Aspirin, NSAIDs, and Acetaminophen and Ovarian Cancer Risk
Study site, number of case patients and control subjects, and exposure prevalence for each of the 12 OCAC studies are described in Table 1. Overall, 18% of the study population reported regular use (at least once per week) of aspirin, 24% reported regular use of nonaspirin NSAIDs, and 16% reported regular use of acetaminophen.
Figure 1A shows the association between aspirin use (regular vs nonregular use) and ovarian cancer risk. Regular aspirin use was associated with a reduced risk of ovarian cancer (OR = 0.91; 95% CI = 0.84 to 0.99; I = 5.2%). Among seven studies that reported information on frequency of use, daily use was associated with a 20% reduction in ovarian cancer risk (OR = 0.80; 95% CI = 0.67 to 0.96) (Table 2). Among three studies that reported information on dose, low-dose aspirin use (<100mg/day) was associated with a 34% reduction in ovarian cancer risk (OR = 0.66; 95% CI = 0.53 to 0.83) (Table 2). In analyses of combined categories of frequency and dose of aspirin use, the reduced risk was apparent for daily users of aspirin regardless of dose (low dose: OR = 0.64, 95% CI = 0.50 to 0.81; high dose: OR = 0.78, 95% CI = 0.62 to 0.97) (Table 3).
(Enlarge Image)
Figure 1.
The summary odds ratios (ORs) and 95% confidence intervals (CIs) for the association between regular (at least once per week) use of aspirin (A), nonaspirin nonsteroidal anti-inflammatory drugs (NSAIDs) (B), and acetaminophen (C) and ovarian cancer risk. Summary odds ratios and 95% confidence intervals were estimated using a random-effect meta-analytic model. All statistical tests were two-sided. I is the percentage of variation across studies due to heterogeneity rather than chance. % Weight describes the weight (inverse variance) each study contributed to the summary odds ratio, and the size of the surrounding square is an illustrative representation of study weighting. The horizontal lines represent study-specific confidence intervals; if ending in an arrow, this indicates that the interval transcends the region plotted. The diamond represents the summary odds ratio and 95% confidence interval. Studies are presented in order of median year of case accrual from earliest to most recent. AUS = Australian Ovarian Cancer Study, Australian Cancer Study; CON = Connecticut Ovary Study; DOV = Diseases of the Ovary and their Evaluation Study; HAW = Hawaii Ovarian Cancer Study; HOP = Hormones and Ovarian Cancer Prediction Study; MAL = Malignant Ovarian Cancer Study; NCO = North Carolina Ovarian Cancer Study; NEC = New England Case–Control Study of Ovarian Cancer; NJO = New Jersey Ovarian Cancer Study; UCI = University of California, Irvine Ovarian Cancer Study; UKO = United Kingdom Ovarian Cancer Population Study; USC = University of Southern California Study of Lifestyle and Women's Health.
In subtype analyses, regular aspirin use was associated with reduced risks of serous, endometrioid, and mucinous ovarian cancer, but only the results for serous cancer reached statistical significance (OR = 0.89; 95% CI = 0.80 to 0.99) (Table 4). Pairwise comparisons showed no significant differences in risk between the subtypes (P > .05).
Regular nonaspirin NSAID use was associated with a reduced, albeit not statistically significant, risk of ovarian cancer (OR = 0.90; 95% CI = 0.77 to 1.05; I = 73.2%) (Figure 1B). Among the three studies that reported information on dose, high-dose nonaspirin NSAID use (≥500mg/day) was associated with a 24% reduction in ovarian cancer risk (OR = 0.76; 95% CI = 0.64 to 0.91) (Table 2). In analyses of combined categories of frequency and dose, the reduced risk of ovarian cancer was apparent among both categories of high-dose nonaspirin NSAID use (<30 days per month: OR = 0.77, 95% CI = 0.57 to 1.04; daily: OR = 0.75; 95% CI = 0.60 to 0.94), with a weaker association with daily users of low-dose nonaspirin NSAIDs (OR = 0.88; 95% CI = 0.70 to 1.11) (Table 3). The association between nonaspirin NSAIDs and risk was strongest for serous cancers but did not differ across histologic subtypes of ovarian cancer (Table 4).
Acetaminophen use was not associated with ovarian cancer risk (OR = 0.99; 95% CI = 0.88 to 1.12; I: 40.0%) (Figure 1C). No associations were observed when analyzing dose, duration, or frequency of acetaminophen use and ovarian cancer risk (Table 2). Further we observed no association between acetaminophen use and histologic subtypes of ovarian cancer (Table 4).
The association between NSAID use and high-grade serous tumors was not substantially different than the results reported for all serous tumors combined (results not shown). Tumors of low malignant potential (n = 2059) were not associated with analgesic use (data not shown). In analyses stratified by age, body mass index, oral contraception use, and history of endometriosis, similar NSAID use and ovarian cancer associations were observed as in the overall population (results not shown). Based on the adjusted rank correlation and regression asymmetry tests, there was no indication of small study effects (all P > .05) in the summary estimates for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer. Although there was heterogeneity in the definition of nonaspirin NSAID use, individual exclusion of each study did not substantially change the summary odds ratio (results not shown); however, the exclusion of two studies resulted in a decrease in I from 73.2% to 27.8% but no substantial change in the summary odds ratio (results not shown).
In a sensitivity analysis excluding peritoneal and fallopian tube cancers, the pooled summary odds ratios for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer were not substantially different from the odds ratios observed for the overall case group (data not shown). The associations between regular use of NSAIDs and ovarian cancer did not substantially change when the analyses were restricted to non-Hispanic white case patients and control subjects (data not shown). In analyses using women who reported nonregular use of all three NSAIDs as the reference group, a stronger reduced risk was observed for regular use of aspirin (OR = 0.81; 95% CI = 0.68–0.99) and nonaspirin NSAID (OR = 0.86; 95% CI = 0.71–1.05), possibly reflecting reduced "contamination" of the referent group with users of NSAID types other than the medication under examination in each specific analysis (data not shown). In sensitivity analyses restricted to the six studies that specified 6 months or more as the minimum duration or the nine US studies, the pooled summary odds ratios for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer were not substantially different from the odds ratios observed for the overall pooled analysis (data not shown). Finally, in the sensitivity analysis excluding case patients with the most restrictive definition of medication use, the pooled summary odds ratios for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer were not substantially different from the pooled odds ratios observed for all 12 studies (data not shown).
Results
Study site, number of case patients and control subjects, and exposure prevalence for each of the 12 OCAC studies are described in Table 1. Overall, 18% of the study population reported regular use (at least once per week) of aspirin, 24% reported regular use of nonaspirin NSAIDs, and 16% reported regular use of acetaminophen.
Aspirin
Figure 1A shows the association between aspirin use (regular vs nonregular use) and ovarian cancer risk. Regular aspirin use was associated with a reduced risk of ovarian cancer (OR = 0.91; 95% CI = 0.84 to 0.99; I = 5.2%). Among seven studies that reported information on frequency of use, daily use was associated with a 20% reduction in ovarian cancer risk (OR = 0.80; 95% CI = 0.67 to 0.96) (Table 2). Among three studies that reported information on dose, low-dose aspirin use (<100mg/day) was associated with a 34% reduction in ovarian cancer risk (OR = 0.66; 95% CI = 0.53 to 0.83) (Table 2). In analyses of combined categories of frequency and dose of aspirin use, the reduced risk was apparent for daily users of aspirin regardless of dose (low dose: OR = 0.64, 95% CI = 0.50 to 0.81; high dose: OR = 0.78, 95% CI = 0.62 to 0.97) (Table 3).
(Enlarge Image)
Figure 1.
The summary odds ratios (ORs) and 95% confidence intervals (CIs) for the association between regular (at least once per week) use of aspirin (A), nonaspirin nonsteroidal anti-inflammatory drugs (NSAIDs) (B), and acetaminophen (C) and ovarian cancer risk. Summary odds ratios and 95% confidence intervals were estimated using a random-effect meta-analytic model. All statistical tests were two-sided. I is the percentage of variation across studies due to heterogeneity rather than chance. % Weight describes the weight (inverse variance) each study contributed to the summary odds ratio, and the size of the surrounding square is an illustrative representation of study weighting. The horizontal lines represent study-specific confidence intervals; if ending in an arrow, this indicates that the interval transcends the region plotted. The diamond represents the summary odds ratio and 95% confidence interval. Studies are presented in order of median year of case accrual from earliest to most recent. AUS = Australian Ovarian Cancer Study, Australian Cancer Study; CON = Connecticut Ovary Study; DOV = Diseases of the Ovary and their Evaluation Study; HAW = Hawaii Ovarian Cancer Study; HOP = Hormones and Ovarian Cancer Prediction Study; MAL = Malignant Ovarian Cancer Study; NCO = North Carolina Ovarian Cancer Study; NEC = New England Case–Control Study of Ovarian Cancer; NJO = New Jersey Ovarian Cancer Study; UCI = University of California, Irvine Ovarian Cancer Study; UKO = United Kingdom Ovarian Cancer Population Study; USC = University of Southern California Study of Lifestyle and Women's Health.
In subtype analyses, regular aspirin use was associated with reduced risks of serous, endometrioid, and mucinous ovarian cancer, but only the results for serous cancer reached statistical significance (OR = 0.89; 95% CI = 0.80 to 0.99) (Table 4). Pairwise comparisons showed no significant differences in risk between the subtypes (P > .05).
Nonaspirin NSAIDs
Regular nonaspirin NSAID use was associated with a reduced, albeit not statistically significant, risk of ovarian cancer (OR = 0.90; 95% CI = 0.77 to 1.05; I = 73.2%) (Figure 1B). Among the three studies that reported information on dose, high-dose nonaspirin NSAID use (≥500mg/day) was associated with a 24% reduction in ovarian cancer risk (OR = 0.76; 95% CI = 0.64 to 0.91) (Table 2). In analyses of combined categories of frequency and dose, the reduced risk of ovarian cancer was apparent among both categories of high-dose nonaspirin NSAID use (<30 days per month: OR = 0.77, 95% CI = 0.57 to 1.04; daily: OR = 0.75; 95% CI = 0.60 to 0.94), with a weaker association with daily users of low-dose nonaspirin NSAIDs (OR = 0.88; 95% CI = 0.70 to 1.11) (Table 3). The association between nonaspirin NSAIDs and risk was strongest for serous cancers but did not differ across histologic subtypes of ovarian cancer (Table 4).
Acetaminophen
Acetaminophen use was not associated with ovarian cancer risk (OR = 0.99; 95% CI = 0.88 to 1.12; I: 40.0%) (Figure 1C). No associations were observed when analyzing dose, duration, or frequency of acetaminophen use and ovarian cancer risk (Table 2). Further we observed no association between acetaminophen use and histologic subtypes of ovarian cancer (Table 4).
Additional Analyses
The association between NSAID use and high-grade serous tumors was not substantially different than the results reported for all serous tumors combined (results not shown). Tumors of low malignant potential (n = 2059) were not associated with analgesic use (data not shown). In analyses stratified by age, body mass index, oral contraception use, and history of endometriosis, similar NSAID use and ovarian cancer associations were observed as in the overall population (results not shown). Based on the adjusted rank correlation and regression asymmetry tests, there was no indication of small study effects (all P > .05) in the summary estimates for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer. Although there was heterogeneity in the definition of nonaspirin NSAID use, individual exclusion of each study did not substantially change the summary odds ratio (results not shown); however, the exclusion of two studies resulted in a decrease in I from 73.2% to 27.8% but no substantial change in the summary odds ratio (results not shown).
In a sensitivity analysis excluding peritoneal and fallopian tube cancers, the pooled summary odds ratios for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer were not substantially different from the odds ratios observed for the overall case group (data not shown). The associations between regular use of NSAIDs and ovarian cancer did not substantially change when the analyses were restricted to non-Hispanic white case patients and control subjects (data not shown). In analyses using women who reported nonregular use of all three NSAIDs as the reference group, a stronger reduced risk was observed for regular use of aspirin (OR = 0.81; 95% CI = 0.68–0.99) and nonaspirin NSAID (OR = 0.86; 95% CI = 0.71–1.05), possibly reflecting reduced "contamination" of the referent group with users of NSAID types other than the medication under examination in each specific analysis (data not shown). In sensitivity analyses restricted to the six studies that specified 6 months or more as the minimum duration or the nine US studies, the pooled summary odds ratios for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer were not substantially different from the odds ratios observed for the overall pooled analysis (data not shown). Finally, in the sensitivity analysis excluding case patients with the most restrictive definition of medication use, the pooled summary odds ratios for the associations between regular use of aspirin, nonaspirin NSAIDs, or acetaminophen and ovarian cancer were not substantially different from the pooled odds ratios observed for all 12 studies (data not shown).