American Journal of Epidemiology

The Author 2014. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of
Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.

Vol. 180, No. 5

DOI: 10.1093/aje/kwu146
Advance Access publication:
July 9, 2014

Original Contribution
Risks of Endometrial Cancer Associated With Different Hormone Replacement
Therapies in the E3N Cohort, 19922008

Agns Fournier, Laure Dossus, Sylvie Mesrine, Alice Vilier, Marie-Christine Boutron-Ruault,
Franoise Clavel-Chapelon*, and Nathalie Chabbert-Buffet
* Correspondence to Dr. Franoise Clavel-Chapelon, INSERM U108, Team 9, Institut Gustave Roussy, 114 rue Edouard Vaillant,
94805 Villejuif Cedex, France (e-mail: francoise.clavel@gustaveroussy.fr).

We assessed whether different oral progestogens in hormone replacement therapy may differentially affect the
risk of endometrial cancer, using data from the Etude Epidmiologique auprs de femmes de lEducation Nationale
(E3N), a French cohort study (19922008). Hazard ratios and their confidence intervals were derived from Cox
models. Among 65,630 postmenopausal women (mean follow-up: 10.8 years), 301 endometrial cancers occurred.
Compared with never use, ever use of estrogen + micronized progesterone was associated with an increased risk
of endometrial cancer (hazard ratio (HR) = 1.80, 95% confidence interval (CI): 1.38, 2.34) that was significantly
more marked with longer duration of use (for 5 years, HR = 1.39 (95% CI: 0.99, 1.97); for >5 years, HR = 2.66
(95% CI: 1.87, 3.77)). Although use of estrogen + dydrogesterone was not associated overall with endometrial cancer risk (HR = 1.05, 95% CI: 0.76, 1.45), there was a significantly increased risk with long-term use compared with
never use (for >5 years, HR = 1.69, 95% CI: 1.06, 2.70). Users of preparations containing other progesterone derivatives or a norsteroid derivative were not at significantly increased risk (HR = 0.79 (95% CI: 0.60, 1.05) and
HR = 1.30 (95% CI: 0.85, 1.99), respectively). In conclusion, micronized progesterone and, to a lesser extent, dydrogesterone at the doses used in France may not be sufficient to prevent estrogen-induced endometrial cancers.
cohort studies; dydrogesterone; endometrial neoplasms; hormone replacement therapy; progesterone; progestins

Abbreviations: BMI, body mass index; CI, confidence interval; E3N, Etude Epidmiologique auprs de femmes de lEducation
Nationale; EPT, estrogen-progestogen therapy; HR, hazard ratio; HRT, hormone replacement therapy.

To date, the epidemiologic evidence suggests that the

safety of EPTs with regard to endometrial cancer depends
on the number of days per month on which the progestogen
component is used (57). Less is known about the impact of
progestogen type. In 2 observational studies, the risk of endometrial cancer was not signicantly increased (compared
with never users) among users of cyclic combined HRT, regardless of whether it contained medroxyprogesterone acetate, norethisterone acetate, or norgestrel/levonorgestrel, and
the risk was lower among users of continuous combined HRT
than in never users, regardless of whether it contained norethisterone acetate or medroxyprogesterone acetate (8, 9).
In the Etude Epidmiologique auprs de femmes de
lEducation Nationale (E3N) cohort study, we previously
found that the excess risk of breast cancer associated with

Endometrial cancer is one of the factors to be considered

when evaluating the risk-benet ratio associated with hormone replacement therapy (HRT). The estrogen:progesterone
ratio plays a critical role in endometrial carcinogenesis (1), and
endometrial cancer is ranked fourth in terms of incidence and
11th in terms of mortality among malignant disorders of
women in the more developed regions of the world (2).
HRT initially consisted of estrogens. After their use was
shown to be associated with a marked increase in endometrial cancer risk (3, 4), progestogens were added to prevent estrogen-induced endometrial proliferation. Numerous
estrogen-progestogen therapies (EPTs), administered at various doses and frequencies and containing various types of
progestogens, are now available for treating menopausal
symptoms.
508

Am J Epidemiol. 2014;180(5):508517

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

Initially submitted November 27, 2013; accepted for publication May 12, 2014.

HRT and Endometrial Cancer 509

METHODS
The E3N cohort

The E3N, a prospective cohort study, began in 1990 when

half a million women aged 4065 years who were residing in
continental France and were insured by a national health insurance fund that mainly covers teachers and their family
members were invited to participate. In all, 98,995 women
gave written informed consent and completed the rst questionnaire. Thereafter, the women received follow-up questionnaires (available at www.e3n.fr) every 23 years. The
questionnaire mailed in June 2008 is the last one that was
used for this analysis. Response rates were 75% for each
follow-up questionnaire. The French National Commission
for Data Protection and Privacy approved the study protocol.
Definition of cases

The occurrence of endometrial cancer was identied mainly

from self-reports in the questionnaires; a few additional cases
came from next-of-kin reports and the French national causeof-death registry. These analyses included only cases conrmed
by a pathology report (91% of the reported incident cases).
Assessment of HRT use

Information on lifetime use of hormonal treatments was recorded on the 1992 questionnaire. We asked for brand names,
starting dates, and duration of use for each treatment episode.
The information was updated in all subsequent questionnaires. Information about dosage and number of days of use
per month was not requested.
Exposure was classied as: 1) estrogen only (systemic
estradiol or conjugated equine estrogens); 2) EPTs containing
oral micronized progesterone; 3) EPTs containing oral dydrogesterone; 4) EPTs containing another oral progesterone
derivative; 5) EPTs containing an oral norsteroid derivative;
6) weak estrogens (i.e., orally or vaginally administered promestriene or estriol); 7) tibolone; or 8) other (i.e., HRT containing an androgen, or which was intramuscularly administered,
Am J Epidemiol. 2014;180(5):508517

or with the progestogen component delivered through a

patch, or with no specied formulation).
Study population and follow-up

The study population was restricted to postmenopausal

women. Menopausal status and date of menopause were determined from regularly updated data on menstrual periods,
hysterectomy, oophorectomy, HRT use, self-reported menopausal status, and menopausal symptoms, as detailed elsewhere (12).
Follow-up started at the date on which the 1992 questionnaire was returned for postmenopausal women and the date
on which menopause was rst reported for the other women.
Follow-up ended at the date of diagnosis of any cancer, the
date of hysterectomy, the date of the last completed questionnaire, or June 2008, whichever occurred rst.
Of the 98,995 E3N women, we excluded those who had no
follow-up (n = 9,541), were diagnosed with cancer (other than
basal-cell carcinoma) before follow-up started (n = 4,832),
were premenopausal at the end of follow-up (n = 2,763), underwent hysterectomy before follow-up started (n = 13,530),
did not respond to the 1992 questionnaire about lifetime HRT
use while already postmenopausal (n = 2,668), or reported
a diagnosis of endometrial carcinoma for which no pathology report could be obtained (n = 31). Accordingly, 65,630
women remained for analysis.
Statistical analysis

Hazard ratios for endometrial cancer and their 95% condence intervals were estimated using Cox proportional hazards models with age as the time scale.
The known risk factors for endometrial cancer (13) and potential confounders that we included in our models are listed in
Table 1 (in the same categories, except for body mass index
(BMI) and age at menopause, both included as continuous variables). The proportional hazards assumption was not violated
for any of these covariates, except for premenopausal use of
oral contraceptives or progestogens alone. However, the introduction into our models of interaction terms between these variables and the time scale only marginally affected the hazard
ratios associated with HRT exposure (data not shown).
We present only the results from fully adjusted models
and note that the hazard ratios derived from these models
differed quite a bit from those derived from age-adjusted
models (e.g., for ever use of HRT, the hazard ratio was
1.33 (95% condence interval (CI): 1.01, 1.76) in the fully
adjusted model and 1.08 (95% CI: 0.83, 1.41) in the ageadjusted model). However, no strong individual confounders
were identied: The removal of any given covariate from the
fully adjusted model resulted in variations in the hazard ratios
associated with HRT ever use of less than 10%.
BMI (calculated as weight (kg)/height (m)2), smoking,
history of benign disease of the uterus (endometriosis, broids,
or polyps), recent gynecological examination, history of diabetes, and history of high blood pressure, all updated during
follow-up, were included in the models as time-varying variables. Models were further stratied by year of birth in 5-year
categories.

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

EPTs was less elevated when they included micronized progesterone or dydrogesterone rather than other progestogens
(10). On the other hand, analyses from the European Prospective Investigation into Cancer and Nutrition suggested that
estrogen-micronized progesterone combinations might be
more deleterious to the endometrium than other EPTs (11).
Exposure was assessed only once in that study, however
8 years before diagnosis, on averageand only 26 cases of
endometrial cancer were diagnosed among women exposed
to estrogen-micronized progesterone (11). The exposed women
were almost exclusively participants in the studys French
component, namely, E3N.
We present here an extended analysis of data from the E3N
cohort. The longer follow-up period, combined with the availability of regularly updated information on HRT use, more than
tripled the number of endometrial cancer cases with exposure to
estrogen-micronized progesterone and allowed us to separately
evaluate EPTs containing different types of progestogens.

510 Fournier et al.

Table 1. Baseline Characteristics of Participants According to HRT Exposure Status at the End of the Follow-up
Period, E3N Study, France, 19922008
Never Users of HRT
(n = 17,107)

All Women
(n = 65,630a)
No.

No.

Ever Users of HRT

(n = 42,460)
No.

Body mass index

20

8,945

13.6

2,059

12.0

6,153

14.5

20.124.9

41,785

63.7

9,808

57.3

28,436

67.0

2529.9

11,963

18.2

3,935

23.0

6,670

15.7

2,937

4.5

1,305

7.6

1,201

2.8

30
Education
No high school diploma

14.8

3,077

18.0

5,390

12.7

85.2

14,030

82.0

37,070

87.3

<27

15,833

24.1

3,864

22.6

10,417

24.5

2739

16,410

25.0

3,970

23.2

10,987

25.9

4057

17,281

26.3

4,422

25.8

11,322

26.7

>57

16,106

24.5

4,851

28.4

9,734

22.9

Never smoker

34,957

53.3

9,740

56.9

22,034

51.9

Ever smoker

30,673

46.7

7,367

43.1

20,426

48.1

No

63,190

96.3

16,483

96.3

40,840

96.2

Yes

2,440

3.7

624

3.7

1,620

3.8

No

50,615

77.1

13,470

78.7

32,339

76.2

Yes

15,015

22.9

3,637

21.3

10,121

23.8

<13

29,443

44.9

7,750

45.3

18,921

44.6

13

36,187

55.1

9,357

54.7

23,539

55.4

Physical activity,
MET-hours/weekc

Smoking status

Familial history of
endometrial cancer
in first-degree relatives

History of benign disease

of the uterus
(endometriosis,
fibroids, or polyps)

Age at menarche, years

Table continues

Information on HRT exposure was updated on the date

of each questionnaire completion and dealt with prospectively: To characterize exposure for the period preceding
the completion of a given questionnaire, we used only the information reported in the previous questionnaires. The same
approach was used for time-varying covariates. At the time of
each questionnaire, a woman was considered a past user of a
given type of HRT when she reported any use (ever) but no
use in the preceding 3 months. For past users at the time of a
given questionnaire, time since last use increased with time
elapsed since questionnaire completion. For current users,
duration of use increased with time elapsed since questionnaire completion, as we made the assumption that HRT use
did not stop until completion of the subsequent questionnaire.

We performed a sensitivity analysis evaluating whether that

assumption was robust to other hypotheses. It appeared to be
so, since our results were only marginally altered when we assumed instead that HRT use was stopped halfway between the
questionnaires (data not shown). If a woman successively took
different types of HRT, she simultaneously contributed to each
of the relevant categories (e.g., past user of estrogen-only therapy and current user of estrogen-micronized progesterone).
When a covariate had a low percentage of missing values
(5%), imputation to the median (for continuous variables)
or mode (for discrete variables) was used to replace missing
values. Missing values exceeded 5% for duration of oral contraceptives use; accordingly, we created an unknown duration category (Table 1). When a woman did not respond to
Am J Epidemiol. 2014;180(5):508517

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

9,691
55,939

High school diploma,

with or without
postsecondary
education

HRT and Endometrial Cancer 511

Table 1. Continued
Never Users of HRT
(n = 17,107)

All Women
(n = 65,630a)
No.

No.

Ever Users of HRT

(n = 42,460)
No.

Age at menopause, years

<48

9,160

14.0

1,763

10.3

6,368

15.0

4851

31,004

47.2

6,483

37.9

21,411

50.4

52

25,466

38.8

8,861

51.8

14,681

34.6

Parity
7,977

12.2

2,495

14.6

4,767

11.2

12

38,406

58.5

9,300

54.4

25,565

60.2

19,247

29.3

5,312

31.0

12,128

28.6

25,025

38.1

8,456

49.4

14,109

33.2

Use of oral contraceptives

Never use
Ever use
14,392

21.9

3,207

18.8

10,079

23.7

11,714

17.9

2,361

13.8

8,473

20.0

Unknown duration

14,499

22.1

3,083

18.0

9,799

23.1

Never

38,308

58.4

11,861

69.3

22,349

52.6

Ever

27,322

41.6

5,246

30.7

20,111

47.4

No

9,221

14.1

4,788

28.0

3,455

8.1

Yes

56,409

85.9

12,319

72.0

39,005

91.9

64,837

98.8

16,769

98.0

42,092

99.1

793

1.2

338

2.0

368

0.9

No

59,734

91.0

15,244

89.1

39,017

91.9

Yes

5,896

9.0

1,863

10.9

3,443

8.1

Premenopausal use of
progestogen alone

Recent gynecological
examinationd

History of diabetes
No
Yes
History of high blood
pressuree

Abbreviations: E3N, Etude Epidmiologique auprs de femmes de lEducation Nationale; HRT, hormone
replacement therapy; MET, metabolic equivalent of task.
a
During the last follow-up cycle, there were 6,063 women in the unknown HRT exposure category.
b
Weight (kg)/height (m)2.
c
From selected recreational and household activities.
d
Papanicolaou smear performed during the previous follow-up period, used as an indicator of gynecological
monitoring.
e
Includes self-reported use of antihypertensive medication and history of high blood pressure.

a questionnaire, her HRT exposure was classied as unknown

for the period between the questionnaires mailing date and the
date of the next questionnaire that she answered. Therefore,
during that period, she contributed person-years neither to
the HRT never-use category nor to any HRT ever-use category, but rather contributed to the unknown exposure status
category. In a sensitivity analysis in which we replaced missing
exposure data with exposure reported on the previous questionnaire, our conclusions were unaltered (data not shown).
For any given type of HRT, we tested BMI as a potential
effect modier by incorporating a cross-product term of ever
use of that type of HRT (yes/no) BMI (25, yes/no).
Am J Epidemiol. 2014;180(5):508517

Model parameters were estimated and compared by means

of likelihood methods and Wald tests. Tests of statistical signicance were 2-sided, and signicance was set at the 0.05
level. SAS software, version 9.3 (SAS Institute, Inc., Cary,
North Carolina), was used to perform the analyses.
RESULTS

A total of 301 histologically conrmed rst primary endometrial cancers were diagnosed during 709,018 person-years
of follow-up (mean = 10.8 years). Mean age at diagnosis was
64.3 years (standard deviation, 6.5). Extension, according to

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

Duration 10 years
Duration >10 years

512 Fournier et al.

Table 2. Characteristics of HRT Use at the End of the Follow-up

Period, E3N Study, France, 19922008
HRT Characteristic

% of Ever
Usersa

Mean Duration
of Use, years (SD)

Estrogen typeb
Estradiol

92.8

Conjugated equine estrogens

Estriol/promestriene
(weak estrogens)
Tibolone

2.2
11.1
5.8

Data not available

10.0

Progestogen typec
Micronized progesterone

40.2

Dydrogesterone

28.8

Other progesterone
derivatived

59.0

Norsteroid derivativee

13.5

Type of preparationf
Estrogen only

1.6 (2.0)

Estrogen + micronized
progesterone

4.2 (3.5)

Estrogen + dydrogesterone

3.9 (4.5)

Estrogen + other
progesterone derivatived

4.5 (3.5)

Estrogen + norsteroid
derivativee

3.1 (3.0)

Abbreviations: E3N, Etude Epidmiologique auprs de femmes

de lEducation Nationale; HRT, hormone replacement therapy;
SD, standard deviation.
a
Totals may exceed 100% because a woman could have used
several different types of HRT.
b
Among ever users of HRT.
c
Among ever users of estrogen-progestogen therapy, with the
progestogen component administered orally.
d
Chlormadinone acetate, cyproterone acetate, demegestone,
medrogestone, medroxyprogesterone acetate, nomegestrol acetate,
or promegestone.
e
Dienogest, drospirenone, ethynodiol, gestodene, levonorgestrel,
norethisterone acetate, or lynestrenol.
f
Among ever users of the considered type of HRT.

associated with a nonsignicantly increased risk (HR = 1.78,

95% CI: 0.87, 3.65; 8 cases) (data not shown).
Table 4 shows the associations for the dichotomous BMI
strata (<25 or 25). Hazard ratios were generally higher
among women with a BMI less than 25, but differences between the two BMI strata did not reach statistical signicance
(Table 4).
DISCUSSION

Few studies have examined the inuence of progestogen

type on the risk of endometrial cancer in HRT users. Both
randomized trials with endometrial hyperplasia as the outcome (16) and observational studies investigating the risk
of endometrial cancer (8, 9) suggest that there is no difference
among EPTs containing norethisterone acetate, containing
Am J Epidemiol. 2014;180(5):508517

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

the International Federation of Gynecology and Obstetrics

classication (14, 15), was mostly local (7 cancers were
in situ, 237 were stage T1, 23 were stage T2, and 21 were
stage T3; information was not available for 13 cancers).
Table 1 gives the baseline characteristics of the study population. Table 2 shows that HRT consisted mostly of estradiol
associated with micronized progesterone, dydrogesterone, or
other progesterone-derived compounds. Many women used
more than 1 type of HRT: Of those who had ever used estrogen only, 85% had also ever used EPTs; of those who had
ever used estrogen-micronized progesterone, estrogendydrogesterone, estrogen-other progesterone derivative, or
estrogen-norsteroid derivative, 68%, 72%, 63%, and 80%, respectively, had also used another EPT or estrogen only.
Any use of HRT was associated with a signicant increase
in the risk of endometrial cancer (hazard ratio (HR) = 1.33,
95% CI: 1.01, 1.76) compared with never use of HRT. However, associations differed signicantly across ever use of
the different types of HRT shown in Table 3 (P for homogeneity < 0.001), including across the different types of progestogens in EPTs (P for homogeneity < 0.001).
Compared with never users, ever users of estrogen only
and ever users of estrogen-micronized progesterone both had
signicantly increased risks of endometrial cancer (HR =
1.80 (95% CI: 1.31, 2.49) and HR = 1.80 (95% CI: 1.38,
2.34), respectively) that also differed signicantly from
the risks associated with any use (ever) of either estrogendydrogesterone or estrogen-other progesterone derivative
(Table 3). Signicant trends of increasing risks with increasing durations of use were observed for estrogen only and
estrogen-micronized progesterone (Table 3). Hazard ratios
associated with any use of EPTs containing dydrogesterone,
other progesterone derivatives, and norsteroid derivatives
did not differ signicantly from unity and did not differ
signicantly from one another (Table 3). However, there
was a signicantly increased risk with long-term use of
estrogen-dydrogesterone (for >5 years, HR = 1.69, 95% CI:
1.06, 2.70).
Risks did not vary signicantly with increasing time since
last use (Table 3). Although ndings were limited by the
small numbers of cases in some strata, additional stratication by duration of use (5 years or >5 years) and time
since last use (<5 years, 510 years, or >10 years) enabled
us to evaluate the persistence of risks several years after
treatment stopped: Signicant increases in risk were still observed 10 or more years after the last use of estrogen only
(HR = 1.76, 95% CI: 1.01, 3.07; 15 cases among exposed
women) and 510 years after the last use of long-term (>5
years) estrogen-micronized progesterone (HR = 2.99, 95%
CI: 1.18, 7.57; 5 cases) (data not shown).
Among women who had ever used estrogen only, we observed no signicant difference in risks between different
routes of estrogen administration: For oral administration (11
cases diagnosed among exposed women), the hazard ratio
was 1.46 (95% CI: 0.78, 2.72), and for gel or a patch (44
cases), it was 1.77 (95% CI: 1.26, 2.50) (data not shown).
Ever use of weak estrogens (i.e., promestriene or estriol)
was not associated with a signicant increase in risk of endometrial cancer (HR = 1.28, 95% CI: 0.87, 1.89; 31 cases diagnosed among exposed women). Ever use of tibolone was

HRT and Endometrial Cancer 513

Table 3. Hazard Ratios for Endometrial Cancer Associated With Different Types of HRT, E3N Study, France,
19922008

HRT Type and Usage

No. of
Casesa

Person-Years
of Follow-up

Hazard
Ratiob

95%
Confidence
Interval

P for Trendc
With Increasing
Duration

P for Trendc
With Increasing
Time Since Last
Use

Estrogen only
Ever used

51

72,645

1.80

1.31, 2.49

Duration 5 years

39

51,918

1.81

1.27, 2.58

Duration >5 years

4,223

3.53

1.44, 8.66

Current use

8,188

3.30

1.61, 6.76

37

53,927

1.67

1.16, 2.41

Ever used

91

149,806

1.80

1.38, 2.34

Duration 5 years

42

76,303

1.39

0.99, 1.97

Past use

<0.001
0.33

Estrogen + micronized
progesterone

46

59,278

2.66

1.87, 3.77

Current use

54

87,366

1.96

1.41, 2.73

Past use

37

61,210

1.44

0.99, 2.08

Ever used

47

117,216

1.05

0.76, 1.45

Duration 5 years

25

66,212

0.87

0.57, 1.32

Duration >5 years

21

39,207

1.69

1.06, 2.70

0.77

Estrogen + dydrogesterone

Current use

11

55,442

0.67

0.36, 1.25

Past use

36

60,390

1.30

0.90, 1.88

0.006
0.96

Estrogen + other
progesterone
derivative
Ever used

79

257,095

0.79

0.60, 1.05

Duration 5 years

49

132,269

0.81

0.59, 1.13

Duration >5 years

28

97,979

0.94

0.62, 1.44

Current use

24

131,653

0.65

0.41, 1.02

Past use

55

123,185

0.89

0.65, 1.23

0.76
0.54
Table continues

medroxyprogesterone acetate, and containing norgestrel/

levonorgestrel. In our study, EPTs containing oral micronized
progesterone or dydrogesterone were the only types of EPTs
associated with duration-dependent increases in endometrial
cancer risk, with signicantly elevated hazard ratios for treatment durations of more than 5 years.
In the Postmenopausal Estrogen/Progestin Interventions
Trial, researchers investigated the risk of atypical hyperplasia
and endometrial cancer among users of 0.625 mg/day of conjugated equine estrogens associated with 200 mg/day of micronized progesterone for the rst 12 days, but during 3 years
of follow-up, only a single case of endometrial cancer was
observed in the placebo group and 1 case of atypia in the
treatment group (17). Studies evaluating short-term treatments suggest that micronized progesterone might not
prevent estrogen-induced endometrial cell proliferation, particularly at 100200 mg/day (18, 19), but that higher doses
might be more effective for that purpose (1820). This may
be due to considerable interindividual variability in the absorption of oral micronized progesterone (21). In France,
Am J Epidemiol. 2014;180(5):508517

the recommended daily doses of micronized progesterone

in EPTs are 100 mg for continuous schemes and 200 mg
for sequential schemes (22, 23), based on evaluation of the
progestational potencies of the different progestogenic compounds available (24).
After micronized progesterone, dydrogesterone is the
progestogen with the lowest bioavailability and potency
(25). In a record-linkage cohort study with limited statistical
power, Jaakkola et al. (26) found that ever use of estrogendydrogesterone was associated with a hazard ratio for endometrial cancer of 1.33 (95% CI: 0.69, 2.33) when used for
less than 5 years and a hazard ratio of 4.81 (95% CI: 0.58,
17.4) when used longer. Among 5 trials that evaluated hyperplasia in estrogen-dydrogesterone users, only 1 case of hyperplasia was reported (2731), perhaps because of the low
number of women included and the short follow-up periods.
Clinical data show that 10 mg of dydrogesterone sequentially
combined with 2 mg of estradiol for 14 days per month or
5 mg of dydrogesterone administered daily (doses typically
used in France) is effective in protecting the endometrium

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

Duration >5 years

<0.001

514 Fournier et al.

Table 3. Continued

HRT Type and Usage

No. of
Casesa

Person-Years
of Follow-up

Hazard
Ratiob

95%
Confidence
Interval

P for Trendc
With Increasing
Duration

P for Trendc
With Increasing
Time Since Last
Use

Estrogen + norsteroid
derivative
Ever used

24

50,698

1.30

0.85, 1.99

Duration 5 years

18

32,795

1.32

0.81, 2.15

Duration >5 years

12,579

0.98

0.31, 3.07

Current use
Past use

20,403

1.10

0.48, 2.49

17

28,955

1.34

0.81, 2.22

0.51
0.38

from estrogenic effects for up to 3 years (32). Data for longer

durations are not available.
Our results suggest that micronized progesterone and longterm dydrogesterone as used in France are not sufcient to
prevent estrogen-induced endometrial cell proliferation. Alternatively, failure to control for various forms of bias may
explain our results. First, although the evidence derived
from trials assessing endometrial hyperplasia yielded no statistically signicant difference between continuous and sequential EPTs including at least 10 days of progestogen per
month (16), observational data suggest that the risk of endometrial cancer increases when the number of days per month
on which the progestogen is taken decreases (57). We could
not control for that parameter in our cohort, in which most
(approximately 80%) EPTs containing progesterone or its
derivatives used 2 distinct brands of estrogen and of progestogen (e.g., estrogen patch + progestogen pill) rather
than single-pill combinations, because the number of days
of hormone use was not recorded on the self-administered
questionnaires used to assess HRT exposure for the present
study. Although this may have changed over time, in 2008,
micronized progesterone and dydrogesterone were not
taken for fewer days than other progestogens: Data from
the questionnaire sent in 2008, which recorded that information, showed that the 5,493 women who were current users of

EPT (either as a single-pill combination or with distinct

brands of estrogen and of oral progestogen) took the progestogen component for a mean of 22.5 days for micronized
progesterone, 23.5 days for dydrogesterone, 18.9 days for
other progesterone derivatives, and 26.8 days for testosterone
derivatives (unpublished data). Thus, if this observation held
true for the entire follow-up of our cohort, it would be unlikely that differences regarding endometrial cancer risk between estrogen-micronized progesterone/dydrogesterone and
other EPTs are explained by the number of days per month on
which the progestogen is taken. Second, a prescription bias
could have affected our results if women at higher risk of
endometrial cancer were more often prescribed micronized
progesterone/dydrogesterone than other progestogens. However, our analyses were systematically adjusted for known endometrial cancer risk factors, and we observed no marked
difference in their distribution according to the progestogen
component of EPT (data not shown). Finally, a diagnostic
bias due to excess bleeding in women using estrogenmicronized progesterone, which may entail an increased
number of endometrial biopsies, could have affected our results, since irregular bleeding has previously been reported
in these women (33). However, this result was not found in
the 2 trials that assessed unscheduled biopsies in these users
(34, 35).
Am J Epidemiol. 2014;180(5):508517

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

Abbreviations: E3N, Etude Epidmiologique auprs de femmes de lEducation Nationale; HRT, hormone
replacement therapy.
a
Eighty-eight cases were diagnosed among never users of HRT (232,726 person-years), 23 among women with
an unknown exposure status at the time of diagnosis, 25 among women ever exposed to another HRT (i.e., HRT
containing an androgen, or intramuscularly administered, or with the progestogen component delivered through a
patch, or with no specified formulation), 31 among ever users of weak estrogens, and 8 among ever users of
tibolone. The total number of endometrial cancer cases exceeds 301 because a woman could contribute personyears to several categories of exposure simultaneously (e.g., past use of estrogen-only therapy and current use of
estrogen-micronized progesterone).
b
Adjusted for age (years; time scale), all variables listed in Table 1, and the equivalent exposure categories for the
other types of HRT. Reference category: never use.
c
Ever use, duration of use (years; continuous), and recency of use (years since last use (continuous); equaled 0
when use was current) were entered simultaneously into the model.
d
Pairwise P values for heterogeneity of hazard ratios associated with ever use were as follows: between estrogen
only and estrogen-micronized progesterone, 0.99; between estrogen only and estrogen-dydrogesterone, 0.03;
between estrogen only and estrogen-other progesterone derivative, <0.001; between estrogen only and estrogennorsteroid derivative, 0.24; between estrogen-micronized progesterone and estrogen-dydrogesterone, 0.01; between
estrogen-micronized progesterone and estrogen-other progesterone derivative, <0.001; between estrogen-micronized
progesterone and estrogen-norsteroid derivative, 0.20; between estrogen-dydrogesterone and estrogen-other
progesterone derivative, 0.19; between estrogen-dydrogesterone and estrogen-norsteroid derivative, 0.44; and
between estrogen-other progesterone derivative and estrogen-norsteroid derivative, 0.07.

HRT and Endometrial Cancer 515

Table 4. Hazard Ratios for Endometrial Cancer Associated With Ever Use of Different Types of HRT, by Body Mass
Indexa, E3N Study, France, 19922008
HRT Type and Body Mass Indexb

No. of
Cases

Person-Years
of Follow-up

Hazard
Ratioc

95% Confidence
Interval

Estrogen only

P for
Homogeneity

0.06

<25

44

59,857

2.16

1.51, 3.10

25

12,788

0.94

0.42, 2.07

Estrogen + micronized
progesterone

0.15

<25

72

122,666

1.99

1.46, 2.71

25

19

27,140

1.39

0.82, 2.37

<25

38

96,877

1.13

0.78, 1.62

25

20,339

0.85

0.42, 1.71

Estrogen + dydrogesterone

0.39

Estrogen + other progesterone

derivative

0.43
63

213,800

0.82

0.59, 1.13

25

16

43,295

0.73

0.41, 1.28

Estrogen + norsteroid derivative

0.53

<25

20

42,760

1.37

0.86, 2.20

25

7,938

0.95

0.34, 2.64

Abbreviations: E3N, Etude Epidmiologique auprs de femmes de lEducation Nationale; HRT, hormone
replacement therapy.
a
Closest record to menopause onset.
b
Weight (kg)/height (m)2.
c
Adjusted for age (years; time scale), all variables listed in Table 1, and ever use of the other types of HRT.
Reference category: never use, among women in each body mass index category.

The epidemiologic evidence shows that estrogen-only

therapy is associated with an increased risk of endometrial
cancer, which is more marked with prolonged duration of
use and persists for several years after treatment discontinuation (36). It also suggests that the increase in risk is conned
to nonobese women (8, 3740). Our ndings agree with these
conclusions.
Most epidemiologic studies suggest an increased risk
of endometrial cancer associated with tibolone (8, 11, 41).
In our study, we observed a nonsignicantly increased risk
in tibolone users, but the number of exposed cases was too
limited to permit a rm interpretation.
Consistent with the reassuring data available about endometrial hyperplasia with the use of vaginal estrogens (42), we
found no signicantly increased risk of endometrial cancer
with the use of weak estrogens (mostly vaginally administered promestriene or estriol in our cohort).
We found no signicant variations in the risk of endometrial cancer according to time since last use of EPT, and the
risk remained signicantly elevated many years after the use
of estrogen-micronized progesterone. Although relatively
scarce, previous epidemiologic data similarly suggest that
the effect (whether deleterious, neutral, or protective) of
EPT on the risk of endometrial cancer persists for several
years after treatment ends (8, 38, 4345).
Strengths of our study include our ability to evaluate different types of HRT and the regular exposure updates. These
Am J Epidemiol. 2014;180(5):508517

updates limited classication bias and allowed us to take into

account possible changes over time in the types of hormones
used, by adjusting our models simultaneously for the different HRT types each woman could have ever used. Hence, it is
unlikely that our results for a given HRT type were instead
due to the previous use of other types of HRT. Finally, exposure was dealt with prospectively: The exposure preceding
the completion of a given questionnaire was characterized
using only the information reported in the previous questionnaires. In particular, this implies for cases that the information on exposure used in our models was only that collected
before any diagnosis of cancer (e.g., for a woman with cancer
diagnosed between the fth and sixth follow-up questionnaires, only the exposure information collected up to the
fth follow-up questionnaire was used), thus eliminating
the possibility that differential recall between cases and noncases may have affected our results.
Some limitations of our study must also be acknowledged.
Exposure assessment was based on self-reports. However,
recall bias was probably limited, since data on HRT use
were updated every 23 years. Furthermore, previous studies
have shown good agreement between self-reported HRT exposure and prescription data, especially for recent use (46,
47). However, the risk of endometrial cancer associated
with estrogen-only therapy might have been underestimated
if some women who were using a combination of progestogen and estrogen as 2 distinct brands rather than a single-pill

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

<25

516 Fournier et al.

ACKNOWLEDGMENTS

Author afliations: Nutrition, Hormones and Womens

Health Team, U1018, Centre for Research in Epidemiology
and Population Health, Institut National de la Sant et de la
Recherche Mdicale (INSERM), Villejuif, France (Agns
Fournier, Laure Dossus, Sylvie Mesrine, Alice Vilier, MarieChristine Boutron-Ruault, Franoise Clavel-Chapelon);
Unit Mixte de Recherche Scientique 1018, Universit
Paris-Sud, Villejuif, France (Agns Fournier, Laure Dossus,
Sylvie Mesrine, Alice Vilier, Marie-Christine BoutronRuault, Franoise Clavel-Chapelon); Institut Gustave
Roussy, Villejuif, France (Agns Fournier, Laure Dossus,
Sylvie Mesrine, Alice Vilier, Marie-Christine BoutronRuault, Franoise Clavel-Chapelon); and Department of
Obstetrics-Gynaecology and Reproductive Medicine, Hpital Tenon APHP, Universit Pierre et Marie Curie, Paris,
France (Nathalie Chabbert-Buffet).
This work was supported by a grant from the Institut de
Recherche en Sant Publique (as part of the Plan Cancer
20092013). The E3N cohort is being studied with the
nancial support of the Mutuelle Gnrale de lEducation
Nationale, the European Community, the Ligue Nationale
Contre le Cancer, the Institut Gustave Roussy, INSERM,
and the Fondation de France.
We thank the medical practitioners for providing pathology
reports. We are grateful to R. Chait, M. Fangon, L. Hoang,
and M. Niravong for technical assistance; to Dr. Alban Fabre,
who performed the preliminary statistical analyses; to the

E3N Study Group; and to JoAnn Cahn for assistance with

translation.
Conict of interest: none declared.

REFERENCES
1. Hecht JL, Mutter GL. Molecular and pathologic aspects of
endometrial carcinogenesis. J Clin Oncol. 2006;24(29):
47834791.
2. Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden
of cancer in 2008: GLOBOCAN 2008. Int J Cancer. 2010;
127(12):28932917.
3. Ziel HK, Finkle WD. Increased risk of endometrial carcinoma
among users of conjugated estrogens. N Engl J Med. 1975;
293(23):11671170.
4. Smith DC, Prentice R, Thompson DJ, et al. Association of
exogenous estrogen and endometrial carcinoma. N Engl J Med.
1975;293(23):11641167.
5. Cogliano V, Grosse Y, Baan R, et al. Carcinogenicity of
combined oestrogen-progestagen contraceptives and
menopausal treatment. Lancet Oncol. 2005;6(8):552553.
6. Santen RJ, Allred DC, Ardoin SP, et al. Postmenopausal
hormone therapy: an Endocrine Society scientic statement.
J Clin Endocrinol Metab. 2010;95(7 suppl 1):s1s66.
7. Brinton LA, Felix AS. Menopausal hormone therapy and risk of
endometrial cancer. J Steroid Biochem Mol Biol. 2014;142:
8389.
8. Beral V, Bull D, Reeves G. Endometrial cancer and
hormone-replacement therapy in the Million Women Study.
Lancet. 2005;365(9470):15431551.
9. Jaakkola S, Lyytinen HK, Dyba T, et al. Endometrial cancer
associated with various forms of postmenopausal hormone
therapy: a case control study. Int J Cancer. 2011;128(7):
16441651.
10. Fournier A, Berrino F, Clavel-Chapelon F. Unequal risks for
breast cancer associated with different hormone replacement
therapies: results from the E3N cohort study. Breast Cancer Res
Treat. 2008;107(1):103111.
11. Allen NE, Tsilidis KK, Key TJ, et al. Menopausal hormone
therapy and risk of endometrial carcinoma among
postmenopausal women in the European Prospective
Investigation into Cancer and Nutrition. Am J Epidemiol. 2010;
172(12):13941403.
12. Fournier A, Fabre A, Mesrine S, et al. Use of different
postmenopausal hormone therapies and risk of histology- and
hormone receptor-dened invasive breast cancer. J Clin Oncol.
2008;26(8):12601268.
13. Amant F, Moerman P, Neven P, et al. Endometrial cancer.
Lancet. 2005;366(9484):491505.
14. Petterson F, ed. Annual Report on the Results of Treatment
in Gynaecologic Cancer. Vol. 20. Stockholm, Sweden:
International Federation of Gynecology and Obstetrics; 1988.
15. Pecorelli S. Revised FIGO staging for carcinoma of the vulva,
cervix, and endometrium. Int J Gynaecol Obstet. 2009;105(2):
103104.
16. Furness S, Roberts H, Marjoribanks J, et al. Hormone therapy in
postmenopausal women and risk of endometrial hyperplasia.
Cochrane Database Syst Rev. 2012;8:CD000402.
17. The Writing Group for the PEPI Trial. Effects of hormone
replacement therapy on endometrial histology in
postmenopausal women. The Postmenopausal Estrogen/
Progestin Interventions (PEPI) Trial. JAMA. 1996;275(5):
370375.
Am J Epidemiol. 2014;180(5):508517

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

combination (approximately 80% of EPT use) neglected to

report the use of the progestogen component; they would
therefore have been misclassied as estrogen-only users. We
also acknowledge that there were relatively few cases in some
subgroups, due to the limited incidence of the disease
in our cohort, 4 cases among 1,000 women over a 10-year
period in HRT never users, far fewer than for invasive
breast cancer (around 34 cases). Finally, the small percentage of overweight or obese women limited our ability to
assess whether the effects of HRT varied across BMI categories, as consistently suggested by previous epidemiologic
studies (reviewed by Brinton and Felix (7)). Our results
should therefore be viewed as pertaining mostly to women
who are neither obese nor overweight, in whom the HRTassociated hazard ratio for endometrial cancer appears to be
highest (7).
In conclusion, our study suggests that estrogen-micronized
progesterone and long-term use (5 years) of estrogendydrogesterone at the doses/frequencies employed in France
are not sufcient to prevent estrogen-induced endometrial
cancers. Such an insufcient progestational potency would
also explain the lack of difference regarding breast cancer
risk we observed in the same cohort between users of estrogen only and users of estrogen-micronized progesterone or
estrogen-dydrogesterone (10). Although endometrial cancer
is far less frequent than breast cancer, further studies are required to dene an optimal HRT regimen that minimizes the
risks of both breast and endometrial cancer.

HRT and Endometrial Cancer 517

Am J Epidemiol. 2014;180(5):508517

33. Di Carlo C, Sammartino A, Di Spiezio Sardo A, et al. Bleeding

patterns during continuous estradiol with different sequential
progestogens therapy. Menopause. 2005;12(5):520525.
34. Lindenfeld EA, Langer RD. Bleeding patterns of the hormone
replacement therapies in the Postmenopausal Estrogen and
Progestin Interventions trial. Obstet Gynecol. 2002;100(5):
853863.
35. Prestwood KM, Kenny AM, Kleppinger A, et al. Ultralow-dose
micronized 17-estradiol and bone density and bone
metabolism in older women: a randomized controlled trial.
JAMA. 2003;290(8):10421048.
36. Grady D, Gebretsadik T, Kerlikowske K, et al. Hormone
replacement therapy and endometrial cancer risk: a
meta-analysis. Obstet Gynecol. 1995;85(2):304313.
37. Brinton LA, Hoover RN. Estrogen replacement therapy and
endometrial cancer risk: unresolved issues. The Endometrial
Cancer Collaborative Group. Obstet Gynecol. 1993;81(2):
265271.
38. Newcomb PA, Trentham-Dietz A. Patterns of postmenopausal
progestin use with estrogen in relation to endometrial cancer
(United States). Cancer Causes Control. 2003;14(2):195201.
39. Lacey JV Jr, Brinton LA, Lubin JH, et al. Endometrial
carcinoma risks among menopausal estrogen plus progestin and
unopposed estrogen users in a cohort of postmenopausal
women. Cancer Epidemiol Biomarkers Prev. 2005;14(7):
17241731.
40. Karageorgi S, Hankinson SE, Kraft P, et al. Reproductive
factors and postmenopausal hormone use in relation to
endometrial cancer risk in the Nurses Health Study cohort
19762004. Int J Cancer. 2010;126(1):208216.
41. Formoso G, Perrone E, Maltoni S, et al. Short and long term
effects of tibolone in postmenopausal women. Cochrane
Database Syst Rev. 2012;2:CD008536.
42. Rees M, Prez-Lpez FR, Ceasu I, et al. EMAS clinical guide:
low-dose vaginal estrogens for postmenopausal vaginal
atrophy. Maturitas. 2012;73(2):171174.
43. Weiderpass E, Adami HO, Baron JA, et al. Risk of endometrial
cancer following estrogen replacement with and without
progestins. J Natl Cancer Inst. 1999;91(13):11311137.
44. Heiss G, Wallace R, Anderson GL, et al. Health risks and
benets 3 years after stopping randomized treatment with
estrogen and progestin. JAMA. 2008;299(9):10361045.
45. Phipps AI, Doherty JA, Voigt LF, et al. Long-term use of
continuous-combined estrogen-progestin hormone therapy and
risk of endometrial cancer. Cancer Causes Control. 2011;
22(12):16391646.
46. Banks E, Beral V, Cameron R, et al. Agreement between
general practice prescription data and self-reported use of
hormone replacement therapy and treatment for various
illnesses. J Epidemiol Biostat. 2001;6(4):357363.
47. Sandini L, Pentti K, Tuppurainen M, et al. Agreement of
self-reported estrogen use with prescription data: an
analysis of women from the Kuopio Osteoporosis Risk
Factor and Prevention Study. Menopause. 2008;15(2):
282289.

Downloaded from http://aje.oxfordjournals.org/ by guest on December 22, 2015

18. Lane G, Siddle NC, Ryder TA, et al. Dose dependent effects of
oral progesterone on the oestrogenised postmenopausal
endometrium. Br Med J (Clin Res Ed). 1983;287(6401):
12411245.
19. Moyer DL, Felix JC, Kurman RJ, et al. Micronized
progesterone regulation of the endometrial glandular cycling
pool. Int J Gynecol Pathol. 2001;20(4):374379.
20. Kim S, Korhonen M, Wilborn W, et al. Antiproliferative effects
of low-dose micronized progesterone. Fertil Steril. 1996;65(2):
323331.
21. McAuley JW, Kroboth FJ, Kroboth PD. Oral administration of
micronized progesterone: a review and more experience.
Pharmacotherapy. 1996;16(3):453457.
22. Gillet JY, Andre G, Faguer B, et al. Induction of amenorrhea
during hormone replacement therapy: optimal micronized
progesterone dose. A multicenter study. Maturitas. 1994;19(2):
103115.
23. de Lignires B. Endometrial hyperplasia. Risks, recognition
and the search for a safe hormone replacement regimen.
J Reprod Med. 1999;44(2 suppl):191196.
24. King RJ, Whitehead MI. Assessment of the potency of orally
administered progestins in women. Fertil Steril. 1986;46(6):
10621066.
25. Stanczyk FZ, Hapgood JP, Winer S, et al. Progestogens used in
postmenopausal hormone therapy: differences in their
pharmacological properties, intracellular actions, and clinical
effects. Endocr Rev. 2013;34(2):171208.
26. Jaakkola S, Lyytinen H, Pukkala E, et al. Endometrial cancer in
postmenopausal women using estradiol-progestin therapy.
Obstet Gynecol. 2009;114(6):11971204.
27. Delzanno G, Paoletti R, Gaudiano L, et al. Dihydrogesterone
vs. medroxyprogesterone acetate in association with
transdermal estradiol in postmenopausal substitution treatment
[in Italian]. Minerva Ginecol. 1994;46(4):179182.
28. Gelfand MM, Fugere P, Bissonnette F, et al. Conjugated
estrogens combined with sequential dydrogesterone or
medroxyprogesterone acetate in postmenopausal women:
effects on lipoproteins, glucose tolerance, endometrial
histology, and bleeding. Menopause. 1997;4(1):1018.
29. van de Weijer PH, Scholten PC, van der Mooren MJ, et al.
Bleeding patterns and endometrial histology during
administration of low-dose estradiol sequentially combined
with dydrogesterone. Climacteric. 1999;2(2):101109.
30. Ferenczy A, Gelfand MM, van de Weijer PH, et al. Endometrial
safety and bleeding patterns during a 2-year study of 1 or 2 mg
17 beta-estradiol combined with sequential 520 mg
dydrogesterone. Climacteric. 2002;5(1):2635.
31. Chang TC, Chen M, Lien YR, et al. Comparison of the
difference in histopathology and cell cycle kinetics among the
postmenopausal endometrium treated with different progestins
in sequential-combined hormone replacement therapy.
Menopause. 2003;10(2):172178.
32. Schindler AE. Progestational effects of dydrogesterone in vitro,
in vivo and on the human endometrium. Maturitas. 2009;
65(suppl 1):S3S11.