IJC International Journal of Cancer
Menopausal hormone therapy and central nervous system tumor risk: Large UK prospective study and meta-analysis Victoria S. Benson, Oksana Kirichek, Valerie Beral and Jane Green Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Richard Doll Building, Roosevelt Drive, Oxford OX3 7LF, United Kingdom
Use of some types of hormone therapy (HT) for the menopause has been associated with an increased risk of cancers of the breast, ovary, and endometrium and decreased risk of cancers of the gastrointestinal tract.1,2 Some, but not all, have reported an increase in the risk of certain central nervous system (CNS) tumors in HT users.3–13 Evidence is limited by the lack of systematic reporting of findings for all CNS tumors combined, as well as by different tumor subtypes and by specific HT preparations. The findings for all CNS tumors combined are of public health relevance. Reliable assessment of the Key words: estrogen, progestin, glioma, meningioma, central nervous system tumors, menopause, hormone therapy, meta-analysis Additional Supporting Information may be found in the online version of this article. Grant sponsor: Cancer Research UK; Grant number: C570/A16491; Grant sponsor: UK Medical Research Council; Grant number: MR/ K02700X DOI: 10.1002/ijc.29274 History: Received 28 Apr 2014; Accepted 1 Oct 2014; Online 21 Oct 2014 Correspondence to: Dr. Jane Green, Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Richard Doll Building, Roosevelt Drive, Oxford, OX3 7LF, UK, E-mail: [email protected], Tel.: 144 (0) 1865 289659, Fax: 144 (0) 1865 289610
C 2014 UICC Int. J. Cancer: 00, 00–00 (2014) V
association between HT use and CNS tumors requires careful control of potential sources of appreciable bias, such as from selective participation or recall of HT use within studies. Differential reporting of HT use in studies where information was recorded retrospectively, i.e., after women had been diagnosed with CNS tumors, is an important source of such bias. To study the association between use of different types of HT and the risk of CNS tumors we used prospectively recorded prescribing information for HT in a case-control study nested within the population-based UK General Practice Research Database (GPRD). We studied all CNS tumors combined and separately tumors specified as glioma, meningioma, acoustic neuroma and pituitary tumors. We also conducted a meta-analysis combining our findings with those from other published studies with prospectively collected information on HT use.
Materials and Methods The General Practice Research Database
The GPRD was established in 1987 and is a computerised database containing linked anonymised patient records for about 10 million people in the UK registered with a National Health Service (NHS) primary care physician [general practitioner (GP)]. All GP consultations, prescriptions issued by the GP, test results and diagnoses from primary and secondary care, referrals to outpatients clinics, hospital admissions
Epidemiology
Female sex hormones are thought to affect women’s risk of developing central nervous system (CNS) tumors. Some have reported an increased risk in users of menopausal hormone therapy (HT) but evidence is limited. In the UK General Practice Research Database we compared prospectively collected information on HT prescriptions in women aged 50–79 years with CNS tumors diagnosed in 1987–2011 with that in matched controls (four per case). Relative risks (RRs) in relation to prescribed HT were calculated overall and by CNS tumor subtype. Statistical tests are two-sided. For all CNS tumors (n 5 3,500), glioma (n 5 689), meningioma (n 5 1,197), acoustic neuroma (n 5 439), and pituitary tumors (n 5 273) adjusted RRs for women prescribed HT versus not were, respectively, 1.21 (95% confidence intervals (CI) 5 1.10–1.32, p < 0.0001), 1.14 (0.93– 1.40, p 5 0.2), 1.30 (1.11–1.51, p 5 0.001), 1.37 (1.06–1.75, p 5 0.01), and 1.35 (0.99–1.85, p 5 0.06). There was no significant difference in risk by tumor subtype (pheterogeneity 5 0.6). A meta-analysis was conducted, combining our results with those from other published studies with prospectively collected exposure information. The meta-analyses yielded significantly increased risks for all CNS tumors, glioma and meningioma in users of estrogen-only [1.35 (1.22–1.49), 1.23 (1.06–1.42) and 1.31 (1.20–1.43), respectively] but not estrogen-progestin HT [1.09 (0.99–1.19), 0.92 (0.78–1.08) and 1.05 (0.95–1.16), respectively]; these differences were statistically significant (p < 0.005 for each tumor type). There was no significant difference between glioma and meningioma risk in users of estrogen-only HT. The totality of the available evidence suggests an increased risk of all CNS tumors (and of glioma and meningioma separately) in users of estrogen-only HT. Absolute excess risk (2 per 10,000 users over 5 years) is small.
2
Menopausal hormone therapy and central nervous system tumor risk
Epidemiology
What’s new? Hormone therapy can ameliorate the symptoms of menopause, but may also increase risk of various cancers. This study examined whether hormone therapy affects a woman’s risk of central nervous system tumors. The authors plumbed data from the UK (GPRD) to compare how often women with CNS tumors had been prescribed hormone therapy versus age-matched controls and conducted the first meta-analysis of CNS tumor risk by hormone therapy type. Overall, women taking estrogen-only therapy had an increased risk of all CNS tumors, and of glioma and meningioma; women taking estrogen-progestin therapy showed no increase in risk.
and deaths are automatically coded by the GP and entered onto the GPRD. For each individual, basic demographic and some lifestyle data are also recorded. Information on prescriptions and clinical diagnosis has been shown to be reliable.14–16 Site-specific cancers, but not tumor histology, have been validated. Patients are recorded as entering the GPRD when they are registered with a participating GP (and the research-active start date is this date or, if later, the date at which the participating practice is declared up to standard), and leave the database when they move to a nonparticipating GP, leave the NHS (e.g., emigrate) or die. The database thus consists of longitudinal medical records with varying lengths of research-active follow-up. Cases were identified from records between January 1, 1987 and August 31, 2011 as any eligible female patient aged 50–79 years with an incident primary malignant or benign tumor of the CNS, using the GPRD codes supplied. The tumors identified were further classified as glioma, meningioma, acoustic neuroma or pituitary tumors. Controls were eligible for inclusion if they had no record of a CNS tumor and if their follow-up within the GPRD included the start and end dates of the observation period of their matched case. For all analyses, the observation period of the controls was set to match that of the cases exactly. Controls were matched to cases in a ratio of up to 4:1 by year of birth (1/21 year), general practice, and observation period. Cases and their matched controls were required to have at least one year of follow-up prior to the index date of tumor diagnosis. The index date for cases was defined as the date of the first CNS tumor diagnosis in the patient record. The observation period for cases is the time from the start of the research-active record to the index date. Women were defined as exposed to HT if they had a record within the observation period of at least one (11) prescription for HT, as classified by the British National Formulary: Sections 6.4.1.1 (estrogens for HT, including tibolone and excluding raloxifene) and 6.4.1.2 (progestins). Women who had a prescription for HT within the 12 months prior to the index date were classified as having a current HT prescription. Duration of use of HT was estimated as the time between the first and last HT prescription during the observation period, as fewer than 10% of GPRD records include the number of months that HT was prescribed. For some analyses, HT was subdivided by the main HT prescriptions estrogen-only or estrogen and progestin.
Statistical analysis
Conditional logistic regression was used to calculate relative risks (RRs) and 95% confidence intervals (CIs) for incidence of all CNS tumors and for each tumor subtype in relation to HT prescriptions. All analyses were conditioned on the matching factors (year of birth, GP practice, and observation period) and were adjusted for smoking status (latest record before index date: never, past, current), alcohol intake (latest record before index date: nondrinker, drinker), body mass index (latest record at least 12 months before index date:
Menopausal hormone therapy and central nervous system tumor risk: Large UK prospective study and meta-analysis Victoria S. Benson, Oksana Kirichek, Valerie Beral and Jane Green Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Richard Doll Building, Roosevelt Drive, Oxford OX3 7LF, United Kingdom
Use of some types of hormone therapy (HT) for the menopause has been associated with an increased risk of cancers of the breast, ovary, and endometrium and decreased risk of cancers of the gastrointestinal tract.1,2 Some, but not all, have reported an increase in the risk of certain central nervous system (CNS) tumors in HT users.3–13 Evidence is limited by the lack of systematic reporting of findings for all CNS tumors combined, as well as by different tumor subtypes and by specific HT preparations. The findings for all CNS tumors combined are of public health relevance. Reliable assessment of the Key words: estrogen, progestin, glioma, meningioma, central nervous system tumors, menopause, hormone therapy, meta-analysis Additional Supporting Information may be found in the online version of this article. Grant sponsor: Cancer Research UK; Grant number: C570/A16491; Grant sponsor: UK Medical Research Council; Grant number: MR/ K02700X DOI: 10.1002/ijc.29274 History: Received 28 Apr 2014; Accepted 1 Oct 2014; Online 21 Oct 2014 Correspondence to: Dr. Jane Green, Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Richard Doll Building, Roosevelt Drive, Oxford, OX3 7LF, UK, E-mail: [email protected], Tel.: 144 (0) 1865 289659, Fax: 144 (0) 1865 289610
C 2014 UICC Int. J. Cancer: 00, 00–00 (2014) V
association between HT use and CNS tumors requires careful control of potential sources of appreciable bias, such as from selective participation or recall of HT use within studies. Differential reporting of HT use in studies where information was recorded retrospectively, i.e., after women had been diagnosed with CNS tumors, is an important source of such bias. To study the association between use of different types of HT and the risk of CNS tumors we used prospectively recorded prescribing information for HT in a case-control study nested within the population-based UK General Practice Research Database (GPRD). We studied all CNS tumors combined and separately tumors specified as glioma, meningioma, acoustic neuroma and pituitary tumors. We also conducted a meta-analysis combining our findings with those from other published studies with prospectively collected information on HT use.
Materials and Methods The General Practice Research Database
The GPRD was established in 1987 and is a computerised database containing linked anonymised patient records for about 10 million people in the UK registered with a National Health Service (NHS) primary care physician [general practitioner (GP)]. All GP consultations, prescriptions issued by the GP, test results and diagnoses from primary and secondary care, referrals to outpatients clinics, hospital admissions
Epidemiology
Female sex hormones are thought to affect women’s risk of developing central nervous system (CNS) tumors. Some have reported an increased risk in users of menopausal hormone therapy (HT) but evidence is limited. In the UK General Practice Research Database we compared prospectively collected information on HT prescriptions in women aged 50–79 years with CNS tumors diagnosed in 1987–2011 with that in matched controls (four per case). Relative risks (RRs) in relation to prescribed HT were calculated overall and by CNS tumor subtype. Statistical tests are two-sided. For all CNS tumors (n 5 3,500), glioma (n 5 689), meningioma (n 5 1,197), acoustic neuroma (n 5 439), and pituitary tumors (n 5 273) adjusted RRs for women prescribed HT versus not were, respectively, 1.21 (95% confidence intervals (CI) 5 1.10–1.32, p < 0.0001), 1.14 (0.93– 1.40, p 5 0.2), 1.30 (1.11–1.51, p 5 0.001), 1.37 (1.06–1.75, p 5 0.01), and 1.35 (0.99–1.85, p 5 0.06). There was no significant difference in risk by tumor subtype (pheterogeneity 5 0.6). A meta-analysis was conducted, combining our results with those from other published studies with prospectively collected exposure information. The meta-analyses yielded significantly increased risks for all CNS tumors, glioma and meningioma in users of estrogen-only [1.35 (1.22–1.49), 1.23 (1.06–1.42) and 1.31 (1.20–1.43), respectively] but not estrogen-progestin HT [1.09 (0.99–1.19), 0.92 (0.78–1.08) and 1.05 (0.95–1.16), respectively]; these differences were statistically significant (p < 0.005 for each tumor type). There was no significant difference between glioma and meningioma risk in users of estrogen-only HT. The totality of the available evidence suggests an increased risk of all CNS tumors (and of glioma and meningioma separately) in users of estrogen-only HT. Absolute excess risk (2 per 10,000 users over 5 years) is small.
2
Menopausal hormone therapy and central nervous system tumor risk
Epidemiology
What’s new? Hormone therapy can ameliorate the symptoms of menopause, but may also increase risk of various cancers. This study examined whether hormone therapy affects a woman’s risk of central nervous system tumors. The authors plumbed data from the UK (GPRD) to compare how often women with CNS tumors had been prescribed hormone therapy versus age-matched controls and conducted the first meta-analysis of CNS tumor risk by hormone therapy type. Overall, women taking estrogen-only therapy had an increased risk of all CNS tumors, and of glioma and meningioma; women taking estrogen-progestin therapy showed no increase in risk.
and deaths are automatically coded by the GP and entered onto the GPRD. For each individual, basic demographic and some lifestyle data are also recorded. Information on prescriptions and clinical diagnosis has been shown to be reliable.14–16 Site-specific cancers, but not tumor histology, have been validated. Patients are recorded as entering the GPRD when they are registered with a participating GP (and the research-active start date is this date or, if later, the date at which the participating practice is declared up to standard), and leave the database when they move to a nonparticipating GP, leave the NHS (e.g., emigrate) or die. The database thus consists of longitudinal medical records with varying lengths of research-active follow-up. Cases were identified from records between January 1, 1987 and August 31, 2011 as any eligible female patient aged 50–79 years with an incident primary malignant or benign tumor of the CNS, using the GPRD codes supplied. The tumors identified were further classified as glioma, meningioma, acoustic neuroma or pituitary tumors. Controls were eligible for inclusion if they had no record of a CNS tumor and if their follow-up within the GPRD included the start and end dates of the observation period of their matched case. For all analyses, the observation period of the controls was set to match that of the cases exactly. Controls were matched to cases in a ratio of up to 4:1 by year of birth (1/21 year), general practice, and observation period. Cases and their matched controls were required to have at least one year of follow-up prior to the index date of tumor diagnosis. The index date for cases was defined as the date of the first CNS tumor diagnosis in the patient record. The observation period for cases is the time from the start of the research-active record to the index date. Women were defined as exposed to HT if they had a record within the observation period of at least one (11) prescription for HT, as classified by the British National Formulary: Sections 6.4.1.1 (estrogens for HT, including tibolone and excluding raloxifene) and 6.4.1.2 (progestins). Women who had a prescription for HT within the 12 months prior to the index date were classified as having a current HT prescription. Duration of use of HT was estimated as the time between the first and last HT prescription during the observation period, as fewer than 10% of GPRD records include the number of months that HT was prescribed. For some analyses, HT was subdivided by the main HT prescriptions estrogen-only or estrogen and progestin.
Statistical analysis
Conditional logistic regression was used to calculate relative risks (RRs) and 95% confidence intervals (CIs) for incidence of all CNS tumors and for each tumor subtype in relation to HT prescriptions. All analyses were conditioned on the matching factors (year of birth, GP practice, and observation period) and were adjusted for smoking status (latest record before index date: never, past, current), alcohol intake (latest record before index date: nondrinker, drinker), body mass index (latest record at least 12 months before index date:
