Hormone replacement therapy and incidence of central nervous system tumours in the Million Women Study§


  • Million Women Study Steering Committee: Joan Austoker, Emily Banks, Valerie Beral, Judith Church, Ruth English, Jane Green, Julietta Patnick, Richard Peto, Gillian Reeves, Martin Vessey and Matthew Wallis.

  • Million Women Study Coordinating Centre Staff: Simon Abbott, Miranda Armstrong, Krys Baker, Angela Balkwill, Vicky Benson, Valerie Beral, Judith Black, Anna Brown, Diana Bull, Benjamin Cairns, James Chivenga, Barbara Crossley, Dave Ewart, Sarah Ewart, Lee Fletcher, Laura Gerrard, Adrian Goodill, Isobel Green, Jane Green, Elizabeth Hilton, Joy Hooley, Sau Wan Kan, Carol Keene, Oksana Kirichek, Nicky Langston, Bette Liu, Maria-Jose Luque, Maria MacGregor, Lynn Pank, Kirstin Pirie, Gillian Reeves, Emma Sherman, Evie Sherry-Starmer, Moya Simmonds, Elizabeth Spencer, Helena Strange, Siân Sweetland, Alison Timadjer, Sarah Tipper, Ruth Travis, Xiaosi Wang, Joanna Watson, Stephen Williams, Lucy Wright, Tienyu Yang and Heather Young.

  • §

    Collaborating UK NHS Breast Screening Centres: Avon, Aylesbury, Barnsley, Basingstoke, Bedfordshire and Hertfordshire, Cambridge and Huntingdon, Chelmsford and Colchester, Chester, Cornwall, Crewe, Cumbria, Doncaster, Dorset, East Berkshire, East Cheshire, East Devon, East of Scotland, East Suffolk, East Sussex, Gateshead, Gloucestershire, Great Yarmouth, Hereford and Worcester, Kent, Kings Lynn, Leicestershire, Liverpool, Manchester, Milton Keynes, Newcastle, North Birmingham, North East Scotland, North Lancashire, North Middlesex, North Nottingham, North of Scotland, North Tees, North Yorkshire, Nottingham, Oxford, Portsmouth, Rotherham, Sheffield, Shropshire, Somerset, South Birmingham, South East Scotland, South East Staffordshire, South Derbyshire, South Essex, South Lancashire, South West Scotland, Surrey, Warrington Halton St. Helens and Knowsley, Warwickshire Solihull and Coventry, West Berkshire, West Devon, West London, West Suffolk, West Sussex, Wiltshire, Winchester, Wirral, Wycombe.


We examined the relation between the use of hormone replacement therapy (HRT) and the incidence of central nervous system (CNS) tumours in a large prospective study of 1,147,894 postmenopausal women. Women were aged 56.6 years on average at entry, and HRT use was recorded at recruitment and updated, where possible, about 3 years later. During a mean follow-up of 5.3 years per woman, 1,266 CNS tumours were diagnosed, including 557 gliomas, 311 meningiomas and 117 acoustic neuromas. Compared with never users of HRT, the relative risks (RRs) for all incident CNS tumours, gliomas, meningiomas and acoustic neuromas in current users of HRT were 1.20 (95% CI: 1.05–1.36), 1.09 (95% CI: 0.89–1.32), 1.34 (95% CI: 1.03–1.75) and 1.58 (95% CI: 1.02–2.45), respectively, and there was no significant difference in the relative risks by tumour type (heterogeneity p = 0.2). In past users of HRT the relative risk was 1.07 (95% CI: 0.93–1.24) for all CNS tumours. Among current users of HRT, there was significant heterogeneity by the type of HRT with the users of oestrogen-only HRT at higher risk of all CNS tumours than users of oestrogen–progestagen HRT (RR = 1.42, 95% CI: 1.21–1.67 versus RR = 0.97, 95% CI: 0.82–1.16) (heterogeneity p < 0.001). Among current users of oestrogen-only and oestrogen–progestagen HRT, there was no significant heterogeneity by duration of use, hormonal constituent or mode of administration of HRT.

Female hormones are thought to play a role in the aetiology of central nervous system (CNS) tumours. Meningiomas, for example, have been shown to be two to three times more common in women than in men,1 increase in size during pregnancy2, 3 and are more likely to develop in pre-menopausal women than post-menopausal women of a similar age.4, 5

The use of certain types of hormone replacement therapy (HRT) for the menopause is associated with an increased incidence of cancers of the breast, ovary and endometrium.6 For CNS tumours, which are relatively uncommon, only two prospective studies have published results on the association with HRT use. In one study, the current use of HRT was reported to be associated with a significant increase in the risk of meningioma,7 and in another study, no significant association was observed for the risk of glioma.8 The results from retrospective studies of glioma or meningioma and the use of HRT are mixed.5, 9–15

Here, we examine the relation between the use of HRT and the incidence of CNS tumours overall, and separately for specified glioma, meningioma and acoustic neuroma, in a large prospective study of middle-aged women in the United Kingdom.


BMI: body mass index; CI: confidence interval; CNS: central nervous system; HRT: hormone replacement therapy; ICD: International Classification of Diseases; NHS: National Health Service; RR: relative risk

Material and Methods

Data collection and definitions

Study design

During 1996 to 2001, 1.3 million women were recruited through the National Health Service (NHS) Breast Screening Programme into the Million Women Study cohort, completing a recruitment questionnaire about sociodemographic and lifestyle factors and other personal characteristics including detailed information on the use of HRT. The cohort was resurveyed about three years after recruitment (65% response rate), when updated information on menopausal status and on use of HRT was obtained. Full details of the study design and methods are described elsewhere,16 and both questionnaires can be viewed at http://www.millionwomenstudy.org. All study participants gave written consent to take part in the study, and ethical approval was provided by the Oxford and Anglia Multi-Centre Research Ethics Committee.

All participants have been flagged on the NHS Central Registers so that cancer registrations and deaths are routinely notified to the study investigators. This information includes the date of each such event and codes the tumour site and morphology using the 10th revision of the International Classification of Diseases (ICD-10).17 Follow-up is complete for more than 99% of the study population.


Incident invasive and non-invasive tumours of the CNS were included from the following sites: ICD-10 C70, C71, C72.0, C75.1-3, D32, D33, D35.2-4, D42, D43 and D44.3-5. Incident cases of glioma (ICD-O 9380-9481) and meningioma (ICD-O 9530-9539) were identified within these sites, where appropriate. Acoustic neuromas were defined as those tumours coded as ICD-10 D33.3 with morphology code ICD-O 9560/0.

Menopausal status

Women were classified at recruitment as pre-menopausal (women who reported still having regular periods), perimenopausal (reported having irregular periods) or post-menopausal (reported that their periods had stopped naturally or after bilateral oophorectomy). For some women, menopause was masked by hysterectomy or by the use of HRT before the natural menopause. Women without hysterectomy who had stopped taking HRT at least two years before recruitment and had not had a period since stopping HRT were classified as post-menopausal. The remaining women whose menopausal status was still unknown were classified as post-menopausal after they turned 53 years old (as 96% of the cohort aged ≥53 years who had not had a hysterectomy or used HRT were postmenopausal). Women classified as pre-menopausal or perimenopausal at recruitment had their menopausal status updated at resurvey, where possible.

Hormone replacement therapy

Women were classified as either current, past or never users of HRT, as reported at the time of last contact. Women were initially classified using information obtained at recruitment and were reclassified at resurvey, where possible, using updated information on HRT use. In some analyses, current users of HRT were further classified according to the type of preparation last used (oestrogen-only, oestrogen–progestagen combinations and other/unknown), by last reported total duration of use (<5 years or ≥5 years), by the mode of administration in oestrogen-only users (oral or transdermal) and by the type of regimen for progestagen in oestrogen–progestagen users (continuous or sequential). Women contributed woman-years to the analyses within the appropriate category of HRT use, as most recently reported. To reduce misclassification due to changes in HRT use, follow-up was censored 48 months after last contact. A sensitivity analysis was done to explore the effect of possible misclassification of HRT use: updating at resurvey and censoring follow-up at the earliest of either 48 months or December 31, 2002. Censoring on December 31, 2002, was done because prescribing of HRT in the United Kingdom is known to have decreased sharply after the 2002 publication of results from the Women's Health Initiative randomised trial.18, 19

Statistical analysis

We considered all CNS tumours and each main tumour type separately (glioma, meningioma, acoustic neuroma and “other” CNS tumours), using Cox proportional hazards models, taking attained age as the underlying time variable to obtain relative risks (RRs) and 95% confidence intervals (CIs) for various measures of HRT use.

Analyses were restricted to post-menopausal women. Women were excluded if at the time of entry into the analyses they had been diagnosed with any type of invasive tumour [other than non-melanoma skin cancer (C44)] or any non-invasive tumour of the CNS, or if there was no information on HRT use. Women were also excluded if they reported having the inherited disorder neurofibromatosis (Q85.0) at recruitment, which is associated with a high risk of neurological tumours.

Eligible women contributed woman-years until the date of registration of the CNS tumour of interest, date of death or the end of follow-up, whichever was earliest. In addition, women diagnosed with any cancer other than the cancer of interest (except non-melanoma skin cancer) during the follow-up period were censored at the date of diagnosis of that cancer. The end of follow-up for cancer incidence was December 31, 2007 for East Anglia, South West and North West (Mersey) regions, June 30, 2007 for Oxford, Thames, West Midlands and Trent and December 31, 2006 for North Yorkshire, North West (Manchester/Lancashire) and Scotland.

All exposure information was taken as reported at recruitment, except for the use of HRT and menopausal status, which were updated where possible, as previously described using information reported at resurvey. All analyses were routinely stratified by quintiles of socioeconomic status (using the Townsend deprivation index20) and by geographical region of residence (10 regions corresponding to the areas covered by the cancer registries). Analyses were further adjusted for height (<160, 160–164.9, ≥165 cm) and body mass index (BMI) (< 25, 25–29.9, ≥30 kg/m2): the risk factors previously found to influence CNS tumour incidence in this cohort.21 Further adjustment for age at menopause (<45, 45–49, 50–54, ≥55 years of age), smoking status (never, past, current smoker), alcohol intake (never, <70, ≥70 g/week), strenuous exercise (< once, once, ≥ twice per week), age at first birth (<20, 20–24, ≥25 years old), parity (0, 1–2, ≥3 full-term pregnancies) and oral contraceptive use (never, ever; <5 years, ≥5 years duration) was also explored. Women with missing values for any of the adjustment variables were assigned to a separate category for that variable. A sensitivity analysis was carried out to assess the effect of excluding women with missing values for any of the adjustment factors.


In total, 1,147,894 post-menopausal women were included in the analyses. Table 1 shows the characteristics of the study population at recruitment, woman-years of follow-up and the number of incident CNS tumours by tumour type, according to a woman's last reported use of HRT. When demographic, social, health and lifestyle characteristics of never, past and current HRT users are compared, the groups do not differ appreciably for most factors: the main difference is that ever users of HRT are more likely to have ever taken oral contraceptives and to drink slightly more alcohol than never users of HRT.

Table 1. Characteristics of the study population by the use of hormone replacement therapy (HRT) recorded at the time of last contact and the details of follow-up
inline image

At the time of last recorded HRT use, 620,490 (54%) women had ever used HRT and 358,252 (31%) were current users. Among current users of HRT, 140,797 (39%) women were using oestrogen-only preparations, 176,709 (49%) were using an oestrogen–progestagen HRT and 40,746 (11%) were using other or unknown preparations (56% of these women were using tibolone). Based on a pre-defined calculation,22 the average duration of HRT use in current users among cases at time of diagnosis was estimated to be 7.8 years.

During a total of 6.1 million woman-years of follow-up (a mean of 5.3 years per woman), 1,266 incident CNS tumour cases were reported. Of these, 557 were specified glioma, 311 were specified meningioma and 117 were specified acoustic neuroma. The remaining 281 CNS tumours were predominantly pituitary tumours (n = 98) and tumours of unspecified type.

When compared to never users of HRT, ever users had a small but significantly increased risk of all CNS tumours (RR = 1.14, 95% CI: 1.02–1.28, p = 0.02). The relative risks did not differ significantly across tumour types for meningiomas (RR = 1.32, 95% CI: 1.05–1.66, p = 0.02) acoustic neuromas (RR = 1.64, 95% CI: 1.11–2.41, p = 0.01) or gliomas (RR = 1.05, 95% CI: 0.89–1.25, p = 0.5) (heterogeneity p = 0.06). Table 2 shows the results for incident CNS tumours in current and past use of HRT and by type of HRT last used in current users. The RR for all CNS tumours combined was 1.20 (95% CI: 1.05–1.36) for current users and 1.07 (95% CI: 0.93–1.24) for past users when compared to never users; there was no significant difference between these RRs (heterogeneity p = 0.2) or by specified tumour type within current and past users (heterogeneity p = 0.2 and 0.1 for current and past users, respectively). In current users, the risk of all CNS tumours significantly varied by the type of HRT used, with the users of oestrogen-only HRT (RR = 1.42, 95% CI: 1.21–1.67) at a significantly higher risk of developing a tumour than women using oestrogen–progestagen HRT (RR = 0.97, 95% CI: 0.82–1.16; heterogeneity p < 0.001). These results were not substantially changed in the sensitivity analyses: RRs for all CNS tumours using adjustment factors with only known values were 1.45 (95% CI: 1.23–1.71) for current users of oestrogen-only HRT and 0.98 (95% CI: 0.82–1.17) for oestrogen–progestagen HRT users. For additional censoring (on December 31, 2002), RRs were 1.51 (95% CI: 1.24–1.84) for current users of oestrogen-only HRT and 0.89 (95% CI: 0.72–1.11) for oestrogen–progestagen HRT users.

Table 2. Relative risks (RR) and 95% confidence intervals (CI) for incident tumours of the central nervous system (CNS) by hormone replacement therapy (HRT) last used in postmenopausal women
inline image

The RRs for women using oestrogen-only HRT were greater than for women using oestrogen–progestagen HRT for each specified CNS tumour type (Fig. 1). For those gliomas specified as glioblastoma (ICD-10 C71, ICD-O 9440, n = 347), RRs in current versus never users were similar to those for all glioma [e.g., the RR for oestrogen-only users compared to never users for glioblastoma was 1.19 (95% CI: 0.85–1.67)]. The incidence of all CNS tumours and of each type of specified tumour did not vary substantially by the duration of use of HRT, by the hormonal constituent or by the mode of administration for current use of either oestrogen-only or oestrogen–progestagen HRT, although numbers were small in most groups (Table 3).

Figure 1.

Relative risks (RRs) and 95% confidence intervals (CIs) for central nervous system (CNS) tumours in current users of oestrogen-only and oestrogen–progestagen hormone replacement therapy (HRT) compared to never users, by tumour type.

Relative risks (*) are for current versus never users of HRT stratified by quintiles of socioeconomic status and region of residence, and adjusted for height and body mass index.

Table 3. Relative risks (RR) and 95% confidence intervals (CI) for incident tumours of the central nervous system (CNS) in current users of oestrogen-only and oestrogen-progestagen hormone replacement therapy (HRT) compared to never users by duration of use, constituent hormones and mode of administration
inline image

Further adjustment for age at menopause, smoking status, alcohol intake, strenuous exercise, age at first birth, parity and oral contraceptive use did not appreciably modify the HRT-associated risks for all CNS tumours or for any of the specified tumour types.


In this large cohort study, 1,266 incident tumours occurred among 1.1 million post-menopausal women in the United Kingdom, followed for an average of 5.3 years each. We found that the incidence of CNS tumours was slightly increased in the current users of HRT when compared to never users, with the excess risk largely confined to the users of oestrogen-only HRT. In analyses by tumour type, the pattern of risk associated with the use of HRT did not differ significantly between tumours specified as glioma, meningioma and acoustic neuroma. Among current users of oestrogen-only and oestrogen–progestagen HRT, there was no statistically significant variation in risk by the duration of use of HRT, by the hormonal constituent or by the mode of administration for all CNS tumours and by tumour subtype.

It has been suggested that sex hormones may influence the development of CNS tumours; however, the evidence is weak and largely indirect.2–5, 15, 23 Oestrogen, progesterone and androgen receptors have been shown to be expressed in both meningioma and glioma tumours24–26; however, the significance of these findings is unclear.

As tumours of the CNS are rare, the majority of studies investigating risk factors, including the use of HRT, have been retrospective. The potential for misclassification and differential reporting of HRT use, a problem in all retrospective studies, is particularly strong for CNS tumours, as the tumour itself may affect memory, and the poor prognosis of CNS tumours means that the use of proxy respondents is common (e.g., proxy information was used for more than one-third of the cases in one study10). The majority of retrospective studies have shown no association between HRT use and the incidence of glioma9, 11, 12 or of meningioma5, 11, 14, 15; one study reported an increased risk of meningioma in ever users of HRT compared to never users,12 and another reported a decreased risk.10

We identified two previous prospective studies that had examined the relation between the use of HRT and the incidence of CNS tumours. Both had small numbers of cases, and analyses by the type of HRT and the mode of its administration were not presented. One study included 69 cases of meningioma and reported a RR of 1.85 (95% CI: 1.07–3.24) for current users compared to never users of HRT.7 The other prospective study included 59 cases of glioma and reported a RR of 0.92 (95% CI: 0.55–1.56) for ever users of HRT compared to never users.8 Both of these findings are consistent with our results.

The main strengths of this investigation are the prospective study design, with complete and non-differential ascertainment of cancer incidence during follow-up as all participants were routinely linked to the NHS Central Registers; details of each tumour were coded before notification to the study investigators. Furthermore, the women themselves reported HRT use before the diagnosis of a CNS tumour, whereas in retrospective studies, HRT use is reported after diagnosis and often by proxy responders who may not know details of the woman's HRT use. In addition, the large study size allows separate analyses for the main CNS tumour types. Despite this, numbers of tumours in some subgroups were limited. Potential confounding was minimised by adjusting for age, region of residence, socioeconomic group, BMI and height.

Although exposure data was self-reported, we found excellent agreement in this cohort between self-reported use of HRT and general practitioner prescriptions.27 Misclassification of HRT use is possible due to changing habits over time, but by updating HRT use approximately 3 years after recruitment for 65% of women in the cohort and censoring HRT use 48 months after last contact, misclassification should be minimised. Any residual misclassification of HRT use would slightly dilute the estimated effect, but should not produce spurious associations.

Although our findings indicate that the current use of oestrogen-only HRT may increase the incidence of all CNS tumours and of each tumour subtype (glioma, meningioma and acoustic neuroma), tumours of the CNS are rare. If confirmed, this association would have limited public health impact in the context of other known effects of HRT.6 However, these findings may contribute towards a better understanding of the mechanisms of carcinogenesis in the CNS.


We thank all of the women who participated in the Million Women Study.