Jun 28, 2010
An Exciting Study Sparks Questions, Including the Critical One: What Will we Tell our Patients?
Amy Gross, MHS, PhD candidate
Johns Hopkins Bloomberg School of Public Health
Kala Visvanathan, MB, BS, FRACP, MHS
Johns Hopkins Bloomberg School of Public Health and Sidney Kimmel Comprehensive Cancer Center
Further Investigation Needed
July 2010 Issue: Current treatment options for breast cancer are, although effective, in general very expensive, and as the global population continues to age, a steady increase in breast cancer cases can be expected, along with growing health care costs. There are currently more than 2.5 million breast cancer survivors in the United States alone,1 and the risk of recurrence and of dying of the disease remains a significant and longstanding concern for these women. The potential for an effective and low-cost medication that can be added to the arsenal of adjuvant agents to further prolong survival is exciting.
The recent prospective observational study in the Nurses’ Health Study by Holmes et al. (2010) provides support for the use of aspirin after a diagnosis of breast cancer.2 The study population consisted of 4,164 female registered nurses between ages 30-55 at baseline in 1976, who were subsequently diagnosed with stage I-III breast cancer and survived at least one year. Information on aspirin use was assessed by questionnaire every two years following breast cancer diagnosis and included questions on never, past, or current regular-use, and days per week of use. Aspirin assessment was missing for 2,910 women. The primary outcomes were breast cancer mortality (assessed by family report or U.S. Postal Service, National Death Index, and death certificates) and breast cancer recurrence.
Aspirin use of six to seven days per week compared with never-users was associated with a 74% reduction in risk of breast cancer death (relative risk = 0.36, 95% CI 0.24 - 0.54). A similar result was seen when comparing aspirin use of two to five days per week to never-use. This association did not change by stage, body mass index, menopausal status, or estrogen receptor status. Results for distant recurrence showed a 43% risk reduction for six to seven days per week of aspirin use versus never-use (relative risk = 0.57, 95% CI 0.39 - 0.82) and a 60% reduction in risk for two to five days per week. These findings are consistent with two previous studies3,4 that found an association between nonsteroidal anti-inflammatory drug (NSAID) use and breast cancer recurrence or survival.
Biologic plausibility for the associations seen in this study is supported by preclinical models. Breast cancer cells have been shown to produce greater amounts of prostaglandins than normal breast cells in vitro.5 Aspirin, among other NSAIDs, inhibits prostaglandin and COX-2 production. COX-2 knockout mice show decreased tumor growth when given NSAIDs in a variety of cancers.6
Questions of dose and duration
Despite these strong results, in evaluating this study we also need to consider some of its limitations. The first issue is dosage and the question of whether there is a dose effect. Although the authors included questions on indications for aspirin use that might imply dosage (e.g., cardiovascular disease prevention, headache, etc.), actual dosages used were not ascertained. This is important information for assessing potential toxicity, particularly when considering long-term use of such agents. In addition, there was minimal information on toxicity of aspirin treatment. Further clarification of the role of both timing and duration of aspirin use following diagnosis would also be helpful, particularly since in the present study, past users (defined as those who had used aspirin after breast cancer diagnosis but had subsequently stopped) did not show a reduced risk of breast cancer death or recurrence compared to never users.
This suggests that current aspirin use compared with prior use may be most important in order to obtain benefit. This also could be clarified by further mechanistic studies. Although studies of aspirin use and breast cancer incidence have been mixed,7-11 information on aspirin use prior to breast cancer diagnosis in the current study might have significance and should also be investigated. In addition, five-year duration of use (combining past and current users) was associated with only a 5% reduction in risk of breast cancer death. It would also be valuable to learn if the effects of aspirin use on breast cancer mortality differ by breast cancer subtype.
In summary, we have good data from epidemiologic studies that support a role for aspirin use in preventing breast cancer recurrence and mortality. What is now needed is a study in a cohort with more detailed information, such as a prescription database, to help clarify issues and confirm findings. A randomized trial would also help answer some of these questions.However, the feasibility of such a study, particularly in the United States where aspirin intake is not uncommon, would have to be assessed first. In terms of translating these findings to clinical practice, at present we feel that the data provide reassurance to women with breast cancer—who are already taking aspirin for another purpose—that aspirin may also have an effect on their mortality. Given aspirin is not without toxicity, we currently would not recommend that women should begin using aspirin long-term for reducing their risk of breast cancer recurrence or mortality. We would await clarification on issues such as duration and dose to be obtained from more detailed trials, which will hopefully be undertaken in a timely fashion given these exciting results.
- American Cancer Society. Breast Cancer Facts & Figures 2007-2008. Atlanta: American Cancer Society, Inc.
- Holmes MD, Chen WY, Li L, et al. J Clin Oncol. 2010;28:1467-1472.
- Kwan ML, Habel LA, Slattery ML, et al. Cancer Causes Control. 2007;18:613-620.
- Blair CK, Sweeney C, Anderson KE. et al. Breast Cancer Res Treat. 2007;101:191-197.
- Bennett A, Charlier EM, McDonald AM, et al. Lancet. 1977;2:624-626.
- Williams CS, Tsujii M, Reese J, et al. J Clin Invest. 2000;105:1589-1594.
- Cook NR, Lee IM, Gaziano, et al. JAMA. 2005;294:47-55.
- Takkouche B, Regueira-Méndez C, Etminan M. J Natl Cancer Inst. 2008;100:1439-1447.
- Mangiapane S, Blettner M, Schlattmann P. Pharmacoepidemiol Drug Saf. 2008;17:115-124.
- Khuder SA, Mutgi AB. Br J Cancer. 2001;84:1188-1192.
- Bosetti C, Gallus S, La Vecchia C. Cancer Causes Control. 2006;17:871-888.
Weighing the Risks vs. Benefits
Kevin Kalinsky, MD, and Dawn L. Hershman, MD, MS
Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center
In 1998, tamoxifen was approved by the U.S. Food and Drug Administration (FDA) for breast cancer (BC) prevention, after being approved for reducing recurrences in patients with early-stage BC. It is therefore not surprising that a drug that appears efficacious for primary prevention of BC, such as aspirin, may also offer benefit as adjuvant treatment.
Over 300 papers have been published evaluating the association between aspirin and other nonsteroidal anti-inflammatory drug (NSAID) medications and cancer prevention. While the results are conflicting, a meta-analysis of 38 studies evaluating more than two million patients found a reduced BC risk with aspirin use (relative risk = 0.87).1 Through prostaglandin reduction, these agents, along with COX-2 inhibitors, inhibit cell growth and decrease cell invasiveness in preclinical models.2 It has been shown that prostaglandin increases aromatase gene expression and, subsequently, estrogen production.3 Thus, inhibitors of prostaglandin synthesis may particularly benefit the treatment and prevention of hormone-sensitive cancer.4
Based on the preventive benefits for BC and other cancers, such as colon cancer, a prospective randomized trial was initiated evaluating the role of aromatase inhibition with or without celecoxib for the adjuvant treatment of hormone-sensitive BC in post-menopausal women (MA.27). This study was terminated early when the cardiovascular toxicities of COX-2 inhibitors were reported. Given the widespread use of aspirin and other NSAIDs for multiple medical indications, it is unclear if a prospective randomized trial evaluating the efficacy of these drugs for adjuvant BC therapy is feasible.
To help us delineate the potential benefit of these medications, we can turn to the observational data from the Nurses’ Health Study. Holmes et al. evaluated the risk of BC mortality in 4,164 women diagnosed with stage I-III BC who never used, previously used, and currently used aspirin.5 Past use was defined as use prior to the diagnosis of BC. In a multivariate model, controlling for confounding factors such as stage, estrogen receptor status, menopausal status, body mass index, use of oral contraceptive pills/hormone replacement therapy, and treatments received (radiation therapy, chemotherapy, and anti-estrogen treatment), women using aspirin two to five days per week had a 71% lower risk of dying from BC as compared to nonusers, and women using aspirin six days or more per week had a 64% lower risk. Similar associations were found with BC recurrence.
Causation or correlation?
As with all observational data, we are left with uncertainty with regard to causality. Is aspirin use associated with better health behavior and possibly superior compliance? Are the indications for aspirin use related to BC outcome leaving the drug as an innocent bystander? While these data are subject to bias, the study has strengths that may lead us to conclude the results are likely valid. The data was collected prospectively, the quality of data has been well validated, and the follow-up time was long. The findings are biologically plausible and are in accordance with recent findings in colon cancer, in which aspirin use during and after adjuvant chemotherapy has been shown to decrease cancer-specific mortality.6,7
However, in the study, the number of deaths was low, there was insufficient data collected to evaluate for a dose response, and no predictive biomarkers were assessed, including COX-2 expression on the primary tumor. Patients using aspirin may also be using other medications that have been implicated as beneficial in BC treatment and prevention, such as metformin and statins. In future studies with more power, it would be useful to perform stratified analyses to determine if similar effects are observed in hormone-sensitive compared with insensitive tumors.
What will we tell our patients? Since it is unlikely that there will be a prospective randomized trial, we depend on large observational data-sets to further characterize this promising association. Until then, it remains appropriate to discuss this medical uncertainty with patients, weighing the known risks with the potential benefits.
- Takkouche B, Regueira-Mendez C, Etminan M. J Natl Cancer Inst. 2008;100:1439-47.
- Sheng H, Shao J, Kirkland SC, et al. J Clin Invest. 1997;99:2254-9.
- Zhao Y, Agarwal VR, Mendelson CR, et al. Endocrinology. 1996;137:5739-42.
- Terry MB, Gammon MD, Zhang FF, et al. JAMA. 2004;291:2433-40.
- Holmes MD, Chen WY, Li L, et al. J Clin Oncol. 2010;28:1467-72.
- Chan AT, Ogino S, Fuchs CS. JAMA. 2009;302:649-58.
- Fuchs C, Meyerhardt JA, Heseltine DL, et al. Proc Am Soc Clin Oncol. 2005;23(16s). Abstract 3530.
Current Controversies in Oncology is a forum for the exchange of views on topical issues in the field of oncology. The views and opinions expressed therein are those of the authors alone. They do not necessarily reflect the views or positions of the Editor or of the American Society of Clinical Oncology.