Boxing Bladder Cancer with COX-2–specific Inhibition

Perspective on Sabichi et al., p. 1580

The body of evidence pointing to the COX-2 pathway playing an important role in the development of bladder neoplasia is compelling and supported by several epidemiologic studies (1). Paralleling early development in colorectal cancer prevention, where the COX-2–specific inhibitor celecoxib [a nonsteroidal anti-inflammatory drug (NSAID)] is clinically effective, preclinical studies confirm that celecoxib can prevent neoplastic transformation of urothelium and reduce urothelial tumor burden in vivo as a single agent or in combination with the immunomodulator Bacillus Calmette-Guérin (BCG; refs. 2, 3), which is standard adjuvant treatment for preventing recurrence of non–muscle-invasive bladder cancer (NMIBC). It is therefore not surprising that celecoxib is being tested in clinical trials in patients with NMIBC.

Sabichi and colleagues report the results of such a trial in this issue of the journal (4), but do the trial's results paint a clear picture of the role of COX-2 inhibition in the setting of bladder cancer prevention? They conducted a phase IIb randomized placebo-controlled trial of celecoxib (200 mg twice a day) for a minimum of 12 months to prevent recurrence in 146 patients following transurethral resection of established NMIBC and adjuvant BCG. Time to recurrence was the primary endpoint, and the statistical design powered the trial to detect a 53% relative reduction in the risk of recurrence at 12 months. The trial was initiated in an era when an effect of celecoxib may have seemed assured, but designing a trial to detect a halving of the recurrence rate in a heavily pretreated population was overly optimistic. It is not surprising that the primary endpoint did not reach statistical significance—the results are based on just 48 events and it is important to keep in mind that the estimated treatment effect was overly ambitious. Nevertheless, the results do support the possibility of a clinically beneficial effect of celecoxib. The adjusted HR was 0.69 [95% confidence interval (CI), 0.37–1.29)], and the absolute 12-month recurrence-free rates were 88% (95% CI, 81%–96%) in the celecoxib group and 78% (95% CI, 69%–89%) in the placebo group (log-rank test: P = 0.17). The study did not detect a difference in the number of adverse cardiovascular (CV) events between the celecoxib and placebo arms, although based on the known CV event rate of celecoxib at the trial's dose and duration, a measurable effect would not have been expected (5).

Certainly, the trends in this study suggesting that celecoxib may well be active in patients with NMIBC will be interpreted with cautious optimism by some, scepticism by others. Fortunately for the ultimate interpretation of these results, the ongoing phase III randomized controlled Bladder COX Inhibition Trial (BOXIT; ISRCTN 84681538; CR UK/07/004) of the same daily dose of celecoxib (albeit 400 mg once a day in BOXIT and 200 mg twice a day in Sabichi and colleagues) versus placebo for 24 months is being conducted in a similar patient population (intermediate- and high-risk NMIBC). Supported by the United Kingdom National Cancer Research Institute, BOXIT will complete accrual of more than 400 patients in early 2012.

The known risk–benefit profile for COX-2 inhibition strategies has changed since the present and BOXIT trials were conceived; indeed, the trial by Sabichi and colleagues (4) was halted for a period following the discovery that adverse CV events were linked to COX-2 inhibitors in randomized controlled trials. The interruption does not seem to have affected the results, and a weighing up of the risk versus benefit for celecoxib in the bladder cancer setting suggests that patients with established NMIBC at a high risk of recurrence potentially will be the most appropriate group for celecoxib treatment. An exploratory analysis of Sabichi and colleagues suggests that the treatment effect was larger in the higher-risk subgroup of patients with multiple recurrences. Although the trial was not powered to look at this effect, the finding is interesting nonetheless. Currently, these higher-risk patients have limited treatment options, and radical cystectomy, which is associated with considerable morbidity, is a standard of care following BCG failure. No oral or intravesical agent heretofore has produced a durable recurrence-free interval that would make it a candidate for standard of care in this group of patients.

What level of benefit will persuade patients and physicians to accept celecoxib for preventing recurrence of NMIBC? A study of celecoxib or aspirin for cancer prevention in the setting of Barrett's esophagus found that patient acceptance of both agents was high, although a preference for aspirin was clear (6). A number of clinical trials testing celecoxib in patients with established cancer have not reported issues relating to acceptance. The field awaits results of a large ongoing CV outcome trial in arthritis patients comparing outcomes among celecoxib, naproxen, and ibuprofen (7) and results of the coxib and traditional NSAID trialists' (CNT) collaboration meta-analysis of data on serious vascular complications associated with coxibs (8), which will inform the risk–benefit discussion; other important data already indicate that celecoxib does not increase CV risk in individuals with a low baseline CV risk (5) or with a low level of C-reactive protein (CRP; ref. 9). The presumption is that the use of COX-2 inhibitors in high-risk cancer-prevention settings seems reasonable, as is born out by successful recruitment to the trial of Sabichi and colleagues in NMIBC. Whether this presumption would hold up in a trial powered to detect a smaller effect size than that assumed by Sabichi and colleagues is not clear.

BOXIT is powered to detect a 15% absolute reduction in 3-year recurrence rates, which translates to an HR of 0.62 (versus placebo); this effect-size estimate is not inconsistent with the encouraging reduction reported by Sabichi and colleagues. The similarities across the 2 trials in terms of celecoxib dose and patients' risk profile would support a future meta-analysis of their combined results.

In conclusion, the present results of Sabichi and colleagues fan interest in continued primary research to advance our understanding of the roles of COX-2 and COX-2 inhibitors in bladder tumorigenesis. Broader efforts to dissect the interaction among tumor cells, the prostaglandin-rich environment, and the inflammatory process continue to focus on the role of COX-2 as an important mediator of neoplastic transformation and cancer progression. There is also continued potential for cross-fertilization between tumor types and between studies to understand COX-2 in the context of genetic susceptibility to cancer development and to explore pharmacogenomic factors with the potential for guiding selection of patients who may benefit most from COX-2 inhibitors. The research community can optimistically view the present results of the phase II trial of Sabichi and colleagues as a positive contribution to a work in progress. The phase III BOXIT is the capstone of this work in NMIBC, with results expected in 2014.

John Kelly is the Chief Investigator and Emma Hall is the Senior Statistician of BOXIT.