Aspirin and COX-2 Inhibitor Use in Patients With Stage III Colon Cancer

Randomized trials support the efficacy of aspirin and COX-2 (cyclooxygenase; prostaglandin-endoperoxide synthase-2 [PTGS2]) inhibitors in reducing adenoma and cancer risk in patients with familial colorectal cancer (CRC) syndromes (1–9). Meta-analyses of randomized cardiovascular disease (CVD) prevention trials confirm the protective effect of aspirin against CRC (10–12), and observational studies report improved survival with postdiagnosis aspirin use (13–16). To test the hypothesis that aspirin and COX-2 inhibitors may be effective in the adjuvant setting, we conducted a prospective analysis of aspirin and COX-2 inhibitor use in stage III colon cancer patients enrolled in CALGB 89803 (1999–2001) (17). In an abstract, we reported improved recurrence-free (RFS), disease-free (DFS), and overall survival (OS) associated with these medications with a median follow-up of 2.7 years (18). These findings led to two ongoing phase III trials, Alliance for Clinical Trials in Oncology study CALGB 80702 and the Aspirin for Dukes C and High Risk Dukes B Colorectal Cancers study (ASCOLT) (19), but results are not expected for many years. Herein we report updated findings from CALGB 89803 with mature follow-up. CALGB 89803 compared fluorouracil (FU) and leucovorin (LV) with irinotecan, FU, and LV for adjuvant treatment of American Joint Committee on Cancer stage III colon cancer and found no statistically significant difference in outcome (17). A self-administered questionnaire assessing diet, lifestyle, and medication use was conducted midway through chemotherapy (Q1) and six months after chemotherapy (Q2). Consistent aspirin use was defined as any aspirin use reported on both Q1 and Q2, and COX-2 inhibitor use as any use reported on Q2 (Supplementary Figure 1, available online). All patients signed informed consent, approved by each institution’s review board. RFS was calculated as the time from Q2 completion to tumor recurrence, death with recurrence, or development of a new invasive colon cancer (n = 1). DFS was defined as time from Q2 to tumor recurrence, occurrence of a new colon cancer, or death from any cause. OS was defined as time from Q2 to death from any cause. Survival was examined using Kaplan-Meier curves (20) and the log-rank test (21). Cox proportional hazards regression was used to simultaneously adjust for potential confounders (22); proportionality of hazards assumption was satisfied by time-dependent covariables and the Schoenfeld residuals method. Because colon cancer recurrences and deaths are rare after five years (23), we conducted a secondary analysis with DFS and OS events censored at five years to minimize misclassification because of noncancer deaths. Statistical significance was at the .05 level with two-sided P values. Among 799 patients who responded to the aspirin question, 75 (9.4%) reported use both during and after chemotherapy. Consistent aspirin users were older and more likely to be male. Among 843 patients with data on COX-2 inhibitor use, 59 (7.0%) reported use after chemotherapy. COX-2 inhibitor users were less likely to have a family history of cancer, had higher body mass index, and reported more acetaminophen use (Supplementary Table 1, available online). After a median follow-up of 6.5 years, consistent aspirin use was associated with improved RFS (83.1% vs 74.9% at five years, adjusted hazard ratio [HR] = 0.51, 95% confidence interval [CI] = 0.28 to 0.95) (Figure 1A), DFS (78.4% vs 71.1%, HR = 0.68, 95% CI = 0.42 to 1.11), and OS (87.6% vs 80.9%, HR = 0.63, 95% CI = 0.35 to 1.12) (Table 1). HRs for DFS and OS censored at five years were 0.61 (95% CI = 0.36 to 1.04) and 0.48 (95% CI = 0.23 to 0.99) (Table 1; Figure 1, ​,BB and ​andC).C). Patients who used five or more tablets/week had an HR of 0.69 (95% CI = 0.41 to 1.18) compared with nonusers (P trend = .15). COX-2 inhibitor use was also associated with improved RFS (85.6% vs 74.5% at five years, HR = 0.53, 95 % CI = 0.27 to 1.04) (Figure 1D), DFS (83.8% vs 70.6%, HR = 0.60, 95% CI = 0.33 to 1.08), and OS (94.3% vs 80.3%, HR = 0.50, 95% CI = 0.23 to 1.07) (Table 1). Results were stronger when events were censored at five years (HR for DFS = 0.47, 95% CI = 0.24 to 0.91; HR for OS = 0.26, 95% CI = 0.08 to 0.81) (Table 1; Figure 1, ​,EE and ​andF).F). Patients who used one to four tablets/week had an HR of 0.59 (95% CI = 0.22 to 1.59) and those who used five or more tablets/week had an HR of 0.50 (95% CI = 0.21 to 1.23) compared with nonusers (P trend = .14). No statistically significant interactions between aspirin and COX-2 inhibitor use and other covariables were seen for cancer recurrence (Supplementary Tables 2 and 3, available online). There was also no statistically significant increase in cardiovascular events or grade 3 or higher toxicities (17) with medication use, except for leukopenia (P = .008), for unclear reasons (Supplementary Table 4, available online). Table 1. Relationship between aspirin and COX-2 inhibitor use and colon cancer recurrence and mortality Figure 1. Recurrence-free, disease-free, and overall survival by aspirin or COX-2 inhibitor use. Survival curves were generated by the Kaplan-Meier method and two-sided P values calculated using the log-rank test. A) Recurrence-free survival according to consistent ... This analysis of enrolled in a chemotherapy clinical trial is an important addition to the literature, which supports a benefit for aspirin use after CRC diagnosis (13–16,24). The exact dose and duration of aspirin or COX-2 inhibitors required for a potential protective effect remains unclear, however. Although the P trend was not statistically significant, our study suggests a dose-response relationship with increased frequency of aspirin use, while any amount of COX-2 inhibitors was associated with decreased recurrence, consistent with previous reports (3,10,11,15,24–28). Current knowledge of the biological pathways affected by these medications provides strong support for our findings. Aspirin inhibits PTGS (COX), which converts arachidonic acid to prostaglandins, which modulate tumor growth through alteration of stem cell gene expression (29), hypermethylation of genes involved in proliferation and differentiation (29,30), promotion of angiogenesis and WNT/CTNNB1 signaling (31,32), and inhibition of apoptosis (33–35), among others. Molecular pathological epidemiology studies also report differential benefit depending on BRAF (36) or PIK3CA mutation status (37–39) and tumoral PTGS2 (13) or HLA class I antigen expression (39); results are conflicting and require further study. Unfortunately, we had insufficient power to explore these interactions. Other limitations of our study include misclassification of the exposure from self-reported data, however, prior studies have demonstrated the reliability of such data (40). Moreover, medication use was recorded before any knowledge of cancer outcomes, thus minimizing reporting biases. Also, we were unable to assess prediagnostic use and total duration of use in this study. Patients who enroll in clinical trials often engage in other healthful behaviors, so we controlled for physical activity, body mass index, and performance status, but residual confounding from unknown variables is possible. In conclusion, this observational study of stage III colon cancer patients found statistically significant associations between aspirin and COX-2 inhibitor use and reduced cancer recurrence and mortality. Results from the ongoing CALGB 80702 and ASCOLT trials are eagerly awaited. Further exploration of predictive biomarkers of aspirin and COX-2 inhibitor activity is warranted.