Regular Aspirin Use and Mortality in Patients with Multiple Myeloma

Multiple myeloma is an incurable, and indeed, lethal malignancy of permanently differentiated B cells. The introduction of novel therapies after 2000 has improved the estimated life expectancy of patients with multiple myeloma, from a 5-year survival rate of 35% in 2000 to just over 50% today (1). However, uptake and utilization of these new therapies vary considerably across different sociographic subsets of the population (2–4), and the vast majority of patients eventually stop responding to treatment and relapse (5). Additional strategies available to clinicians and their patients for extending survival time are limited.

It is known that inflammation is important in multiple myeloma pathogenesis, and both mechanistic and population-based data suggest that upregulation of inflammatory pathways, including those mediated by NF-κB and IL6, contributes to multiple myeloma pathogenesis (6–8). Aspirin is an NSAID that mediates inflammation, in part by downregulating NF-κB and several of its downstream targets, such as COX and its production of prostaglandins. Aspirin can irreversibly inactivate COX-1 and COX-2 through covalent bond formation, although the effect of aspirin on COX-2 may be particularly relevant to multiple myeloma prevention. Specifically, COX-2 is frequently expressed in multiple myeloma cells (9, 10), and expression has been shown to predict poor outcome in patients with multiple myeloma (9). Therefore, investigations of the impact of pharmacologic agents such as aspirin, which can reduce COX-2 expression, is warranted in patients with multiple myeloma.

In the current study, we investigated aspirin-use patterns in relation to multiple myeloma–specific and all-cause mortality among men and women diagnosed with multiple myeloma between 1980 and 2016 in two large cohort studies: the Nurses' Health Study (NHS) and Health Professionals Follow-up Study (HPFS). We chose to focus on aspirin as opposed to a broader category of NSAIDs in the current manuscript because the interval of exposure assessment for aspirin began much earlier during cohort follow-up and was more complete than for other NSAIDs; available data also did not support a sufficiently detailed or statistically powered examination of the other analgesics as main effects. In addition, aspirin or other blood thinners are often recommended for patients with multiple myeloma who take modern immunomodulatory agents (which have been incorporated into multiple myeloma first-line treatment regimens since 2006), to prevent venous thromboembolism. This may introduce confounding by indication to studies based solely in the modern era, and thus we sought to leverage the early inception and long follow-up of the NHS and HPFS to examine associations between aspirin use and survival of multiple myeloma both before and after the widespread availability of such novel therapies. To our knowledge, this is the first epidemiologic investigation of aspirin-use patterns and mortality in patients with multiple myeloma.

The NHS began in 1976 when 121,700 female U.S. registered nurses ages 30 to 55 years returned an enrollment questionnaire (11). The HPFS was established in 1986 as a parallel cohort of 51,529 US men who were dentists, optometrists, osteopathic physicians, podiatrists, pharmacists, and veterinarians ages 40 to 75 years at entry (12). In both cohorts, biennial follow-up questionnaires were used to update information on lifestyle and disease history, including whether participants received a diagnosis of multiple myeloma. For any report of multiple myeloma, participants gave written permission to obtain hospital medical records and pathology reports pertaining to their diagnosis, and trained study personnel, blinded to exposure data, reviewed the records to confirm the diagnosis. When the original medical records were unavailable, case confirmation was pursued via linkage to state tumor registries.

The current analysis includes individuals who were diagnosed with multiple myeloma who reported their aspirin-use patterns on one or more follow-up periods and had no history of cancer (except nonmelanoma skin) prior to the first aspirin intake assessment (1980 in NHS; 1986 in HPFS). The study protocol was approved by the Institutional Review Boards of the Brigham and Women's Hospital and Harvard T.H. Chan School of Public Health, and those of participating registries as required. Informed consent was implied by return of the questionnaires.

Deaths were identified by next of kin, the postal system or routine searches of the National Death Index (13, 14). Mortality follow-up in these cohorts has been shown to be more than 98% complete (13). Individuals blinded to exposure information ascertained cause of death from death certificates, which were supplemented with medical records or, for cancer deaths, tumor registry linkage when possible. Survival time was assessed as the interval of time from multiple myeloma diagnosis to death or January 2016, whichever came first.

Clinical disease characteristics were not available on the full sample of multiple myeloma cases in these cohorts; however, we manually abstracted select disease characteristics recorded at the time of diagnosis (prior to therapy) from available hospital medical records and pathology reports, as previously published (15).

Aspirin was first assessed in NHS in 1980 and every two years thereafter, except 1986. Men in HPFS were first asked about aspirin use at baseline (1986). Early in the cohort follow-up periods, the questions about aspirin use did not distinguish between low and standard-dose tablets, and frequency of intake was not asked until 1992. From 1992 through 1998, participants were asked to report their weekly use by converting baby aspirin intake to adult strength equivalents via the question, “On average, how many aspirin tablets do you take per week? (4 baby aspirin = 1 tablet).” From 2000 onward, the questionnaire was modified to allow participants to select their dose (e.g., 81 vs. 325 mg). The primary indications for aspirin use were determined using a survey among randomly selected self-reported aspirin users (16). In NHS, top reasons for use included heart disease prevention (35%), muscle or joint pain (16%), and headache (13%); in HPFS, top reasons included cardiovascular disease prevention (58%), joint or musculoskeletal pain (33%), cardiovascular disease (25%), and headaches (25%; ref. 17). Individual-level indication(s) for aspirin use were not available in either cohort.

Aspirin intake patterns were subdivided into pre- and postdiagnosis exposures. The prediagnosis aspirin exposures of interest included aspirin use status (use vs. nonuse) and average weekly 325 mg aspirin intake (i.e., quantity) reported on the questionnaire returned before multiple myeloma diagnosis, as well as years of continuous aspirin use (i.e., duration) before diagnosis. We used 325 mg as the unit of exposure for aspirin, because information on “low dose” aspirin could not be distinguished from adult strength use before the year 2000. Participants' average prediagnosis quantity was computed by averaging the number of adult strength tablets taken weekly, as reported on all questionnaires up to the last returned questionnaire prior to the diagnosis of multiple myeloma (18). Prediagnosis duration of use was calculated by summing the consecutive years in which a participant reported regular aspirin use up to the last questionnaire that was returned before the diagnosis of multiple myeloma (18). If aspirin use information was missing on a given questionnaire, data from the previous follow-up interval were carried forward for one interval; the exposure variables were set to missing thereafter. This exposure captures the years of continuous duration of use most proximal to the multiple myeloma diagnosis. The postdiagnosis aspirin exposures of interest included aspirin use status (use vs. nonuse) as well as the weekly 325 mg aspirin intake (i.e., quantity) reported on the first questionnaire that was returned after diagnosis. We did not update aspirin-use information after the first questionnaire cycle after diagnosis because fewer than half the participants with multiple myeloma survived long enough to return two questionnaires (median survival was <4 years and the follow-up cycles biennial).

Individuals were classified as regular users of acetaminophen and ibuprofen if they reported use at least twice per week. In NHS, this exposure was derived from a question on the regular use of other nonsteroidal analgesics asked in 1980, as well as questions specifically pertaining to acetaminophen and ibuprofen use from 1990 onward. In HPFS, this exposure was derived from separate questions on the use of other analgesics beginning at baseline.

In both cohorts, height and weight were self-reported at baseline and current weight was updated on follow-up questionnaires. Height and weight measurements have been validated in these cohorts (r = 0.94, recalled vs. previously measured height; 0.97 for recalled vs. technician measured weight; ref. 19). We used self-reported height and weight to calculate the body mass index (BMI) of participants at each follow-up period. Medical comorbidities were also assessed in NHS and HPFS on follow-up questionnaires. A comorbidity index score was calculated by summing the number of cardiovascular-related comorbidities reported on each follow-up questionnaire (15, 20). Comorbidities of interest included high blood pressure, diabetes, elevated cholesterol, myocardial infraction, angina pectoris, coronary artery surgery or angioplasty, stroke, pulmonary embolism, paroxysmal atrial tachycardia, or other heart-rhythm disturbance.

Cox proportional hazards regression models with time since diagnosis as the underlying time scale were used to estimate the HRs and 95% confidence intervals (CI) for multiple myeloma–specific and overall mortality for the exposures of pre- and postdiagnosis aspirin use. Models adjusted for the main covariates of age at diagnosis and year of diagnosis, BMI and comorbidity index; the postdiagnosis models also adjusted for aspirin use status self-reported prior to diagnosis (user/nonuser). We could not adjust for clinically presenting characteristics of the multiple myeloma diagnosis, or first-line therapy, but adjustment for year of diagnosis may partially control for first-line therapy in these cohorts. We considered additional adjustment for acetaminophen and/or other NSAID use (yes/no/unknown) by adding the corresponding term to the multivariable models. All analyses were conducted on a pooled data set with stratification for cohort (sex), after finding no consistent statistically significant interaction between aspirin-use patterns and sex in relation to multiple myeloma–specific or all-cause mortality. Models of prediagnosis aspirin use were run with and without a two-year exposure lag. To illustrate the postdiagnosis findings, we plotted survival curves by stratum of postdiagnosis aspirin use status (user vs. nonuser) using Kaplan–Meier methods, testing their statistical significance with the log-rank test. We also performed competing risk analyses for causes of death: multiple myeloma–specific mortality versus other causes of death using the Fine–Gray method (21).

Given that some multiple myeloma treatments could influence a participant's likelihood of taking aspirin, we explored whether the associations between postdiagnosis aspirin-use patterns and mortality were substantially different across time periods of diagnosis. This was done by stratifying models of aspirin intake exposures by time period of diagnosis (before 2000 vs. after 2000). In exploratory analyses, we examined joint associations of pre- and postdiagnosis aspirin use in models that adjusted for the same main covariates described above. Finally, we examined differences in multiple myeloma clinical characteristics by aspirin-use patterns in a subset of participants with available clinical data.

All statistical tests were two-sided, and the proportional hazards assumption was tested and satisfied in all models by including time-dependent covariates in the Cox model.

In this cohort of 436 patients with multiple myeloma who were a mean age of 72 years at diagnosis, we identified 321 multiple myeloma–specific and 383 total deaths over a median of 43 months of follow-up. Compared with individuals who did not take aspirin after a multiple myeloma diagnosis, individuals who regularly took 6 or more aspirin tablets per week after multiple myeloma diagnosis were slightly older and more racially/ethnically diverse, with more comorbidities (Table 1). The majority of individuals who reported taking aspirin after diagnosis also reported a history of aspirin use prior to the diagnosis of multiple myeloma.

No statistically significant associations were observed between prediagnosis aspirin use exposures with multiple myeloma–specific or all-cause mortality (Table 2). These results were not materially different in models that included a two-year exposure lag.

Aspirin use postdiagnosis was inversely associated with multiple myeloma–specific and all-cause mortality (Fig. 1A and B). In multivariable models, compared with nonusers, participants who used aspirin after diagnosis had a HR for multiple myeloma–specific mortality of 0.61 (95% CI, 0.46–0.79) after adjustment for age at diagnosis, year of diagnosis, BMI, prediagnosis aspirin use, and number of comorbidities (Table 3). The effect estimates for overall mortality was of similar magnitude, suggesting a 37% reduction in risk among postdiagnosis aspirin users compared with nonusers (HR, 0.63; 95% CI, 0.49–0.80). For postdiagnosis aspirin quantity, we observed a modest trend of reduction in multiple myeloma–specific and all-cause mortality with increasing number of 325-mg tablets of aspirin taken per week (or equivalent weekly use of low-dose tablets), although it is notable that the CIs for 1 to <6 and ≥6 tablets per week overlapped. There were no marked differences in the effect estimates observed in models stratified by time period of diagnosis (before vs. after 2000), although the effect estimates for the benefit of postdiagnosis aspirin use were slightly stronger in the models for the period after 2000 (Supplementary Tables S1 and S2). We also confirmed that the time from diagnosis to the return of the first postdiagnosis questionnaire did not materially change the results reported herein. In addition, adjustment for concurrent use of other analgesics/NSAIDs did not materially change the effect estimates reported herein.

The analysis of joint associations of pre- and postdiagnosis aspirin supports the finding that postdiagnosis aspirin use was associated with enhanced survival irrespective of prediagnosis aspirin use. Compared with individuals who reported no aspirin use pre- or postdiagnosis, individuals who only reported aspirin use after diagnosis had a HR for multiple myeloma–specific mortality of 0.50 (95% CI, 0.30–0.83) and a HR for all-cause mortality of 0.56 (95% CI, 0.36–0.87). As indicated by the sample sizes per category in Table 4, we observed that only 7% of participants began taking aspirin after multiple myeloma diagnosis, whereas a much larger proportion of participants that took aspirin after diagnosis had also taken it before multiple myeloma diagnosis. Joint models also suggest that risk of early mortality was not significantly different for individuals who discontinued aspirin after diagnosis compared with individuals who never took aspirin (Table 4).

Exploratory analyses of differences in clinical characteristics of postdiagnosis aspirin use were limited by considerable missing data in the retrospective medical record review; however no marked differences in the clinically presenting features at diagnosis were apparent across postdiagnosis aspirin-use pattern categories.

We previously reported an almost 40% reduction of multiple myeloma risk in individuals with higher average quantity or longer duration of regular aspirin use compared with nonusers in the combined populations of the NHS and HPFS (18). In the current study, we observed that regular aspirin use after a diagnosis of multiple myeloma was associated with an almost 40% reduction in both disease-specific and overall survival, independent of prediagnosis use. Notably, the association was similar during the time periods before and after the availability of novel multiple myeloma therapies in the year 2000. While we are not able to infer causation from this observational study design, the strong association between postdiagnosis aspirin use and multiple myeloma–specific mortality, the directionally consistent (albeit nonsignificant) evidence of a survival benefit for individuals who reported aspirin use on the questionnaire cycle before diagnosis, and the observation that multiple myeloma–specific deaths comprised the majority of all deaths suggest the benefit may be, in part, through antineoplastic properties in multiple myeloma.

There are limited data on the association of aspirin and prognosis in patients with multiple myeloma. Nevertheless, our findings are consistent with trends observed in trials of aspirin for the prevention of vascular disease. In an intention-to-treat meta-analysis of eight randomized prevention trials, Rothwell and colleagues observed a reduced risk of death from hematologic cancer among those randomized to the aspirin group and who used it daily for at least 5 years (22). However, it is notable that findings from that study were based on a small number of patients in the hematologic cancer subgroup, and the association did not reach statistical significance; it is also possible that the modest improvement in hematologic cancer survival in that analysis was driven by an effect of aspirin on reduction in hematologic cancer incidence. In addition, a post hoc analysis of data from a phase II trial of aspirin for the prevention of thrombotic events in patients with multiple myeloma receiving thalidomide (23) suggested better survival among aspirin users compared with nonusers (24). Specifically, 89% of patients with multiple myeloma taking low-dose aspirin (81 mg daily) were alive at year 1 compared with 68% of patients in the no-aspirin group (P = 0.03). Despite the encouraging supportive evidence, we must also acknowledge that data from the Aspirin in Reducing Events in the Elderly (ASPREE) trial, which randomized 19,114 individuals to aspirin or a placebo, suggest an increased risk of cancer mortality among individuals in the aspirin group on the order of magnitude of 1.6 excess deaths per 1,000 person-years for all cancers; however, the excess risk of cancer mortality in the aspirin group was largely driven by gastrointestinal cancers, and risk was not statistically significantly increased for hematologic cancers, although the hematologic cancer subgroup was small and multiple myeloma–specific data were unavailable (25). Although other primary prevention trials of aspirin have not demonstrated similar adverse results (22, 26), results of the ASPREE trial do underscore the need to interpret the data we report with caution until more definitive studies have been conducted.

Recent population-based data consistently indicate that survival time after a diagnosis of multiple myeloma has been steadily increasing over the past several decades due to the rapid expansion of standard-of-care therapies available. Indeed, the 5-year relative survival rate increased from 34.5% in 2000 to more than 50% today (1). However, these improvements in survival have not been observed across all segments of the population equally, with older and minority patients receiving fewer gains for reasons that are thought to include treatment delays and unequal access to novel therapies (27–30). The finding that aspirin may enhance multiple myeloma survival is particularly relevant in light of these disparities, given its affordability and accessibility.

Strengths of the current study include the large prospective design, with sufficient endpoints occurring before availability of novel multiple myeloma therapies, which enabled us to examine heterogeneity in the effects we report by time period. In addition, participants provided biennially updated exposure information that enabled us to investigate pre- and postdiagnosis exposures separately.

The major limitation of the current study is the lack of comprehensive data on treatment or multiple myeloma clinically presenting features with which to adjust for disease severity at diagnosis or first-line therapy. However, in the subsample of participants with clinical data available, we did not observe statistically significant differences in the clinically presenting features of multiple myeloma by postdiagnosis aspirin use, providing preliminary (albeit crude) indication that selection bias or residual confounding by clinical characteristics may not be strongly influencing the findings we report. We were also not able to specifically investigate the potential benefit of low-dose aspirin use (81 mg) on multiple myeloma outcomes, which is the dose commonly used for cardiovascular disease prevention. In addition, our study sample comprised a relatively homogeneous population of nurses and health professionals who we can infer may have had comparable access to medical care and multiple myeloma therapies. Our sample is also comprised of predominantly white individuals and therefore these exposures warrant evaluation in more diverse cohorts. It is also notable that our results regarding an apparent benefit of aspirin in the postdiagnosis interval may only be generalizable to patients with multiple myeloma who have lived long enough after diagnosis to report aspirin use on a postdiagnosis follow-up questionnaire; 79 individuals in our sample did not return a questionnaire after their multiple myeloma diagnosis. Further, we did not see any evidence of confounding or effect measure modification by time interval between multiple myeloma diagnosis and the return of the first postdiagnosis questionnaire, suggesting that questionnaire return was not related to the severity of illness. Finally, given the relatively short survival time for patients with multiple myeloma in these cohorts, we did not use repeated measures of postdiagnosis aspirin use and instead relied on a single self-report assessment. Therefore, this exposure may not reflect participants' true long-run average postdiagnosis aspirin intake.

Our study provides preliminary support for the hypothesis that regular aspirin use may be associated with extended survival among patients with multiple myeloma. We urge others to seek confirmation in more diverse patient samples that have comprehensive clinical and treatment information and an ability to examine aspirin use by tablet dose so that stronger conclusions can be drawn. Until additional confirmation is achieved, the most conservative interpretation of our data is that aspirin use in the postdiagnosis interval appears to do no harm. If confirmed in other patient samples and randomized trials, regular aspirin use could be considered a safe, inexpensive, and complementary strategy to enhance survival in individuals living with multiple myeloma.

C.R. Marinac reports grants from NCI, Stand Up to Cancer, and American Cancer Society during the conduct of the study. G.A. Colditz reports grants from NCI during the conduct of the study. B. Rosner reports grants from NIH during the conduct of the study. M. Bustoros reports personal fees from Takeda Pharmaceutical Company, Bristol Myers Squibb, and Janssen Pharmaceuticals outside the submitted work. I.M. Ghobrial reports honoraria from Celgene, Bristol-Myers Squibb, Takeda, Amgen, and Janssen; consulting/advisory role with Bristol-Myers Squibb, Novartis, Amgen, Takeda, Celgene, Cellectar, Sanofi, Janssen, Pfizer, Menarini Silicon Biosystems Oncopeptides, The Binding Site, GlaxoSmithKline, AbbVie, and Adaptive; and travel and accommodations, and expenses from Bristol-Myers Squibb, Novartis, Celgene, Takeda, and Janssen Oncology. Her spouse, William Savage, MD, PhD, is also the Chief Medical Officer at Disc Medicine and holds equity in the company. B.M. Birmann reports grants from NIH/NCI during the conduct of the study as well as grants from American Institute for Cancer Research, NIH/NIAAA, and NIH/NCI outside the submitted work. No disclosures were reported by the other authors.

The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH. Opinions, interpretations, conclusions, and recommendations are those of the author(s) and are not necessarily endorsed by Stand Up To Cancer, the Entertainment Industry Foundation, or the American Association for Cancer Research.

C.R. Marinac: Conceptualization, resources, data curation, formal analysis, writing–original draft, writing–review and editing. D.H. Lee: Formal analysis, writing–review and editing. G.A. Colditz: Conceptualization, supervision, funding acquisition, writing–review and editing. T.R. Rebbeck: Conceptualization, supervision, funding acquisition, writing–review and editing. B. Rosner: Supervision, methodology, writing–review and editing. M. Bustoros: Data curation, supervision, methodology, writing–review and editing. I.M. Ghobrial: Resources, supervision, writing–review and editing. B.M. Birmann: Conceptualization, data curation, formal analysis, supervision, funding acquisition, methodology, writing–review and editing.

This study was funded in part by the NCI of the NIH under award numbers K07 CA115687 (to B.M. Birmann), R01 CA127435 (to G.A. Colditz), P01 CA87969 (to B.M. Birmann, B. Rosner), UM1 CA186107 (to B.M. Birmann, B. Rosner), U01 CA167552 (to B.M. Birmann, B. Rosner), R21 CA198239 (to B.M. Birmann), F32 CA220859 (to C.R. Marinac), R03 CA204825 (to D.H. Lee), and K22 CA251648 (to C.R. Marinac). This research was also supported by the American Cancer Society under award number PF-17–231–01-CCE (to C.R. Marinac), Clinical Research Professorship (to G.A. Colditz), as well as institutional funds from the Dana–Farber Cancer Institute, and Stand Up To Cancer under award number SU2C-AACR-DT-28-18 (to I.M. Ghobrial). Stand Up To Cancer is a program of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the scientific partner of Stand Up To Cancer. Finally, we would like to thank the participants and staff of the HPFS and NHS for their valuable contributions as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, and WY. The authors assume full responsibility for analyses and interpretation of these data.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.