1397
LETTERS to the EDITOR
Tamoxifen to prevent breast
cancer
SiR,—There is great interest in evaluating
the long-term use of prevent breast cancer. One randomised trial is underway in the UKl and another is under discussion in the US.2 The UK study includes premenopausal and postmenopausal women at increased risk while the US trial will probably be limited to postmenopausal women at some increased risk. An issue of concern is the control arm. In both trials a placebo group has been established or suggested. It may be, however, that for a significant proportion of postmenopausal women, the control should be
tamoxifen
to
oestrogen replacement therapy (ERT). Postmenopausal women on ERT are at lower risk of osteoporosis and cardiovascular disease (CVD).3 The risks of ERT are an increased incidence of endometrial cancer and, we believe, of breast cancer;4,5 increased mortality from these two diseases is, however, small when set beside the ERT-associated decrease in mortality from CVD. If tamoxifen is to be tried in the prevention of breast cancer in postmenopausal women total mortality must be a major end-point. So should the control arm be placebo or "best available therapy"? Best available therapy may be ERT. Tamoxifen has beneficial effects on blood cholesterol and, probably, on bone mineral content.2,6 Whether these will translate into reduced mortality from CVD and fractures remains to be established. A slight increased risk of endometrial cancer with
long-term tamoxifen use seems likely.7 To explore a tamoxifen versus ERT comparison further we have used the estimated changes in annual mortality in women aged 65-74 produced by long-term ERT use.4 Within this age range CVD accounts for 690 deaths per 100 000 annually; the equivalent figure for fractures is 55, for endometrial cancer 42, and for breast cancer 102. ERT reduces the figure for CVD by 333 and that for fractures by 33 but increases that for endometrial cancer by 26 and for breast cancer by a small amount (although the incidence of breast cancer is significantly increased). Assuming, as seems likely, that the effect of tamoxifen on mortality from fractures and
endometrial cancer is equivalent to that of ERT the net mortality effect of ERT versus tamoxifen is defined by the equation: D = 333 - (333 x TE) - (102 x RR x Re) where D =net gain or loss in annual mortality per 100 000 women (ERT minus tamoxifen); TE =effect on CVD mortality of tamoxifen relative to ERT; RR = relative risk of breast cancer; and Re proportional reduction in breast cancer mortality from tamoxifen use. When D = 0 the net effect of tamoxifen on mortality would be equal to that of ERT. We have calculated the reduction in breast cancer mortality from tamoxifen use necessary to produce a net equivalent effect on overall mortality for a range of potential effects on CVD mortality of tamoxifen relative to ERT over a range of breast cancer RR (figure). For example, for an RR of 50 and with tamoxifen 60% as effective as ERT in reducing CVD deaths, tamoxifen needs to reduce breast cancer deaths by 26% to be the equal of ERT in total mortality terms. Circumstances which fall to the left of a curve favour the use of ERT, while those to the right favour tamoxifen. If the CVD protection conferred by tamoxifen approaches that of ERT use (TE=80%), as suggested by its effect on serum cholesterol,8 then for a breast cancer RR of 2 0 (which is the relative risk the proposed trials are planned for) tamoxifen needs to reduce breast cancer mortality by 33% to be the equal of ERT. 33% seems close to the best result that can reasonably be expected.8 If the CVD protection afforded by ERT is only partly explained by its effect on cholesterol and the remaining CVD effects are only partly mimicked by tamoxifen (TE 60%, say), then only in women with a breast cancer RR of 4 or more would a 33% protective effect of tamoxifen on breast cancer mortality be sufficient to make tamoxifen "the treatment of choice". If tamoxifen has little effect on CVD (TE 40% or less) ERT would be far more appropriate. The proposed trials have been designed on the assumption that tamoxifen will reduce breast cancer mortality by 25%. The figure shows that for TE 80 %, RR has to be greater than 26 to break even. For TE 60% RR has to be above 5-2. Only if any increase in breast cancer and endometrial cancer cases is intolerable does it seem reasonable not to include an ERT arm in a randomised trial. A placebo arm may be difficult to justify. Estimates of the effect of tamoxifen on CVD must be taken from current trials of adjuvant tamoxifen to guide the design of chemoprevention trials with total mortality as a key end-point. Trials focused on breast cancer incidence may be too narrowly conceived. =
=
Norris Cancer Hospital, University of Southern California, Medical School, Los Angeles, California 90033, USA
D. SPICER M. C. PIKE
TJ, Hardy JR, Ashley SE, et al. A pilot trial to evaluate the acute toxicity and feasibility of tamoxifen for prevention of breast cancer. Br J Cancer 1989; 60:
1 Powles
126-31 2. Love RR. Prospects for antiestrogen chemoprevention of breast cancer. JNatl Cancer Inst 1990; 82: 18-21. 3. Mishell DR, ed. Menopause physiology and pharmacology. Chicago: Year Book
Medical Publishers, 1988. RK, Pike MC, Henderson
4. Ross 5. 6.
Reduction in breast cancer mortality required for tamoxifen to equal ERT in total mortality terms. 20%. 40%, 60%. and 80% curves are for different relative efficacies of tamoxifen vis-a-vis ERT in respect of CVD deaths
7. 8
BE, Mack TM, Lobo RA. Stroke prevention and estrogen replacement therapy. Lancet 1980; i: 505 Pike MC, Bernstein L. Breast cancer and hormone replacement therapy. Lancet 1990; 335: 297. Turken S, Sins E, Seldin D, Flaster E, Hyman G, Lindsay R. Effects of tamoxifen on spinal bone density m women with breast cancer. J Natl Cancer Inst 1980; 81: 1086-88. Fornander T, Cedermark B, Mattsson A, et al. Adjuvant tamoxifen in early breast cancer. occurrence of new primary cancer. Lancet 1989; i: 117-20. Spicer D, Pike MC, Henderson BE "Estrogen replacement therapy" in patients with a prior diagnosis of breast cancer. Oncology (in press)
LETTERS to the EDITOR
Tamoxifen to prevent breast
cancer
SiR,—There is great interest in evaluating
the long-term use of prevent breast cancer. One randomised trial is underway in the UKl and another is under discussion in the US.2 The UK study includes premenopausal and postmenopausal women at increased risk while the US trial will probably be limited to postmenopausal women at some increased risk. An issue of concern is the control arm. In both trials a placebo group has been established or suggested. It may be, however, that for a significant proportion of postmenopausal women, the control should be
tamoxifen
to
oestrogen replacement therapy (ERT). Postmenopausal women on ERT are at lower risk of osteoporosis and cardiovascular disease (CVD).3 The risks of ERT are an increased incidence of endometrial cancer and, we believe, of breast cancer;4,5 increased mortality from these two diseases is, however, small when set beside the ERT-associated decrease in mortality from CVD. If tamoxifen is to be tried in the prevention of breast cancer in postmenopausal women total mortality must be a major end-point. So should the control arm be placebo or "best available therapy"? Best available therapy may be ERT. Tamoxifen has beneficial effects on blood cholesterol and, probably, on bone mineral content.2,6 Whether these will translate into reduced mortality from CVD and fractures remains to be established. A slight increased risk of endometrial cancer with
long-term tamoxifen use seems likely.7 To explore a tamoxifen versus ERT comparison further we have used the estimated changes in annual mortality in women aged 65-74 produced by long-term ERT use.4 Within this age range CVD accounts for 690 deaths per 100 000 annually; the equivalent figure for fractures is 55, for endometrial cancer 42, and for breast cancer 102. ERT reduces the figure for CVD by 333 and that for fractures by 33 but increases that for endometrial cancer by 26 and for breast cancer by a small amount (although the incidence of breast cancer is significantly increased). Assuming, as seems likely, that the effect of tamoxifen on mortality from fractures and
endometrial cancer is equivalent to that of ERT the net mortality effect of ERT versus tamoxifen is defined by the equation: D = 333 - (333 x TE) - (102 x RR x Re) where D =net gain or loss in annual mortality per 100 000 women (ERT minus tamoxifen); TE =effect on CVD mortality of tamoxifen relative to ERT; RR = relative risk of breast cancer; and Re proportional reduction in breast cancer mortality from tamoxifen use. When D = 0 the net effect of tamoxifen on mortality would be equal to that of ERT. We have calculated the reduction in breast cancer mortality from tamoxifen use necessary to produce a net equivalent effect on overall mortality for a range of potential effects on CVD mortality of tamoxifen relative to ERT over a range of breast cancer RR (figure). For example, for an RR of 50 and with tamoxifen 60% as effective as ERT in reducing CVD deaths, tamoxifen needs to reduce breast cancer deaths by 26% to be the equal of ERT in total mortality terms. Circumstances which fall to the left of a curve favour the use of ERT, while those to the right favour tamoxifen. If the CVD protection conferred by tamoxifen approaches that of ERT use (TE=80%), as suggested by its effect on serum cholesterol,8 then for a breast cancer RR of 2 0 (which is the relative risk the proposed trials are planned for) tamoxifen needs to reduce breast cancer mortality by 33% to be the equal of ERT. 33% seems close to the best result that can reasonably be expected.8 If the CVD protection afforded by ERT is only partly explained by its effect on cholesterol and the remaining CVD effects are only partly mimicked by tamoxifen (TE 60%, say), then only in women with a breast cancer RR of 4 or more would a 33% protective effect of tamoxifen on breast cancer mortality be sufficient to make tamoxifen "the treatment of choice". If tamoxifen has little effect on CVD (TE 40% or less) ERT would be far more appropriate. The proposed trials have been designed on the assumption that tamoxifen will reduce breast cancer mortality by 25%. The figure shows that for TE 80 %, RR has to be greater than 26 to break even. For TE 60% RR has to be above 5-2. Only if any increase in breast cancer and endometrial cancer cases is intolerable does it seem reasonable not to include an ERT arm in a randomised trial. A placebo arm may be difficult to justify. Estimates of the effect of tamoxifen on CVD must be taken from current trials of adjuvant tamoxifen to guide the design of chemoprevention trials with total mortality as a key end-point. Trials focused on breast cancer incidence may be too narrowly conceived. =
=
Norris Cancer Hospital, University of Southern California, Medical School, Los Angeles, California 90033, USA
D. SPICER M. C. PIKE
TJ, Hardy JR, Ashley SE, et al. A pilot trial to evaluate the acute toxicity and feasibility of tamoxifen for prevention of breast cancer. Br J Cancer 1989; 60:
1 Powles
126-31 2. Love RR. Prospects for antiestrogen chemoprevention of breast cancer. JNatl Cancer Inst 1990; 82: 18-21. 3. Mishell DR, ed. Menopause physiology and pharmacology. Chicago: Year Book
Medical Publishers, 1988. RK, Pike MC, Henderson
4. Ross 5. 6.
Reduction in breast cancer mortality required for tamoxifen to equal ERT in total mortality terms. 20%. 40%, 60%. and 80% curves are for different relative efficacies of tamoxifen vis-a-vis ERT in respect of CVD deaths
7. 8
BE, Mack TM, Lobo RA. Stroke prevention and estrogen replacement therapy. Lancet 1980; i: 505 Pike MC, Bernstein L. Breast cancer and hormone replacement therapy. Lancet 1990; 335: 297. Turken S, Sins E, Seldin D, Flaster E, Hyman G, Lindsay R. Effects of tamoxifen on spinal bone density m women with breast cancer. J Natl Cancer Inst 1980; 81: 1086-88. Fornander T, Cedermark B, Mattsson A, et al. Adjuvant tamoxifen in early breast cancer. occurrence of new primary cancer. Lancet 1989; i: 117-20. Spicer D, Pike MC, Henderson BE "Estrogen replacement therapy" in patients with a prior diagnosis of breast cancer. Oncology (in press)
