Background: There is increasing evidence that aspirin reduces the incidence and mortality of several cancers. The results of epidemiologic studies of aspirin in relation to non-Hodgkin lymphoma (NHL) however, are inconsistent, and the association appears to vary by NHL histologic subtype. We aimed to prospectively assess the association between aspirin and major subtypes of NHL in a cohort with detailed data on aspirin use.

Methods: The Nurses’ Health Study (NHS) comprises 121,701 female United States licensed registered nurses who were ages 30-55 years at enrollment in 1976 and have been followed thereafter with biennial questionnaires to update medical history and personal and lifestyle information. Aspirin use was first queried in 1980. We incorporated a two-year exposure lag to minimize influence of reverse causation. Thus, “baseline” was 1982 for this analysis. We followed participants from baseline through June 2010, including all who had no baseline history of cancer or rheumatoid arthritis (RA)—a risk factor for NHL that may also affect aspirin use habits—and who completed the first questions on aspirin use. Participants were censored at the date of a cancer or RA diagnosis, or at date of death. We identified incident primary diagnoses of NHL via the follow-up questionnaires or by National Death Index follow-up. We confirmed the diagnoses and determined the histologic subtype by medical record review. Aspirin use was quantified both as cumulative average quantity (adult-strength tablets/week), and as duration of continuous regular use (years). To estimate the relative risk of NHL associated with a given quantity or duration of regular aspirin use, we calculated hazard ratios (HRs) and 95% confidence intervals (CIs) in Cox models that were stratified on calendar period and age (months) and adjusted for geographic region of residence and height. Trend tests were based on comparable multivariable models in which the medians of aspirin use variable categories were modeled as ordinal variables.

Results: In 1,749,512 person-years of follow-up we identified 738 cases of NHL with available information about regular aspirin use. In the 2-year lagged analyses of regular aspirin quantity, regular aspirin use showed a suggestive positive association with risk of follicular lymphoma. In particular, women who used a cumulative average of 1 to <2, 2 to <5, or ≥5 tablets/week had a modest non-significant increased risk of follicular lymphoma compared to non-users (HR [95% CI] were 1.56 [0.84-2.91], 1.28 [0.69-2.35] and 1.64 [0.91-2.94], respectively; p-trend=0.03, based on 135 cases). No association was suggested for regular aspirin quantity with diffuse large B-cell lymphoma (DLBCL; p-trend=0.72, 114 cases), chronic lymphocytic leukemia (CLL; p-trend=0.13, 219 cases) or with NHL overall (p-trend=0.14). Duration of regular aspirin use was not associated significantly with follicular lymphoma (p-trend=0.70) or with DLBCL (p-trend=0.52), CLL (p-trend=0.32) or NHL overall (p-trend=0.45).

Conclusions: Our findings of no association between regular aspirin use and NHL, other than a suggestive positive association between cumulative average quantity of use and follicular lymphoma risk, were contrary to expectation but consistent with an earlier published cohort study. The present analysis was limited in part by sparse data, especially in subtype-specific analyses. Further evaluation in a comparable cohort of men and in the two cohorts combined (where statistically appropriate) may lend further insights. If confirmed in other large populations, the findings in the NHS suggest that aspirin use may be only associated with risk of follicular lymphoma, consistent with the view that there is some etiologic heterogeneity across major subtypes of NHL.


Zhang:Takeda Pharmaceuticals Inc: Employment.

Author notes


Asterisk with author names denotes non-ASH members.