Background
Breast cancer is the most common malignancy among women in high-income countries. About 8000 new cases are diagnosed each year in Sweden [
1], 255,000 cases in the USA [
2] and 55,000 cases in the UK [
3]. Early detection of breast cancer with mammographic screening and more effective adjuvant treatment has gradually improved the prognosis in breast cancer [
4,
5]. However, a considerable number of women die from the disease [
3]. Hence, additional cost-effective therapies are still needed. Several studies have indicated that low-dose aspirin use around the time of a breast cancer diagnosis may reduce the risk of both breast cancer-specific and all-cause mortality [
6‐
9], but results are inconsistent. A few studies have also reported no associations between aspirin use after breast cancer diagnosis and breast cancer deaths [
10,
11]. In a recent meta-analysis, pooled results found evidence of a reduction in breast cancer-specific death following aspirin use versus no use: RR of 0.73 (95% CI, 0.54–0.98,
p = 0.04) [
12]. However, there was heterogeneity among the included studies and following exclusion of one deviant study, the significant association between postdiagnostic aspirin use and breast cancer-specific mortality disappeared.
There are several plausible biological mechanisms of action for a potential beneficial effect of aspirin use in breast cancer initiation and progression, involving inflammation, hormonal alterations and platelet inhibition [
13,
14]. Aspirin irreversibly inhibits cyclooxygenase (COX)-1 and COX-2 which are crucial for synthesis of prostaglandins involved in cellular migration and proliferation. Prostaglandins are present at elevated levels in breast cancer tissue, where they are believed to also stimulate angiogenesis and inhibit apoptosis, and inhibition of COX-1 and COX-2 has reduced growth of breast cancer cell lines [
15]. Prostaglandins further stimulate aromatase activity, which subsequently increase estrogen levels, and lower levels have been reported in postmenopausal aspirin users compared to nonusers [
16]. Aspirin could perhaps also inhibit platelet-induced adhesion of circulating tumor cells from initiating metastases [
17,
18].
Because of the mixed results from previous studies and a lack of randomized trials, large observational studies remain important to understand whether aspirin has the potential of altering breast cancer prognosis, and, if so, among which subgroups of women. Two randomized trials on aspirin use are ongoing, one in the USA and one in the UK, to evaluate aspirin use and disease-free survival in women with early-stage breast cancer [
19], but results will be not be available for many years (preliminarily in 2026). In the present study, we have used detailed clinical information from Swedish population-based breast cancer quality-of-care registers and the national drug prescription register to address the possible associations between low-dose aspirin use and outcomes in subgroups of women with breast cancer. We tested the hypothesis that aspirin use is associated with improved breast cancer-specific outcomes overall or in certain clinical subgroups of breast cancer.
Discussion
In this large Swedish population-based cohort study we found no evidence that low-dose aspirin use before or after breast cancer diagnosis reduces the risk of breast cancer-specific death in breast cancer patients overall. There were no indications of dose-response by dose or duration of aspirin use. However, among women with stage I tumors, aspirin use after diagnosis was associated with a reduced risk of breast cancer death. We also found a possible reduced risk in women with ER+ tumors who were treated with low-dose aspirin before breast cancer diagnosis. Even though these associations could also have arisen by chance, further subgroup-specific investigations in larger datasets are warranted to confirm or refute these findings. Aspirin did not reduce the risk of metastases among stage I–III breast cancer patients, nor prolong the time to breast cancer-specific death in stage IV disease.
Our results corroborate findings from some, but not all, earlier studies. HRs for aspirin use assessed before diagnosis have shown no association in two studies [
11,
34] and an increased risk of breast cancer death in one study [
8]. Results for aspirin use assessed after diagnosis (and risk of breast cancer death) range from no association in five studies [
10,
11,
34‐
36] to a protective association in three studies [
6‐
8]. In a meta-analysis of five randomized trials of daily aspirin (≥ 75 mg) versus placebo for prevention of cardiovascular events in the UK, a reanalysis of risk for cancer metastases indicated a lower rate of metastases in breast cancer patients although the association was not statistically significant [
37].
From the Iowa Women’s health study, Blair et al. [
7] reported that breast cancer patients using aspirin after diagnosis were at a significantly lower risk for breast cancer death compared with nonusers. The aspirin exposure in this study was self-reported through a questionnaire and focused on current use of aspirin, and modifications in risk by clinical factors such as tumor size were found. Similarly, Holmes et al. [
6] reported current aspirin use to be associated with a substantially decreased risk of breast cancer death, with an adjusted RR of 0.36 (0.24–0.54) for aspirin users 6–7 days per week when compared with nonusers from the Nurses’ Health Study. However, past aspirin use was not associated. Results did not differ appreciably when stratified by stage, BMI, menopausal status or ER status.
In a Swedish register-based study with a nested case–control design, aspirin use after breast cancer diagnosis was not associated with risk of breast cancer death, except when aspirin use had been terminated close to death/end of follow-up. However, clinical information such as tumor stage at diagnosis was not available. It cannot be excluded that this finding reflects confounding by indication and/or reverse causation, since patients may stop taking aspirin and other drugs due to a worsening general condition in the end-of-life period. Also, patients admitted to hospital or palliative care in Sweden generally do not receive their drugs by prescription and are therefore not recorded in the prescription registry. In our study, aspirin exposure was handled as a time-varying exposure with a 180-day lag to avoid bias due to change in medication close to death [
29‐
31]. This may explain why our results differ from the previous Swedish study and the two American studies.
In a recent cohort study from the Scottish Cancer Register including 15,140 stage I–III breast cancer patients, Mc Menamin et al. [
36] reported a HR of 0.92 (95% CI 0.75–1.14) for breast cancer death among women using aspirin after breast cancer diagnosis compared with nonusers. They also examined low-dose aspirin use before breast cancer diagnosis with no association with cancer-specific mortality (adjusted HR 0.95, 95% CI 0.81–1.11). With a median follow-up of 4 years, the Scottish study also lagged medication use after diagnosis by 6 months to avoid bias due to change in medication. They had detailed data on clinical factors such as stage and ER status, but could not find any protective effect on breast cancer-specific mortality in these subgroups. However, the HR point estimate was lower among stage I patients than in other groups (HR 0.74, 95% CI 0.35–1.54).
Other epidemiological studies have also reported no association for aspirin use after diagnosis and risk of breast cancer death. In an Irish cohort study of 4540 women aged 50–80 years with stage I–III breast cancer who were nonaspirin users before diagnosis, aspirin initiation after diagnosis meant no reduction of breast cancer-specific mortality. Aspirin exposure was identified from linked national prescription data, and the analysis was adjusted for clinical characteristics. There was no evidence of effect modification by tumor size, lymph node status or ER status, but subgroup-specific results were not presented [
10].
In summary, several recent studies do not provide support for an association between aspirin use and breast cancer progression overall, but they have had limited power to investigate potential differences across patient subgroups. Our study represents one of the largest so far, and therefore the finding of a potential association among stage I patients, and perhaps also in the ER
+ group, warrants further investigation. Since aspirin inhibits prostaglandins which in turn leads to inhibition of angiogenesis and stimulation of apopotosis, as well as to lower estrogen levels through an impact on aromatase activity, there are several biologically plausible mechanisms that could explain a putative association [
13,
14,
17,
18]. It is perhaps not unfeasible that biological effects, if at all present, could be confined to early-stage ER
+ tumors given putative anti-hormonal effects. Alternatively, early-stage tumors with slow progression may be more susceptible than aggressive ones. In a few patient subgroups (users of high-dose aspirin (> 75 mg) and patients with stage III tumors), moderately increased risks of breast cancer deaths were noted. Although we cannot exclude true subgroup-specific increased risks, we believe that the most likely explanations for these results include chance and/or potential residual confounding by the indications for aspirin use.
Strengths of our study include the use of information from a large population-based unselected cohort of women with breast cancer identified from regional prospective quality-of-care registers. These databases provide information on patient and tumor characteristics as well as treatment intentions allowing for detailed subgroup analyses and adjustments. By use of data on drug dispensing from a national register, we avoided potential bias by self-reported data. Swedish health-care register data are generally of high quality and completeness. Data available from the National Patient Register allowed for adjustments for comorbidity. Several limitations need mentioning. Since aspirin is sold over the counter in Sweden, we cannot exclude that some aspirin users were misclassified as nonusers. However, low-dose aspirin represents 90% of all aspirin sold, and this form is only available by prescription [
24]. Another limitation is that the prescription register provides information on dispensed drugs only, without taking into account patient adherence and actual use. Confounding by indication is a challenge in pharmaco-epidemiological studies [
38] and might also have affected our study since the prescribed dose of aspirin differs by indication. However, a majority of women (85.2%) were prescribed 75 mg daily dose of aspirin, the standard dose for prevention of heart disease in Sweden. Only a small fraction (6.3%) of women had their dose changed during follow-up. In the analysis of the association between aspirin use before breast cancer diagnosis and breast cancer mortality, there is potential selection bias/collider stratification bias when stratifying by tumor stage, which may change the association between aspirin use before diagnosis and breast cancer mortality in either direction. However, by adjusting for potential confounders between stage and breast cancer mortality this bias should have been reduced. Still, the results for aspirin use before breast cancer diagnosis should be interpreted assuming that there is no residual confounding. We did not have data on potentially important confounders such as BMI, smoking habits or physical activity. However, findings from a few earlier studies have not found evidence of any large confounding of the association between aspirin and breast cancer progression by these factors [
6,
7,
39].