SEARCH

SEARCH BY CITATION

Keywords:

  • Analgesics;
  • ovarian cancer;
  • histological types;
  • case–control study;
  • gyne-oncology

Abstract

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References

Objective. The role of analgesic drug use in development of ovarian cancer is not fully understood. We examined the association between analgesic use and risk of ovarian cancer. In addition, we examined whether the association differed according to histological types. Design. Population-based case–control study. Setting. Denmark in the period 1995–1999. Population. We included 756 women with epithelial ovarian cancer and 1564 randomly selected control women aged 35–79 years. Methods. Information on analgesic drug use was collected from personal interviews. Analgesic drugs were divided into the following categories: any analgesics; aspirin; non-aspirin non-steroidal anti-inflammatory drugs; paracetamol; and other analgesic drugs. The association between analgesic drug use and ovarian cancer risk was analysed using multiple logistic regression models. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated. Main outcome measures. Epithelial ovarian cancer. Results. Women with a regular use of any analgesics (OR = 0.79; 95% CI 0.62 − 1.01) or aspirin (OR = 0.68; 95% CI 0.46 − 1.02) had a decreased risk of ovarian cancer, although not statistically significant. Regular use of non-aspirin non-steroidal anti-inflammatory drugs, paracetamol or other analgesics did not decrease ovarian cancer risk. Use of any analgesics (OR = 0.72; 95% CI 0.53–0.98) or aspirin (OR = 0.60; 95% CI 0.36–1.00) resulted in a statistically significant decreased risk of serous ovarian cancer but not mucinous or other ovarian tumors. Conclusion. In accordance with most previous studies, our results indicate a possible inverse association between analgesic use, particularly aspirin, and ovarian cancer risk.


Abbreviations: 
CI

confidence interval

COX

cyclo-oxygenase

NSAID

non-steroidal anti-inflammatory drug

OR

odds ratio

Key Message

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References

We found that regular use of any analgesics and of aspirin decreased the risk of overall and of serous ovarian cancer. Use of non-aspirin non-steroidal anti-inflammatory drugs, paracetamol or other analgesics was not related to ovarian cancer risk.

Introduction

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References

Ovarian cancer is the most lethal gynecological cancer, with an overall five year survival of approximately 40% (1). Denmark has one of the world's highest incidence and mortality rates (11.0 and 7.0 per 100 000 women, respectively) of ovarian cancer (World standard population; 2).

Several factors may be associated with ovarian cancer, such as age, hormonal, genetic and environmental factors (3–5), which have led to different hypotheses on the etiology of ovarian cancer. According to the incessant ovulation hypothesis, ovarian cancer occurs through repeated cycles of ovulation-induced trauma and repair of the ovarian surface epithelium (6). Another hypothesis is that hormones and reproductive factors may raise the levels of the hormone gonadotropin, which increases the risk of ovarian cancer because excess gonadotropin secretion leads to increased estrogen stimulation of the ovarian epithelium (7). Lastly, chronic inflammation is suspected to play a role in carcinogenesis, possibly acting in concert with ovulation (8). Analgesic drugs with anti-inflammatory effects are thought to exhibit chemopreventive effects through inhibition of cyclo-oxygenase enzymes (COX-1 and COX-2 enzymes) and their prostaglandin products, particularly prostaglandin E2. Therefore, use of analgesic drugs with anti-inflammatory effects has been proposed to reduce the risk of several types of cancer, particularly colorectal cancer (9).

In 1993, Tzonou et al. (10) first reported an inverse association between analgesic use and ovarian cancer risk, but no information on the type of analgesic was available. Since then, numerous epidemiological studies have investigated the relation between specific types of analgesics and risk of ovarian cancer. Several studies indicate that analgesic use reduces the risk of ovarian cancer (11–16), but some do not find evidence of an association (17–21). However, the definitions and measurement methods of regular analgesic use vary considerably between studies, and some studies (15,16,21) were not able to control for potential confounders. Furthermore, only few studies (14,19,20,22) have examined the effect of analgesic use in relation to histological types of ovarian cancer and they are limited by a small number of analgesic users.

Although several studies have investigated the association between analgesic use and ovarian cancer risk, the findings are inconsistent. In the present study, we examined the association between analgesic use and risk of ovarian cancer using data from a large population-based case–control study among Danish women. In addition, we examined the association between analgesic use and risk of histological types of ovarian cancer.

Material and methods

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References

The study is based on data from the Danish MALOVA (MALignant OVArian cancer) study, a population-based case–control study on ovarian cancer. The study was approved by the national as well as the local ethical committees, and written informed consent was obtained from all study participants. The design has been described in detail elsewhere (5).

Briefly, from January 1995 to May 1999 women aged 35–79 years who were scheduled for an explorative laparotomy or laparoscopy on the suspicion of an ovarian tumor were requested to participate in the study with blood and tissue samples and a personal interview. The study area consisted of 16 gynecological hospital departments in Denmark. To ensure that all eligible cases in the study area were included the study database was linked to the Danish Cancer Registry every second month. The Danish Cancer Registry contains nationwide information about all incident cancer cases diagnosed in Denmark since 1943. It is supplemented by linkage to the Causes of Death Registry and the National Patient Registry to ensure complete data (23). The questionnaire was designed as a person-to-person interview and provided information on social, reproductive, medical and gynecological history, use of hormones, use of different types of medicine including analgesics, family history of cancer, and lifestyle factors. The interview took place at the hospital immediately or as soon as possible after the diagnosis and was conducted by trained personnel.

In total, 1235 women with histologically verified ovarian cancer were identified in the study area. Inclusion and exclusion criteria for ovarian cancer cases are shown in Figure 1. For analysis, we included 756 women with epithelial ovarian cancer (hereafter denoted ovarian cancer). Of these, there were 554 women with invasive ovarian cancer and 202 women with borderline ovarian tumors. Concerning histological types, 447 were serous adenocarcinomas, 138 were mucinous adenocarcinomas, and 171 were other types of adenocarcinomas (including endometrioid, undifferentiated and papillary adenocarcinomas and clear-cell neoplasms).

image

Figure 1. Overview of included cases and controls for analysis.

Download figure to PowerPoint

A random sample of women aged 35–79 years from the general female population in the study area was drawn using the unique Danish personal identification number as the key identifier. Controls were included simultaneously with the cases and frequency matched in five year intervals using the age distribution of women with ovarian cancer registered in the Danish Cancer Registry from 1987 to 1992. In all, 3839 women were invited to participate as controls with a personal interview and a blood sample. Figure 1 outlines the inclusion and exclusion criteria for the controls. As shown, we included 1564 women as controls. The telephone interview was less comprehensive, so the present study is based exclusively on personal interviews.

In the personal interview, study participants were asked whether they had ever taken medicine on a regular basis, i.e. two times or more per week for more than one month, when they started and stopped using the preparation, how often they used it, and what the name of the preparation was. According to the common classification of analgesic drugs, we divided analgesics into the following four categories: aspirin; non-aspirin non-steroidal anti-inflammatory drugs (non-aspirin NSAIDs); paracetamol; and other analgesic drugs, including morphine-like drugs, antidepressants, migraine medicine and drugs that were stated as “analgesic drugs” without further specification (24). We also calculated any analgesic use as the sum of each category of analgesics.

Statistical analysis

The association between regular analgesic drug use and ovarian cancer risk was analysed using multiple logistic regression analysis by estimating odds ratios (ORs) and corresponding 95% confidence intervals (CIs). We estimated the association between regular use of analgesics (any analgesics, aspirin, non-aspirin NSAIDs, paracetamol, and other analgesic drugs, respectively) and ovarian cancer, as well as histological types of ovarian cancer. In addition, we investigated the effect of duration of regular analgesic use. In all analyses, women who never had a regular use of the analgesic drug in question served as the reference group, i.e. women in the reference group could have a regular use of the other categories of analgesic drugs not in question. To reduce the possibility that early cancer symptoms might have influenced the use of analgesic drugs we only considered analgesic use that began at least one year before diagnosis for cases and one year before the interview for controls. Lastly, we performed a separate analysis of the association between regular use of selective COX-2 inhibitors [as defined by Cronin-Fenton et al. (25)] and ovarian cancer.

Concerning the frequency of regular analgesic use (number of pills per week), the majority of analgesic users had taken the preparation seven times per week, and a few analgesic users had taken the preparation for less or more than seven times per week. Therefore, no analyses of the effect of frequency of analgesic use were performed. Information about the dose of analgesics was not available. All analyses were repeated separately for women with invasive ovarian cancer and borderline ovarian tumors. As we found virtually no differences in the risk estimates, we only report the results for invasive ovarian cancer and borderline ovarian tumors combined.

All analyses were adjusted for age in five year categories corresponding to the sampling of controls. Furthermore, all analyses were adjusted for pregnancy (ever/never and number of additional pregnancies as a linear variable) and oral contraceptive use (ever/never and additional years of oral contraceptive use as a linear variable). These potential confounders were selected based on a priori knowledge of their possible causal role in the development of ovarian cancer. We also considered the following other potential confounders as adjustment variables: hysterectomy; menopausal status; age at menarche; age at first and last pregnancy; breastfeeding; hormone replacement therapy use; family history of breast and/or ovarian cancer; education; body mass index; and smoking. They were defined as relevant adjustment variables if they altered the risk estimate by 10% or more. As this was not the case, they were not included in the final logistic regression model. Linearity for all quantitative variables except age was tested by comparison with a model with spline effects, with knots placed at the tertiles. No significant deviations from linearity were found for any of the variables. The level of statistical significance was chosen at a p-value of <0.05. All analyses were two sided and were performed using the PROC GENMOD procedure in the SAS software package (version 9.2; SAS institute, Cary, NC, USA).

Results

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References

The mean age was 57.8 years (SD 10.8 years) for cases and 57.1 years (SD 11.3 years) for controls. Cases were more likely than controls to be nulliparous. Eighty-five per cent of the ovarian cancer cases had ever been pregnant compared with 94% of the control women. Both ever pregnancy (OR = 0.39; 95% CI 0.29 − 0.52) and additional number of pregnancies (OR = 0.89; 95% CI 0.83 − 0.95) were associated with a significantly reduced risk of ovarian cancer. Oral contraceptives were used by 46% of women with ovarian cancer and 55% of the control women. Ever use of oral contraceptives was associated with a significantly reduced risk of ovarian cancer (OR = 0.65; 95% CI 0.53 − 0.81). Furthermore, a dose–response relation between oral contraceptive use and ovarian cancer was observed, where each extra year of oral contraceptive use significantly decreased the risk of ovarian cancer (OR = 0.96; 95% CI 0.94 − 0.98; data not shown).

Table 1 presents ORs for ovarian cancer in relation to analgesic use. Women who ever had a regular use of any analgesics had a decreased risk of ovarian cancer (OR = 0.79; 95% CI 0.62 − 1.01; p= 0.06) compared with women who never used any analgesics, although the association was not statistically significant. Likewise, regular users of aspirin had a statistically non-significant decreased risk of ovarian cancer (OR = 0.68; 95% CI 0.46 − 1.02; p= 0.06) compared with non-users. Aspirin use for less than five years was associated with a decreased, but not statistically significant, risk of ovarian cancer compared with non-users (OR = 0.59; 95% CI 0.34 − 1.02 p= 0.06), whereas no association was found with aspirin use for five years or more. However, when the linearity of duration was evaluated we found that, among ever users, each extra year of aspirin use resulted in an increased risk of ovarian cancer (OR = 1.04; 95% CI 1.00–1.09, p= 0.06), although not statistically significant. We did not find evidence of a reduced risk of ovarian cancer with use of non-aspirin NSAIDs, paracetamol, and other analgesics. The separate analysis of the relation between use of selective COX-2 inhibitors and ovarian cancer risk was based on sparse data, because only eight cases and 12 controls had a regular use of COX-2 inhibitors. We found no evidence of an association between use of COX-2 inhibitors and risk of ovarian cancer (OR = 1.32; 95% CI 0.52–3.33; data not shown). It was not possible to perform meaningful analyses of duration of use of COX-2 inhibitors or risk of histological types of ovarian cancer in relation to use of COX-2 inhibitors.

Table 1.  Association between regular use of analgesics and ovarian cancer.
 Controls (n= 1564)Cases (n= 756)OR95% CI*
n Percentage n Percentage
  1. Note: Abbreviations: CI, confidence interval; and OR, odds ratio.

  2. *Adjusted for age, pregnancy (ever/never), number of pregnancies (linear), oral contraceptive use (ever/never) and duration of oral contraceptive use (linear).

  3. Due to missing values, the numbers do not sum up to the total.

Any analgesic use 
  Regular use 
  Never127481.563884.41.0
  Ever29018.511815.60.790.62–1.01
 Duration of use (years)†
  <51157.4506.60.890.62–1.27
  ≥51539.8628.20.780.57–1.08
  Per year used    0.990.97–1.01
Aspirin
 Regular use
  Never146193.471995.11.0
  Ever1036.6374.90.680.46–1.02
 Duration of use (years)†
  <5624.0182.40.590.34–1.02
  ≥5211.3162.11.360.69–2.70
  Per year used    1.041.00–1.09
Non-aspirin non-steroidal anti-inflammatory drugs
 Regular use
  Never145192.870092.61.0
  Ever1137.2567.41.060.76–1.50
 Duration of use (years)†
  <5583.7293.81.130.71–1.81
  ≥5291.9192.51.380.75–2.54
 Per year used    1.000.95–1.07
Paracetamol
 Regular use
  Never151296.773897.61.0
  Ever523.3182.40.730.42–1.26
 Duration of use (years)†
  <5231.5121.61.250.61–2.57
 ≥5130.840.50.530.17–1.67
  Per year used    0.970.90–1.05
Other analgesics
 Regular use
  Never148094.672696.01.0
  Ever845.4304.00.730.47–1.13
 Duration of use (years)†
  <5412.6141.90.760.40–1.42
  ≥5231.591.20.780.35–1.74
  Per year used    0.980.92–1.04

In Table 2, ORs for histological types of ovarian cancer in relation to analgesic use are presented. In general, owing to a smaller number of cases, the risk estimates for histological types were less precise than the risk estimates presented in Table 1. As serous ovarian cancer is the most frequent type of ovarian cancer, the association observed for this histological type resembled those for overall ovarian cancer the most. Accordingly, we found a statistically significant decreased risk of serous ovarian cancer among women who ever had a regular use of any analgesics (OR = 0.72; 95% CI 0.53–0.98) and aspirin (OR = 0.60; 95% CI 0.36–1.00). In general, we found no association between analgesic use and risk of mucinous and other ovarian tumours. However, due to a limited number of analgesic users among mucinous and other ovarian cases, these results should be interpreted with caution.

Table 2.  Association between regular use of analgesics and histological types of ovarian cancer.
 ControlsSerousMucinousOther
(n= 1564)(n= 447)(n= 138)(n= 171)
n Percentage n PercentageOR95% CI* n PercentageOR95% CI* n PercentageOR95% CI*
  1. Note: Abbreviations: CI, confidence interval; and OR, odds ratio.

  2. *Adjusted for age, pregnancy (ever/never), number of pregnancies (linear), oral contraceptive use (ever/never) and duration of oral contraceptive use (linear).

Any analgesic use
 Never127481.538385.71.011684.11.013981.31.0
 Ever29018.56414.30.720.53 − 0.982215.90.900.56 − 1.503218.70.890.58 − 1.36
Aspirin
 Never146193.442895.81.013094.21.016194.21.0
 Ever1036.6194.30.600.36 − 1.0085.80.950.45 − 2.01105.90.660.33 − 1.37
Non-aspirin non-steroidal anti-inflammatory drugs
 Never145192.842294.41.012691.31.015288.91.0
 Ever1137.2255.60.810.51 − 1.29128.71.300.69 − 2.431911.11.590.93 − 2.72
Paracetamol
 Never151296.743196.41.013799.31.017099.41.0
 Ever523.3163.61.090.61 − 1.9610.70.230.03 − 1.6510.60.170.02 − 1.22
Other analgesics
 Never148094.642795.51.013497.11.016596.51.0
 Ever845.4204.50.830.50 − 1.4042.90.550.20 − 1.5466.70.640.27 − 1.52

We also investigated whether the decreased risk of ovarian cancer associated with analgesic use varied according to pregnancy and oral contraceptive use, i.e. whether there was a possible interaction between analgesic use and these potential effect modifiers. However, our results provided no evidence that the decreased risk of ovarian cancer associated with analgesic use was modified by pregnancy and oral contraceptive use, and none of the interaction terms was statistically significant (i.e. all p-values >0.05; data not shown).

Discussion

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References

In this population-based case–control study of the association between analgesic use and ovarian cancer we found a decreased risk of overall and serous ovarian cancer among regular users of any analgesics and aspirin. In contrast, use of non-aspirin NSAIDs, paracetamol, and other analgesics was not associated with ovarian cancer risk. Furthermore, no relation between analgesic use and risk of mucinous and other ovarian tumors was observed.

Only one previous case–control study has examined the association between any analgesic use and ovarian cancer risk (10). In accordance with our study, it revealed an inverse association between regular use of analgesics (twice per week or more) and ovarian cancer (relative risk = 0.51; 95% CI 0.26 − 1.02).

The inverse association between regular aspirin use and risk of ovarian cancer observed in this study is consistent with findings from seven studies (11–14,26–28). A meta-analysis conducted in 2005 by Bonovas et al. (29) and six other studies (17–21,30) found no evidence of an association between aspirin use and ovarian cancer risk, while one study (22) found that use of aspirin increased the risk.

We found no association between use of non-aspirin NSAIDs and ovarian cancer risk, which is in agreement with the meta-analysis by Bonovas et al. (29) and four other studies (14,18,26,27). By contrast, six studies (11–13,16,19,20) found a decreased risk with non-aspirin NSAIDs use and one study found an increased risk between use of non-aspirin NSAIDs and ovarian cancer risk (22).

The chemopreventive effect of aspirin and non-aspirin NSAIDs found in some studies is thought to be mediated through inhibition of the COX enzymes, which account for the synthesis of prostaglandins that may lead to increased estrogen stimulation (9). Cyclo-oxygenase enzymes consist of the classical COX-1 enzyme, which is constitutively expressed in many tissues, and a second enzyme, COX-2, which is induced by various stimuli, such as mitogens and cytokines, and is involved in many inflammatory reactions (24). The protective effect of aspirin, but not non-aspirin NSAIDs, found in this study may reflect differences in the way these drugs induce their effect on inflammation. Aspirin is known to be relatively selective for COX-1, while most non-aspirin NSAIDs are inhibitors of both COX-1 and COX-2, although they vary in the degree of inhibition of each enzyme. In addition, aspirin inactivates COX enzymes irreversibly, whereas non-aspirin NSAIDs inactivate COX-enzymes reversibly (24). Moreover, unlike aspirin, non-aspirin NSAIDS constitute a heterogeneous group of medications with different properties and effects on COX inhibition. Furthermore, the concentration of the respective drugs that is required to induce anti-proliferative effects may differ. Lastly, in contrast to non-aspirin NSAIDs, aspirin may also exert its inhibitory effect through COX-independent mechanisms, such as modulation of the estrogen synthesis and antioxidant effects (31).

A protective effect of paracetamol has been suggested in a meta-analysis from 2006 by Bonovas et al. (32) and three other studies (11,14,30), while one study (22) found that use of paracetamol increased the risk. In accordance with other studies (12,15,18,20,26), we did not find sufficient evidence of an association between paracetamol use and risk of ovarian cancer. The anticancer mechanism proposed for paracetamol differs from the mechanism proposed for aspirin and non-aspirin NSAIDs, because paracetamol has only a weak anti-inflammatory effect. It is unlikely, therefore, that paracetamol reduces the risk of ovarian cancer through a prostaglandin inhibitor pathway. Instead, paracetamol may reduce ovarian cancer risk through an antigonadotropic effect (14) or by induction of specific reproductive atrophy due to its sex-steroid resembling phenolic ring (32). However, more studies are needed to verify these hypotheses.

No previous studies have examined the effect of other types of analgesics and therefore our results for this category could not be compared with other studies. Most studies investigating the effect of duration of analgesic use found that the longer the use, the stronger the protective effect (11,12,14,30). In contrast, for aspirin we found an increased risk of ovarian cancer with increasing use. This finding is not easily explainable and may be due to chance. For the other types of analgesics we found no effect of duration on ovarian cancer risk.

An interaction between COX-2 and prostaglandin E2 has been found to play an important role in carcinogenesis (33). To our knowledge, only two epidemiological studies (11,34) have specifically investigated the relation between selective COX-2 inhibitors and ovarian cancer risk, both indicating a slightly decreased risk. Owing to small numbers, we were not able to carry out a meaningful evaluation of the association between selective COX-2 inhibitors and ovarian cancer risk. Larger studies are warranted to provide further insights into the various effects and mechanisms of COX-1 and COX-2 inhibitors. Furthermore, for colorectal cancer, laboratory studies have indicated common pathways for COX-2 and epidermal growth factor receptor, implying that COX-2 may induce epidermal growth factor receptor expression or vice versa. Epidermal growth factor receptor has been shown to play an important role in cell proliferation, adhesion, invasion, survival and angiogenesis in colorectal cancer (35). A similar interaction between COX-2 and epidermal growth factor receptor for ovarian cancer may have important implications for the tumorigenesis of ovarian cancer and thus be relevant for the chemopreventive effect of COX-2 inhibitors against ovarian cancer.

When we performed separate analyses for histological types of ovarian cancer we found a decreased risk of serous ovarian cancer with use of any analgesics and aspirin, whereas no sufficient evidence of a decreased risk of mucinous or other ovarian tumours was observed with analgesic use. To our knowledge, no previous studies have found a decreased risk of serous ovarian cancer with analgesic use. Given the number of tests performed, this may be a chance finding. However, recent literature suggests that the Fallopian tube may be the source of both low- and high-grade serous ovarian tumors (36). As inflammation of the Fallopian tube is common, our finding may be biologically plausible (37). Given the evidence of a strong association between analgesic use and risk of colon cancer and the resemblance of mucinous ovarian tumors to the cells of the intestines (38), a decreased risk of mucinous ovarian cancer with analgesic use was expected. However, as our analyses of mucinous ovarian tumors were based on a limited number of cases, no firm conclusions can be made regarding the association between analgesic use and mucinous ovarian tumors. Concerning the degree of invasiveness, we found no differences in the risk estimates for invasive ovarian cancer and borderline ovarian tumors, which is consistent with previous studies (14,22). Hence, based on these sparse findings there are no major indications of differences in the risk profile for ovarian cancer in relation to invasiveness and histological types with regard to analgesic use, but further studies are certainly needed to address this issue. Such studies should also take into consideration the new dualistic distinction between type I and type II ovarian cancer. Type I consists of low-grade serous, low-grade endometrioid, clear cell, mucinous and Brenner tumors, while type II comprises high-grade serous, high-grade endometrioid, malignant mixed mesodermal tumors and undifferentiated carcinomas (36). Owing to a limited number of analgesic users in our study, we were not able to investigate analgesic use in relation to risk of type I and type II ovarian cancer, and we are not familiar with other epidemiological studies that have investigated this association; therefore, we can only hypothesize about the link between the inflammatory theory and the new paradigm for the pathogenesis and origin of ovarian cancer. It has been proposed that analgesic drugs inhibit the progression of ovarian cancer but that they are not critical for its initiation (27). As type I tumors are thought to develop in a stepwise fashion from precursor lesions, while type II tumors grow rapidly (36), analgesic use may affect the development of type I and type II lesions in different ways. Furthermore, it has been suggested that COX-1 and COX-2 expression differs with regard to tumor type. Tumors expressing COX-2 are characterized as aggressive and with a poor prognosis, and COX-2 expression is higher in invasive than in borderline ovarian tumors (39).

The differing findings in previous epidemiological studies of the association between analgesic use and ovarian cancer could reflect chance findings due to a limited number of observations or differences in study design and analytical strategies. Most importantly, definitions of regular analgesic use vary considerably. Most studies, including the present study, assessed analgesic use by self-reporting. These studies rely on the subjects’ ability to recall and may be biased as a result of this. By contrast, studies using prescription databases provide detailed information about types of drugs and dates of use. However, these studies lack information about over-the-counter use and are based on the assumption that prescribed medicine is taken as it is prescribed and not only when needed. Moreover, studies based on prescription databases often do not have information on potential confounders.

The strengths of our study include the population-based design, the relatively high number of ovarian cancer cases compared with several other case–control studies, reasonably high participation rates, the ability to control for known risk factors for ovarian cancer, and the ability to examine invasiveness and specific histological types of ovarian cancer. In addition, by eliminating analgesic use one year before diagnosis/interview, we tried to avoid misclassification introduced by analgesic use among cases caused by symptoms of an undiagnosed ovarian cancer. The study also has some limitations that should be considered. An important limitation is the retrospective nature of the study and the possible introduction of recall bias, which is a source of error in all case–control studies. Our information on analgesic use was based on in-person interviews rather than self-completed questionnaires, which may have reduced recall bias. In addition, some degree of selection bias cannot entirely be excluded, because the women who did not participate in the study may have differed from the participants. Moreover, as information about analgesic use was self-reported, some misclassification cannot be excluded. Several features of analgesic use, such as change of brand names, recency of use, switching between different types of analgesic drugs, or using several types of analgesics simultaneously, may influence the accuracy of reporting. We had no information about specific doses of the preparations used, and the effect of frequency of analgesic use could not be examined. Furthermore, our information about reasons for analgesic use reflects the fact that most regular analgesic users had multiple reasons for lifetime regular analgesic use, and thus we were not able to investigate confounding by indications for analgesic use. Finally, in spite of the relatively large numbers of cases and controls, some of the analyses in our study are limited by small numbers.

In conclusion, the results of this case–control study indicate a possible protective effect of analgesic use in relation to development of ovarian cancer. Thus, our results support findings from most previous studies. We found that regular use of aspirin is inversely associated with overall and serous ovarian cancer risk. In contrast, we found no evidence of a decreased risk of ovarian cancer with use of non-aspirin NSAIDs, paracetamol or other analgesics. Moreover, use of analgesics did not decrease the risk of mucinous or other ovarian tumors in our study. To further understand the role of analgesic use in ovarian cancer development, more large epidemiological studies with accurate assessment of dosage, frequency and duration of analgesics are needed. Furthermore, future studies should distinguish between the effect of COX-1 and COX-2 inhibitors and take into consideration the new dualistic division of ovarian cancer into type I and type II. Finally, the potential protective effect of analgesic use on ovarian cancer risk should be balanced against possible adverse effects of analgesics, such as risk of bleeding and peptic ulcers.

Funding

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References

Financial support: The National Cancer Institute (grant RO1 CA 61107) and The Danish Cancer Society.

References

  1. Top of page
  2. Abstract
  3. Key Message
  4. Introduction
  5. Material and methods
  6. Results
  7. Discussion
  8. Funding
  9. References