In this prospective study among male health professionals, twice or more weekly aspirin use was not associated with risk of total lung cancer or NSCLC, even for long-term consistent users. Also, no dose–response association was observed for frequency of aspirin use and lung cancer incidence.
Several epidemiologic studies have reported an increased incidence of lung cancer associated with several inflammatory-related lung diseases including asthma and chronic bronchitis, suggesting that inflammation may play an important etiologic role (Wu et al, 1995
; Mayne et al, 1999
; Brownson and Alavanja, 2000
; Brenner et al, 2001
). The reduction of inflammatory prostaglandins by inhibition of COX-2 by NSAIDs, including aspirin, might have a chemopreventive effect. However, in a recent study of hydroxytoluene-induced early lung adenocarcinoma in mice, aspirin attenuated pulmonary inflammation but was ineffective at preventing lung tumorigenesis (Kisley et al, 2002
). In other animal studies, COX-2 was also consistently expressed in normal bronchoalveolar and alveolar epithelium of the lung (Bauer et al, 2000
; Wardlaw et al, 2000
), casting some doubt on a specific role of COX-2 in the development of lung cancer.
Results from previous clinical and epidemiological studies of aspirin use and the risk of lung cancer have been mixed. An early trial among British physicians found a statistically nonsignificant 36% lower mortality rate from lung cancer among users of aspirin compared to nonusers (Peto et al, 1988
). A prospective study based on data from the National Health and Nutrition Examination Survey found among men a statistically significant 46% lower lung cancer risk among users of aspirin during the 30-day period preceding recruitment into the cohort compared with nonusers (Schreinemachers and Everson, 1994
). Two recent case–control studies (Harris et al, 2002
; Moysich et al, 2002
) found statistically significant reductions in lung cancer risk for aspirin users compared to nonusers. In contrast, some prospective studies reported no overall differences in lung cancer mortality (Thun et al, 1993
) or incidence (Paganini-Hill et al, 1989
) between aspirin users and nonusers, consistent with our results. Furthermore, in two case–control studies, lung cancer risk was not significantly lower for regular aspirin use defined as 4 days per week for 3 months compared to nonuse (Rosenberg et al, 1991
) or seven prescriptions of aspirin and other NSAIDs received in 13–36 months before diagnosis compared to no prescriptions (Langman et al, 2000
). In a recent case–control study, a nonsignificant reduction in total lung cancer risk (odds ratio (OR)=0.66; 95% CI=0.34–1.28, adjusted for smoking (never, past, current) and educational status (attended college, attended graduate school)) was seen among those who reported aspirin use 3 or more times per week for at least 6 months. Similar reductions in total lung cancer risk were observed for 5 or more years of aspirin use (OR=0.68; 95% CI=0.31–1.51) (Akhmedkhanov et al, 2002
). We used a roughly similar exposure definition of aspirin use (i.e. aspirin use two or more times per week), and our findings on duration of aspirin use are in fair agreement. However, we found no significant association between aspirin use at baseline and total lung cancer incidence (RR=1.13; 95% CI=0.89–1.43, adjusted for current age, age at starting to smoke regularly, and smoking status (includes dose and time since quitting)); the risk of total lung cancer associated with consistent aspirin use for 6 years was slightly inverse (RR=0.89; 95% CI=0.47–1.67) but the results were not statistically significant.
No association was evident between regular aspirin use and NSCLC in the current study. To date, two case–control studies have examined the association between regular aspirin use and NSCLC. In one study, results for NSCLC among women taking aspirin three or more times per week for at least 6 months (adjusted OR=0.39; 95% CI=0.16–0.96) were stronger that those when all histological types combined were considered (adjusted OR=0.66; 95% CI=0.34–1.28) (Akhmedkhanov et al, 2002
). In contrast, a second study observed similar significant risk reductions for NSCLC among men and women who were regular users of aspirin (adjusted OR=0.62; 95% CI=0.45–0.86) and total lung cancer risk (adjusted OR=0.57; 95% CI=0.41–0.78) (Moysich et al, 2002
We note a number of potential limitations in the epidemiologic studies evaluating the role of aspirin and other NSAIDs in lung cancer risk, including (1) recall bias and selection bias potential in case–control studies; (2) inefficient control for smoking history; (3) small numbers of cases; and (4) misclassification due to the insufficient assessment of long-term aspirin use or lack of detailed information on aspirin dose. Although we acknowledge the dosage limitation in the current study, the lack of information on aspirin dose will result in any potential misclassification of the risk estimate towards the null. Furthermore, the current study has strengths regarding the other issues raised. The potential for recall or selection bias is greatly minimised, if not eliminated, due to the prospective study design. To minimise confounding by smoking, smoking habits were modeled to best predict lung cancer. The relative homogeneity of this population of male health professionals decreases the likelihood of residual confounding by smoking or other factor. With repeated measurements of aspirin use we account for changes in aspirin exposure over time and reduce the potential for misclassification of aspirin use. The lack of association observed between aspirin use and lung cancer risk in the present study might reflect measurement error in aspirin use. However, using our definition of regular aspirin use, we observed a markedly decreased risk of colorectal cancer and adenoma in this same population (Giovannucci et al, 1994
), consistent with other studies. This suggests that we are able to detect important associations with aspirin use in this study. We were unable to assess the long-term use of acetaminophen and other NSAIDs due to the insufficient number of cases among users of these compounds (range of cases among users=0–8).
Our findings do not support the hypothesis that regular aspirin use is associated with a decreased risk of lung cancer. Prevention of lung cancer should be practiced primarily through smoking prevention and cessation.