- Split View
-
Views
-
Cite
Cite
Claudia M. Greiser, Eberhard M. Greiser, Martina Dören, Menopausal hormone therapy and risk of breast cancer: a meta-analysis of epidemiological studies and randomized controlled trials, Human Reproduction Update, Volume 11, Issue 6, November/December 2005, Pages 561–573, https://doi.org/10.1093/humupd/dmi031
- Share Icon Share
Abstract
We conducted meta-analyses to assess the impact of menopausal hormone therapy (MHT) on the risk of incident invasive breast cancer (BC) in cohort studies (CS), case–control studies (CCS) and randomized controlled trials (RCTs) published 1989–2004. We used published data providing information upon unopposed estrogen therapy (ET), estrogen–progestin therapy (EPT) or all MHT combined. Major outcomes were MHT-associated overall risk of BC and change of risk per year used. There is a linear increase of overall risk by midterm year of case ascertainment based upon data of all study types for MHT and to a larger extent for EPT, not for ET. Effects are larger in CS than in CCS. Meta-analyses stratified by <1992 versus $1992 as midterm year of case ascertainment indicate larger summary risks for the latter period for all MHT analysed, in particular for EPT. Annual increases in BC risk for EPT across study types are 0–9%, for ET 0–3%. In conclusion, there is evidence that relative risks for BC risks by MHT, in particular EPT, have been increasing in recent years. Given the widespread use of MHT, and often long duration, more detailed knowledge about differential BC risks of both estrogens and progestins are necessary to minimize BC risk in symptomatic women who consider MHT.
Submitted on February 3, 2005; revised on June 14, 2005; accepted on July 18, 2005
Introduction
The relation between risk of breast cancer (BC) and use of menopausal hormone therapy (MHT) has been the subject of many epidemiological studies and several meta-analyses (Armstrong, 1988; Dupont and Page, 1991; Steinberg et al., 1991; Grady et al., 1992; Sillero-Arenas et al., 1992; Colditz et al., 1993; Steinberg et al., 1994; Collaborative Group on Hormonal Factors in Breast Cancer, 1997; Hemminki and McPherson, 1997; Humphrey, 2002; National Heart, Lung, and Blood Institute, 2002; U.S. Preventive Services Task Force, 2005). Most information derives from a reanalysis of individual data from 51 epidemiological studies. These results indicate that risk of BC is elevated in women using MHT, risk increases with duration of use, and decreases after cessation of therapy (Collaborative Group on Hormonal Factors in Breast Cancer, 1997). However, results largely depend on studies with unopposed conjugated equine estrogens (CEE), and most studies were conducted in the United States. Combined estrogen–progestin therapy (EPT), a treatment modality used to reduce the excess risk of endometrial cancer associated with unopposed estrogen use, may further increase the risk of BC as suggested by some studies (Colditz et al., 1995; La Vecchia et al., 1995; Magnusson et al., 1999; Ross et al., 2000; Schairer et al., 2000; Newcomb et al., 2000). Two previous meta-analyses suggested differences regarding the magnitude of risk of BC attributable to MHT, when studies from the US and few European studies were compared (Steinberg et al., 1991; Colditz et al., 1993). We conducted a systematic search of the literature and performed a meta-analysis of cohort studies (CS), case–control studies (CCS) and randomized controlled trials (RCTs) to assess (i) the association between specified groups of hormone regimens and overall risk of BC and (ii) the magnitude of time-dependent risk as major prespecified outcomes.
Materials and methods
Identification of studies
We conducted a topic-specific search using Medline and the Cochrane Controlled Trials Register (http://www.cochranelibrary.com) for the time period 1989 until August 2004; publications retrieved for the year 1989 were restricted to those not already included in a previous meta-analysis (Steinberg et al., 1991). The search was restricted to studies conducted in the United States, Canada and European countries. We used the keywords ‘hormone replacement therapy’, ‘hormone therapy’, ‘(o)estrogen therapy’, ‘(o)estradiol therapy’, ‘(o)estrogen and progest* therapy’, ‘HRT’, ‘ERT’, ‘breast cancer’, ‘case control study’, ‘cohort study’ and ‘randomized/randomised/controlled clinical trial’ in several combinations. We used recent systematic reviews to potentially identify further studies (Humphrey, 2002; National Heart, Lung, and Blood Institute, 2002; Nelson et al., 2002) and reference lists of pertinent studies, topic-specific reviews (mostly in English, additionally in German), editorials, supplements, conference proceedings and abstract books. In studies with multiple publications from the same population, we included only data from the most recent publication. In the case of double publication, we included only the data sets of the first publication.
We included CS, CCS and RCTs, if information upon unopposed estrogen therapy (ET), or EPT or all hormone therapy combined (MHT: including unspecified/unknown preparations) was provided. We acknowledged a large variation of reporting types of MHT; earlier studies are more likely to predominantly include ET, more recent studies to progressively include EPT. Eligible studies had to include ever/never use of MHT, risk by duration of use or increase of risk within a given time interval. Studies were eligible if confidence intervals (CIs) or standard errors of risk estimates and dates on conduct of the study were provided.
Meta-analyses
Major outcomes were (i) the association between specified groups of hormone regimens and life-time risk of BC and (ii) the magnitude of time-dependent risk. Data were abstracted and statistical analyses performed independently (C. Greiser, E. Greiser) using two different approaches. To summarize effects of MHT on BC risk irrespective of duration or dosage, point estimates and CI were used in a fixed-effects model applying the general variance-based method (Petitti, 2000). In this method, the variance for each risk parameter is derived from published CIs (for the formula used see Appendix). To estimate summary slopes to determine increase of risk per year of use, slopes for both individual studies and summary slopes were calculated using inverse variance-weighted least squares estimates as suggested (Berlin et al., 1993). In studies providing both risk estimates for various time periods and annual change estimates, the latter were used assuming that they provide a more precise estimate than values derived from tables in publications. We included two studies, which provided only slope estimates (Colditz and Rosner, 1998; Ross et al., 2000), in the summary slope estimations. We used midperiod time points to determine duration values for time-periods. When durations were reported as ‘greater than’ a specific number of months or years, we added 20% to that duration. We examined heterogeneity across studies by applying the general variance-based method (Petitti, 2000) and providing for Cochrane’s Q for individual substrata and for various totals of substrata. Cochrane’s Q is a measure of heterogeneity and can be tested, using the chi-square distribution. P values less than 0.05 indicate heterogeneity. Analyses are stratified by type of study (CS, RCT and CCS), type of hormone therapy (ET, EPT and MHT), and midterm of years of case ascertainment (before 1992 versus 1992 and later). We chose the latter stratification level to improve analysis of time trends of MHT. We calculated weighted linear regressions with midterm date of case ascertainment as independent variable and the natural logarithm of risk parameters (odds ratio, relative risk and hazard ratio) as well as untransformed risk parameters as dependent variables. For these meta-regressions the inverse of variances were used as weights. We used SAS version 6.12 (SAS Institute) for all analyses and STATA-8 (Stata Corporation) for bubble plots.
Results
We retrieved 94 publications, 44 CCS, 37 CS, 12 RCTs and one cross-sectional study. Included studies are listed in Table I (15 CS and 6 RCTs) and Table II (21 CCS) and excluded studies are listed in Table III. Briefly, included CS and RCTs varied widely regarding number of BC cases, women-years, age, longest reported duration of MHT use, type of MHT and time of follow-up. Of 21 studies, 12 were conducted in North America. The same variation was found in CCS; 14 of 21 were conducted in the US or Canada. The majority of studies had to be excluded for a wide spectrum of reasons, apart from not meeting our inclusion criteria.
First author (year), region . | Number of cases . | Women (women-years) . | Age/source . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Mills (1989), USA | 215 | 20 341 (115 619) | PMP† Seventh-day-Adventist women | ≥10 | MHT | 1976 to December 1982 (6 years) | ||
Risch (1994), Canada | 742 | 3279 (448 716) | 43–49 years in 1976 Saskatchewan Health Prescription Drug Plan | ≥3 | MHT, ET | 1976–1990 | ||
Colditz (1995), USA | 1935 | 69 586 (725 550) | 30–55 years in 1976, PMP†, Nurses’ Health Study | ≥10 | ET, EPT | 1978–1992 | ||
Folsom (1995), USA | 468 | 41 070 (129 149) | 55–69 years in 1985, Iowa Women’s Health Study | MHT | 1985–1991 | |||
Schuurman (1995), The Netherlands | 471 | 62 573 (not given) | 55–69 years, Netherlands Cohort Study (NLCS) | ≥15 | MHT | September 1986 to December 1989 (3.3 years) | ||
Colditz (1998), USA | 2035 | Not given (980 000) | PMP†, Nurses’ Health Study | 10 | ET, EPT | 16 years | ||
Lando (1999), USA | 219 | 5761 (73 253) | PMP†, NHANES I Epidemiologic Follow-up Study | ≥10 | MHT | 1971–1992 (22 years) | ||
Persson (1999), Sweden | 198 | 10 472 (60 298) | Born after 1918, original cohort, Sweden 1977–1980 Uppsala Health Care Region | ≥6 | ET, EPT | 1987–1993 | ||
Schairer (2000), USA | 2082 | 46 355 (473 687) | PMP†, Breast Cancer Detection Demonstration Project | ≥16 | ET, EPT | ET, EPT | 1980–1995 (16 years) | |
Manjer (2001), Sweden | 141 | 5755 (57 199) | PMP†, Department of Preventive Medicine at Malmö University Hospital | MHT | June 1983 to April 1997 (average 9.8 years) | |||
Viscoli (2001), RCT, USA | 10 | 664 (not given) | 71 years‡ from 21 hospitals in USA for stroke trial | ET | 2.8 years‡ | |||
ESPRIT Team (2002), RCT, UK | 8 | 1017 (not given) | 50–69 years, 62.5 years‡ | 2 | ET | July 1996 to February 2000 (trial duration 2 years) | ||
Hulley (2002), RCT, USA | 88 | 2763 (17 308§) | 67 years‡, PMP†, HERS and HERS II | EPT | EPT | 1993–2000 (6.8 years) | ||
de Lignières (2002), France | 105 | 3175 (28 367) | PMP†, cohort women, attending from January 1975 to December 1987, Department of Endocrinology and Reproductive Medicine, Necker Hospital, Paris | ≥10 | MHT, EPT | January 1975 to December 1995 | ||
Waters (2002), RCT, USA, Canada | 4 | 423 (3160) | 65 years‡, WAVE Trial, seven clinical sites | MHT | July 1997 to January 2002 (trial duration 2 years‡) | |||
Writing Group for the Women’s Health Initiative Investigators (2002), RCT, USA | 290 | 16 608 (85 364) | 50–79 years, PMP†, 40 clinical centres | EPT | EPT | 5.2 years‡ | ||
Jernström (2003), Sweden | 101 | 6586 | 50–64 years, WHILA Study | MHT | December 1995 to December 2001 | |||
Million Women Study Collaborators (2003), UK | 9364 | 828 923 | 50–64 years | MHT, ET, EPT | ET, EPT | May 1996 to March 2001 | ||
Tjönneland (2004), Denmark | 423 | 23 618 | 50–64 years, PMP†, Diet, Cancer and Health Study, Danish cancer society | MHT | December 1993 to December 2000 | |||
Stahlberg (2004), Denmark | 244 | 10 874 (68 912) | >44 years | MHT | EPT¶ | 1993 to April 2001 | ||
PMP† | ET | |||||||
Danish Nurse Cohort | ||||||||
Women’s Health Initiative Steering Committee (2004), RCT, USA | 218 | 10 739 (71 522) | 50–79 years, PMP†, 40 clinical centres | ET | ET | 1993 to November 2003 (average follow up of 6.8 years‡) | ||
Total | 17 138 | 1 129 512 |
First author (year), region . | Number of cases . | Women (women-years) . | Age/source . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Mills (1989), USA | 215 | 20 341 (115 619) | PMP† Seventh-day-Adventist women | ≥10 | MHT | 1976 to December 1982 (6 years) | ||
Risch (1994), Canada | 742 | 3279 (448 716) | 43–49 years in 1976 Saskatchewan Health Prescription Drug Plan | ≥3 | MHT, ET | 1976–1990 | ||
Colditz (1995), USA | 1935 | 69 586 (725 550) | 30–55 years in 1976, PMP†, Nurses’ Health Study | ≥10 | ET, EPT | 1978–1992 | ||
Folsom (1995), USA | 468 | 41 070 (129 149) | 55–69 years in 1985, Iowa Women’s Health Study | MHT | 1985–1991 | |||
Schuurman (1995), The Netherlands | 471 | 62 573 (not given) | 55–69 years, Netherlands Cohort Study (NLCS) | ≥15 | MHT | September 1986 to December 1989 (3.3 years) | ||
Colditz (1998), USA | 2035 | Not given (980 000) | PMP†, Nurses’ Health Study | 10 | ET, EPT | 16 years | ||
Lando (1999), USA | 219 | 5761 (73 253) | PMP†, NHANES I Epidemiologic Follow-up Study | ≥10 | MHT | 1971–1992 (22 years) | ||
Persson (1999), Sweden | 198 | 10 472 (60 298) | Born after 1918, original cohort, Sweden 1977–1980 Uppsala Health Care Region | ≥6 | ET, EPT | 1987–1993 | ||
Schairer (2000), USA | 2082 | 46 355 (473 687) | PMP†, Breast Cancer Detection Demonstration Project | ≥16 | ET, EPT | ET, EPT | 1980–1995 (16 years) | |
Manjer (2001), Sweden | 141 | 5755 (57 199) | PMP†, Department of Preventive Medicine at Malmö University Hospital | MHT | June 1983 to April 1997 (average 9.8 years) | |||
Viscoli (2001), RCT, USA | 10 | 664 (not given) | 71 years‡ from 21 hospitals in USA for stroke trial | ET | 2.8 years‡ | |||
ESPRIT Team (2002), RCT, UK | 8 | 1017 (not given) | 50–69 years, 62.5 years‡ | 2 | ET | July 1996 to February 2000 (trial duration 2 years) | ||
Hulley (2002), RCT, USA | 88 | 2763 (17 308§) | 67 years‡, PMP†, HERS and HERS II | EPT | EPT | 1993–2000 (6.8 years) | ||
de Lignières (2002), France | 105 | 3175 (28 367) | PMP†, cohort women, attending from January 1975 to December 1987, Department of Endocrinology and Reproductive Medicine, Necker Hospital, Paris | ≥10 | MHT, EPT | January 1975 to December 1995 | ||
Waters (2002), RCT, USA, Canada | 4 | 423 (3160) | 65 years‡, WAVE Trial, seven clinical sites | MHT | July 1997 to January 2002 (trial duration 2 years‡) | |||
Writing Group for the Women’s Health Initiative Investigators (2002), RCT, USA | 290 | 16 608 (85 364) | 50–79 years, PMP†, 40 clinical centres | EPT | EPT | 5.2 years‡ | ||
Jernström (2003), Sweden | 101 | 6586 | 50–64 years, WHILA Study | MHT | December 1995 to December 2001 | |||
Million Women Study Collaborators (2003), UK | 9364 | 828 923 | 50–64 years | MHT, ET, EPT | ET, EPT | May 1996 to March 2001 | ||
Tjönneland (2004), Denmark | 423 | 23 618 | 50–64 years, PMP†, Diet, Cancer and Health Study, Danish cancer society | MHT | December 1993 to December 2000 | |||
Stahlberg (2004), Denmark | 244 | 10 874 (68 912) | >44 years | MHT | EPT¶ | 1993 to April 2001 | ||
PMP† | ET | |||||||
Danish Nurse Cohort | ||||||||
Women’s Health Initiative Steering Committee (2004), RCT, USA | 218 | 10 739 (71 522) | 50–79 years, PMP†, 40 clinical centres | ET | ET | 1993 to November 2003 (average follow up of 6.8 years‡) | ||
Total | 17 138 | 1 129 512 |
HERS, Heart Estrogen/Progestin Replacement Study; NHANES, National Health and Nutrition Examination Survey; RCT, randomized controlled trial; PMP, postmenopausal; WHILA Study, Women’s Health in the Lund Area Study; WAVE, Women’s Angiographic Vitamins and Estrogen.
Menopausal hormone therapy (MHT) [unspecified/unknown hormone therapy or estrogen therapy (ET) plus estrogen–progestin therapy (EPT) combined], as utilized for meta-analysis.
Only postmenopausal women included.
Mean.
Women-years derived from Grady et al. (2002), as Hulley et al. (2002) did not provide women-years.
Estrogen plus testosterone-like cyclical progestin.
First author (year), region . | Number of cases . | Women (women-years) . | Age/source . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Mills (1989), USA | 215 | 20 341 (115 619) | PMP† Seventh-day-Adventist women | ≥10 | MHT | 1976 to December 1982 (6 years) | ||
Risch (1994), Canada | 742 | 3279 (448 716) | 43–49 years in 1976 Saskatchewan Health Prescription Drug Plan | ≥3 | MHT, ET | 1976–1990 | ||
Colditz (1995), USA | 1935 | 69 586 (725 550) | 30–55 years in 1976, PMP†, Nurses’ Health Study | ≥10 | ET, EPT | 1978–1992 | ||
Folsom (1995), USA | 468 | 41 070 (129 149) | 55–69 years in 1985, Iowa Women’s Health Study | MHT | 1985–1991 | |||
Schuurman (1995), The Netherlands | 471 | 62 573 (not given) | 55–69 years, Netherlands Cohort Study (NLCS) | ≥15 | MHT | September 1986 to December 1989 (3.3 years) | ||
Colditz (1998), USA | 2035 | Not given (980 000) | PMP†, Nurses’ Health Study | 10 | ET, EPT | 16 years | ||
Lando (1999), USA | 219 | 5761 (73 253) | PMP†, NHANES I Epidemiologic Follow-up Study | ≥10 | MHT | 1971–1992 (22 years) | ||
Persson (1999), Sweden | 198 | 10 472 (60 298) | Born after 1918, original cohort, Sweden 1977–1980 Uppsala Health Care Region | ≥6 | ET, EPT | 1987–1993 | ||
Schairer (2000), USA | 2082 | 46 355 (473 687) | PMP†, Breast Cancer Detection Demonstration Project | ≥16 | ET, EPT | ET, EPT | 1980–1995 (16 years) | |
Manjer (2001), Sweden | 141 | 5755 (57 199) | PMP†, Department of Preventive Medicine at Malmö University Hospital | MHT | June 1983 to April 1997 (average 9.8 years) | |||
Viscoli (2001), RCT, USA | 10 | 664 (not given) | 71 years‡ from 21 hospitals in USA for stroke trial | ET | 2.8 years‡ | |||
ESPRIT Team (2002), RCT, UK | 8 | 1017 (not given) | 50–69 years, 62.5 years‡ | 2 | ET | July 1996 to February 2000 (trial duration 2 years) | ||
Hulley (2002), RCT, USA | 88 | 2763 (17 308§) | 67 years‡, PMP†, HERS and HERS II | EPT | EPT | 1993–2000 (6.8 years) | ||
de Lignières (2002), France | 105 | 3175 (28 367) | PMP†, cohort women, attending from January 1975 to December 1987, Department of Endocrinology and Reproductive Medicine, Necker Hospital, Paris | ≥10 | MHT, EPT | January 1975 to December 1995 | ||
Waters (2002), RCT, USA, Canada | 4 | 423 (3160) | 65 years‡, WAVE Trial, seven clinical sites | MHT | July 1997 to January 2002 (trial duration 2 years‡) | |||
Writing Group for the Women’s Health Initiative Investigators (2002), RCT, USA | 290 | 16 608 (85 364) | 50–79 years, PMP†, 40 clinical centres | EPT | EPT | 5.2 years‡ | ||
Jernström (2003), Sweden | 101 | 6586 | 50–64 years, WHILA Study | MHT | December 1995 to December 2001 | |||
Million Women Study Collaborators (2003), UK | 9364 | 828 923 | 50–64 years | MHT, ET, EPT | ET, EPT | May 1996 to March 2001 | ||
Tjönneland (2004), Denmark | 423 | 23 618 | 50–64 years, PMP†, Diet, Cancer and Health Study, Danish cancer society | MHT | December 1993 to December 2000 | |||
Stahlberg (2004), Denmark | 244 | 10 874 (68 912) | >44 years | MHT | EPT¶ | 1993 to April 2001 | ||
PMP† | ET | |||||||
Danish Nurse Cohort | ||||||||
Women’s Health Initiative Steering Committee (2004), RCT, USA | 218 | 10 739 (71 522) | 50–79 years, PMP†, 40 clinical centres | ET | ET | 1993 to November 2003 (average follow up of 6.8 years‡) | ||
Total | 17 138 | 1 129 512 |
First author (year), region . | Number of cases . | Women (women-years) . | Age/source . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Mills (1989), USA | 215 | 20 341 (115 619) | PMP† Seventh-day-Adventist women | ≥10 | MHT | 1976 to December 1982 (6 years) | ||
Risch (1994), Canada | 742 | 3279 (448 716) | 43–49 years in 1976 Saskatchewan Health Prescription Drug Plan | ≥3 | MHT, ET | 1976–1990 | ||
Colditz (1995), USA | 1935 | 69 586 (725 550) | 30–55 years in 1976, PMP†, Nurses’ Health Study | ≥10 | ET, EPT | 1978–1992 | ||
Folsom (1995), USA | 468 | 41 070 (129 149) | 55–69 years in 1985, Iowa Women’s Health Study | MHT | 1985–1991 | |||
Schuurman (1995), The Netherlands | 471 | 62 573 (not given) | 55–69 years, Netherlands Cohort Study (NLCS) | ≥15 | MHT | September 1986 to December 1989 (3.3 years) | ||
Colditz (1998), USA | 2035 | Not given (980 000) | PMP†, Nurses’ Health Study | 10 | ET, EPT | 16 years | ||
Lando (1999), USA | 219 | 5761 (73 253) | PMP†, NHANES I Epidemiologic Follow-up Study | ≥10 | MHT | 1971–1992 (22 years) | ||
Persson (1999), Sweden | 198 | 10 472 (60 298) | Born after 1918, original cohort, Sweden 1977–1980 Uppsala Health Care Region | ≥6 | ET, EPT | 1987–1993 | ||
Schairer (2000), USA | 2082 | 46 355 (473 687) | PMP†, Breast Cancer Detection Demonstration Project | ≥16 | ET, EPT | ET, EPT | 1980–1995 (16 years) | |
Manjer (2001), Sweden | 141 | 5755 (57 199) | PMP†, Department of Preventive Medicine at Malmö University Hospital | MHT | June 1983 to April 1997 (average 9.8 years) | |||
Viscoli (2001), RCT, USA | 10 | 664 (not given) | 71 years‡ from 21 hospitals in USA for stroke trial | ET | 2.8 years‡ | |||
ESPRIT Team (2002), RCT, UK | 8 | 1017 (not given) | 50–69 years, 62.5 years‡ | 2 | ET | July 1996 to February 2000 (trial duration 2 years) | ||
Hulley (2002), RCT, USA | 88 | 2763 (17 308§) | 67 years‡, PMP†, HERS and HERS II | EPT | EPT | 1993–2000 (6.8 years) | ||
de Lignières (2002), France | 105 | 3175 (28 367) | PMP†, cohort women, attending from January 1975 to December 1987, Department of Endocrinology and Reproductive Medicine, Necker Hospital, Paris | ≥10 | MHT, EPT | January 1975 to December 1995 | ||
Waters (2002), RCT, USA, Canada | 4 | 423 (3160) | 65 years‡, WAVE Trial, seven clinical sites | MHT | July 1997 to January 2002 (trial duration 2 years‡) | |||
Writing Group for the Women’s Health Initiative Investigators (2002), RCT, USA | 290 | 16 608 (85 364) | 50–79 years, PMP†, 40 clinical centres | EPT | EPT | 5.2 years‡ | ||
Jernström (2003), Sweden | 101 | 6586 | 50–64 years, WHILA Study | MHT | December 1995 to December 2001 | |||
Million Women Study Collaborators (2003), UK | 9364 | 828 923 | 50–64 years | MHT, ET, EPT | ET, EPT | May 1996 to March 2001 | ||
Tjönneland (2004), Denmark | 423 | 23 618 | 50–64 years, PMP†, Diet, Cancer and Health Study, Danish cancer society | MHT | December 1993 to December 2000 | |||
Stahlberg (2004), Denmark | 244 | 10 874 (68 912) | >44 years | MHT | EPT¶ | 1993 to April 2001 | ||
PMP† | ET | |||||||
Danish Nurse Cohort | ||||||||
Women’s Health Initiative Steering Committee (2004), RCT, USA | 218 | 10 739 (71 522) | 50–79 years, PMP†, 40 clinical centres | ET | ET | 1993 to November 2003 (average follow up of 6.8 years‡) | ||
Total | 17 138 | 1 129 512 |
HERS, Heart Estrogen/Progestin Replacement Study; NHANES, National Health and Nutrition Examination Survey; RCT, randomized controlled trial; PMP, postmenopausal; WHILA Study, Women’s Health in the Lund Area Study; WAVE, Women’s Angiographic Vitamins and Estrogen.
Menopausal hormone therapy (MHT) [unspecified/unknown hormone therapy or estrogen therapy (ET) plus estrogen–progestin therapy (EPT) combined], as utilized for meta-analysis.
Only postmenopausal women included.
Mean.
Women-years derived from Grady et al. (2002), as Hulley et al. (2002) did not provide women-years.
Estrogen plus testosterone-like cyclical progestin.
Author (year), region . | Number of cases . | Number /type of controls . | Age (years), Type of cancer . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Kaufmann (1991) USA and Canada | 1686, H | 2077, H | 40–69 PMP† | ≥15 | ET, EPT | ET | January 1980 to September 1988 | |
Palmer (1991), Canada | 607, P | 1214, P | <70 | ≥15 | ET, EPT | ET | 1982–1986 | |
Harris (1992), USA | 412, H | 336, H | <30 to ≥80 PMP† | ≥5 | MHT | January 1987 to December 1989 | ||
La Vecchia (1992), Italy | 3037, H | 2569, H | ≥3 | MHT | 1983–1990 | |||
Yang (1992), Canada | 699, P | 685, P | <75 PMP† | ≥10 | ET, EPT | ET | June 1988 to June 1989 | |
Weinstein (1993), USA | 837, P | 860, P | <79 PMP† | ≥20 | MHT | January 1984 to December 1986 | ||
La Vecchia (1995), Italy | 2569, H | 2588, H | 20–74 | ≥5 | MHT, EPT | June 1991 to February 1994 | ||
Lipworth (1995), Greece | 820, H | 795, 753, H | Mean: 56.4 (cases) and 54.4 (controls) PMP† | ≥3 | MHT | January 1989 to December 1991 | ||
Newcomb (1995), USA | 3130, P | 3698, P | <75 PMP† | ≥15 | MHT, ET, EPT | ET, EPT | April 1989 to December 1991 | |
Stanford (1995), USA | 537, P | 492, P | 50–64, 16.2% in situ | ≥20 | ET, EPT | ET, EPT | January 1988 to June 1990 | |
Levi (1996), Switzerland | 230, H | 507, H | 24–75 | ≥10 | MHT | January 1990 to July 1995 | ||
Persson (1997), Sweden | 435, P | 1770, P | 40–74, 13% in situ | ≥11 | MHT, ET, EPT | ET, EPT | February 1990 to June 1995 | |
Henrich (1998), USA | 109, P | 545, P | 45 + PMP†, 29.4% in situ | ≥11 | MHT, ET, EPT | July 1987 to March 1992 | ||
Magnusson (1999), Sweden | 2563, P | 2845, P | 50–74, PMP† | >10 | MHT, ET, EPT | ET, EPT | October 1993 to March 1995 | |
Moorman (2000), USA | 397, P | 425, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | 1993–1996 | ||
Ross (2000), USA | 1897, P | 1637, P | 55–72, PMP† | ≥15 | ET, EPT | March 1987 to December 1989, January 1992 to December 1992, January 1995 to April 1996 | ||
Kirsh (2002), Canada | 404, P | 403, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | ET, EPT | April 1995 to March 1996 | |
Newcomb (2002), USA | 5298, P | 5571, P | 50–79 PMP† | ≥5 | MHT, ET, EPT | ET, EPT | January 1992 to December 1994 | |
Weiss (2002), USA | 1870, P | 1953, P | 55 to <65, PMP† | ET, EPT | July 1994 to April 1998 | |||
Wirfält (2002), Sweden | 237, P | 673, P | >49 | MHT | May 1995 to December 1992 | |||
Li (2003), USA | 975, P | 1007, P | 65–79 | ET, EPT | ET, EPT | April 1997 to May 1999 | ||
Total | 28 749 | 33, 403 |
Author (year), region . | Number of cases . | Number /type of controls . | Age (years), Type of cancer . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Kaufmann (1991) USA and Canada | 1686, H | 2077, H | 40–69 PMP† | ≥15 | ET, EPT | ET | January 1980 to September 1988 | |
Palmer (1991), Canada | 607, P | 1214, P | <70 | ≥15 | ET, EPT | ET | 1982–1986 | |
Harris (1992), USA | 412, H | 336, H | <30 to ≥80 PMP† | ≥5 | MHT | January 1987 to December 1989 | ||
La Vecchia (1992), Italy | 3037, H | 2569, H | ≥3 | MHT | 1983–1990 | |||
Yang (1992), Canada | 699, P | 685, P | <75 PMP† | ≥10 | ET, EPT | ET | June 1988 to June 1989 | |
Weinstein (1993), USA | 837, P | 860, P | <79 PMP† | ≥20 | MHT | January 1984 to December 1986 | ||
La Vecchia (1995), Italy | 2569, H | 2588, H | 20–74 | ≥5 | MHT, EPT | June 1991 to February 1994 | ||
Lipworth (1995), Greece | 820, H | 795, 753, H | Mean: 56.4 (cases) and 54.4 (controls) PMP† | ≥3 | MHT | January 1989 to December 1991 | ||
Newcomb (1995), USA | 3130, P | 3698, P | <75 PMP† | ≥15 | MHT, ET, EPT | ET, EPT | April 1989 to December 1991 | |
Stanford (1995), USA | 537, P | 492, P | 50–64, 16.2% in situ | ≥20 | ET, EPT | ET, EPT | January 1988 to June 1990 | |
Levi (1996), Switzerland | 230, H | 507, H | 24–75 | ≥10 | MHT | January 1990 to July 1995 | ||
Persson (1997), Sweden | 435, P | 1770, P | 40–74, 13% in situ | ≥11 | MHT, ET, EPT | ET, EPT | February 1990 to June 1995 | |
Henrich (1998), USA | 109, P | 545, P | 45 + PMP†, 29.4% in situ | ≥11 | MHT, ET, EPT | July 1987 to March 1992 | ||
Magnusson (1999), Sweden | 2563, P | 2845, P | 50–74, PMP† | >10 | MHT, ET, EPT | ET, EPT | October 1993 to March 1995 | |
Moorman (2000), USA | 397, P | 425, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | 1993–1996 | ||
Ross (2000), USA | 1897, P | 1637, P | 55–72, PMP† | ≥15 | ET, EPT | March 1987 to December 1989, January 1992 to December 1992, January 1995 to April 1996 | ||
Kirsh (2002), Canada | 404, P | 403, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | ET, EPT | April 1995 to March 1996 | |
Newcomb (2002), USA | 5298, P | 5571, P | 50–79 PMP† | ≥5 | MHT, ET, EPT | ET, EPT | January 1992 to December 1994 | |
Weiss (2002), USA | 1870, P | 1953, P | 55 to <65, PMP† | ET, EPT | July 1994 to April 1998 | |||
Wirfält (2002), Sweden | 237, P | 673, P | >49 | MHT | May 1995 to December 1992 | |||
Li (2003), USA | 975, P | 1007, P | 65–79 | ET, EPT | ET, EPT | April 1997 to May 1999 | ||
Total | 28 749 | 33, 403 |
H, hospital-based cases resp. controls; P, population-based cases resp. controls; PMP, postmenopausal.
Menopausal hormone therapy (MHT) [unspecified/unknown hormone therapy or estrogen therapy (ET) plus estrogen–progestin therapy (EPT) combined], as utilized for meta-analysis.
Only postmenopausal women included.
Author (year), region . | Number of cases . | Number /type of controls . | Age (years), Type of cancer . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Kaufmann (1991) USA and Canada | 1686, H | 2077, H | 40–69 PMP† | ≥15 | ET, EPT | ET | January 1980 to September 1988 | |
Palmer (1991), Canada | 607, P | 1214, P | <70 | ≥15 | ET, EPT | ET | 1982–1986 | |
Harris (1992), USA | 412, H | 336, H | <30 to ≥80 PMP† | ≥5 | MHT | January 1987 to December 1989 | ||
La Vecchia (1992), Italy | 3037, H | 2569, H | ≥3 | MHT | 1983–1990 | |||
Yang (1992), Canada | 699, P | 685, P | <75 PMP† | ≥10 | ET, EPT | ET | June 1988 to June 1989 | |
Weinstein (1993), USA | 837, P | 860, P | <79 PMP† | ≥20 | MHT | January 1984 to December 1986 | ||
La Vecchia (1995), Italy | 2569, H | 2588, H | 20–74 | ≥5 | MHT, EPT | June 1991 to February 1994 | ||
Lipworth (1995), Greece | 820, H | 795, 753, H | Mean: 56.4 (cases) and 54.4 (controls) PMP† | ≥3 | MHT | January 1989 to December 1991 | ||
Newcomb (1995), USA | 3130, P | 3698, P | <75 PMP† | ≥15 | MHT, ET, EPT | ET, EPT | April 1989 to December 1991 | |
Stanford (1995), USA | 537, P | 492, P | 50–64, 16.2% in situ | ≥20 | ET, EPT | ET, EPT | January 1988 to June 1990 | |
Levi (1996), Switzerland | 230, H | 507, H | 24–75 | ≥10 | MHT | January 1990 to July 1995 | ||
Persson (1997), Sweden | 435, P | 1770, P | 40–74, 13% in situ | ≥11 | MHT, ET, EPT | ET, EPT | February 1990 to June 1995 | |
Henrich (1998), USA | 109, P | 545, P | 45 + PMP†, 29.4% in situ | ≥11 | MHT, ET, EPT | July 1987 to March 1992 | ||
Magnusson (1999), Sweden | 2563, P | 2845, P | 50–74, PMP† | >10 | MHT, ET, EPT | ET, EPT | October 1993 to March 1995 | |
Moorman (2000), USA | 397, P | 425, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | 1993–1996 | ||
Ross (2000), USA | 1897, P | 1637, P | 55–72, PMP† | ≥15 | ET, EPT | March 1987 to December 1989, January 1992 to December 1992, January 1995 to April 1996 | ||
Kirsh (2002), Canada | 404, P | 403, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | ET, EPT | April 1995 to March 1996 | |
Newcomb (2002), USA | 5298, P | 5571, P | 50–79 PMP† | ≥5 | MHT, ET, EPT | ET, EPT | January 1992 to December 1994 | |
Weiss (2002), USA | 1870, P | 1953, P | 55 to <65, PMP† | ET, EPT | July 1994 to April 1998 | |||
Wirfält (2002), Sweden | 237, P | 673, P | >49 | MHT | May 1995 to December 1992 | |||
Li (2003), USA | 975, P | 1007, P | 65–79 | ET, EPT | ET, EPT | April 1997 to May 1999 | ||
Total | 28 749 | 33, 403 |
Author (year), region . | Number of cases . | Number /type of controls . | Age (years), Type of cancer . | Longest reported duration . | Type of hormone therapy* . | . | Time period follow-up . | |
---|---|---|---|---|---|---|---|---|
. | . | . | . | . | Ever/never . | Duration . | . | |
Kaufmann (1991) USA and Canada | 1686, H | 2077, H | 40–69 PMP† | ≥15 | ET, EPT | ET | January 1980 to September 1988 | |
Palmer (1991), Canada | 607, P | 1214, P | <70 | ≥15 | ET, EPT | ET | 1982–1986 | |
Harris (1992), USA | 412, H | 336, H | <30 to ≥80 PMP† | ≥5 | MHT | January 1987 to December 1989 | ||
La Vecchia (1992), Italy | 3037, H | 2569, H | ≥3 | MHT | 1983–1990 | |||
Yang (1992), Canada | 699, P | 685, P | <75 PMP† | ≥10 | ET, EPT | ET | June 1988 to June 1989 | |
Weinstein (1993), USA | 837, P | 860, P | <79 PMP† | ≥20 | MHT | January 1984 to December 1986 | ||
La Vecchia (1995), Italy | 2569, H | 2588, H | 20–74 | ≥5 | MHT, EPT | June 1991 to February 1994 | ||
Lipworth (1995), Greece | 820, H | 795, 753, H | Mean: 56.4 (cases) and 54.4 (controls) PMP† | ≥3 | MHT | January 1989 to December 1991 | ||
Newcomb (1995), USA | 3130, P | 3698, P | <75 PMP† | ≥15 | MHT, ET, EPT | ET, EPT | April 1989 to December 1991 | |
Stanford (1995), USA | 537, P | 492, P | 50–64, 16.2% in situ | ≥20 | ET, EPT | ET, EPT | January 1988 to June 1990 | |
Levi (1996), Switzerland | 230, H | 507, H | 24–75 | ≥10 | MHT | January 1990 to July 1995 | ||
Persson (1997), Sweden | 435, P | 1770, P | 40–74, 13% in situ | ≥11 | MHT, ET, EPT | ET, EPT | February 1990 to June 1995 | |
Henrich (1998), USA | 109, P | 545, P | 45 + PMP†, 29.4% in situ | ≥11 | MHT, ET, EPT | July 1987 to March 1992 | ||
Magnusson (1999), Sweden | 2563, P | 2845, P | 50–74, PMP† | >10 | MHT, ET, EPT | ET, EPT | October 1993 to March 1995 | |
Moorman (2000), USA | 397, P | 425, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | 1993–1996 | ||
Ross (2000), USA | 1897, P | 1637, P | 55–72, PMP† | ≥15 | ET, EPT | March 1987 to December 1989, January 1992 to December 1992, January 1995 to April 1996 | ||
Kirsh (2002), Canada | 404, P | 403, P | 20–74, PMP† | ≥10 | MHT, ET, EPT | ET, EPT | April 1995 to March 1996 | |
Newcomb (2002), USA | 5298, P | 5571, P | 50–79 PMP† | ≥5 | MHT, ET, EPT | ET, EPT | January 1992 to December 1994 | |
Weiss (2002), USA | 1870, P | 1953, P | 55 to <65, PMP† | ET, EPT | July 1994 to April 1998 | |||
Wirfält (2002), Sweden | 237, P | 673, P | >49 | MHT | May 1995 to December 1992 | |||
Li (2003), USA | 975, P | 1007, P | 65–79 | ET, EPT | ET, EPT | April 1997 to May 1999 | ||
Total | 28 749 | 33, 403 |
H, hospital-based cases resp. controls; P, population-based cases resp. controls; PMP, postmenopausal.
Menopausal hormone therapy (MHT) [unspecified/unknown hormone therapy or estrogen therapy (ET) plus estrogen–progestin therapy (EPT) combined], as utilized for meta-analysis.
Only postmenopausal women included.
Author . | Year of publication . | Type of study . | Reason for exclusion . |
---|---|---|---|
Bergkvist | 1989 | Cohort | Included in meta-analysis by Steinberg et al. (1991) |
Dupont | 1989 | Cohort | Included in meta-analysis as above |
Colditz | 1990 | Cohort | Data included in Colditz et al. (1995) |
Colditz | 1992 | Cohort | Data included in Colditz et al. (1995) |
Nachtigall | 1992 | RCT | Neither person-years nor CI provided |
Yuen | 1993 | Cohort | Investigated breast cancer mortality only; same population as in Li et al. (2003) |
Squitieri | 1994 | CCS | Compared breast cancer women with and without MHT |
Schairer | 1994 | Cohort | More recent publication (Schairer et al., 2000) with identical cohort included |
Willis | 1996 | Cohort | Investigated breast cancer mortality only |
Franceschi | 1997 | CCS | Identical study as in La Vecchia et al. (1995), published in Italian |
Tavani | 1997a | CCS | Combined results of two separate studies (La Vecchia et al., 1992, 1995) |
Tavani | 1997b | CCS | Same data as La Vecchia et al. (1995) analysed with respect to attributable risk factors for breast cancer |
Grodstein | 1997 | Cohort | End point: breast cancer mortality risk |
Sellers | 1997 | Cohort | Restricted to women with family history of breast cancer |
Bonnier | 1998 | CCS | Analysis restricted to women with breast cancer by stage |
Brinton | 1998 | CCS | Included premenopausal women (<55 years) only |
Sourander | 1998 | Cohort | Analyses restricted to past and current versus never users; lagtime until follow-up 7 years |
Hulley | 1998 | RCT | Subsequent publication (Hulley et al., 2002) with prolonged follow-up included |
Fioretti | 1999 | CCS | Data of two previous CC studies: La Vecchia et al. (1987, 1995); latter study included |
Newcomer | 1999 | CCS | Analysed for current use (within 2 years of reference date) only |
Cobleigh | 1999 | Cohort | Only women with breast cancer included, analyses performed according to estrogen receptor status |
Dupont | 1999 | Cohort | Restricted to women with a history of proliferative breast disease |
Gapstur | 1999 | Cohort | No parameter suitable for meta-analysis |
Lam | 2000 | CCS | Analysed for impact of weight increase, BMI, and breast tissue density on breast cancer risk |
Li | 2000a | CCS | Identical data in Stanford et al. (1995), analysed for impact of anthropometric variables |
Li | 2000b | CCS | Analysed MHT in relation to risk of lobular and ductal breast cancer, no risk coefficients by duration of use given |
Newcomb | 2000 | CCS | Data included in more recent publication (Newcomb et al., 2002) |
Trentham-Dietz | 2000 | CCS | Data overlap with case–control study (Newcomb et al., 1995), study evaluates risk factors of carcinoma in situ versus invasive breast cancer |
Reeves | 2000 | CCS | Risk factors analysed in women with fatal breast cancer risk |
Stallard | 2000 | Cross-sectional | Included women with breast cancer, analysing the effect of MHT on stage of breast cancer |
Claus | 2001 | CCS | Evaluated risk factors of breast cancer in situ |
Joffe | 2001 | Cohort | Data from the Nurses’ Health Study (1988–1994) with mammography screening as potential bias (data included in Colditz et al., 1995) |
Olsson | 2001 | Cohort | No parameter suitable for this meta-analysis |
Pukkala | 2001 | Cohort | Analyses restricted to long cycle versus monthly MHT use |
Clarke | 2002 | RCT | One case of breast cancer in two study groups each; excluded as dates of study not provided |
Chen | 2002a | Cohort | Data from Nurses’ Health Study with shorter observation period than in Colditz and Rosner (1998) |
Chen | 2002b | CCS | Restricted analyses (current/past MHT use) |
Daling | 2002 | CCS | Analysed MHT in relation to histology |
Li | 2002 | CCS | Reference group short-term users of MHT, not never-users |
Hedblad | 2002 | Cohort | Identical study as Manjer et al. (2001), but shorter follow-up |
Lumachi | 2002 | CCS | Control group partially self-selected |
Porch | 2002 | Cohort | Only current users at baseline included, past users excluded |
Ursin | 2002 | CCS | Analysed MHT in relation to histology |
Cotterchio | 2003 | CCS | Analysed MHT in relation to estrogen/progestin receptor status |
van der Klift | 2003 | Cohort | No data upon MHT use provided |
Newcomer | 2003 | CCS | Analysed MHT in relation to histology |
Norman | 2003 | CCS | Identical data as Weiss et al. (2002), analysed impact of OC |
Olsson | 2003 | Cohort | No parameter suitable for meta-analysis |
Feigelson | 2004 | Cohort | Analyses stratified exclusively by BMI |
Author . | Year of publication . | Type of study . | Reason for exclusion . |
---|---|---|---|
Bergkvist | 1989 | Cohort | Included in meta-analysis by Steinberg et al. (1991) |
Dupont | 1989 | Cohort | Included in meta-analysis as above |
Colditz | 1990 | Cohort | Data included in Colditz et al. (1995) |
Colditz | 1992 | Cohort | Data included in Colditz et al. (1995) |
Nachtigall | 1992 | RCT | Neither person-years nor CI provided |
Yuen | 1993 | Cohort | Investigated breast cancer mortality only; same population as in Li et al. (2003) |
Squitieri | 1994 | CCS | Compared breast cancer women with and without MHT |
Schairer | 1994 | Cohort | More recent publication (Schairer et al., 2000) with identical cohort included |
Willis | 1996 | Cohort | Investigated breast cancer mortality only |
Franceschi | 1997 | CCS | Identical study as in La Vecchia et al. (1995), published in Italian |
Tavani | 1997a | CCS | Combined results of two separate studies (La Vecchia et al., 1992, 1995) |
Tavani | 1997b | CCS | Same data as La Vecchia et al. (1995) analysed with respect to attributable risk factors for breast cancer |
Grodstein | 1997 | Cohort | End point: breast cancer mortality risk |
Sellers | 1997 | Cohort | Restricted to women with family history of breast cancer |
Bonnier | 1998 | CCS | Analysis restricted to women with breast cancer by stage |
Brinton | 1998 | CCS | Included premenopausal women (<55 years) only |
Sourander | 1998 | Cohort | Analyses restricted to past and current versus never users; lagtime until follow-up 7 years |
Hulley | 1998 | RCT | Subsequent publication (Hulley et al., 2002) with prolonged follow-up included |
Fioretti | 1999 | CCS | Data of two previous CC studies: La Vecchia et al. (1987, 1995); latter study included |
Newcomer | 1999 | CCS | Analysed for current use (within 2 years of reference date) only |
Cobleigh | 1999 | Cohort | Only women with breast cancer included, analyses performed according to estrogen receptor status |
Dupont | 1999 | Cohort | Restricted to women with a history of proliferative breast disease |
Gapstur | 1999 | Cohort | No parameter suitable for meta-analysis |
Lam | 2000 | CCS | Analysed for impact of weight increase, BMI, and breast tissue density on breast cancer risk |
Li | 2000a | CCS | Identical data in Stanford et al. (1995), analysed for impact of anthropometric variables |
Li | 2000b | CCS | Analysed MHT in relation to risk of lobular and ductal breast cancer, no risk coefficients by duration of use given |
Newcomb | 2000 | CCS | Data included in more recent publication (Newcomb et al., 2002) |
Trentham-Dietz | 2000 | CCS | Data overlap with case–control study (Newcomb et al., 1995), study evaluates risk factors of carcinoma in situ versus invasive breast cancer |
Reeves | 2000 | CCS | Risk factors analysed in women with fatal breast cancer risk |
Stallard | 2000 | Cross-sectional | Included women with breast cancer, analysing the effect of MHT on stage of breast cancer |
Claus | 2001 | CCS | Evaluated risk factors of breast cancer in situ |
Joffe | 2001 | Cohort | Data from the Nurses’ Health Study (1988–1994) with mammography screening as potential bias (data included in Colditz et al., 1995) |
Olsson | 2001 | Cohort | No parameter suitable for this meta-analysis |
Pukkala | 2001 | Cohort | Analyses restricted to long cycle versus monthly MHT use |
Clarke | 2002 | RCT | One case of breast cancer in two study groups each; excluded as dates of study not provided |
Chen | 2002a | Cohort | Data from Nurses’ Health Study with shorter observation period than in Colditz and Rosner (1998) |
Chen | 2002b | CCS | Restricted analyses (current/past MHT use) |
Daling | 2002 | CCS | Analysed MHT in relation to histology |
Li | 2002 | CCS | Reference group short-term users of MHT, not never-users |
Hedblad | 2002 | Cohort | Identical study as Manjer et al. (2001), but shorter follow-up |
Lumachi | 2002 | CCS | Control group partially self-selected |
Porch | 2002 | Cohort | Only current users at baseline included, past users excluded |
Ursin | 2002 | CCS | Analysed MHT in relation to histology |
Cotterchio | 2003 | CCS | Analysed MHT in relation to estrogen/progestin receptor status |
van der Klift | 2003 | Cohort | No data upon MHT use provided |
Newcomer | 2003 | CCS | Analysed MHT in relation to histology |
Norman | 2003 | CCS | Identical data as Weiss et al. (2002), analysed impact of OC |
Olsson | 2003 | Cohort | No parameter suitable for meta-analysis |
Feigelson | 2004 | Cohort | Analyses stratified exclusively by BMI |
BMI, body mass index; CCS, case control study; CI, confidence interval; OC, oral contraceptives; MHT, menopausal hormone therapy; RCT, randomized controlled trial.
Author . | Year of publication . | Type of study . | Reason for exclusion . |
---|---|---|---|
Bergkvist | 1989 | Cohort | Included in meta-analysis by Steinberg et al. (1991) |
Dupont | 1989 | Cohort | Included in meta-analysis as above |
Colditz | 1990 | Cohort | Data included in Colditz et al. (1995) |
Colditz | 1992 | Cohort | Data included in Colditz et al. (1995) |
Nachtigall | 1992 | RCT | Neither person-years nor CI provided |
Yuen | 1993 | Cohort | Investigated breast cancer mortality only; same population as in Li et al. (2003) |
Squitieri | 1994 | CCS | Compared breast cancer women with and without MHT |
Schairer | 1994 | Cohort | More recent publication (Schairer et al., 2000) with identical cohort included |
Willis | 1996 | Cohort | Investigated breast cancer mortality only |
Franceschi | 1997 | CCS | Identical study as in La Vecchia et al. (1995), published in Italian |
Tavani | 1997a | CCS | Combined results of two separate studies (La Vecchia et al., 1992, 1995) |
Tavani | 1997b | CCS | Same data as La Vecchia et al. (1995) analysed with respect to attributable risk factors for breast cancer |
Grodstein | 1997 | Cohort | End point: breast cancer mortality risk |
Sellers | 1997 | Cohort | Restricted to women with family history of breast cancer |
Bonnier | 1998 | CCS | Analysis restricted to women with breast cancer by stage |
Brinton | 1998 | CCS | Included premenopausal women (<55 years) only |
Sourander | 1998 | Cohort | Analyses restricted to past and current versus never users; lagtime until follow-up 7 years |
Hulley | 1998 | RCT | Subsequent publication (Hulley et al., 2002) with prolonged follow-up included |
Fioretti | 1999 | CCS | Data of two previous CC studies: La Vecchia et al. (1987, 1995); latter study included |
Newcomer | 1999 | CCS | Analysed for current use (within 2 years of reference date) only |
Cobleigh | 1999 | Cohort | Only women with breast cancer included, analyses performed according to estrogen receptor status |
Dupont | 1999 | Cohort | Restricted to women with a history of proliferative breast disease |
Gapstur | 1999 | Cohort | No parameter suitable for meta-analysis |
Lam | 2000 | CCS | Analysed for impact of weight increase, BMI, and breast tissue density on breast cancer risk |
Li | 2000a | CCS | Identical data in Stanford et al. (1995), analysed for impact of anthropometric variables |
Li | 2000b | CCS | Analysed MHT in relation to risk of lobular and ductal breast cancer, no risk coefficients by duration of use given |
Newcomb | 2000 | CCS | Data included in more recent publication (Newcomb et al., 2002) |
Trentham-Dietz | 2000 | CCS | Data overlap with case–control study (Newcomb et al., 1995), study evaluates risk factors of carcinoma in situ versus invasive breast cancer |
Reeves | 2000 | CCS | Risk factors analysed in women with fatal breast cancer risk |
Stallard | 2000 | Cross-sectional | Included women with breast cancer, analysing the effect of MHT on stage of breast cancer |
Claus | 2001 | CCS | Evaluated risk factors of breast cancer in situ |
Joffe | 2001 | Cohort | Data from the Nurses’ Health Study (1988–1994) with mammography screening as potential bias (data included in Colditz et al., 1995) |
Olsson | 2001 | Cohort | No parameter suitable for this meta-analysis |
Pukkala | 2001 | Cohort | Analyses restricted to long cycle versus monthly MHT use |
Clarke | 2002 | RCT | One case of breast cancer in two study groups each; excluded as dates of study not provided |
Chen | 2002a | Cohort | Data from Nurses’ Health Study with shorter observation period than in Colditz and Rosner (1998) |
Chen | 2002b | CCS | Restricted analyses (current/past MHT use) |
Daling | 2002 | CCS | Analysed MHT in relation to histology |
Li | 2002 | CCS | Reference group short-term users of MHT, not never-users |
Hedblad | 2002 | Cohort | Identical study as Manjer et al. (2001), but shorter follow-up |
Lumachi | 2002 | CCS | Control group partially self-selected |
Porch | 2002 | Cohort | Only current users at baseline included, past users excluded |
Ursin | 2002 | CCS | Analysed MHT in relation to histology |
Cotterchio | 2003 | CCS | Analysed MHT in relation to estrogen/progestin receptor status |
van der Klift | 2003 | Cohort | No data upon MHT use provided |
Newcomer | 2003 | CCS | Analysed MHT in relation to histology |
Norman | 2003 | CCS | Identical data as Weiss et al. (2002), analysed impact of OC |
Olsson | 2003 | Cohort | No parameter suitable for meta-analysis |
Feigelson | 2004 | Cohort | Analyses stratified exclusively by BMI |
Author . | Year of publication . | Type of study . | Reason for exclusion . |
---|---|---|---|
Bergkvist | 1989 | Cohort | Included in meta-analysis by Steinberg et al. (1991) |
Dupont | 1989 | Cohort | Included in meta-analysis as above |
Colditz | 1990 | Cohort | Data included in Colditz et al. (1995) |
Colditz | 1992 | Cohort | Data included in Colditz et al. (1995) |
Nachtigall | 1992 | RCT | Neither person-years nor CI provided |
Yuen | 1993 | Cohort | Investigated breast cancer mortality only; same population as in Li et al. (2003) |
Squitieri | 1994 | CCS | Compared breast cancer women with and without MHT |
Schairer | 1994 | Cohort | More recent publication (Schairer et al., 2000) with identical cohort included |
Willis | 1996 | Cohort | Investigated breast cancer mortality only |
Franceschi | 1997 | CCS | Identical study as in La Vecchia et al. (1995), published in Italian |
Tavani | 1997a | CCS | Combined results of two separate studies (La Vecchia et al., 1992, 1995) |
Tavani | 1997b | CCS | Same data as La Vecchia et al. (1995) analysed with respect to attributable risk factors for breast cancer |
Grodstein | 1997 | Cohort | End point: breast cancer mortality risk |
Sellers | 1997 | Cohort | Restricted to women with family history of breast cancer |
Bonnier | 1998 | CCS | Analysis restricted to women with breast cancer by stage |
Brinton | 1998 | CCS | Included premenopausal women (<55 years) only |
Sourander | 1998 | Cohort | Analyses restricted to past and current versus never users; lagtime until follow-up 7 years |
Hulley | 1998 | RCT | Subsequent publication (Hulley et al., 2002) with prolonged follow-up included |
Fioretti | 1999 | CCS | Data of two previous CC studies: La Vecchia et al. (1987, 1995); latter study included |
Newcomer | 1999 | CCS | Analysed for current use (within 2 years of reference date) only |
Cobleigh | 1999 | Cohort | Only women with breast cancer included, analyses performed according to estrogen receptor status |
Dupont | 1999 | Cohort | Restricted to women with a history of proliferative breast disease |
Gapstur | 1999 | Cohort | No parameter suitable for meta-analysis |
Lam | 2000 | CCS | Analysed for impact of weight increase, BMI, and breast tissue density on breast cancer risk |
Li | 2000a | CCS | Identical data in Stanford et al. (1995), analysed for impact of anthropometric variables |
Li | 2000b | CCS | Analysed MHT in relation to risk of lobular and ductal breast cancer, no risk coefficients by duration of use given |
Newcomb | 2000 | CCS | Data included in more recent publication (Newcomb et al., 2002) |
Trentham-Dietz | 2000 | CCS | Data overlap with case–control study (Newcomb et al., 1995), study evaluates risk factors of carcinoma in situ versus invasive breast cancer |
Reeves | 2000 | CCS | Risk factors analysed in women with fatal breast cancer risk |
Stallard | 2000 | Cross-sectional | Included women with breast cancer, analysing the effect of MHT on stage of breast cancer |
Claus | 2001 | CCS | Evaluated risk factors of breast cancer in situ |
Joffe | 2001 | Cohort | Data from the Nurses’ Health Study (1988–1994) with mammography screening as potential bias (data included in Colditz et al., 1995) |
Olsson | 2001 | Cohort | No parameter suitable for this meta-analysis |
Pukkala | 2001 | Cohort | Analyses restricted to long cycle versus monthly MHT use |
Clarke | 2002 | RCT | One case of breast cancer in two study groups each; excluded as dates of study not provided |
Chen | 2002a | Cohort | Data from Nurses’ Health Study with shorter observation period than in Colditz and Rosner (1998) |
Chen | 2002b | CCS | Restricted analyses (current/past MHT use) |
Daling | 2002 | CCS | Analysed MHT in relation to histology |
Li | 2002 | CCS | Reference group short-term users of MHT, not never-users |
Hedblad | 2002 | Cohort | Identical study as Manjer et al. (2001), but shorter follow-up |
Lumachi | 2002 | CCS | Control group partially self-selected |
Porch | 2002 | Cohort | Only current users at baseline included, past users excluded |
Ursin | 2002 | CCS | Analysed MHT in relation to histology |
Cotterchio | 2003 | CCS | Analysed MHT in relation to estrogen/progestin receptor status |
van der Klift | 2003 | Cohort | No data upon MHT use provided |
Newcomer | 2003 | CCS | Analysed MHT in relation to histology |
Norman | 2003 | CCS | Identical data as Weiss et al. (2002), analysed impact of OC |
Olsson | 2003 | Cohort | No parameter suitable for meta-analysis |
Feigelson | 2004 | Cohort | Analyses stratified exclusively by BMI |
BMI, body mass index; CCS, case control study; CI, confidence interval; OC, oral contraceptives; MHT, menopausal hormone therapy; RCT, randomized controlled trial.
The time trend of BC risk in conjunction with MHT use (hormone or unspecified/unknown hormone therapy) accrued at date of study, is shown in Figure 1. MHT regimens show a steady linear increase of BC risk within the studied time period when CS, RCTs and CCS are combined in a linear regression. The increase amounts to 4.4% per year from 1979 onwards, where ‘year’ reflects the midterm year of case ascertainment for respective studies. Figure 2 shows the results of a meta-analysis to assess risks in these women stratified by study type and two case ascertainment periods. There are larger risk increases in CS and the one RCT combined than in CCS when the latter time periods are compared; there is no difference between study types for the first time period. There is major heterogeneity present except in the first time period of CS. The time trend of BC risk in conjunction with EPT is shown in Figure 3. The regression analysis of time-dependent risk change shows a larger increase than in MHT, i.e. 5.8% per year, as derived from the regression coefficient for the untransformed risk. Figure 4 shows the results of a further meta-analysis to assess risks in EPT users, stratified by study type and time periods; increasing risks after EPT are evident in the latter time periods. All effects are homogenous except for the last period in CS, where the Million Women Study results with its large weight dominate the overall effect of CS with midterm of case ascertainment year after 1991. Figure 5 shows BC risks per year in EPT users, with stratifications as before. Risk increases by year of EPT use amount to 5–9% per year in those strata where effects are not heterogenous, i.e. CCS of the second time period and CS/RCTS irrespective of time periods. The first time period in CCS (three studies only) shows a wide range from a significant decrease of 7% per year of use to a significant increase of 5% per year. Again, as in risks associated with ever use (Figure 4), effects are larger in CS than in CCS. A regression of overall risk dependent on ET by midterm year of case ascertainment in different studies provided nonsignificant regression coefficients (data not shown). Figure 6 shows stratified risks in ET users. Summary risks derived from a meta-analysis differ between CCS and CS: a nil effect in the period before 1992 in CCS compared to a 19% risk increase in CS and increased risks of 18% (CCS) versus 27% (CS) in the later time period. However, effects were heterogenous for both time periods across study types. Figure 7 shows stratified BC risks per year in ET users. There is no risk increase per year of use in early CCS, minimal increase of risk (1–3% in CS, 0.9% in CCS in 2nd period) with major differences in effect sizes when comparing different studies.
Discussion
The main findings are that overall risk of BC is increased in ever users of MHT, in particular EPTs. This difference was evident in all types of studies assessed. There is also an apparent time trend towards larger risks in recent studies. The increasing risk of intake of EPT (ever use) is indicative of increasing periods of use in subsequent years covered by different studies, as the analyses of annual increase of risk (Figure 5) show only a moderate increase from the first to second time period. The larger risk increases we found in CS could be due to the fact that there is generally a lag time of several years between the last follow-up of exposure data and the end of follow-up for BC incidence. It can be safely assumed that in the years before publication of end results of recent major studies (Million Women Study Collaborators, 2003; Writing Group for the Womens’ Health Initiative Investigators, 2002; Women’s Health Initiative Steering Committee, 2004), a majority of women using MHT will have continued use until the end of morbidity assessment.
There have been secular trends in the use of both estrogens and progestins, including type, dose, and administration. Since it was acknowledged in the 1970s that use of ET increases endometrial cancer risk, the use of EPT has increased (Wysowski et al., 1995; Brett and Madans, 1997; Hemminki et al., 1988). The increasing duration of progestin use, alongside the extended duration of MHT use, may contribute to our finding that risks increase in ET and EPT users according to studies conducted more recently. The progestin largely used in North America is medroxyprogesterone acetate (MPA), in European countries various progestins, apart from MPA (Sitruk-Ware, 2002), including some with the potential to increase insulin-like growth factor 1 activity are used (Warren, 2004). Differences in BC risk may be attributable to differences in the functional profile of the various estrogens and progestins, apart from many other factors including environmental, genetic and life-style factors. A recent key issue is whether (different) progestins exert (more) proliferative effects on the postmenopausal breast. In a French cohort study, there were also differences regarding risk in EPT users according to the type of progestin (Fournier et al., 2005).
Our results regarding overall risk, derived from ever use of hormones, and annual change are consistent with some, but not all major findings of the collaborative reanalysis (Collaborative Group on Hormonal Factors in Breast Cancer, 1997). Our findings suggest a distinct difference between ET and EPT. This difference was also found in a French CS published outside the time frame of our study selection (Fournier et al., 2005). In the reanalysis (Collaborative Group on Hormonal Factors in Breast Cancer, 1997), in only 31% of cases and in 24% of controls, constituents of MHT could be identified. The proportion of those with EPT was apparently 12.5% in cases and 11% in controls (Figure 3, Table II in Collaborative Group on Hormonal Factors in Breast Cancer, 1997); this may have limited the possibility to detect differences between regimens. Thus, results are primarily based on ET with CEE as predominant constituent.
We refrained from including risk estimates for current use of hormones into our meta-analysis although several studies provided these data. In general, these risks are higher than estimates for ever use of hormones. One major recent study (Million Women Study Collaborators, 2003) found elevated risks in current users, but not in past users. Our reluctance is based on vastly differing definitions of ‘current use’, i.e. the range reaches from use within 6 months before the interview (in a case–control study) up to 8 years before last date of follow-up (in a cohort study). In our opinion this diversity could have been a source of major bias.
The same argument applies to the analysis of change of risk after discontinuation of MHT, as for such an analysis ‘current use’ should be the starting point from which a change could be estimated. Therefore we did not present a meta-analysis for risk estimates after discontinuation of MHT. The available studies, however, present an equivocal picture, as some—especially the older studies—show a remarkable decrease of risk, whereas some of the recently published major studies present figures to the contrary. The largest of recent studies, the Million Women Study, shows a statistically increased risk only within the first year after discontinuation of hormone therapy. The problem of risk change after discontinuation is a pertinent one which might be more easily investigated in ongoing studies, as after the publication of the results of the Women’s Health Initiative RCT and the Million Women Study the proportion of women ceasing MHT has considerably increased.
In two studies risk substantially increased after cessation of MHT (Magnusson et al., 1999; Newcomb et al., 2002). Newcomb and coauthors (2002) calculated odds ratios for annual change of risks for ET of 0.99 (95% CI = 0.99–1.00) and for EPT of 0.98 (95% CI = 0.97–0.99). The latter result, however, seems not to be quite compatible with further risk estimates provided by the authors for different time periods after discontinuation, which show a marked increase with elapsed time after therapy was stopped [current use 1.39 (95% CI = 1.12–1.71); past use <5 years ago 1.71 (95% CI = 0.92–3.18); past use >5 years ago 2.38 (95% CI = 0.82–6.87); Table III of Newcomb et al., 2002]. In a Swiss CCS (Levi et al., 1996) risks remained elevated for 10–15 years after stopping MHT. In a large Swedish CCS an annual increase of 8% per year of MHT 10 and more years after discontinuation was observed. After adjustment for recency, there was no attenuation of risk increase by duration of use (Magnusson et al., 1999). Ross and coauthors (2000) noted that for at least 2 years after cessation of MHT, risks for women taking EPT remained elevated. In contrast, Olsson and coauthors (2001) found increased risks for current users only and Weiss and coauthors (Weiss et al., 2002) concluded that risk dissipated once use discontinued. The discrepancy between studies which show a risk decrease shortly after discontinuation (Yang et al., 1992; Stanford et al., 1995; Colditz et al., 1995; Weiss et al., 2002) and two major recent studies (Magnusson et al., 1999; Newcomb et al., 2002) with no decrease of risk even after prolonged periods without use of MHT can possibly be attributed to lack of power of smaller studies. In these studies the absolute number of women who stopped for defined periods probably was too small to achieve small CIs and statistical significance.
There are several limitations of our analyses. It has to be considered, that a relevant part of heterogeneity might be due to the mix of different regimens that constitutes ‘hormone therapy’. In studies conducted in earlier years with ‘menopausal hormones’ it is likely that predominantly ET was prescribed whereas in later years this term reflects a varying mix of ET and EPT. It was not possible to control for the varying degree of adjustment for confounding factors in the studies included. Women who use MHT differ in many ways from nonusers (Matthews et al., 1996). User may be more likely to undergo mammography, possibly leading to an earlier diagnosis of BC. This argument, however, seems to be contradicted by results of the WHI randomised controlled trial, where every woman underwent screening and annual mammograms, irrespective of allocation to hormone or placebo group. We did not address the impact of body mass index and age of menopause on BC risk because of the paucity of studies providing appropriate results for pooling. Owing to the use of published, not raw data, we could not include the duration of exogenous hormone exposure into our calculations of changes of risk after cessation of MHT, as it has been done by authors of one CCS (Magnusson et al., 1999). Finally, given the lack of specific information on hormonal constituents, subanalyses addressing the issue of potential differences between CEE-based and non-CEE based preparations were not possible, likewise we could not differentiate between various progestins.
In conclusion, there is evidence that relative risks for BC risks by MHT, in particular EPT, and risks have been increasing in recent years. Given the widespread use of MHT, and often long duration, more detailed knowledge about differential BC risks of both estrogens and progestins are necessary to minimize BC risk in symptomatic women who consider MHT. We think that a potential lack of decrease of risk in past users of MHT calls for a pooled reanalysis of recent major epidemiological studies.
Appendix
Acknowledgements
This study was conducted at the Institute for Public Health and Nursing Research, Faculty of Health Sciences, Bremen University; Epi. Consult GmbH, Bremen and the Charité Universitätsmedizin Berlin, Campus Benjamin Franklin, Berlin, Germany. This manuscript is based to a greater extent on the MPH thesis of CMG.
References
Collaborative Group on Hormonal Factors in Breast Cancer (
ESPRIT Team (
Million Women Study Collaborators (
National Heart, Lung, and Blood Institute (
Preventive Services Task Force (May
U.S. Preventive Services Task Force (May
Women’s Health Initiative Steering Committee (
Writing Group for the Women’s Health Initiative Investigators (
Author notes
1Epi. Consult GmbH, 2Institute for Public Health and Nursing Research, Bremen University, Bremen and 3Charité Universitätsmedizin Berlin, Campus Benjamin Franklin, Clinical Research Center of Women’s Health, Berlin, Germany