Endometrial cancer associated with various forms of postmenopausal hormone therapy: A case control study

Authors


Abstract

This study evaluates the effect of different modes of estradiol-progestagen therapy (EPT) regimens on the postmenopausal endometrial cancer risk in Finland. Women diagnosed with endometrial cancer in 1995–2007 at the age of 50–80 years were identified from the Finnish Cancer Registry (N = 7,261). For each case, three age-matched controls were retrieved from the Finnish Population Register. The use of EPT since 1994 was ascertained from the national Medical Reimbursement Register. Odds ratios (ORs) for different EPT regimens were calculated with conditional logistic regression analysis, adjusted for parity and ages at the deliveries. For use of <5 years, the OR for sequential EPT was 0.67 (95% confidence interval 0.52–0.86), for continuous EPT 0.45 (0.27–0.73), and for estradiol plus levonorgestrel-releasing intrauterine device system (LNG-IUS) 0.39 (0.17–0.88). A decreased risk persisted for the use of continuous EPT and estradiol plus LNG-IUS of up to 10 years. The use of long-cycle EPT showed a tendency toward an elevated risk both for exposure of <5 years (1.40; 0.82–2.38) and for estimated use of >5 years (1.63; 1.12–2.38). For an estimated exposure of >10 years, the risk for endometrial cancer was elevated for both users of long-cycle EPT (2.95; 2.40–3.62) and sequential EPT (1.38; 1.15–1.66). Norethisterone acetate and medroxyprogesterone acetate as parts of EPT did not differ in their endometrial cancer risk. The use of tibolone showed no endometrial risk. The use of sequential and long-cycle EPT is associated with an increased risk of endometrial cancer, whereas the use of continuous EPT or estradiol plus LNG-IUS shows a decreased risk.

The incidence of postmenopausal endometrial cancer is increasing in all Nordic countries except Denmark,1 evidently because of lowered parity and increasing body weight. One of the determinants, the use and content of postmenopausal estradiol-progestagen therapy (EPT), varies remarkably between populations. Therefore, the impact of EPT on the risk for endometrial cancer should be studied separately in each country.2

It is established that different modes of postmenopausal EPT regimens have opposite influences on the risk for endometrial hyperplasia and cancer.3, 4 Continuous EPT has been reported to be protective against cancer in many European countries5–7 and in the United States.8, 9 Sequential EPT (monthly progestagen course) and especially long-cycle EPT use (progestagen course every three months) are associated with increased risk for endometrial cancer.6, 9–12 However, no data exist on the effect of various progestagens with different progestin potency in long-term exposure on endometrial safety.13 Tibolone—a synthetic steroid used as EPT—has shown controversial results from increased risk6 to protective effects up to three years of use.14, 15 The levonorgestrel-releasing intrauterine system (LNG-IUS)—as a part of postmenopausal EPT—may be the most logical route of administration to protect the endometrium,16, 17 because it acts locally and causes only a minimal rise in the progestin level in the blood.

Our research team previously made a nationwide study on the risk of endometrial cancer in a cohort of EPT users whose cancer risk was compared with the background population of all Finnish postmenopausal women.18 It was evident from this study that continuous EPT was protective for endometrial cancer, whereas sequential and long-cycle EPT use associated with an increased risk after 5 years of use; the oral and transdermal routes showed similar risks. In that study setting, there was no access to data on tibolone and LNG-IUS users or to any important confounding factors known to be significant for endometrial cancer. Therefore, we designed the present case-control study to further assess the endometrial cancer risks associated with different modes of EPT, including tibolone and LNG-IUS in Finland.

Abbreviations:

CI: confidence interval; EPT: estradiol-progestagen therapy; HT: postmenopausal hormone therapy; LNG-IUS: levonorgestrel-releasing intrauterine system; MPA: medroxyprogesterone acetate; NETA: norethisterone acetate; OR: odds ratio

Material and Methods

Subjects

All Finnish women from 50 to 80 years diagnosed with endometrial cancer between January 1, 1995 and December 31, 2007 (N = 7,261) were identified from the Finnish Cancer Registry (Table 1). For each case, three control women born at the same time (+/−one month) as the case, alive and without endometrial cancer at the time of endometrial cancer diagnosis of the case were randomly selected from the Finnish National Population Register, which also provided data on dates of birth of their children born on the October 1, 1953 or later (n = 21,783). The Hospital Inpatient Register of the National Institute of Health and Welfare provided data on hysterectomies in Finland between 1986–2003 with almost 100% coverage.19 With the aid of this registry, we excluded all control women who had had a hysterectomy before the diagnosis of endometrial cancer in the respective case at that time (n = 2,293). Thus, the final number of control women was 19,490 (Table 1). Nulliparity was more common in women with endometrial cancer than in the controls. The proportion of women with three or more children was smaller in the cases than in the controls (28.1% vs. 33.8%) (Table 1).

Table 1. Distribution of selected characteristics of the all endometrial cancer cases and controls
inline image

Hormone therapy exposure

The data about EPT use by cases and controls were obtained from the National Medical Reimbursement Registry, which contains data on purchases of systemic EPT since 1994 (Fig. 1). In Finland, all systemic hormonal therapies (HTs) are only available with a doctor's prescription and are partly reimbursed directly in the pharmacy by the national health-care system.

Figure 1.

Lexis diagram of the study population and hormone therapy (HT) exposure time.

The only estrogen used in systemic HTs in Finland is estradiol. According to our present national guidelines, only hysterectomized women are allowed to use estradiol-only therapy, and women with an intact uterus should use sequential or continuous progestagen as a supplement to estradiol. Apart from this, we encountered 307 endometrial cancer cases (4.2%) and 1,291 controls (6.6%) who had used estrogen-only therapy. A larger proportion of the controls using such a monotherapy derive evidently from the fact that they were older women who had undergone a hysterectomy before 1986 (when the hysterectomy registry was opened). Therefore, we were not able to make analyses on the cancer risk related to estrogen-only therapy.

The fixed commercial EPT regimens are presented in Table 2.20 The daily doses for oral estradiol are 1 or 2 mg and for transdermal estradiol between 25 and 100 μg from a patch and 0.5–1 mg from gel. The sequential progestagen course included norethisterone acetate (NETA) 5–10 mg/d, medroxyprogesterone acetate (MPA) 5–10 mg/d, dydrogesterone 10–20 mg/d, lynestrenol 5 mg/d, progesterone 200–300 mg/d or megestrole acetate 10 mg/d. LNG-IUS releases 20 μg per 24 hr and the dose of tibolone is 2.5 mg/d.

Table 2. Type and dose of progestagens and estradiol in fixed commercial products analyzed in this study1
inline image

In this study, the EPT regimens are classified as sequential, long-cycle or continuous. Women using EPT orally or transdermally were lumped together and analyzed as one group. A sequential EPT regimen is defined as progestagen for 10–14 days once a month (“monthly sequential”) or every three months (“long-cycle”). Some women formed individual long-cycle regimens by adding 10–14 day progestin courses at 2–3 months intervals to daily continuous estradiol. In continuous EPT regimens, progestagen is administrated daily, orally, transdermally or with LNG-IUS. All routes of systemic administration were accepted.

The mode of regimen used for >50% of the registered total exposure time since 1994 determined the exposure category for a given woman. HT use of <6 months was classified as no use. Progestagen-only therapy has been used by some perimenopausal women without hot flushes, evidently to control menstrual cycle disturbances. Uses of tibolone or LNG-IUS, the latter with or without estradiol, were also considered as separate categories in the analyses. Women who could not be classified into former categories were included into the mixed therapy category. Endometrial cancer cases had used HT more often than the controls (Table 1). The use of EPT, mixed and progestagen-only therapies, was more common in the cases than the controls, whereas the LNG-IUS use, with or without estradiol, was more common among the controls. Tibolone was used equally by both cases and controls (Table 1).

For LNG-IUS users, the exposure was estimated to last for 5 years from the purchase date. Only LNG-IUS use among women beyond 50 years of age was regarded to be truly postmenopausal. For EPT or tibolone, the exposure was calculated to have started from the date of first purchase.

Statistical methods

Information on hormone exposure is complete only for the years after 1994 (Fig. 1). Earlier use has been estimated to be initiated at the age of 52. Women who were HT users only before 1994 have been classified as nonusers. Mode of regimen for estimated use (before 1994) was classified according to registered use since 1994 onwards.

To estimate the effect of EPT exposure on the risk for endometrial cancer, we divided the case-control sets into three categories. First, the risk for EPT exposure of less than 5 years was calculated from those women who turned 50 years old during 1994–2007 and whose entire EPT history was known to us (Fig. 1). For the older birth cohorts, the duration of EPT use was estimated on the basis of age at the last known purchase during 1994–2007. If a woman bought her last EPT regimen between the ages of 57–61, it can be estimated that her exposure had lasted 5–10 years, assuming that the EPT therapy was started at the age of 52. If a woman bought her last EPT regimen at the age of 62 or higher, her exposure was estimated to be at least 10 years (Fig. 1).

A multivariate conditional logistic regression model was used to estimate, by means of the odds ratio (OR), the relative risk for endometrial cancer associated with each category of HT use. The ages at first and last delivery, and parity, were considered as possible confounders. Testing for possible interactions between the variables was based on the likelihood ratio statistics. Estimates of the model parameters and 95% confidence intervals (CIs) were computed by the maximum likelihood technique. The statistical analyses were performed with STATA software, release 9.

Results

Uses of sequential EPT, continuous EPT and continuous estradiol plus LNG-IUS for less than 5 years were associated with a significantly decreased risk for endometrial cancer (Table 3). When the estimated exposure time exceeded 5 years but remained below 10 years, continuous EPT and estradiol plus LNG-IUS were associated with decreased risk, whereas long-cycle EPT and mixed therapy were associated with a significantly increased risk for endometrial cancer. Among EPT users with an estimated use of 10 years or more, the risk for endometrial cancer rose to three-fold among long-cycle EPT users (2.95; 2.40–3.62) and mixed therapy users (3.28; 2.92–3.68). The use of sequential EPT for >10 years carried a significantly increased risk for endometrial cancer (1.38; 1.15–1.66). The risks for tibolone use for any exposure time were close to unity, but the confidence intervals were wide (Table 3).

Table 3. Odds ratio (OR) of endometrial cancer for postmenopausal use of estradiol-progestagen therapy, by estimated duration1 and mode of regimen
inline image

The risk of endometrial cancer related to different progestagens was studied among women with a completely known history of postmenopausal EPT use, i.e., women born 1944–1957 (Fig. 1). The use of sequential MPA, NETA and other progestagens for less than 5 years did not differ from each other in the risk for endometrial cancer (Table 4). Among users of sequential EPT, there was no difference related to progestagen in the regimen, but a tendency toward an increased risk for endometrial cancer after 5 years of use was seen for all progestagens (Table 4). Continuous EPT regimens showed a similar protective association against endometrial cancer with NETA (0.45; 0.28–0.73), MPA (0.49; 0.20–1.21) and LNG-IUS (0.34; 0.16–0.72). LNG-IUS without estradiol also showed a decreased risk for endometrial cancer (0.27; 0.13–0.56).

Table 4. Relative risk of endometrial cancer among postmenopausal women aged 50 to 62 using sequential1 estradiol-progestagen therapy by type of progestagen and by duration
inline image

In all models, increased parity and a higher age at the last delivery were associated with a decreased risk of endometrial cancer, while age at the first delivery had only a minor effect. In a model for any duration of use of all HT types for >6 months, the OR for women with 1 or 2 children was 0.83 (0.77–0.89), with 3 or 4 children 0.73 (0.67–0.79) and with 5 children or more 0.58 (0.50–0.66), as compared with nulliparous women. Among women with 1–2 children, the OR for those with the last delivery at the age of ≥35 years was 0.70 (0.62–0.80) and for those with last delivery at the age of <35 years 0.85 (0.79–0.92). Among women with ≥3 children, the ORs were 0.62 (0.55–0.69) and 0.73 (0.67–0.80), respectively.

Discussion

This large nationwide case-control study on postmenopausal Finnish women showed that sequential EPT, continuous EPT or estradiol plus LNG-IUS for less than 5 years inversely associated with diagnosis of endometrial cancer. This association was seen for up to 10 of use years with continuous EPT and estradiol plus LNG-IUS users. The use of long-cycle EPT was associated with an increased risk of endometrial cancer in relation to the duration of the use. An elevated risk of endometrial cancer was also seen for sequential EPT when used longer than 10 years. Tibolone use did not associate with endometrial cancer risk, but the prevalence of exposure is still low and the risk estimates are based on few cases.

The strengths of our study are the large number of cases, no losses to follow-up, the exact follow-up since the first EPT purchase (13 years) and detailed information about tibolone and different EPT regimens purchased during 1994–2007. Since only a small proportion of the EPT price is reimbursed and women have to use their own money to buy the drugs, it is likely that they truly used the EPT they had bought. We likely captured nearly all EPT users in Finland, because reimbursement is collected at the time of the purchase in the pharmacy i.e., it does not require any extra effort.

Limitation to our exposure data is EPT purchases before 1994. We had to assume that older women who were still using EPT in 1994 had started its use at the age of 52; the average menopausal age in Finland is 51 years.21 Women who were HT users only before 1994 were classified as nonusers because of the lack of information. This may dilute our risk estimation for endometrial cancer. The types of EPT regimens used among these women were also approximated from EPT used in 1994–2007. The changes of type of EPT regimen typically happens in the beginning of the use. Therefore, the classification based on the later part of use is justified as we did for the cases with partially estimated EPT use.

We were able to control for parity and ages at deliveries, which are associated with endometrial cancer risk.1 Nulliparity is an established risk factor for endometrial cancers,22 and in line with this, nulliparous women also showed the highest risk in the present study. An advanced age at the last delivery may be protective for endometrial cancer,23 and in our study, the endometrial cancer risk was smallest when the woman's age at the last delivery was more than 35 and she had three children or more. We did not have data on potential confounders such as the duration of reproductive period, oral contraceptive use or body weight.24 Based on the characterization of Finnish hormone users,25 there is no difference in socioeconomic status in HT users and nonusers in Finland i.e., using of contraceptives and bodyweight are similar. Therefore, we have no reason to believe that these factors would strongly modify our conclusions.

We were able to exclude the majority of women who had had a hysterectomy from the set of control women. Failure to do so would have diluted our results remarkably, because up to one-fifth of Finnish women are hysterectomized between 50–64 years age.26 We missed hysterectomies done before 1986 and after 2003. Therefore, some of the controls may not have been at risk of getting endometrial cancer. Fortunately, the estimated number of missed hysterectomies is so low that it does not markedly affect our results.

Women using EPT visit regularly their physician, and chance of endometrial assessment and diagnosis of endometrial cancer is higher than among women without such visits. This is supported by the findings indicating that the stage of endometrial cancer tends to be less advanced among sequential EPT users than in controls.18, 27 We like to emphasize that some precancerous and perhaps even cancerous changes could have been present before the initiation of EPT or tibolone, because endometrial biopsies are recommended only for women with bleeding disturbances. Therefore, it is possible in clinical practice that symptoms of early cancer may have been controlled with cyclic progestagen only, LNG-IUS or EPT courses. However, it is likely that these symptoms persisted and the diagnosis of endometrial cancer was made within the first six months, and such a short use was not counted as EPT use in our study. This reduces the chance of diagnosis bias.

The highest relative risk for endometrial cancer was seen among long-cycle EPT users already after 5 years of use; this is in line with previous data.11, 12, 18 In the present study, the use of sequential EPT for <5 years showed a decreased risk for endometrial cancer, and the use of 5–10 years showed a nonsignificant excess of 11%. There was a significant 38% excess risk after 10 years of use. This is in line with some9, 18, 28, 29 but not with all reports.30–32 Our previous cohort study18 showed a higher risk increase (69%) among sequential EPT users for at least 5 years, but in that study, a very large proportion of cases were diagnosed at higher ages, and they had to have been using EPT for a very long time if they were still using it in 1994 and were hence included in the cohort, i.e., in practice they had been using EPT for much longer than the 5 years.

The Women's Health Initiative Study in the United States,8 a British study6 and other previous studies33 have rather uniformly indicated the protectiveness of continuous EPT against endometrial cancer, but the longest follow-up time so far has been 6 years of use. Our study shows endometrial protection for at least 10 years of use or longer.

Affinity of different progestagens to their own and other steroid receptors vary.34 We could compare the cancer risks related to the most commonly used progestagens, NETA and MPA, in sequential and continuous use. We could not demonstrate any difference between these progestagen in this regard; this novel finding is important for clinicians. Our data on endometrial protection in LNG-UIS confirms many previous suggestions and further show that such an EPT is protective for 10 years. Prevalence of LNG-IUS with estradiol exposure of more than 10 years in the present study is low, and finding that protection is not afforded beyond 10 years is based on few cases. In Finland, the indication and the reimbursement for being a part of postmenopausal EPT is since 1997.

To our knowledge, our study is the first to follow estradiol plus LNG-IUS up to 10 years and over. LNG-IUS was first launched in Finland 1990, then in the United Kingdom 1995 and in the United States in 2000.35 Earlier studies on estradiol plus LNG-IUS have reported high endometrial safety.16, 35–37 In the present study, we encountered only 16 women with estradiol plus LNG-IUS users and endometrial cancer. However, when they were compared to control women with estradiol plus LNG-IUS, the significant protective effect was present for 10 years. It seems that the LNG-IUS has the potential to prevent endometrial cancer.

Information about the endometrial safety of tibolone has been controversial. A British study6 reported an almost two-fold increased risk of endometrial cancer for tibolone users (average use of 3.4 years). Randomized, double-blind trials comparing tibolone (2.5 or 1.25 mg) with continuous EPT for 2–3 years among women with a mean age of 54–59 years14, 15 concluded that tibolone does not induce endometrial hyperplasia or carcinoma. A statistically modest increase was reported in endometrial cancer among elderly (mean age 68 years) osteoporotic women after an average of 3 years of treatment of 1.25 mg of tibolone, and the number of hyperplasia in endometrial polyps was two-fold compared with the placebo group.38 Power of our study was low to evaluate the possible influence of tibolone to endometrial cancer risk, because we had only 19 women with endometrial cancer who had used tibolone for more than 50% of their whole exposure time, and the risk for endometrial cancer was close to unity irrespective of the duration of use.

To sum up, this large case-control study shows that sequential EPT does not significantly increase the endometrial cancer risk until the use exceeds 10 years, but even rather short-term use of long-cycle EPT increases the risk. Both continuous combined EPT and systemic estradiol combined to LNG-IUS are associated with a decreased risk for endometrial cancer in long-term use. The effects of HT regimens on the endometrium should be balanced with information about positive life quality effects, cardiovascular diseases and breast cancer related to the use of hormonal therapy.

Ancillary