Pharmacoepidemiology
Drug Safety 5 (6): 436-446, 1990 0114-5916/90/00 11-0436/$05.50/0 © Adis International Limited All rights reserved. MEDT03313
The Relationship Between Oral Contraceptive Use, Cancer and Vascular Disease Carlo La Vecchia, Silvia Franceschi, Paolo Bruzzi, Fabio Parazzini and Peter Boyle Istituto di Ricerche Farmacologiche 'Mario Negri', Milan, Italy, Institute of Social and Preventive Medicine, University of Lausanne, Switzerland, Servizio di Epidemiologia, Centro di Riferimento Oncologico, Aviano, Italy, Department of Clinical Epidemiology, Istituto Nazionale per la Ricerca sui Cancro, Genoa, Italy, and Unit of Analytical Epidemiology, International Agency for Research on Cancer, Lyons, France
Contents
Summary .................................................................................................................................... 436 1. Incidence Of and Mortality From Major Groups of Diseases ............................ ............. .437 1.1 Vascular Diseases ............................................................................................................ 438 1.2 Breast Cancer .................................................................................................................. 439 1.3 Cervical Neoplasia .......................................................................................................... 440 1.4 Endometrial Cancer ...... .................................................... .. ............................................ 441 1.5 Ovarian Cancer ............................................................................................................... 441 1.6 Primary Liver Cancer ..................................................................................................... 441 1.7 Cutaneous Malignant Melanoma .................................................................................. .443 2. Changes in Oral Contraceptive Formulations ................................................................... .443 3. Conclusions ............................................................................................................................444
Summary
An overview of the available epidemiological evidence on the connection between oral contraceptives and cancer or vascular disease is presented, including the observation that epidemiological studies have produced important indications for changing both the composition and prescription patterns of oral contraceptives to avoid a large proportion of vascular side effects. Further, the evidence is remarkably clear and consistent in .relation to the elevated risks of cervical neoplasms and, although based on a limited number of small studies, of primary liver cancer; protection against endometrial and ovarian cancers up to middle age; and the absence of association with malignant melanoma. There are still uncertainties regarding breast cancer, mainly related to the role of time factors, and the potential persisting risk related to long term use at younger age: published studies, in fact, show elevated risks for long term use in women below age 35 or perhaps up to age 45, but no evidence of association in middle age. Since breast cancer and ovarian cancer account for most of the mortality burden in women up to age 50 or 55 in developed countries, a clarification of the risk relationship for these 2 neoplasms will determine most of the quantitative evaluation on positive or adverse effects of oral contraceptives. The impact of other neoplasms and of cardiovascular disease, on the basis of current oral contraceptive composition, is comparatively limited, if not negligible.
437
Cancer, Vascular Disease and Oral Contraceptives
The adverse health risks related to oral contraceptive use have been widely publicised, and often have a great emotional impact on both the medical profession and, more particularly, the general public. It is important, however, that the issue be viewed in perspective: in terms not only of relative risks and epidemiological inferences, but also of absolute risks and relevance to public health. Thus, critical attention should certainly be paid to criteria for the definition of causality and strength of the associations, but equally important is the incidence of each disease in the population. Furthermore, a key element of both the evaluation and the uncertainty is the temporal factors in the risk relationship for each disease, i.e. the relation to age at first use, duration of use and, in particular, time since first and last use. Time since first use provides information on the length of time which may be necessary before any effect becomes apparent, whereas time since last use is important in showing how long the (adverse) effect is likely to persist. These temporal factors are reasonably well defined for vascular disease but still largely undetermined for other conditions, particularly neoplastic diseases. The overall risk-benefit evaluation of oral contraceptives may change substantially depending on which assumption is made about these factors.
1. Incidence Of and Mortality From Major Groups of Diseases Information on total mortality and death rates from 3 major groups of diseases is provided by the 3 large cohort studies of oral contraceptives and health (Ramcharan et al. 1981; RCGP 1974; Vessey et al. 1976). In these populations of young and middle-aged women, cancer mortality accounted for approximately 50% of all deaths, while diseases of the circulatory system made up only 10 to 20% of deaths among women who had never used oral contraceptives (table I). In these studies, a significant excess mortality from vascular diseases was observed among oral contraceptive users [overall relative risk (RR) = 2.5]. This was the principal, or in 2 studies the sole
Table I. Mortality from major causes among oral contraceptive users in cohort studies (after Prentice & Thomas 1987) Study
Death rate8
Total mortality RCGpb Oxford-FPAc Walnut Creekd
0.64 0.62 1.53
Valcula, dllealel RCGP Oxford-FPA Walnut Creek
0.07 0.05 0.33
Cancers RCGP Oxford-FPA Walnut Creek
0.32 0.39 0.69
All other mortality RCGP Oxford-FPA Walnut Creek
0.24 0.17 0.44
RR
"! ! "! "! 0.9 1.1
~2 2.5
0.8
0.6 1.4
0.9 1.2
CRR(CI)
1.2 (1.0-1.4)
2.5 (1.6-4.0)
1.0 (0.8-1-.3)
1.1 (0.8-1.5)
a
Rate per 1000 woman-years among women who have never used oral contraceptives. b Royal College of General Practitioners (1974). c Vessey et al. (1976). d Ramcharan et al. (1981). Abbreviations: RR = relative risk estlmata for oral contraceptive users; CRR (CI) = combined relative risk (95% confidence 'Interval given in parentheses).
cause, of the approximately 20% increase (of borderline statistical significance) in total mortality in pill users compared with nonusers. The pooled relative risk estimate for total cancer mortality was 1.0, with a 95% confidence interval between 0.8 and 1.3. Studies conducted in the 1970s (Ramcharan etal. 1981: RCGP 1974; Vessey et al. 1976} were thus able to show significant associations between oral contraceptives (in the preparations and pattern of use of that period) and vascular disease~, particularly ischaemic heart diseases, while they were largely reassuring with reference 10 cancer. However, it is clear that, in terms of absOlute risk in the population, the impotjance of cancer can be substantially greater than that of vascular diseases, since cancer is a much more frequent cause of dis-
Drug Safety 5 (6) 1990
438
Table II. Incidence of vascular disease and oral contraceptives in cohort studies (after Prentice & Thomas 1987) Study
Ischaemic heart disease RCGpb Walnut Creekc
Incidence rate8
RR
CRR(CI)
0.54 0.79
1.4 1.4
1.4 (1.0-1.9)
Fatal Ischaemlc heart disease RCGP 0.02 Oxford-FPAd 0.02
2.0 5.78
Acute myocardial Infarction 0.15 RCGP Walnut Creek 0.23
2.0 1.1
1.9 (1.1-3.3)
All cerebrovascular disease 0.20 RCGP Oxford-FPA 0.18 Walnut Creek 0.28
3.1 2.2 3.4
2.9 (2.0-4.1)
a
Rate per 1000 woman-years among women who have never used oral contraceptives. b Royal College of General Practitioners (1974). c Ramcharan et al. (1981). d Vessey et al. (1976). e Current plus ex-users. Abbreviations; see table I.
ease and death in younger women, and even small changes in the relative risk would have a relevant impact on the overall risk/benefit evaluation. Among female hormone-related neoplasms, furthermore, 2 sites (breast and ovary) account for most of the mortality burden in young and middle.aged women in developed countries. The riskrelationship for the 2 neoplasms, therefore, will have a substantial impact on the overall evaluation ofthe positive or adverse effects of oral contraceptives on mortality. 1.1 Vascular Diseases
Table II summarises the incidence of ischaemic heart disease and cerebrovascular disease among women who have never used oral contraceptives, and the corresponding relative risks in current users, derived from cohort studies (Ramcharan et al. 1981; RCGP 1974; Vessey et al. 1976). Overall, the relative risk was increased about 2-fold in current oral
contraceptive users both for cardiovascular and cerebrovascular diseases, and a similar increased ratio was observed in relation to mortality from ischaemic heart disease. Although the issue has been somewhat controversial in the past (Slone et al. 1981), there now appears to be convincing evidence that the elevated risk is restricted to current users, and no material association is observed in ex-users. In the American Nurses' Health Study cohort (Stampfer et al. 1988), the relative risks for past users were 0.8 for major coronary disease, 1.0 for stroke and 0.9 for all cardiovascular cases, and there was no evidence of a trend in the risk with duration or time since last use. Case-control studies of myocardial intarction (Jick et al. 1978; Mann et al. 1976; Shapiro et al. 1979; Vessey & Doll 1968) have produced comparable results and, being based on substantially greater numbers of cases, have permitted more detailed analysis of interactions with other risk factors, such as smoking, hypertension or hyperlipidaemia. The overall evidence indicates that the role of oral contraceptives is apparently independent from that of other factors, i.e. mUltiplicative on the relative risk. Thus, the relative risk (although clearly not the absolute excess risk) was similar in subsequent age groups. For instance, the relative risk (calculated from published data) from a large multicentre American study (Shapiro et al. 1979) was 4.1 under age 40 versus 3.8 at age 40 to 49. In the same study, the relative risk rose to 39 in oral contraceptive users/heavy smokers compared with nonsmokers/non-oral contraceptive users, thus indicating a multiplicative interaction on the relative risk. Such grossly elevated risk estimates are clearly of major concern for women exposed simultaneously to both risk factors but, on the other hand, they offer extremely important indications for prevention since, in those women, avoidance of at least 1 of the risk factors would lead to a substantial reduction of risk. In terms of practical implications for prescription, this indicates that the avoidance of oral contraceptive use in women above age 35 or 40 or/and at high risk of ischaemic heart disease for other reasons would
Cancer, Vascular Disease and Oral Contraceptives
avoid, and has indeed avoided (Stadel 1981) most of the excess risk of cardiovascular diseases associated with oral contraceptives. From the point of view of public health, these indications for selective prescription, based on earlier epidemiological findings, have probably been even more important for reducing vascular risk than the changes in composition (in particular the reduced oestrogen content) of newer formulations with their reduced steroid hormone levels. It is possible, indeed, that the paucity in epidemiological data on newer pills and cardiovascular events is, at least in part, due to the decreased prevalence of vascular events related to oral contraceptives, and to the consequently greater difficulties in conducting studies of acceptable statistical power. Thus, knowledge produced by earlier epidemiological research has certainly contributed to a substantial reduction in the frequency of cardiovascular adverse effects of oral contraceptives. 1.2 Breast Cancer Despite the large number of published studies, the debate on the possible association between oral contraceptives and breast cancer is far from settled. The overall evidence to date is reassuring: ecological comparisons and temporal trends indicate that there is as yet no consistent evidence that breast cancer rates have been substantially influenced in any age group by the widespread use of oral contraceptives in Western countries. Mortality is stable or declining in younger women in several countries, but death rates may reflect both risk and prognostic implications of oral contraceptives, as well as the effect of other risk factors or interventions, on breast cancer rates as well. Most casecontrol and cohort studies, as well as a formal overview Wrentice & Thomas 1987) based on 16 studies and over 12 000 cases, indicate that the risk of breast cancer is not higher among women who have used oral contraceptives than in those who have not. The summary RR was 1.0, with a very narrow confidence interval (0.9 to 1.1), and no significant heterogeneity between studies. There is, however, a subgroup of women in
439
whom positive association between oral contraceptives and breast cancer cannot easily be dismissed as due to chance or bias alone. These are younger (i.e. below age 35 or perhaps up to age 45), long term oral contraceptive users, among whom all of the studies included in table III (16 of 16) found relative risks above unity (Jick et al. 1989; Kay & Hannaford 1988; La Vecchia et aI. 1986; Lubin et al. 1982; McPherson et al. 1987; Meirik et aI: 1986; Miller et aI. 1989; Olsson et al. 1989; Paul et aI. 1986; Pike et aI. 1983; Ravnihar et al. 1988; Romieu et al. 1989; Rosenberg et al. 1984; Stadel et al. 1985; Stanford et al. 1989; UK National Case-Control Study Group 1989). This has led to 2 different types of speculations: (1) that long term (e.g. >5 years) use of oral contraceptives at early ages induces unfavourable modifications in the breast, or (2). that the impact of these modifications (or, more generally, of oral contraceptives on breast carcinogenesis) may become appreciable only after a long 'latent period' and, hence, that we are now observing only the start of a future oral contraceptiveinduced breast cancer epidemic. At present, however, this is essentially a theoretical speculation, with many others being equally plausible. For instance, the observation of vital statistics and subsequent analyses of data from prospective and retrospective studies have shown that the effect of parity on breast carcinogenesis is different in different age groups: below age 40, parous women have a higher breast cancer risk than the multiparous, while the reverse is true in older age groups. Indeed, it has been shown that a full-term pregnancy is followed by a transient increase in the risk of breast cancer, which for some time counteracts the eventually favourable effect of pregnancy (Bruzzi et al. 1988). Such a dual effect may also pertain to oral contraceptives. Besides pharmacological and biological considerations, there is direct epidemiological evidence that the effect of oral contraceptives on various female hormone-related neoplasms is similar to that of pregnancy: in fact, these drugs lower the risk of ovarian and endometrial cancer, and probably increase that of cervical cancer (see below). Along the same lines, the elevated breast cancer risk for
Drug Safety 5 (6) 1990
440
Table III. Long term oral contraceptive use and breast cancer risk in young women from selected studies Reference
Age group
RR
Lubin et al. (1982) Pike et al. (1983) Rosenberg et al. (1984) Stadel et al. (1985) Meirik et al. (1986) La Vecchia et al. (1986) Paul et al. (1986) McPherson et al. (1987) Ravnihar et al. (1988) Kay & Hannaford (1988) Miller et al. (1989) Jick et al. (1989) UK National CaseControl Study Group (1989) Olsson et al. (1989) Romieu et al. (1989)
30-44
Drug Safety 5 (6): 436-446, 1990 0114-5916/90/00 11-0436/$05.50/0 © Adis International Limited All rights reserved. MEDT03313
The Relationship Between Oral Contraceptive Use, Cancer and Vascular Disease Carlo La Vecchia, Silvia Franceschi, Paolo Bruzzi, Fabio Parazzini and Peter Boyle Istituto di Ricerche Farmacologiche 'Mario Negri', Milan, Italy, Institute of Social and Preventive Medicine, University of Lausanne, Switzerland, Servizio di Epidemiologia, Centro di Riferimento Oncologico, Aviano, Italy, Department of Clinical Epidemiology, Istituto Nazionale per la Ricerca sui Cancro, Genoa, Italy, and Unit of Analytical Epidemiology, International Agency for Research on Cancer, Lyons, France
Contents
Summary .................................................................................................................................... 436 1. Incidence Of and Mortality From Major Groups of Diseases ............................ ............. .437 1.1 Vascular Diseases ............................................................................................................ 438 1.2 Breast Cancer .................................................................................................................. 439 1.3 Cervical Neoplasia .......................................................................................................... 440 1.4 Endometrial Cancer ...... .................................................... .. ............................................ 441 1.5 Ovarian Cancer ............................................................................................................... 441 1.6 Primary Liver Cancer ..................................................................................................... 441 1.7 Cutaneous Malignant Melanoma .................................................................................. .443 2. Changes in Oral Contraceptive Formulations ................................................................... .443 3. Conclusions ............................................................................................................................444
Summary
An overview of the available epidemiological evidence on the connection between oral contraceptives and cancer or vascular disease is presented, including the observation that epidemiological studies have produced important indications for changing both the composition and prescription patterns of oral contraceptives to avoid a large proportion of vascular side effects. Further, the evidence is remarkably clear and consistent in .relation to the elevated risks of cervical neoplasms and, although based on a limited number of small studies, of primary liver cancer; protection against endometrial and ovarian cancers up to middle age; and the absence of association with malignant melanoma. There are still uncertainties regarding breast cancer, mainly related to the role of time factors, and the potential persisting risk related to long term use at younger age: published studies, in fact, show elevated risks for long term use in women below age 35 or perhaps up to age 45, but no evidence of association in middle age. Since breast cancer and ovarian cancer account for most of the mortality burden in women up to age 50 or 55 in developed countries, a clarification of the risk relationship for these 2 neoplasms will determine most of the quantitative evaluation on positive or adverse effects of oral contraceptives. The impact of other neoplasms and of cardiovascular disease, on the basis of current oral contraceptive composition, is comparatively limited, if not negligible.
437
Cancer, Vascular Disease and Oral Contraceptives
The adverse health risks related to oral contraceptive use have been widely publicised, and often have a great emotional impact on both the medical profession and, more particularly, the general public. It is important, however, that the issue be viewed in perspective: in terms not only of relative risks and epidemiological inferences, but also of absolute risks and relevance to public health. Thus, critical attention should certainly be paid to criteria for the definition of causality and strength of the associations, but equally important is the incidence of each disease in the population. Furthermore, a key element of both the evaluation and the uncertainty is the temporal factors in the risk relationship for each disease, i.e. the relation to age at first use, duration of use and, in particular, time since first and last use. Time since first use provides information on the length of time which may be necessary before any effect becomes apparent, whereas time since last use is important in showing how long the (adverse) effect is likely to persist. These temporal factors are reasonably well defined for vascular disease but still largely undetermined for other conditions, particularly neoplastic diseases. The overall risk-benefit evaluation of oral contraceptives may change substantially depending on which assumption is made about these factors.
1. Incidence Of and Mortality From Major Groups of Diseases Information on total mortality and death rates from 3 major groups of diseases is provided by the 3 large cohort studies of oral contraceptives and health (Ramcharan et al. 1981; RCGP 1974; Vessey et al. 1976). In these populations of young and middle-aged women, cancer mortality accounted for approximately 50% of all deaths, while diseases of the circulatory system made up only 10 to 20% of deaths among women who had never used oral contraceptives (table I). In these studies, a significant excess mortality from vascular diseases was observed among oral contraceptive users [overall relative risk (RR) = 2.5]. This was the principal, or in 2 studies the sole
Table I. Mortality from major causes among oral contraceptive users in cohort studies (after Prentice & Thomas 1987) Study
Death rate8
Total mortality RCGpb Oxford-FPAc Walnut Creekd
0.64 0.62 1.53
Valcula, dllealel RCGP Oxford-FPA Walnut Creek
0.07 0.05 0.33
Cancers RCGP Oxford-FPA Walnut Creek
0.32 0.39 0.69
All other mortality RCGP Oxford-FPA Walnut Creek
0.24 0.17 0.44
RR
"! ! "! "! 0.9 1.1
~2 2.5
0.8
0.6 1.4
0.9 1.2
CRR(CI)
1.2 (1.0-1.4)
2.5 (1.6-4.0)
1.0 (0.8-1-.3)
1.1 (0.8-1.5)
a
Rate per 1000 woman-years among women who have never used oral contraceptives. b Royal College of General Practitioners (1974). c Vessey et al. (1976). d Ramcharan et al. (1981). Abbreviations: RR = relative risk estlmata for oral contraceptive users; CRR (CI) = combined relative risk (95% confidence 'Interval given in parentheses).
cause, of the approximately 20% increase (of borderline statistical significance) in total mortality in pill users compared with nonusers. The pooled relative risk estimate for total cancer mortality was 1.0, with a 95% confidence interval between 0.8 and 1.3. Studies conducted in the 1970s (Ramcharan etal. 1981: RCGP 1974; Vessey et al. 1976} were thus able to show significant associations between oral contraceptives (in the preparations and pattern of use of that period) and vascular disease~, particularly ischaemic heart diseases, while they were largely reassuring with reference 10 cancer. However, it is clear that, in terms of absOlute risk in the population, the impotjance of cancer can be substantially greater than that of vascular diseases, since cancer is a much more frequent cause of dis-
Drug Safety 5 (6) 1990
438
Table II. Incidence of vascular disease and oral contraceptives in cohort studies (after Prentice & Thomas 1987) Study
Ischaemic heart disease RCGpb Walnut Creekc
Incidence rate8
RR
CRR(CI)
0.54 0.79
1.4 1.4
1.4 (1.0-1.9)
Fatal Ischaemlc heart disease RCGP 0.02 Oxford-FPAd 0.02
2.0 5.78
Acute myocardial Infarction 0.15 RCGP Walnut Creek 0.23
2.0 1.1
1.9 (1.1-3.3)
All cerebrovascular disease 0.20 RCGP Oxford-FPA 0.18 Walnut Creek 0.28
3.1 2.2 3.4
2.9 (2.0-4.1)
a
Rate per 1000 woman-years among women who have never used oral contraceptives. b Royal College of General Practitioners (1974). c Ramcharan et al. (1981). d Vessey et al. (1976). e Current plus ex-users. Abbreviations; see table I.
ease and death in younger women, and even small changes in the relative risk would have a relevant impact on the overall risk/benefit evaluation. Among female hormone-related neoplasms, furthermore, 2 sites (breast and ovary) account for most of the mortality burden in young and middle.aged women in developed countries. The riskrelationship for the 2 neoplasms, therefore, will have a substantial impact on the overall evaluation ofthe positive or adverse effects of oral contraceptives on mortality. 1.1 Vascular Diseases
Table II summarises the incidence of ischaemic heart disease and cerebrovascular disease among women who have never used oral contraceptives, and the corresponding relative risks in current users, derived from cohort studies (Ramcharan et al. 1981; RCGP 1974; Vessey et al. 1976). Overall, the relative risk was increased about 2-fold in current oral
contraceptive users both for cardiovascular and cerebrovascular diseases, and a similar increased ratio was observed in relation to mortality from ischaemic heart disease. Although the issue has been somewhat controversial in the past (Slone et al. 1981), there now appears to be convincing evidence that the elevated risk is restricted to current users, and no material association is observed in ex-users. In the American Nurses' Health Study cohort (Stampfer et al. 1988), the relative risks for past users were 0.8 for major coronary disease, 1.0 for stroke and 0.9 for all cardiovascular cases, and there was no evidence of a trend in the risk with duration or time since last use. Case-control studies of myocardial intarction (Jick et al. 1978; Mann et al. 1976; Shapiro et al. 1979; Vessey & Doll 1968) have produced comparable results and, being based on substantially greater numbers of cases, have permitted more detailed analysis of interactions with other risk factors, such as smoking, hypertension or hyperlipidaemia. The overall evidence indicates that the role of oral contraceptives is apparently independent from that of other factors, i.e. mUltiplicative on the relative risk. Thus, the relative risk (although clearly not the absolute excess risk) was similar in subsequent age groups. For instance, the relative risk (calculated from published data) from a large multicentre American study (Shapiro et al. 1979) was 4.1 under age 40 versus 3.8 at age 40 to 49. In the same study, the relative risk rose to 39 in oral contraceptive users/heavy smokers compared with nonsmokers/non-oral contraceptive users, thus indicating a multiplicative interaction on the relative risk. Such grossly elevated risk estimates are clearly of major concern for women exposed simultaneously to both risk factors but, on the other hand, they offer extremely important indications for prevention since, in those women, avoidance of at least 1 of the risk factors would lead to a substantial reduction of risk. In terms of practical implications for prescription, this indicates that the avoidance of oral contraceptive use in women above age 35 or 40 or/and at high risk of ischaemic heart disease for other reasons would
Cancer, Vascular Disease and Oral Contraceptives
avoid, and has indeed avoided (Stadel 1981) most of the excess risk of cardiovascular diseases associated with oral contraceptives. From the point of view of public health, these indications for selective prescription, based on earlier epidemiological findings, have probably been even more important for reducing vascular risk than the changes in composition (in particular the reduced oestrogen content) of newer formulations with their reduced steroid hormone levels. It is possible, indeed, that the paucity in epidemiological data on newer pills and cardiovascular events is, at least in part, due to the decreased prevalence of vascular events related to oral contraceptives, and to the consequently greater difficulties in conducting studies of acceptable statistical power. Thus, knowledge produced by earlier epidemiological research has certainly contributed to a substantial reduction in the frequency of cardiovascular adverse effects of oral contraceptives. 1.2 Breast Cancer Despite the large number of published studies, the debate on the possible association between oral contraceptives and breast cancer is far from settled. The overall evidence to date is reassuring: ecological comparisons and temporal trends indicate that there is as yet no consistent evidence that breast cancer rates have been substantially influenced in any age group by the widespread use of oral contraceptives in Western countries. Mortality is stable or declining in younger women in several countries, but death rates may reflect both risk and prognostic implications of oral contraceptives, as well as the effect of other risk factors or interventions, on breast cancer rates as well. Most casecontrol and cohort studies, as well as a formal overview Wrentice & Thomas 1987) based on 16 studies and over 12 000 cases, indicate that the risk of breast cancer is not higher among women who have used oral contraceptives than in those who have not. The summary RR was 1.0, with a very narrow confidence interval (0.9 to 1.1), and no significant heterogeneity between studies. There is, however, a subgroup of women in
439
whom positive association between oral contraceptives and breast cancer cannot easily be dismissed as due to chance or bias alone. These are younger (i.e. below age 35 or perhaps up to age 45), long term oral contraceptive users, among whom all of the studies included in table III (16 of 16) found relative risks above unity (Jick et al. 1989; Kay & Hannaford 1988; La Vecchia et aI. 1986; Lubin et al. 1982; McPherson et al. 1987; Meirik et aI: 1986; Miller et aI. 1989; Olsson et al. 1989; Paul et aI. 1986; Pike et aI. 1983; Ravnihar et al. 1988; Romieu et al. 1989; Rosenberg et al. 1984; Stadel et al. 1985; Stanford et al. 1989; UK National Case-Control Study Group 1989). This has led to 2 different types of speculations: (1) that long term (e.g. >5 years) use of oral contraceptives at early ages induces unfavourable modifications in the breast, or (2). that the impact of these modifications (or, more generally, of oral contraceptives on breast carcinogenesis) may become appreciable only after a long 'latent period' and, hence, that we are now observing only the start of a future oral contraceptiveinduced breast cancer epidemic. At present, however, this is essentially a theoretical speculation, with many others being equally plausible. For instance, the observation of vital statistics and subsequent analyses of data from prospective and retrospective studies have shown that the effect of parity on breast carcinogenesis is different in different age groups: below age 40, parous women have a higher breast cancer risk than the multiparous, while the reverse is true in older age groups. Indeed, it has been shown that a full-term pregnancy is followed by a transient increase in the risk of breast cancer, which for some time counteracts the eventually favourable effect of pregnancy (Bruzzi et al. 1988). Such a dual effect may also pertain to oral contraceptives. Besides pharmacological and biological considerations, there is direct epidemiological evidence that the effect of oral contraceptives on various female hormone-related neoplasms is similar to that of pregnancy: in fact, these drugs lower the risk of ovarian and endometrial cancer, and probably increase that of cervical cancer (see below). Along the same lines, the elevated breast cancer risk for
Drug Safety 5 (6) 1990
440
Table III. Long term oral contraceptive use and breast cancer risk in young women from selected studies Reference
Age group
RR
Lubin et al. (1982) Pike et al. (1983) Rosenberg et al. (1984) Stadel et al. (1985) Meirik et al. (1986) La Vecchia et al. (1986) Paul et al. (1986) McPherson et al. (1987) Ravnihar et al. (1988) Kay & Hannaford (1988) Miller et al. (1989) Jick et al. (1989) UK National CaseControl Study Group (1989) Olsson et al. (1989) Romieu et al. (1989)
30-44
