and other non-steroidal anti-inflammatory drugs in cancer prevention: a critical review of non-selective COX-2 blockade (review) Oncol Rep. 2005;13:559–83. [PubMed]
4. Ulrich CM, Bigler J, Potter JD. Non-steroidal anti-inflammatory drugs for cancer prevention: promise, perils and pharmacogenetics. Nat Rev Cancer. 2006;6:130–40. [PubMed]
5. Harris RE, Beebe-Donk J, Alshafie GA. Similar reductions in the risk of human colon cancer by selective and nonselective cyclooxygenase-2 (COX-2) inhibitors. BMC Cancer. 2008;8:237. [PMC free article] [PubMed]
6. Pruthi RS, Derksen E, Gaston K. Cyclooxygenase-2 as a potential target in the prevention and treatment of genitourinary tumors: a review. J Urol. 2003;169:2352–9. [PubMed]
7. Khwaja F, Allen J, Lynch J, et al. Ibuprofen inhibits survival of bladder cancer cells by induced expression of the p75NTR tumor suppressor protein. Cancer Res. 2004;64:6207–13. [PubMed]
8. Tegeder I, Pfeilschifter J, Geisslinger G. Cyclooxygenase-independent actions of cyclooxygenase inhibitors. FASEB J. 2001;15:2057–72. [PubMed]
9. Castelao JE, Yuan JM, Gago-Dominguez M, et al. Non-steroidal anti-inflammatory drugs and bladder cancer prevention. Br J Cancer. 2000;82:1364–9. [PMC free article] [PubMed]
10. Fortuny J, Kogevinas M, Garcia-Closas M, et al. Use of analgesics and nonsteroidal anti-inflammatory drugs, genetic predisposition, and bladder cancer risk in Spain. Cancer Epidemiol Biomarkers Prev. 2006;15:1696–702. [PubMed]
11. Fortuny J, Kogevinas M, Zens MS, et al. Analgesic and anti-inflammatory drug use and risk of bladder cancer: a population based case control study. BMC Urol. 2007;7:13. [PMC free article] [PubMed]
12. Ratnasinghe LD, Graubard BI, Kahle L, et al. Aspirin use and mortality from cancer in a prospective cohort study. Anticancer Res. 2004;24:3177–84. [PubMed]
13. Genkinger JM, De Vivo I, Stampfer MJ, et al. Nonsteroidal antiinflammatory drug use and risk of bladder cancer in the health professionals follow-up study. Int J Cancer. 2007;120:2221–5. [PubMed]
14. Sorensen HT, Friis S, Norgard B, et al. Risk of cancer in a large cohort of nonaspirin NSAID users: a population-based study. Br J Cancer. 2003;88:1687–92. [PMC free article] [PubMed]
15. Daugherty SE, Pfeiffer RM, Sigurdson AJ, et al. Nonsteroidal antiinflammatory drugs and bladder cancer: a pooled analysis. Am J Epidemiol. 2011;173:721–30. [PMC free article] [PubMed]
16. Silverman DT, Devesa SS, Moore LL, Rothman N. Bladder cancer. In: Schottenfeld D, Fraumeni JF Jr, editors. Cancer epidemiology and prevention. New York: Oxford University Press; 2006.
17. Sobin LH. The WHO histological classification of urinary bladder tumours. Urol Res. 1978;6:193–5. [PubMed]
18. Epstein JI, Amin MB, Reuter VR, et al. The World Health Organization/International Society of Urological Pathology consensus classification of urothelial (transitional cell) neoplasms of the urinary bladder. Bladder Consensus Conference Committee. Am J Surg Pathol. 1998;22:1435–48. [PubMed]
19. Rothman N, Garcia-Closas M, Chatterjee N, et al. A multi-stage genome-wide association study of bladder cancer identifies multiple susceptibility loci. Nat Genet. 2010;42:978–84. [PMC free article] [PubMed]
20. Koutros S, Silverman DT, Baris D, et al. Hair dye use and risk of bladder cancer in the new england bladder cancer study. Int J Cancer. 2011;129:2894–904. [PMC free article] [PubMed]
21. Baris D, Karagas MR, Verrill C, et al. A case–control study of smoking and bladder cancer risk: emergent patterns over time. J Natl Cancer Inst. 2009;101:1553–61. [PMC free article] [PubMed]
22. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Statist Soc Ser B. 1995;57:289–300.
23. Han S, Rosenberg P, Chatterjee N. Testing for gene-environment and gene–gene interactions under monotonicity constraints. 2012 In press.
24. Chatterjee N, Kalaylioglu Z, Carroll RJ. Exploiting gene-environment independence in family-based case–control studies: increased power for detecting associations, interactions and joint effects. Genet Epidemiol. 2005;28:138–56. [PubMed]
25. Mukherjee B, Chatterjee N. Exploiting gene-environment independence for analysis of case-control studies: an empirical Bayes-type shrinkage estimator to trade-off between bias and efficiency. Biometrics. 2008;64:685–94. [PubMed]
26. Klein RD, Van Pelt CS, Sabichi AL, et al. Transitional cell hyperplasia and carcinomas in urinary bladders of transgenic mice with keratin 5 promoter-driven cyclooxygenase-2 overexpression. Cancer Res. 2005;65:1808–13. [PubMed]
27. Shariat SF, Kim JH, Ayala GE, et al. Cyclooxygenase-2 is highly expressed in carcinoma in situ and T1 transitional cell carcinoma of the bladder. J Urol. 2003;169:938–42. [PubMed]
28. Shirahama T, Arima J, Akiba S, et al. Relation between cyclooxygenase-2 expression and tumor invasiveness and patient survival in transitional cell carcinoma of the urinary bladder. Cancer. 2001;92:188–93. [PubMed]
29. Mohammed SI, Knapp DW, Bostwick DG, et al. Expression of cyclooxygenase-2 (COX-2) in human invasive transitional cell carcinoma (TCC) of the urinary bladder. Cancer Res. 1999;59:5647–50. [PubMed]
30. Grubbs CJ, Lubet RA, Koki AT, et al. Celecoxib inhibits N-butyl-N-(4-hydroxybutyl)-nitrosamine-induced urinary bladder cancers in male B6D2F1 mice and female Fischer-344 rats. Cancer Res. 2000;60:5599–602. [PubMed]
31. Ratliff TL. Aspirin, ibuprofen, and other non-steroidal anti-inflammatory drugs in cancer prevention: a critical review of non-selective COX-2 blockade (review) J Urol. 2005;174:787–8. [PubMed]
32. Khwaja F, Djakiew D. Inhibition of cell-cycle effectors of proliferation in bladder tumor epithelial cells by the p75NTR tumor suppressor. Mol Carcinog. 2003;36:153–60. [PubMed]
33. Lim JT, Piazza GA, Han EK, et al. Sulindac derivatives inhibit growth and induce apoptosis in human prostate cancer cell lines. Biochem Pharmacol. 1999;58:1097–107. [PubMed]
34. Sabichi AL, Lee JJ, Grossman HB, et al. A randomized controlled trial of celecoxib to prevent recurrence of nonmuscle-invasive bladder cancer. Cancer Prev Res (Phila) 2011;4:1580–9. [PubMed]
35. Niederberger E, Tegeder I, Vetter G, et al. Celecoxib loses its anti-inflammatory efficacy at high doses through activation of NF-kappaB. FASEB J. 2001;15:1622–4. [PubMed]
36. Karashima T, Sweeney P, Kamat A, et al. Nuclear factor-kappaB mediates angiogenesis and metastasis of human bladder cancer through the regulation of interleukin-8. Clin Cancer Res. 2003;9:2786–97. [PubMed]
37. Gruenenfelder J, McGuire EJ, Faerber GJ. Acute urinary retention associated with the use of cyclooxygenase-2 inhibitors. J Urol. 2002;168:1106. [PubMed]
38. Verhamme KM, Dieleman JP, Van Wijk MA, et al. Nonsteroidal anti-inflammatory drugs and increased risk of acute urinary retention. Arch Intern Med. 2005;165:1547–51. [PubMed]
39. Silverman DT, Alguacil J, Rothman N, et al. Does increased urination frequency protect against bladder cancer? Int J Cancer. 2008;123:1644–8. [PubMed]
40. Steineck G, Wiholm BE, Gerhardsson de Verdier M. Acetaminophen, some other drugs, some diseases and the risk of transitional cell carcinoma. A population-based case-control study. Acta Oncol. 1995;34:741–8. [PubMed]
41. Paganini-Hill A, Chao A, Ross RK, Henderson BE. Aspirin use and chronic diseases: a cohort study of the elderly. BMJ. 1989;299:1247–50. [PMC free article] [PubMed]
42. Thun MJ, Henley SJ, Patrono C. Nonsteroidal anti-inflammatory drugs as anticancer agents: mechanistic, pharmacologic, and clinical issues. J Natl Cancer Inst. 2002;94:252–66. [PubMed]
43. Angervall L, Bengtsson U, Zetterlund CG, et al. Renal pelvic carcinoma in a Swedish district with abuse of a phenacetin-containing drug. Br J Urol. 1969;41:401–5. [PubMed]
44. IARC. Overall evaluation of carcinogenecity: an updating IARC monographs volumes 1 to 42. Vol. 7. Lyon, France: International Agency for Research on Cancer; 1987.
45. Derby LE, Jick H. Acetaminophen and renal and bladder cancer. Epidemiology. 1996;7:358–62. [PubMed]
46. Xie HG, Wood AJ, Kim RB, et al. Genetic variability in CYP3A5 and its possible consequences. Pharmacogenomics. 2004;5:243–72. [PubMed]
47. Chang SY, Li W, Traeger SC, et al. Confirmation that cytochrome P450 2C8 (CYP2C8) plays a minor role in (S)-(+)- and (R)-(−)-ibuprofen hydroxylation in vitro. Drug Metab Dispos. 2008;36:2513–22. [PubMed]
48. Hamman MA, Thompson GA, Hall SD. Regioselective and stereoselective metabolism of ibuprofen by human cytochrome P450 2C. Biochem Pharmacol. 1997;54:33–41. [PubMed]
49. Kaufman DW, Kelly JP, Rosenberg L, et al. Recent patterns of medication use in the ambulatory adult population of the United States: the Slone survey. J Am Med Assoc. 2002;287:337–44. [PubMed]
50. Gago-Dominguez M, Yuan JM, Castelao JE, et al. Regular use of analgesics is a risk factor for renal cell carcinoma. Br J Cancer. 1999;81:542–8. [PMC free article] [PubMed]