- Email: [email protected]
First results from the International Breast Cancer Intervention Study (IBIS-I): a randomised prevention trial
ARTICLES
Articles
First results from the International Breast Cancer Intervention Study (IBIS-I): a randomised prevention trial IBIS investigators*
Summary Background Three clinical trials on the use of tamoxifen to prevent breast cancer have reported mixed results. The overall evidence supports a reduction in the risk of breast cancer, but whether this benefit outweighs the risks and side-effects associated with tamoxifen is unclear. Methods We undertook a double-blind placebo-controlled randomised trial of tamoxifen, 20 mg/day for 5 years, in 7152 women aged 35–70 years, who were at increased risk of breast cancer. The primary outcome measure was the frequency of breast cancer (including ductal carcinoma in situ). Analyses were by intention to treat after exclusion of 13 women found to have breast cancer at baseline mammography. Findings After median follow-up of 50 months (IQR 32–67), 69 breast cancers had been diagnosed in 3578 women in the tamoxifen group and 101 in 3566 in the placebo group (risk reduction 32% [95% CI 8–50]; p=0·013). Age, degree of risk, and use of hormone-replacement therapy did not affect the reduction. Endometrial cancer was nonsignificantly increased (11 vs 5; p=0·2) and thromboembolic events were significantly increased with tamoxifen (43 vs 17; odds ratio 2·5 [1·5–4·4], p=0·001), particularly after surgery. There was a significant excess of deaths from all causes in the tamoxifen group (25 vs 11, p=0·028). Interpretation Prophylactic tamoxifen reduces the risk of breast cancer by about a third. Temporary cessation of tamoxifen should be considered and the use of appropriate antithrombotic measures is recommended during and after major surgery or periods of immobilisation. Prophylactic use of tamoxifen is contraindicated in women at high risk of thromboembolic disease. The combined evidence indicates that mortality from non-breast-cancer causes is not increased by tamoxifen. The overall risk to benefit ratio for the use of tamoxifen in prevention is still unclear, and continued follow-up of the current trials is essential. Lancet 2002; 360: 817–24 See Commentary page 813
*Investigators listed at end of paper Correspondence to: Prof J Cuzick, IBIS Trials Centre, Cancer Research UK, PO Box 123, London WC2A 3PX, UK (e-mail: [email protected])
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
Introduction The belief that oestrogen is the primary promotional factor for breast cancer has a long history and is now well established.1–4 Attempts to control the incidence of breast cancer by decreasing the oestrogenic stimulus are more recent. The marked reduction in the rate of new contralateral tumours found when tamoxifen was used to treat early cancer,5 and the drug’s apparently low side-effect profile, led to the proposal that tamoxifen prophylaxis might be a suitable approach for reduction of the risk of breast cancer in high-risk women.6 A decrease of about 50% in the rate of new contralateral tumours with 5 years of tamoxifen treatment has been noted in several studies, which have been summarised in the Early Breast Cancer Trialists’ Collaborative Group overview.7 These early results led to the initiation of a tamoxifen prevention pilot trial at the Royal Marsden Hospital, London, UK, in 1986.8,9 That trial has now evolved into a full-scale prevention study. Three other studies started in 1992: the National Surgical Adjuvant Breast and Bowel Project P-1 Study (NSABP-P1),10 the Italian National trial,11 and the International Breast Cancer Intervention Study (IBIS) in the UK, Australia, New Zealand, and some European countries.12 Three of these studies have now published initial findings, with mixed results. A 50% decrease in breast-cancer incidence was observed in the North American trial, but little or no reduction was seen in the two European trials. The possible reasons for the differences in results have been the subject of much discussion, but no clear explanation is available. Results from all the trials are statistically compatible with a 30–40% decrease in incidence,13 but opinions vary substantially as to whether this effect is large enough to outweigh the established side-effects of tamoxifen, notably an increase in menopausal symptoms, vascular events, and endometrial abnormalities, including cancer. Initial results are also available on the effect on breastcancer incidence of another specific oestrogen-receptor modulator, raloxifene.14 These early results suggest a reduction of as much as 70%, and the side-effect profile is similar to that of tamoxifen, except possibly for endometrial cancer. A direct comparison of these two specific oestrogenreceptor modulators is the subject of a second North American trial. Here we report the initial results from IBIS-I, comparing tamoxifen with placebo. The primary endpoint was frequency of breast cancer (including ductal carcinoma in situ). Predefined subgroups were oestrogen receptor status of the cancer, use of hormonal replacement therapy, and age (<50, 50 years) Secondary endpoints were other cancers, thromboembolic events, cardiovascular events, and cause-specific mortality.
Methods Participants Women aged 35–70 years were entered into the trial from April, 1992, until March, 2001 (figure 1). Eligible women 817
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
70
7152 randomised Placebo
Tamoxifen
3574 assigned placebo 8 excluded (breast cancer at entry) 3566 in primary analysis
3578 assigned tamoxifen 5 excluded (breast cancer at entry) 3573 in primary analysis
Proportion of women (%)
60
Placebo Tamoxifen
50 40 30 20 10
3528 (98·9%) began treatment
3523 (98·6%) began treatment
959 (26·9%) completed 5 years of treatment
837 (23·4%) completed 5 years of treatment
0 <2
3–5 5–10 2–3 Predicted absolute risk (%)
>10
Figure 2: Distribution of predicted absolute risk of breast cancer in the 10 years after randomisation according to allocated treatment Based on a program by J Tyrer (unpublished).
Figure 1: Trial profile
had risk factors for breast cancer indicating at least a twofold relative risk for ages 45–70 years, a four-fold relative risk for ages 40–44 years, and a roughly ten-fold relative risk for ages 35–39 years. Specific entry criteria are shown in table 1. Reasons for exclusion were any previous invasive cancer (except non-melanoma skin cancer), a previous deep-vein thrombosis or pulmonary embolism, current use of anticoagulants, or a life expectancy judged to be less than 10 years. Women who were pregnant or wished to become pregnant were also excluded. Women were recruited through attendance at family history clinics, relatives of women with breast cancer, breast screening centres, general practitioners, and the media. Approval of the local ethics committee from each centre was obtained, and written informed consent was required for all women. The trial was done under the auspices of the UK Coordinating Committee for Cancer Research (now part of the National Cancer Research Network) and was supported by the Imperial Cancer Research Fund, the Cancer Research Campaign (now merged as Cancer Research UK), and the Department of Health, under a Clinical Trials Certificate from the UK Medicines Control Agency. In Australia and New Zealand, the trial was done by the Australia New Zealand Breast Cancer Trials Group. An independent data-monitoring committee reviewed data regularly and reported to the IBIS Working Party, who were responsible for the overall conduct of the trial. At their
January, 2002, meeting the data-monitoring committee decided that the data were sufficiently mature to be published. Design and procedures All women had had a baseline mammogram within the previous 12 months to exclude pre-existing cancer, and a blood sample was requested for cholesterol assays and marker studies. Women were randomly assigned 5 years of treatment with tamoxifen 20 mg/day or matching placebo. For logistic reasons, the baseline mammogram was taken at the time of randomisation for some women, and 13 were subsequently excluded because breast cancer was diagnosed on their baseline mammogram (figure 1). Women were followed up every 6 months during the 5 years of active treatment and by annual questionnaire or clinical visit thereafter for up to 5 years. Details of any side-effects were collected at every visit, both as predefined items and free text and coded according to the NHS Read codes.15 Symptoms, diagnoses, and procedures were each recorded separately. All concomitant medications were recorded. Mammography was done every 12–18 months, and blood samples were requested again at 1 year and 5 years. Randomisation was done centrally by telephone or fax to a central office in London, UK for European centres and in Sydney, Australia for Australia and New Zealand. Randomisation was stratified by centre and balanced in
Risk factor
Placebo (n=3566)
Tamoxifen (n=3573)
First-degree relative who developed breast cancer at or before age 50 First-degree relative with bilateral breast cancer* Two or more first-degree or second-degree relatives with breast cancer† Lobular carcinoma in situ Atypical hyperplasia Nulliparous and a first-degree relative who developed breast cancer Benign biopsy and first-degree relative who developed breast cancer Risk equivalent‡
1744 (48·9%) 601 (16·9%) 2206 (61·9%) 44 (1·2%) 104 (2·9%) 325 (9·1%) 132 (3·7%) 143 (4·0%)
1689 (47·3%) 579 (16·2%) 2204 (61·7%) 44 (1·2%) 97 (2·7%) 314 (8·8%) 123 (3·4%) 177 (5·0%)
All criteria permit entry from age 45 years. Atypical hyperplasia permits entry from age 40 and lobular carcinoma in situ from age 35. Total number adds up to 148% of total entry because some women met several entry criteria. *Eligible from age 40 if relative had cancer before age 50 and at age 35 if relative’s cancer was diagnosed before age 40. †Eligible from age 40 if both relatives developed breast cancer before age 50 and from age 35 if both relatives were first degree and developed breast cancer before age 50. ‡Risk-equivalent women were those with a strong family history, not fitting specific categories, but judged to be at higher risk than the minimum eligibility category by the study chairman.
Table 1: Entry criteria and distribution by treatment group
818
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Proportion of women (%)
ARTICLES
40 35 30 25 20 15 10 5 40
Number of cancers
Placebo Tamoxifen
Placebo
Tamoxifen
101
69
0·68 (0·50–0·92)
Age (years) <50 50
39 62
25 44
0·64 (0·39–1·05) 0·70 (0·48–1·04)
HRT use During the trial Before trial only Never
38 21 42
29 9 31
0·76 (0·47–1·23) 0·43 (0·20–0·91) 0·73 (0·46–1·17)
Invasiveness Ductal carcinoma in situ Invasive
16 85
5 64
0·31 (0·12–0·82) 0·75 (0·54–1·04)
Grade* Low Intermediate High Unknown
14 37 27 7
14 27 21 2
1·00 (0·48–2·06) 0·73 (0·44–1·19) 0·77 (0·44–1·36) ··
Nodal status* Positive Negative Not done
23 59 3
16 44 4
0·69 (0·37–1·30) 0·74 (0·50–1·10) ··
ER status* Positive Negative Unknown
63 19 3
44 19 1
0·69 (0·47–1·02) 1·00 (0·53–1·87) ··
Tumour size* 1 cm 1–2 cm 2–5 cm >5 cm Metastatic only
32 36 17 0 0
23 25 10 5 1
0·72 (0·42–1·22) 0·69 (0·42–1·15) 0·59 (0·27–1·26) ·· ··
Total
<40
40–44 45–49 50–54 55–59 60–64 65 Age (years)
Figure 3: Age distribution of participants
blocks of eight. The lists were then randomly permuted again in blocks of six to ten (chosen randomly) to ensure that the last member of each block was not predictable. Endpoints and deaths were externally reviewed and coded with masking of treatment allocation. Primary cause rather than underlying cause was used (eg, deaths from pulmonary embolism after cancer surgery were coded as from pulmonary embolism). Statistical analysis The trial was powered to detect a 40% complianceadjusted reduction in the rate of breast cancer including ductal carcinoma in situ. For 90% power, 164 events would be required. Risk profiles were computed retrospectively by means of a specially developed program that included detailed family-history variables and other standard risk factors (J Tyrer et al, unpublished). Analyses were mainly based on comparison of proportions by odds ratios, and Fisher’s exact values were used where appropriate. Major comparisons were expressed as odds ratios, which are very good approximations to relative risks or hazard ratios for rare events and equal follow-up. All p values are two-sided and confidence intervals were based on a normal approximation. Analyses were by intention to treat, after exclusion of the 13 women found to have breast cancer at baseline. Role of the funding source The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The manufacturers (AstraZeneca) supplied tamoxifen and matching placebo without charge and provided technical advice, but were not involved in the conduct or analysis of the trial. Placebo (n=3566) Tamoxifen (n=3573) Demography Mean (SD) age, years Postmenopausal
50·8 (6·7) 1740 (48·8%)
50·7 (7·0) 1761 (49·3%)
HRT use Before entry During trial Ever
1443 (40·5%) 1399 (39·2%) 1783 (50·0%)
1469 (41·1%) 1445 (40·4%) 1849 (51·7%)
162·9 (6·4) 71·4 (14·0) 26·9 (5·1)
162·8 (6·6) 71·7 (14·5) 27·0 (5·3)
1283 (36·0%) 737 (20·7%) 207 (5·8%) 327 (9·2%)
1232 (34·5%) 711 (19·9%) 229 (6·4%) 281 (7·9%)
Anthropometry Mean (SD) height, cm Mean (SD) weight, kg Mean (SD) body-mass index, kg/m2 Hysterectomy All With both ovaries retained One ovary removed Both ovaries removed
Data are number of women unless otherwise stated.
Table 2: Baseline characteristics and HRT use
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
Odds ratio (95% CI) tamoxifen vs placebo
*For invasive cancers only (n=85 placebo, n=64 tamoxifen). None of the interactions was significant.
Table 3: Breast cancers and characteristics by treatment allocation
Results A total of 7139 women were included in the analysis (figure 1, table 1). The largest risk group was women who had two or more first-degree or second-degree relatives with breast cancer (62%). In all, 97% of the participants reported some family history, and 8% had a benign lesion associated with increased risk of breast cancer. The projected 10-year risk based on family history and other established risk factors is shown in figure 2. The yearly frequency of breast cancer in the absence of treatment was projected to be 7·50 per 1000; the actual frequency in the placebo group was 6·74 per 1000, which does not differ significantly from the projected frequency. The mean age was 50·8 years (SD 6·9); 54·7% of the women were between the ages of 45 and 54 (figure 3). 60% were from the UK, 37% from Australia or New Zealand, and 3% from the rest of Europe. 49% were postmenopausal and 41% had previously used hormone-replacement therapy (HRT). Further demographic details are shown in table 2. The cut-off date of follow-up for the analysis was Jan 1, 2002. Median follow-up was 50 months (IQR 32–67) and a total of 29 967 woman-years of follow-up have been accrued (14 969 in the placebo group and 14 998 in the tamoxifen group). The numbers of woman-years of treatment were 11 539 and 10 370, respectively. Full compliance to 5 years was estimated to be 64% in the tamoxifen group and 74% in the placebo group (p<0·001). 40% of women used HRT at some time during the active treatment. At the time of data lock, 25% of women had completed a full 5 years of treatment (959 [26·9%] placebo vs 837 [23·4%] tamoxifen) and a further 47% were still on treatment (1760 [49·4%] vs 1574 [44·0%]).
819
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
1·00 0·92
8 vs 10
12 vs 15
Odds ratio
17 vs 25
14 vs 19
95% CI
0·68 0·50
8 vs 14
10 vs 18
0·25
0 1
2
3 4 Year of follow-up
5
>5
Figure 4: Risk reduction with tamoxifen by year of follow-up Numbers on curve are the number of cancers in the tamoxifen group versus the number in the placebo group.
170 breast cancers (including ductal carcinoma in situ) have been recorded (table 3). The rate is 32% (95% CI 8–50) lower in the tamoxifen group than in the placebo group (tamoxifen 69, placebo 101; p=0·01). A reduction in risk with tamoxifen was apparent for both invasive disease (64 vs 85; reduction 25%) and non-invasive disease (five vs 16; reduction 69%). The difference between these two reductions was not significant. Of the invasive cancers, 69% were node negative, 72% were oestrogen-receptor (ER) positive, and 78% were 2 cm or less in diameter. Nodal status, size, and grade were similar in both study groups. The reduction in risk of confirmed ER-positive invasive tumours with tamoxifen was 31%. There was no reduction in the risk of ER-negative invasive tumours. There were five large tumours (diameter more than 5 cm) in the tamoxifen group and none in the placebo group. Age and use of HRT did not significantly affect the risk reduction (table 3). There was no evidence that the degree of protection changed over the 5 years of treatment (figure 4). There are as yet insufficient data for assessment of whether protection continues after treatment has stopped. Only four deaths from breast cancer have been reported (two in each study group), a number that is well below the 9·5 deaths expected from the observed tumour characteristics and survival predicted from the Early Breast Cancer Trialists Collaborative Group overview. Event
Number of cases
p
Placebo Tamoxifen Venous thromboembolism (excluding superficial thrombophlebitis) All thromboembolism 17 43 0·001 Occurring within 3 months of leg surgery 5 20 0·004 or fracture Spontaneous 12 23 0·09 Pulmonary embolism 10 13 0·68 Deep-vein thrombosis 5 24 0·0005 Thrombosis (other) 2 6 0·29 Thrombophlebitis
9
27
0·004
Cerebrovascular All Stroke or cerebrovascular accident Transient ischaemic attack
17 11 6
16 13 3
0·86 0·84 0·34
Cardiac All Myocardial infarction Coronary revascularisation Other revascularisation Other cardiovascular events Angina
63 5 5 2 17 34
73 5 5 8 16 39
0·44 1·00 1·00 0·11 0·86 0·64
Table 4: Thromboembolic, cerebrovascular, and cardiac events according to treatment
820
However, the median follow-up from diagnosis is still short (27 months [IQR 15–42]), and all women were having regular mammography. A non-significant two-fold excess of endometrial cancer was found in the tamoxifen group (11 vs five in the placebo group; odds ratio 2·20 [95% CI 0·80–6·06], p=0·2). Most of these cancers were in women who were older than 50 years at randomisation (ten tamoxifen group vs three placebo group), and all the women affected were postmenopausal at diagnosis. All but one of these cancers (a low-grade sarcoma in the placebo group) were adenocarcinomas, and all but one (a stage II case in the placebo group) were of FIGO stage I. Most of the tumours were of low (four vs two) or intermediate (six vs one) grade. There was no evidence that endometrial cancer was linked to HRT use or an interaction between HRT and tamoxifen. Ten of the women with endometrial cancers had never used HRT (seven vs three), only four had used it during the trial (two vs two), and two in the tamoxifen group had used it before the trial. There have been no deaths from endometrial cancer. The rate of endometrial cancer in the placebo group (34 per 100 000 woman-years) was similar to population rates for the UK and Australia. Overall, cancers other than those of the breast and endometrium were equally distributed between the two study groups (39 in each group), and no cancer differed significantly in frequency between the tamoxifen and placebo groups (colorectal nine vs six; stomach one vs three; liver none vs two; pancreas one vs none; lung three vs none; larynx one vs none; melanoma eight vs seven; bladder or kidney two vs three; ovary six in each group; endocrine two vs seven; meningioma two in each group; haemological or lymphatic two vs three; and two primary unknown cancers in the tamoxifen group). The rate of venous thromboembolic events was about 2·5 times higher in the tamoxifen group than in the placebo group (95% CI 1·5–4·4; p=0·001; table 4). 25 (42%) of these events occurred within 3 months of major surgery or after long-duration immobility; 20 were in the tamoxifen group (p=0·004). This feature was the major determinant of thromboembolic events. There was no indication of any synergy between tamoxifen and HRT and some evidence for a negative interaction. There were almost twice as many spontaneous venous thromboembolic events with tamoxifen than with placebo, but this difference was not significant (p=0·09). An increase in superficial thrombophlebitis was also noted. There were no differences in the numbers of cerebrovascular accidents, myocardial infarctions, or other vascular events. A detailed exploration of factors affecting vascular events will be published elsewhere. Details of any reported side-effects were collected at each follow-up visit. Although very large numbers of Number of events (%)
Side-effect Gynaecological or vasomotor Headaches and migraines All fractures Osteoporotic fractures (hip, spine, wrist, or forearm) Breast complaints Nail changes Eye (excluding cataracts) Cataracts
p
Placebo (n=3566)
Tamoxifen (n=3573)
2414 (67·7%) 1067 (29·9%) 127 (3·6%) 40 (1·1%)
2922 (81·8%) <0·0001 997 (27·9%) 0·13 116 (3·3%) 0·52 45 (1·3%) 0·66
675 (18·9%) 96 (2·7%) 376 (10·5%) 37 (1·0%)
525 (14·7%) <0·0001 148 (4·1%) 0·001 373 (10·4%) 0·94 38 (1·0%) 1·00
Table 5: Side-effects reported at any time and of any severity, according to allocated treatment
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Number of events (%)
p
Premenopausal Placebo (n=1826) Procedures Hysteroscopy Pelvic ultrasonography Dilation and curettage Hysterectomy Oophorectomy Symptoms Vasomotor symptoms Vaginal discharge Vaginal dryness Abnormal bleeding Endometrial polyps Uterine fibroids Amenorrhoea Thrush/candida Prolapse Ovarian cysts and lumps Endometriosis
Postmenopausal Tamoxifen (n=1810)
Placebo (n=1740)
Tamoxifen (n=1763)
107 (5·9%) 89 (4·9%)
136 (7·5%) 136 (7·5%)
31 (1·8%) 43 (2·5%)
92 (5·2%) 73 (4·1%)
<0·0001 <0·0001
76 (4·2%)
106 (5·9%)
18 (1·0%)
72 (4·1%)
<0·0001
76 (4·2%) 53 (2·9%)
118 (6·5%) 76 (4·2%)
28 (1·6%) 14 (0·8%)
36 (2·0%) 27 (1·5%)
0·002 0·006
988 (54·1%)
1233 (68·0%)
849 (48·8%)
1219 (69·0%)
<0·0001
266 (14·6%) 329 (18·0%) 565 (30·9%) 43 (2·4%) 71 (3·9%) 77 (4·2%) 36 (1·9%) 36 (1·9%) 33 (1·8%)
517 (28·6%) 379 (20·9%) 669 (36·9%) 69 (3·8%) 84 (4·6%) 171 (9·4%) 103 (5·7%) 33 (1·8%) 86 (4·8%)
237 (13·6%) 394 (22·6%) 113 (6·5%) 22 (1·3%) 11 (0·6%) ·· 28 (1·6%) 45 (2·6%) 9 (0·5%)
509 (28·9%) 403 (22·9%) 173 (9·8%) 61 (3·5%) 22 (1·2%) ·· 84 (4·8%) 52 (2·9%) 15 (0·9%)
<0·0001 0·10 <0·0001 <0·0001 0·09 <0·0001 <0·0001 0·81 <0·0001
16 (0·9%)
14 (0·8%)
5 (0·3%)
6 (0·3%)
0·88
p values are for premenopausal and postmenopausal women combined.
Table 6: Gynaecological procedures and gynaecological and vasomotor symptoms in premenopausal and postmenopausal women (defined at randomisation) according to allocated treatment
Patient Cause of death
Age at randomisation (years)
Time on treatment (months)
Time to diagnosis (months)
HRT use
Tamoxifen group 1 Breast cancer 2 Breast cancer 3 Laryngeal cancer 4 Stomach cancer 5 Colorectal cancer 6 Colorectal cancer 7 Colorectal cancer 8 Colorectal cancer 9 Carcinomatosis 10 Carcinomatosis 11 Myeloma 12 Myeloma 13 Pulmonary embolism 14 Pulmonary embolism 15 Deep-vein thrombosis 16 Subarachnoid haemorrhage 17 Cerebral aneurysm 18 Dissecting aortic aneurysm 19 Myocardial infarction 20 Myocardial infarction 21 Pericarditis 22 Cardiac failure 23 Cardiac failure 24 Road traffic accident 25 COAD
48 53 49 53 47 61 63 64 52 62 54 68 56 47 58 69 50 49 60 49 43 47 69 45 49
20 6 3 19 3 44 18 60 42 9 60 10 42 5 42 1 2 10 7 17 6 22 9 16 12
20 26 64 19 3 44 20 72 42 9 36 6 42 5 42 1 47 31 to death 7 17 39 22 41 16 22
Never 96 months before and 12 months during trial 1 month before and 2 months during trial 56 months before and 12 months during trial Never 6 months before and 11 months during trial Never 84 months before and 36 months during trial Never Never Never 96 months before and 18 months duirng trial 5 months before and 42 months during trial Never 84 months before and 23 months during trial 144 months before and 1 month during trial Never 42 months before and 12 months during trial Never Never Never Never Never 5 months before trial 84 months before and 10 months during trial
Placebo group 26 Breast cancer 27 Breast cancer 28 Ovarian cancer 29 Ovarian cancer 30 Colorectal cancer 31 Burkitt’s lymphoma 32 Pulmonary embolism 33 Pulmonary embolism 34 Sub-arachnoid haemorrhage 35 Ischaemic bowel disease 36 Air crash
50 54 54 59 52 49 47 62 56 56 38
8 48 6 44 12 12 48 1 3 12 5
8 48 15 44 12 32 48 1 3 77 5
Never Never 60 months before trial Never 72 months before and 12 months during trial Never Never 192 months before and 1 month during trial Never Never Never
COAD=chronic obstructive airway disease.
Table 7: Detailed breakdown of deaths
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
821
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Marsden NSABP-P1 Italian IBIS-I All trials
0·3
0·5
0·9 1·0
2·0
5·0
Odds ratio Figure 5: Death from any cause in the tamoxifen prevention trials Error bars=95% CI.
side-effects were recorded (table 5), the only major grouping that showed differences was vasomotor and gynaecological reports, which were about 21% higher in the tamoxifen than the placebo group, and breast complaints, which were 22% lower. As expected, there were higher proportions of hot flushes, vaginal discharge, and abnormal vaginal bleeding in the tamoxifen group (table 6). Rates of pelvic ultrasonography, dilation and curettage, and hysteroscopy were also significantly higher, and there were more women with endometrial polyps. The hysterectomy rate was 2·7% in the placebo group and 4·2% in the tamoxifen group (p=0·002). In premenopausal women, ovarian cysts and amenorrhoea were more than twice as common on tamoxifen. Reports of vaginal thrush were substantially increased in the tamoxifen group in both premenopausal and postmenopausal women. A new finding was an increase in reports of brittle nails (both fingernails and toenails) in the tamoxifen group. To date, no differences have been found for osteoporotic or non-osteoporotic fractures. No difference was found in the frequency of cataracts or other eye complaints. By contrast, the rate of benign breast disease was 31% lower in the tamoxifen group than in the placebo group (182 vs 262, p=0·0001). The largest reductions were in cysts (57 vs 123, p<0·0001), fibroadenomas (nine vs 19, p=0·06), and fibrocystic disease (16 vs 25, p=0·16). The effect was mostly seen in premenopausal women. Reports of breast pain were also reduced by 32% (70 vs 103, p=0·01) in the tamoxifen group. At the 1-year follow-up, mean weight had increased by 0·60 kg in the placebo group and 0·04 kg in the tamoxifen group (p<0·0001 for comparison of mean differences). At year 5, the mean weight gain was similar for the two groups (2·00 kg placebo vs 2·20 kg tamoxifen, p=0·8). Fuller details on the timing and severity of side-effects will be the subject of a separate report. The death rate from all causes was significantly higher in the tamoxifen group than in the placebo group (25 vs 11; p=0·028; table 7). Increases are seen for cancers other than breast cancer, pulmonary embolisms, other vascular causes, and cardiac deaths. The cancers were at a range of different sites, and the lack of an increase in cancer frequency suggests that this may be a chance finding. The cardiac and vascular causes of death are also varied, but the increase in thromboembolic events indicates that the extra deaths from this cause may be attributable to tamoxifen.
822
Updated data predict a 38% (95% CI 28–47) reduction in breast cancer incidence, and our value of 32% is compatible with this estimate. Also, the doubling of the rate of endometrial cancer and the 2·5-fold relative risk for vascular events are similar to findings of the other trials. What is unexpected is the increased number of deaths. This increase was not seen in the other prevention trials; lower, though not significantly so, numbers of deaths with tamoxifen were seen in NSABP-P1 and the Italian trial. The only cause for which there is any consistent evidence for an increase in mortality is thromboembolic disease. This conclusion is supported by the clear excess in the thromboembolicevent rate and the small but significant increased mortality from pulmonary embolism in the overview of tamoxifen use in the adjuvant setting. However, there is no evidence for an increase in deaths from other cancers, cardiac causes, or other vascular causes, and overall mortality is not increased when all of the evidence is combined (figure 5).
Discussion This study has confirmed that tamoxifen can reduce the risk of breast cancer in healthy women during the active treatment phase. The 32% reduction in risk seen in the study is consistent with the cumulative results when all of the chemoprevention trials are combined. The increased frequency of gynaecological problems, particularly the increased requirement for hysterectomy and oophorectomy, is of some concern. It may be a result of the observed higher proportions of women with abnormal vaginal bleeding, endometrial polyps, endometrial thickening, and ovarian cysts. The indications for hysterectomy and oophorectomy in these circumstances have not been clearly established. New findings were increases in vaginal thrush and brittle nails. The latter has been associated with decreased oestrogen concentrations at the menopause; however, both these side-effects were uncommon and caused only minor concern. The low rate of death from breast cancer highlights the importance of a full assessment of side-effects. The increased mortality in the tamoxifen group in this trial is of concern, but it may be due partly to statistical variability. The cancers were at a range of sites, and in two cases there had been only 3 months of treatment. The cardiovascular deaths are also varied. However, the increase in thromboembolic events seen in all trials and the increase, though smaller, in thromboembolic deaths seen in IBIS-I and the NSABP-P1 study indicate that thromboembolism is the most important complication of tamoxifen use, and that every effort should be taken to reduce this risk. The risk seems to be similar to that for women taking HRT,16 and similar precautions should be adopted.17 In particular, most of the risk and all of the deaths in the tamoxifen group occurred after surgery, and a wise precaution would be to discontinue tamoxifen before any surgery and to ensure that appropriate antithrombotic measures are provided during surgery and that tamoxifen is not restarted until full mobility has returned. Similar recommendations should be followed for women who become immobile for any reason. Although tamoxifen is unquestionably valuable in the adjuvant setting, some of its oestrogen-agonist properties restrict its ultimate usefulness—for both efficacy and side-effects. Early results18 show that anastrozole is more effective than tamoxifen in reducing recurrence (by about 25% in ER-positive tumours), and this drug seems to be even more promising in reducing the risk of new
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
contralateral tumours (58% reduction compared with tamoxifen). Furthermore, the agonist properties of tamoxifen account for its main side-effects of thromboembolic disease and endometrial cancer, which do not occur with aromatase inhibitors. However, new issues in relation to bone loss and other potential effects of long-term oestrogen deprivation might arise. The use of an aromatase inhibitor in the preventive setting is attractive and will be considered as an option in the forthcoming IBIS-II trial. Although when used as adjuvant therapy for breast cancer, tamoxifen can clearly reduce the risk of recurrence and death, at present the overall risk to benefit ratio in the preventive setting is still unclear. Further long-term follow-up to study breast-cancer incidence and mortality, other causes of death, and side-effects in the current trials remains essential. Writing Committee—J Cuzick, J Forbes, R Edwards, M Baum, S Cawthorn, A Coates, H Hamed, A Howell, T Powles.
Data Monitoring Committee—G Clunie (Melbourne, Australia); R Collins (CTSU, Oxford, UK); N Day (Strangeways Laboratories, Cambridge, UK; Chairman); J Northover (St Mark’s Hospital, London, UK).
IBIS Working Party and Principal Investigators—E Abdi (Bendigo Hospital, Bendigo, Australia); E Anderson (Breast Screening Centre, Edinburgh, UK); C Atkinson (Christchurch Hospital, Christchurch, New Zealand); M Baum (University College London, London, UK; Clinical Adviser); J Beith (Royal Prince Alfred Hospital, Sydney, Australia); R Bell (Geelong Hospital, Geelong, Australia); A Bird (Moorfields Hospital, London, UK; Ophthalmology Adviser); S Birrell (Flinders Medical Centre, Adelaide, Australia); R Blamey (Nottingham City Hospital, Nottingham, UK); J Boyages (Westmead Hospital, Sydney, Australia); K Buser (Engeriedspital, Bern, Switzerland); I Campbell (Waikato Hospital, Hamilton, New Zealand); S Cawthorn (Frenchay Healthcare Trust, Bristol, UK; Clinical Adviser); C Chapman (Oxford Radcliffe Hospital, Oxford, UK); M Chipman (Austin and Repatriation Medical Centre, West Heidelburg, Australia); A Coates (Royal Prince Alfred Hospital, Sydney, Australia; Clinical Adviser); J P Collins (Royal Melbourne Hospital, Melbourne, Australia); P Craft (Canberra Hospital, Canberra, Australia); J Cuzick (Cancer Research UK, London, UK; Chairman); L Denton (Glenfield Hospital NHS Trust, Leicester, UK); J Dewar (Sir Charles Gairdner Hospital, Nedlands, Australia); M Dowsett (Royal Marsden Hospital, London, UK); H Earl (Addenbrooke’s Hospital, Cambridge, UK); D Eccles (Princess Anne Hospital, Southampton, UK); R Edwards (Cancer Research UK, London, UK; Trial Statistician); G Evans (Christie Hospital NHS Trust, Manchester, UK); L Fallowfield (University of Sussex, Brighton, UK); I Fentiman (Guy’s Hospital, London, UK); J Forbes (ANZ BCTG, University of Newcastle, Newcastle Mater Hospital, Australia; Clinical Adviser); M Friedlander (Prince of Wales Hospital, Sydney, Australia); C Furnival (Wesley Medical Centre, Brisbane, Australia); J Garcia (Hospiat Universitario “Principe de Asturias”, Madrid, Spain); W George (Western Infirmary, Glasgow, UK); F J Gilbert (University of Aberdeen, Aberdeen, UK); A Goldhirsch (Oncology Institute of Southern Switzerland, Lugano, Switzerland); H Hamed (Guy’s Hospital, London, UK; Clinical Adviser); A Hanby (St James’s University Hospital, Leeds, UK; Trial Pathologist); S Hart (Monash Medical Centre, Clayton, Australia); J Hearne (NCRN, London, UK; Observer); A Henry (Hopital Jolimont, St Paul, Belgium); C Hirst (Wesley Breast Clinic, Brisbane, Australia); C Holcombe (Royal Liverpool University Hospital, Liverpool, UK); K Holli (Tampere University, and University Hospital, Finland); A Howell (Christie Hospital NHS Trust, Manchester, UK; Clinical Adviser); I Jackson (Astra Zeneca Pharmaceuticals, Macclesfield, UK; Observer); J Kirk (Westmead Hospital, Sydney, Australia); M Lansdown (St James’s University Hospital, Leeds, UK); K Law (Cancer Research UK, London, UK; Observer); M Lee (City Hospital NHS Trust, Birmingham, UK); T Lennard (University of Newcastle upon Tyne, Newcastle Upon Tyne, UK); F MacNeil (Essex County Hospital, Colchester, UK); P Maddox (Royal United Hospital, Bath, UK); R Mansel (University Hospital of Wales, Cardiff, UK); P McAleese (Monaghan General Hospital, Monaghan, Ireland); J MacKay (Addenbrooke’s NHS Trust, Cambridge, UK); K MacMichael (Huddersfield Royal Infirmary, Huddersfield, UK); C Mitine (Hopital Jolimont, St Paul, Belgium); C Normand (London School of Hygiene and Tropical Medicine, London, UK); W Odling-Smee (Belfast City Hospital, Belfast, UK); T Oivanen (Pirkanmaa Cancer Society, Tampere, Finland); C O’Neill (Cancer Research UK, London, UK; Senior Trial Coordinator); O Pagani (Ospedale Regionale della Beata Vergine, Mendrisio, Switzerland); T Powles (Royal Marsden Hospital, London, UK; Clinical Adviser); Z Raytor (Bristol Royal Infirmary, Bristol, UK); B Richmond (St Mary’s Hospital, London, UK); J Robertson
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
(Nottingham City Hospital, Nottingham, UK); R Sainsbury (University College Hospital, London, UK); T Sahmoud (Astra Zeneca Pharmaceuticals, Macclesfield, UK; Observer); P Sauven (Broomfield NHS Trust, Chelmsford, UK); R J Simes (University of Sydney, Sydney, Australia); R Snyder (St Vincent’s Hospital, Melbourne, Australia); R Stewart (Kettering General Hospital, Kettering, UK); A Stotter (Glenfield Hospital, Leicester, UK); A Thompson (Ninewells Hospital and Medical School, Dundee, UK); J Toy (Cancer Research UK, London, UK; Observer); P Twentyman (UKCCCR, London, UK; Observer); C Underhill (Border Medical Oncology, Wodonga, Australia); R Ward (St Vincent’s Hospital, Sydney, Australia); A Wilkinson (Belfast City Hospital, Belfast, UK); S Wilkinson (Royal Hobart Hospital, Hobart, Australia); J Williamson (City Hospital NHS Trust, Birmingham, UK); C Wynne (Christchurch Hospital, Christchurch, New Zealand). IBIS Coordinating Centre—Cancer Research UK, London, UK: C Duggan, R Edwards, S Falk, J Hickman, R Kealy, C O’Neill, E Pinney, S Tilling. IBIS–ANZ BCTG Operations/Statistical Office—H Dhillon, V Gebski, D Lindsay, A Melmeth, A Newton, L Paksec, M Seccombe, R Thornton. Study Coordinators—J Affen (Christie Hospital NHS Trust, Manchester, UK); S Bailey (University Hospital Wales, Cardiff, UK); S Bensted (Geelong Hospital, Geelong, Australia); J Bickerton (Monash Medical Centre, Clayton, Australia); K Brimson (Wesley Medical Centre, Brisbane, Australia); R Buder (Flinders Medical Centre, Adelaide, Australia); M Busteed (Canberra Hospital, Canberra, Australia); A Carnall (Prince of Wales Hospital, Sydney, Australia); O Claber (Northern Genetics Services, Newcastle upon Tyne, UK); S Clark (Bendigo Hospital, Bendigo, Australia); M Cross (Royal South Hants Hospital, Southampton, UK); S Daly (Flinders Medical Centre, Adelaide, Australia); A Deluca (Royal Melbourne Hospital, Melbourne, Australia); A Dowd (Geelong Hospital, Geelong, Australia); S Drummond (Breast Screening Unit, Edinburgh, UK); S Durell (Oxford Radcliffe Hospital, Oxford, UK); E Etherington (Huddersfield Royal Infirmary, Huddersfield, UK); W Fowler (Royal Hobart Hospital, Hobart, Australia); C Gano (Westmead Hospital, Sydney, Australia); D Garratt (Border Medical Oncology, Wodonga, Australia); C Gradige (Royal South Hants Hospital, Southampton, UK); E Granger (Guy’s Hospital, London, UK); J Gray (Belfast City Hospital, Belfast, UK); R Greenhalgh (Christie Hospital NHS Trust, Manchester, UK); L Gunn (Breast Screening Centre, Aberdeen, UK); A Hayes (Austin and Repatriation Medical Centre, West Heidelburg, Australia); J Hepper (St James’s Hospital, Leeds, UK); S Holcombe (Royal Liverpool University Hospital, Liverpool, UK); G Hoogeveen (Royal Melbourne Hospital, Melbourne, Australia); L Hughes (St Vincent’s Hospital, Melbourne, Australia); N Humphreys (Bendigo Hospital, Bendigo, Australia); S Ingelido (Wesley Breast Clinic, Brisbane, Australia); J Innes-Rowe (Sir Charles Gairdner Hospital, Nedlands, Australia); L Jolly (St Vincent’s Hospital, Sydney, Australia); A Jonson (Addenbrooke’s NHS Trust, Cambridge, UK); C Kershaw (Kettering General Hospital, Kettering, UK); T Klau (Flinders Medical Centre, Adelaide, Australia); M Lloyd (Prince of Wales Hospital, Sydney, Australia); J Magnay (Prince of Wales Hospital, Sydney, Australia); K Makinson (Royal Liverpool University Hospital, Liverpool, UK); A Miller (Flinders Medical Centre, Adelaide, Australia); B Murray (Royal Prince Alfred Hospital, Sydney, Australia); S Napier (Bendigo Hospital, Bendigo, Australia); S Pearce (Frenchay Hospital, Bristol, UK); A Pickersgill (University Hospital Wales, Cardiff, UK); N Ranieri (St Vincent’s Hospital, Melbourne, Australia); J Rice (Flinders Medical Centre, Adelaide, Australia); D Ridley (Guy’s Hospital, London, UK); F Richardson (Austin and Repatriation Medical Centre, West Heidelburg, Australia); J Roberts (Ninewells Hospital, Dundee, UK); K Rooke (Chelmsford and Essex Centre, Chelmsford, UK); J Scarlet (Waikato Hospital, Hamilton, New Zealand); N Scott (City Hospital, Nottingham UK); H Shirley (Sir Charles Gairdner Hospital, Nedlands, Australia); E Singleton (Breast Screening Centre, Aberdeen UK); A Smith (Christchurch Hospital, Christchurch, New Zealand); J Stein (Elizabeth Garrett Anderson Hospital, London, UK); B Thompson (Frenchay Hospital, Bristol, UK); R Toivanen (Pirkanmaa Cancer Society, Tampere, Finland); L Walsh (Newcastle Mater Misericordiae Hospital, Newcastle, Australia); C Watt (Western Infirmary, Glasgow, UK); F Way (Newcastle Mater Misericordiae Hospital, Newcastle, Australia); R Winter (Westmead Hospital, Sydney, Australia); A Woollett (Geelong Hospital, Geelong, Australia); M Young (Royal Hobart Hospital, Hobart, Australia).
Conflict of interest statement M Baum, J Cuzick, J Forbes, and A Howell have served as occasional consultants and advisory board members to AstraZeneca. J Forbes, M Baum, and J Cuzick are principal investigators for trials for which their institutions receive funding fron AstraZeneca. T Powles is a member of the advisory board to Eli Lilly and Pfizer and J Forbes has occasionally been a consultant to advisory boards for Pharmacia Oncology.
823
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Acknowledgments We express our gratitude and appreciation to the thousands of women volunteers who have taken part in this study during the past 10 years. We also thank the nurses and clinicians in the local centres for their continuing support and Jane Armitage for her review and coding of the deaths and cardiac and vascular events. The IBIS Trial was supported in the UK by Cancer Research UK. In Australia it was supported by the National Health and Medical Research Council via project grant numbers 209811, 980381, 950319, 920876, awarded to the ANZ Breast Cancer Trials Group, University of Newcastle.
9
10
11
12
References 1
2 3
4
5 6 7
8
Beatson GW. On the treatment of inoperable cases of carcinoma of the mamma: suggestions for a new method of treatment with illustrative cases. Lancet 1896; 2: 104–07. Lacassagne A. Hormonal pathogenesis of adenocarcinoma of the breast. Am J Cancer 1936; 14: 217–25. Pike MC, Krailo MD, Henderson BE, et al. Hormonal risk factors, breast tissue age and the age-incidence of breast cancer. Nature 1983; 303: 767–70. Thomas HV, Key TJ, Allen DS, et al. A prospective study of endogenous serum hormone concentrations and breast cancer risk in post-menopausal women on the island of Guernsey. Br J Cancer 1997; 76: 401–05. Cuzick J, Baum M. Tamoxifen and contralateral breast cancer. Lancet 1985; 1: 282. Cuzick J, Wang DY, Bulbrook RD. The prevention of breast cancer. Lancet 1986; 2: 83–86. Early Breast Cancer Trialists’ Collaborative Group. Tamoxifen for early breast cancer: an overview of the randomised trials. Lancet 1998; 351: 1451–67. Powles TJ, Tillyer CR, Jones AL, et al. Prevention of breast cancer with tamoxifen: an update on the Royal Marsden Hospital pilot programme. Eur J Cancer 1990; 26: 680–84.
824
13
14
15 16
17
18
Powles TJ, Eeles R, Ashley S, et al. Interim analysis of the incidence of breast cancer in the Royal Marsden Hospital tamoxifen randomised chemoprevention trial. Lancet 1998; 352: 98–101. Fisher B, Costantino JP, Wickerham DL, et al. Tamoxifen for prevention of breast cancer: Report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 1998; 90: 1371–87. Veronesi U, Maisonneuve P, Costa A, et al. Prevention of breast cancer with tamoxifen: preliminary findings from the Italian randomised trial among hysterectomised women. Lancet 1998; 352: 93–97. Cuzick J. Chemoprevention of breast cancer with tamoxifen. In: Hakama M, Beral V, Cuiatti E, Faivre J, Parkin DM, eds. Chemoprevention in cancer control. IARC Scientific Publication number 136. Lyon: IARC, 1996: 95–109. Cuzick J. A brief review of the current breast cancer prevention trials and proposals for future trials. Eur J Cancer 2000; 36: 1298–302. Cauley JA, Norton L, Lippman ME, et al. Continued breast cancer risk reduction in postmenopausal women treated with raloxifene: 4-year results from the MORE trial. Breast Cancer Res Treat 2001; 65: 125–34. National Health Service. Clinical terms (the Read codes). London: Stationery Office, 2001. Grady D, Wenger NK, Herrington D, et al. Postmenopausal hormone therapy increases risk for venous thromboembolic disease: the Heart and Estrogen/progestin Replacement Study. Ann Intern Med 2000; 132: 689–96. Mosca L, Collins P, Herrington DM, et al. Hormone replacement therapy and cardiovascular disease: a statement for healthcare professionals from the American Heart Association. Circulation 2001; 104: 499–503. The ATAC (Arimidex, Tamoxifen Alone or in Combination) Trialists’ Group. Anastrozole alone or in combination with tamoxifen versus tamoxifen alone for adjuvant treatment of postmenopausal women with early breast cancer: first results of the ATAC randomised trial. Lancet 2002; 359: 2131–39.
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Articles
First results from the International Breast Cancer Intervention Study (IBIS-I): a randomised prevention trial IBIS investigators*
Summary Background Three clinical trials on the use of tamoxifen to prevent breast cancer have reported mixed results. The overall evidence supports a reduction in the risk of breast cancer, but whether this benefit outweighs the risks and side-effects associated with tamoxifen is unclear. Methods We undertook a double-blind placebo-controlled randomised trial of tamoxifen, 20 mg/day for 5 years, in 7152 women aged 35–70 years, who were at increased risk of breast cancer. The primary outcome measure was the frequency of breast cancer (including ductal carcinoma in situ). Analyses were by intention to treat after exclusion of 13 women found to have breast cancer at baseline mammography. Findings After median follow-up of 50 months (IQR 32–67), 69 breast cancers had been diagnosed in 3578 women in the tamoxifen group and 101 in 3566 in the placebo group (risk reduction 32% [95% CI 8–50]; p=0·013). Age, degree of risk, and use of hormone-replacement therapy did not affect the reduction. Endometrial cancer was nonsignificantly increased (11 vs 5; p=0·2) and thromboembolic events were significantly increased with tamoxifen (43 vs 17; odds ratio 2·5 [1·5–4·4], p=0·001), particularly after surgery. There was a significant excess of deaths from all causes in the tamoxifen group (25 vs 11, p=0·028). Interpretation Prophylactic tamoxifen reduces the risk of breast cancer by about a third. Temporary cessation of tamoxifen should be considered and the use of appropriate antithrombotic measures is recommended during and after major surgery or periods of immobilisation. Prophylactic use of tamoxifen is contraindicated in women at high risk of thromboembolic disease. The combined evidence indicates that mortality from non-breast-cancer causes is not increased by tamoxifen. The overall risk to benefit ratio for the use of tamoxifen in prevention is still unclear, and continued follow-up of the current trials is essential. Lancet 2002; 360: 817–24 See Commentary page 813
*Investigators listed at end of paper Correspondence to: Prof J Cuzick, IBIS Trials Centre, Cancer Research UK, PO Box 123, London WC2A 3PX, UK (e-mail: [email protected])
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
Introduction The belief that oestrogen is the primary promotional factor for breast cancer has a long history and is now well established.1–4 Attempts to control the incidence of breast cancer by decreasing the oestrogenic stimulus are more recent. The marked reduction in the rate of new contralateral tumours found when tamoxifen was used to treat early cancer,5 and the drug’s apparently low side-effect profile, led to the proposal that tamoxifen prophylaxis might be a suitable approach for reduction of the risk of breast cancer in high-risk women.6 A decrease of about 50% in the rate of new contralateral tumours with 5 years of tamoxifen treatment has been noted in several studies, which have been summarised in the Early Breast Cancer Trialists’ Collaborative Group overview.7 These early results led to the initiation of a tamoxifen prevention pilot trial at the Royal Marsden Hospital, London, UK, in 1986.8,9 That trial has now evolved into a full-scale prevention study. Three other studies started in 1992: the National Surgical Adjuvant Breast and Bowel Project P-1 Study (NSABP-P1),10 the Italian National trial,11 and the International Breast Cancer Intervention Study (IBIS) in the UK, Australia, New Zealand, and some European countries.12 Three of these studies have now published initial findings, with mixed results. A 50% decrease in breast-cancer incidence was observed in the North American trial, but little or no reduction was seen in the two European trials. The possible reasons for the differences in results have been the subject of much discussion, but no clear explanation is available. Results from all the trials are statistically compatible with a 30–40% decrease in incidence,13 but opinions vary substantially as to whether this effect is large enough to outweigh the established side-effects of tamoxifen, notably an increase in menopausal symptoms, vascular events, and endometrial abnormalities, including cancer. Initial results are also available on the effect on breastcancer incidence of another specific oestrogen-receptor modulator, raloxifene.14 These early results suggest a reduction of as much as 70%, and the side-effect profile is similar to that of tamoxifen, except possibly for endometrial cancer. A direct comparison of these two specific oestrogenreceptor modulators is the subject of a second North American trial. Here we report the initial results from IBIS-I, comparing tamoxifen with placebo. The primary endpoint was frequency of breast cancer (including ductal carcinoma in situ). Predefined subgroups were oestrogen receptor status of the cancer, use of hormonal replacement therapy, and age (<50, 50 years) Secondary endpoints were other cancers, thromboembolic events, cardiovascular events, and cause-specific mortality.
Methods Participants Women aged 35–70 years were entered into the trial from April, 1992, until March, 2001 (figure 1). Eligible women 817
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
70
7152 randomised Placebo
Tamoxifen
3574 assigned placebo 8 excluded (breast cancer at entry) 3566 in primary analysis
3578 assigned tamoxifen 5 excluded (breast cancer at entry) 3573 in primary analysis
Proportion of women (%)
60
Placebo Tamoxifen
50 40 30 20 10
3528 (98·9%) began treatment
3523 (98·6%) began treatment
959 (26·9%) completed 5 years of treatment
837 (23·4%) completed 5 years of treatment
0 <2
3–5 5–10 2–3 Predicted absolute risk (%)
>10
Figure 2: Distribution of predicted absolute risk of breast cancer in the 10 years after randomisation according to allocated treatment Based on a program by J Tyrer (unpublished).
Figure 1: Trial profile
had risk factors for breast cancer indicating at least a twofold relative risk for ages 45–70 years, a four-fold relative risk for ages 40–44 years, and a roughly ten-fold relative risk for ages 35–39 years. Specific entry criteria are shown in table 1. Reasons for exclusion were any previous invasive cancer (except non-melanoma skin cancer), a previous deep-vein thrombosis or pulmonary embolism, current use of anticoagulants, or a life expectancy judged to be less than 10 years. Women who were pregnant or wished to become pregnant were also excluded. Women were recruited through attendance at family history clinics, relatives of women with breast cancer, breast screening centres, general practitioners, and the media. Approval of the local ethics committee from each centre was obtained, and written informed consent was required for all women. The trial was done under the auspices of the UK Coordinating Committee for Cancer Research (now part of the National Cancer Research Network) and was supported by the Imperial Cancer Research Fund, the Cancer Research Campaign (now merged as Cancer Research UK), and the Department of Health, under a Clinical Trials Certificate from the UK Medicines Control Agency. In Australia and New Zealand, the trial was done by the Australia New Zealand Breast Cancer Trials Group. An independent data-monitoring committee reviewed data regularly and reported to the IBIS Working Party, who were responsible for the overall conduct of the trial. At their
January, 2002, meeting the data-monitoring committee decided that the data were sufficiently mature to be published. Design and procedures All women had had a baseline mammogram within the previous 12 months to exclude pre-existing cancer, and a blood sample was requested for cholesterol assays and marker studies. Women were randomly assigned 5 years of treatment with tamoxifen 20 mg/day or matching placebo. For logistic reasons, the baseline mammogram was taken at the time of randomisation for some women, and 13 were subsequently excluded because breast cancer was diagnosed on their baseline mammogram (figure 1). Women were followed up every 6 months during the 5 years of active treatment and by annual questionnaire or clinical visit thereafter for up to 5 years. Details of any side-effects were collected at every visit, both as predefined items and free text and coded according to the NHS Read codes.15 Symptoms, diagnoses, and procedures were each recorded separately. All concomitant medications were recorded. Mammography was done every 12–18 months, and blood samples were requested again at 1 year and 5 years. Randomisation was done centrally by telephone or fax to a central office in London, UK for European centres and in Sydney, Australia for Australia and New Zealand. Randomisation was stratified by centre and balanced in
Risk factor
Placebo (n=3566)
Tamoxifen (n=3573)
First-degree relative who developed breast cancer at or before age 50 First-degree relative with bilateral breast cancer* Two or more first-degree or second-degree relatives with breast cancer† Lobular carcinoma in situ Atypical hyperplasia Nulliparous and a first-degree relative who developed breast cancer Benign biopsy and first-degree relative who developed breast cancer Risk equivalent‡
1744 (48·9%) 601 (16·9%) 2206 (61·9%) 44 (1·2%) 104 (2·9%) 325 (9·1%) 132 (3·7%) 143 (4·0%)
1689 (47·3%) 579 (16·2%) 2204 (61·7%) 44 (1·2%) 97 (2·7%) 314 (8·8%) 123 (3·4%) 177 (5·0%)
All criteria permit entry from age 45 years. Atypical hyperplasia permits entry from age 40 and lobular carcinoma in situ from age 35. Total number adds up to 148% of total entry because some women met several entry criteria. *Eligible from age 40 if relative had cancer before age 50 and at age 35 if relative’s cancer was diagnosed before age 40. †Eligible from age 40 if both relatives developed breast cancer before age 50 and from age 35 if both relatives were first degree and developed breast cancer before age 50. ‡Risk-equivalent women were those with a strong family history, not fitting specific categories, but judged to be at higher risk than the minimum eligibility category by the study chairman.
Table 1: Entry criteria and distribution by treatment group
818
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
Proportion of women (%)
ARTICLES
40 35 30 25 20 15 10 5 40
Number of cancers
Placebo Tamoxifen
Placebo
Tamoxifen
101
69
0·68 (0·50–0·92)
Age (years) <50 50
39 62
25 44
0·64 (0·39–1·05) 0·70 (0·48–1·04)
HRT use During the trial Before trial only Never
38 21 42
29 9 31
0·76 (0·47–1·23) 0·43 (0·20–0·91) 0·73 (0·46–1·17)
Invasiveness Ductal carcinoma in situ Invasive
16 85
5 64
0·31 (0·12–0·82) 0·75 (0·54–1·04)
Grade* Low Intermediate High Unknown
14 37 27 7
14 27 21 2
1·00 (0·48–2·06) 0·73 (0·44–1·19) 0·77 (0·44–1·36) ··
Nodal status* Positive Negative Not done
23 59 3
16 44 4
0·69 (0·37–1·30) 0·74 (0·50–1·10) ··
ER status* Positive Negative Unknown
63 19 3
44 19 1
0·69 (0·47–1·02) 1·00 (0·53–1·87) ··
Tumour size* 1 cm 1–2 cm 2–5 cm >5 cm Metastatic only
32 36 17 0 0
23 25 10 5 1
0·72 (0·42–1·22) 0·69 (0·42–1·15) 0·59 (0·27–1·26) ·· ··
Total
<40
40–44 45–49 50–54 55–59 60–64 65 Age (years)
Figure 3: Age distribution of participants
blocks of eight. The lists were then randomly permuted again in blocks of six to ten (chosen randomly) to ensure that the last member of each block was not predictable. Endpoints and deaths were externally reviewed and coded with masking of treatment allocation. Primary cause rather than underlying cause was used (eg, deaths from pulmonary embolism after cancer surgery were coded as from pulmonary embolism). Statistical analysis The trial was powered to detect a 40% complianceadjusted reduction in the rate of breast cancer including ductal carcinoma in situ. For 90% power, 164 events would be required. Risk profiles were computed retrospectively by means of a specially developed program that included detailed family-history variables and other standard risk factors (J Tyrer et al, unpublished). Analyses were mainly based on comparison of proportions by odds ratios, and Fisher’s exact values were used where appropriate. Major comparisons were expressed as odds ratios, which are very good approximations to relative risks or hazard ratios for rare events and equal follow-up. All p values are two-sided and confidence intervals were based on a normal approximation. Analyses were by intention to treat, after exclusion of the 13 women found to have breast cancer at baseline. Role of the funding source The sponsors of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The manufacturers (AstraZeneca) supplied tamoxifen and matching placebo without charge and provided technical advice, but were not involved in the conduct or analysis of the trial. Placebo (n=3566) Tamoxifen (n=3573) Demography Mean (SD) age, years Postmenopausal
50·8 (6·7) 1740 (48·8%)
50·7 (7·0) 1761 (49·3%)
HRT use Before entry During trial Ever
1443 (40·5%) 1399 (39·2%) 1783 (50·0%)
1469 (41·1%) 1445 (40·4%) 1849 (51·7%)
162·9 (6·4) 71·4 (14·0) 26·9 (5·1)
162·8 (6·6) 71·7 (14·5) 27·0 (5·3)
1283 (36·0%) 737 (20·7%) 207 (5·8%) 327 (9·2%)
1232 (34·5%) 711 (19·9%) 229 (6·4%) 281 (7·9%)
Anthropometry Mean (SD) height, cm Mean (SD) weight, kg Mean (SD) body-mass index, kg/m2 Hysterectomy All With both ovaries retained One ovary removed Both ovaries removed
Data are number of women unless otherwise stated.
Table 2: Baseline characteristics and HRT use
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
Odds ratio (95% CI) tamoxifen vs placebo
*For invasive cancers only (n=85 placebo, n=64 tamoxifen). None of the interactions was significant.
Table 3: Breast cancers and characteristics by treatment allocation
Results A total of 7139 women were included in the analysis (figure 1, table 1). The largest risk group was women who had two or more first-degree or second-degree relatives with breast cancer (62%). In all, 97% of the participants reported some family history, and 8% had a benign lesion associated with increased risk of breast cancer. The projected 10-year risk based on family history and other established risk factors is shown in figure 2. The yearly frequency of breast cancer in the absence of treatment was projected to be 7·50 per 1000; the actual frequency in the placebo group was 6·74 per 1000, which does not differ significantly from the projected frequency. The mean age was 50·8 years (SD 6·9); 54·7% of the women were between the ages of 45 and 54 (figure 3). 60% were from the UK, 37% from Australia or New Zealand, and 3% from the rest of Europe. 49% were postmenopausal and 41% had previously used hormone-replacement therapy (HRT). Further demographic details are shown in table 2. The cut-off date of follow-up for the analysis was Jan 1, 2002. Median follow-up was 50 months (IQR 32–67) and a total of 29 967 woman-years of follow-up have been accrued (14 969 in the placebo group and 14 998 in the tamoxifen group). The numbers of woman-years of treatment were 11 539 and 10 370, respectively. Full compliance to 5 years was estimated to be 64% in the tamoxifen group and 74% in the placebo group (p<0·001). 40% of women used HRT at some time during the active treatment. At the time of data lock, 25% of women had completed a full 5 years of treatment (959 [26·9%] placebo vs 837 [23·4%] tamoxifen) and a further 47% were still on treatment (1760 [49·4%] vs 1574 [44·0%]).
819
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
1·00 0·92
8 vs 10
12 vs 15
Odds ratio
17 vs 25
14 vs 19
95% CI
0·68 0·50
8 vs 14
10 vs 18
0·25
0 1
2
3 4 Year of follow-up
5
>5
Figure 4: Risk reduction with tamoxifen by year of follow-up Numbers on curve are the number of cancers in the tamoxifen group versus the number in the placebo group.
170 breast cancers (including ductal carcinoma in situ) have been recorded (table 3). The rate is 32% (95% CI 8–50) lower in the tamoxifen group than in the placebo group (tamoxifen 69, placebo 101; p=0·01). A reduction in risk with tamoxifen was apparent for both invasive disease (64 vs 85; reduction 25%) and non-invasive disease (five vs 16; reduction 69%). The difference between these two reductions was not significant. Of the invasive cancers, 69% were node negative, 72% were oestrogen-receptor (ER) positive, and 78% were 2 cm or less in diameter. Nodal status, size, and grade were similar in both study groups. The reduction in risk of confirmed ER-positive invasive tumours with tamoxifen was 31%. There was no reduction in the risk of ER-negative invasive tumours. There were five large tumours (diameter more than 5 cm) in the tamoxifen group and none in the placebo group. Age and use of HRT did not significantly affect the risk reduction (table 3). There was no evidence that the degree of protection changed over the 5 years of treatment (figure 4). There are as yet insufficient data for assessment of whether protection continues after treatment has stopped. Only four deaths from breast cancer have been reported (two in each study group), a number that is well below the 9·5 deaths expected from the observed tumour characteristics and survival predicted from the Early Breast Cancer Trialists Collaborative Group overview. Event
Number of cases
p
Placebo Tamoxifen Venous thromboembolism (excluding superficial thrombophlebitis) All thromboembolism 17 43 0·001 Occurring within 3 months of leg surgery 5 20 0·004 or fracture Spontaneous 12 23 0·09 Pulmonary embolism 10 13 0·68 Deep-vein thrombosis 5 24 0·0005 Thrombosis (other) 2 6 0·29 Thrombophlebitis
9
27
0·004
Cerebrovascular All Stroke or cerebrovascular accident Transient ischaemic attack
17 11 6
16 13 3
0·86 0·84 0·34
Cardiac All Myocardial infarction Coronary revascularisation Other revascularisation Other cardiovascular events Angina
63 5 5 2 17 34
73 5 5 8 16 39
0·44 1·00 1·00 0·11 0·86 0·64
Table 4: Thromboembolic, cerebrovascular, and cardiac events according to treatment
820
However, the median follow-up from diagnosis is still short (27 months [IQR 15–42]), and all women were having regular mammography. A non-significant two-fold excess of endometrial cancer was found in the tamoxifen group (11 vs five in the placebo group; odds ratio 2·20 [95% CI 0·80–6·06], p=0·2). Most of these cancers were in women who were older than 50 years at randomisation (ten tamoxifen group vs three placebo group), and all the women affected were postmenopausal at diagnosis. All but one of these cancers (a low-grade sarcoma in the placebo group) were adenocarcinomas, and all but one (a stage II case in the placebo group) were of FIGO stage I. Most of the tumours were of low (four vs two) or intermediate (six vs one) grade. There was no evidence that endometrial cancer was linked to HRT use or an interaction between HRT and tamoxifen. Ten of the women with endometrial cancers had never used HRT (seven vs three), only four had used it during the trial (two vs two), and two in the tamoxifen group had used it before the trial. There have been no deaths from endometrial cancer. The rate of endometrial cancer in the placebo group (34 per 100 000 woman-years) was similar to population rates for the UK and Australia. Overall, cancers other than those of the breast and endometrium were equally distributed between the two study groups (39 in each group), and no cancer differed significantly in frequency between the tamoxifen and placebo groups (colorectal nine vs six; stomach one vs three; liver none vs two; pancreas one vs none; lung three vs none; larynx one vs none; melanoma eight vs seven; bladder or kidney two vs three; ovary six in each group; endocrine two vs seven; meningioma two in each group; haemological or lymphatic two vs three; and two primary unknown cancers in the tamoxifen group). The rate of venous thromboembolic events was about 2·5 times higher in the tamoxifen group than in the placebo group (95% CI 1·5–4·4; p=0·001; table 4). 25 (42%) of these events occurred within 3 months of major surgery or after long-duration immobility; 20 were in the tamoxifen group (p=0·004). This feature was the major determinant of thromboembolic events. There was no indication of any synergy between tamoxifen and HRT and some evidence for a negative interaction. There were almost twice as many spontaneous venous thromboembolic events with tamoxifen than with placebo, but this difference was not significant (p=0·09). An increase in superficial thrombophlebitis was also noted. There were no differences in the numbers of cerebrovascular accidents, myocardial infarctions, or other vascular events. A detailed exploration of factors affecting vascular events will be published elsewhere. Details of any reported side-effects were collected at each follow-up visit. Although very large numbers of Number of events (%)
Side-effect Gynaecological or vasomotor Headaches and migraines All fractures Osteoporotic fractures (hip, spine, wrist, or forearm) Breast complaints Nail changes Eye (excluding cataracts) Cataracts
p
Placebo (n=3566)
Tamoxifen (n=3573)
2414 (67·7%) 1067 (29·9%) 127 (3·6%) 40 (1·1%)
2922 (81·8%) <0·0001 997 (27·9%) 0·13 116 (3·3%) 0·52 45 (1·3%) 0·66
675 (18·9%) 96 (2·7%) 376 (10·5%) 37 (1·0%)
525 (14·7%) <0·0001 148 (4·1%) 0·001 373 (10·4%) 0·94 38 (1·0%) 1·00
Table 5: Side-effects reported at any time and of any severity, according to allocated treatment
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Number of events (%)
p
Premenopausal Placebo (n=1826) Procedures Hysteroscopy Pelvic ultrasonography Dilation and curettage Hysterectomy Oophorectomy Symptoms Vasomotor symptoms Vaginal discharge Vaginal dryness Abnormal bleeding Endometrial polyps Uterine fibroids Amenorrhoea Thrush/candida Prolapse Ovarian cysts and lumps Endometriosis
Postmenopausal Tamoxifen (n=1810)
Placebo (n=1740)
Tamoxifen (n=1763)
107 (5·9%) 89 (4·9%)
136 (7·5%) 136 (7·5%)
31 (1·8%) 43 (2·5%)
92 (5·2%) 73 (4·1%)
<0·0001 <0·0001
76 (4·2%)
106 (5·9%)
18 (1·0%)
72 (4·1%)
<0·0001
76 (4·2%) 53 (2·9%)
118 (6·5%) 76 (4·2%)
28 (1·6%) 14 (0·8%)
36 (2·0%) 27 (1·5%)
0·002 0·006
988 (54·1%)
1233 (68·0%)
849 (48·8%)
1219 (69·0%)
<0·0001
266 (14·6%) 329 (18·0%) 565 (30·9%) 43 (2·4%) 71 (3·9%) 77 (4·2%) 36 (1·9%) 36 (1·9%) 33 (1·8%)
517 (28·6%) 379 (20·9%) 669 (36·9%) 69 (3·8%) 84 (4·6%) 171 (9·4%) 103 (5·7%) 33 (1·8%) 86 (4·8%)
237 (13·6%) 394 (22·6%) 113 (6·5%) 22 (1·3%) 11 (0·6%) ·· 28 (1·6%) 45 (2·6%) 9 (0·5%)
509 (28·9%) 403 (22·9%) 173 (9·8%) 61 (3·5%) 22 (1·2%) ·· 84 (4·8%) 52 (2·9%) 15 (0·9%)
<0·0001 0·10 <0·0001 <0·0001 0·09 <0·0001 <0·0001 0·81 <0·0001
16 (0·9%)
14 (0·8%)
5 (0·3%)
6 (0·3%)
0·88
p values are for premenopausal and postmenopausal women combined.
Table 6: Gynaecological procedures and gynaecological and vasomotor symptoms in premenopausal and postmenopausal women (defined at randomisation) according to allocated treatment
Patient Cause of death
Age at randomisation (years)
Time on treatment (months)
Time to diagnosis (months)
HRT use
Tamoxifen group 1 Breast cancer 2 Breast cancer 3 Laryngeal cancer 4 Stomach cancer 5 Colorectal cancer 6 Colorectal cancer 7 Colorectal cancer 8 Colorectal cancer 9 Carcinomatosis 10 Carcinomatosis 11 Myeloma 12 Myeloma 13 Pulmonary embolism 14 Pulmonary embolism 15 Deep-vein thrombosis 16 Subarachnoid haemorrhage 17 Cerebral aneurysm 18 Dissecting aortic aneurysm 19 Myocardial infarction 20 Myocardial infarction 21 Pericarditis 22 Cardiac failure 23 Cardiac failure 24 Road traffic accident 25 COAD
48 53 49 53 47 61 63 64 52 62 54 68 56 47 58 69 50 49 60 49 43 47 69 45 49
20 6 3 19 3 44 18 60 42 9 60 10 42 5 42 1 2 10 7 17 6 22 9 16 12
20 26 64 19 3 44 20 72 42 9 36 6 42 5 42 1 47 31 to death 7 17 39 22 41 16 22
Never 96 months before and 12 months during trial 1 month before and 2 months during trial 56 months before and 12 months during trial Never 6 months before and 11 months during trial Never 84 months before and 36 months during trial Never Never Never 96 months before and 18 months duirng trial 5 months before and 42 months during trial Never 84 months before and 23 months during trial 144 months before and 1 month during trial Never 42 months before and 12 months during trial Never Never Never Never Never 5 months before trial 84 months before and 10 months during trial
Placebo group 26 Breast cancer 27 Breast cancer 28 Ovarian cancer 29 Ovarian cancer 30 Colorectal cancer 31 Burkitt’s lymphoma 32 Pulmonary embolism 33 Pulmonary embolism 34 Sub-arachnoid haemorrhage 35 Ischaemic bowel disease 36 Air crash
50 54 54 59 52 49 47 62 56 56 38
8 48 6 44 12 12 48 1 3 12 5
8 48 15 44 12 32 48 1 3 77 5
Never Never 60 months before trial Never 72 months before and 12 months during trial Never Never 192 months before and 1 month during trial Never Never Never
COAD=chronic obstructive airway disease.
Table 7: Detailed breakdown of deaths
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
821
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Marsden NSABP-P1 Italian IBIS-I All trials
0·3
0·5
0·9 1·0
2·0
5·0
Odds ratio Figure 5: Death from any cause in the tamoxifen prevention trials Error bars=95% CI.
side-effects were recorded (table 5), the only major grouping that showed differences was vasomotor and gynaecological reports, which were about 21% higher in the tamoxifen than the placebo group, and breast complaints, which were 22% lower. As expected, there were higher proportions of hot flushes, vaginal discharge, and abnormal vaginal bleeding in the tamoxifen group (table 6). Rates of pelvic ultrasonography, dilation and curettage, and hysteroscopy were also significantly higher, and there were more women with endometrial polyps. The hysterectomy rate was 2·7% in the placebo group and 4·2% in the tamoxifen group (p=0·002). In premenopausal women, ovarian cysts and amenorrhoea were more than twice as common on tamoxifen. Reports of vaginal thrush were substantially increased in the tamoxifen group in both premenopausal and postmenopausal women. A new finding was an increase in reports of brittle nails (both fingernails and toenails) in the tamoxifen group. To date, no differences have been found for osteoporotic or non-osteoporotic fractures. No difference was found in the frequency of cataracts or other eye complaints. By contrast, the rate of benign breast disease was 31% lower in the tamoxifen group than in the placebo group (182 vs 262, p=0·0001). The largest reductions were in cysts (57 vs 123, p<0·0001), fibroadenomas (nine vs 19, p=0·06), and fibrocystic disease (16 vs 25, p=0·16). The effect was mostly seen in premenopausal women. Reports of breast pain were also reduced by 32% (70 vs 103, p=0·01) in the tamoxifen group. At the 1-year follow-up, mean weight had increased by 0·60 kg in the placebo group and 0·04 kg in the tamoxifen group (p<0·0001 for comparison of mean differences). At year 5, the mean weight gain was similar for the two groups (2·00 kg placebo vs 2·20 kg tamoxifen, p=0·8). Fuller details on the timing and severity of side-effects will be the subject of a separate report. The death rate from all causes was significantly higher in the tamoxifen group than in the placebo group (25 vs 11; p=0·028; table 7). Increases are seen for cancers other than breast cancer, pulmonary embolisms, other vascular causes, and cardiac deaths. The cancers were at a range of different sites, and the lack of an increase in cancer frequency suggests that this may be a chance finding. The cardiac and vascular causes of death are also varied, but the increase in thromboembolic events indicates that the extra deaths from this cause may be attributable to tamoxifen.
822
Updated data predict a 38% (95% CI 28–47) reduction in breast cancer incidence, and our value of 32% is compatible with this estimate. Also, the doubling of the rate of endometrial cancer and the 2·5-fold relative risk for vascular events are similar to findings of the other trials. What is unexpected is the increased number of deaths. This increase was not seen in the other prevention trials; lower, though not significantly so, numbers of deaths with tamoxifen were seen in NSABP-P1 and the Italian trial. The only cause for which there is any consistent evidence for an increase in mortality is thromboembolic disease. This conclusion is supported by the clear excess in the thromboembolicevent rate and the small but significant increased mortality from pulmonary embolism in the overview of tamoxifen use in the adjuvant setting. However, there is no evidence for an increase in deaths from other cancers, cardiac causes, or other vascular causes, and overall mortality is not increased when all of the evidence is combined (figure 5).
Discussion This study has confirmed that tamoxifen can reduce the risk of breast cancer in healthy women during the active treatment phase. The 32% reduction in risk seen in the study is consistent with the cumulative results when all of the chemoprevention trials are combined. The increased frequency of gynaecological problems, particularly the increased requirement for hysterectomy and oophorectomy, is of some concern. It may be a result of the observed higher proportions of women with abnormal vaginal bleeding, endometrial polyps, endometrial thickening, and ovarian cysts. The indications for hysterectomy and oophorectomy in these circumstances have not been clearly established. New findings were increases in vaginal thrush and brittle nails. The latter has been associated with decreased oestrogen concentrations at the menopause; however, both these side-effects were uncommon and caused only minor concern. The low rate of death from breast cancer highlights the importance of a full assessment of side-effects. The increased mortality in the tamoxifen group in this trial is of concern, but it may be due partly to statistical variability. The cancers were at a range of sites, and in two cases there had been only 3 months of treatment. The cardiovascular deaths are also varied. However, the increase in thromboembolic events seen in all trials and the increase, though smaller, in thromboembolic deaths seen in IBIS-I and the NSABP-P1 study indicate that thromboembolism is the most important complication of tamoxifen use, and that every effort should be taken to reduce this risk. The risk seems to be similar to that for women taking HRT,16 and similar precautions should be adopted.17 In particular, most of the risk and all of the deaths in the tamoxifen group occurred after surgery, and a wise precaution would be to discontinue tamoxifen before any surgery and to ensure that appropriate antithrombotic measures are provided during surgery and that tamoxifen is not restarted until full mobility has returned. Similar recommendations should be followed for women who become immobile for any reason. Although tamoxifen is unquestionably valuable in the adjuvant setting, some of its oestrogen-agonist properties restrict its ultimate usefulness—for both efficacy and side-effects. Early results18 show that anastrozole is more effective than tamoxifen in reducing recurrence (by about 25% in ER-positive tumours), and this drug seems to be even more promising in reducing the risk of new
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
contralateral tumours (58% reduction compared with tamoxifen). Furthermore, the agonist properties of tamoxifen account for its main side-effects of thromboembolic disease and endometrial cancer, which do not occur with aromatase inhibitors. However, new issues in relation to bone loss and other potential effects of long-term oestrogen deprivation might arise. The use of an aromatase inhibitor in the preventive setting is attractive and will be considered as an option in the forthcoming IBIS-II trial. Although when used as adjuvant therapy for breast cancer, tamoxifen can clearly reduce the risk of recurrence and death, at present the overall risk to benefit ratio in the preventive setting is still unclear. Further long-term follow-up to study breast-cancer incidence and mortality, other causes of death, and side-effects in the current trials remains essential. Writing Committee—J Cuzick, J Forbes, R Edwards, M Baum, S Cawthorn, A Coates, H Hamed, A Howell, T Powles.
Data Monitoring Committee—G Clunie (Melbourne, Australia); R Collins (CTSU, Oxford, UK); N Day (Strangeways Laboratories, Cambridge, UK; Chairman); J Northover (St Mark’s Hospital, London, UK).
IBIS Working Party and Principal Investigators—E Abdi (Bendigo Hospital, Bendigo, Australia); E Anderson (Breast Screening Centre, Edinburgh, UK); C Atkinson (Christchurch Hospital, Christchurch, New Zealand); M Baum (University College London, London, UK; Clinical Adviser); J Beith (Royal Prince Alfred Hospital, Sydney, Australia); R Bell (Geelong Hospital, Geelong, Australia); A Bird (Moorfields Hospital, London, UK; Ophthalmology Adviser); S Birrell (Flinders Medical Centre, Adelaide, Australia); R Blamey (Nottingham City Hospital, Nottingham, UK); J Boyages (Westmead Hospital, Sydney, Australia); K Buser (Engeriedspital, Bern, Switzerland); I Campbell (Waikato Hospital, Hamilton, New Zealand); S Cawthorn (Frenchay Healthcare Trust, Bristol, UK; Clinical Adviser); C Chapman (Oxford Radcliffe Hospital, Oxford, UK); M Chipman (Austin and Repatriation Medical Centre, West Heidelburg, Australia); A Coates (Royal Prince Alfred Hospital, Sydney, Australia; Clinical Adviser); J P Collins (Royal Melbourne Hospital, Melbourne, Australia); P Craft (Canberra Hospital, Canberra, Australia); J Cuzick (Cancer Research UK, London, UK; Chairman); L Denton (Glenfield Hospital NHS Trust, Leicester, UK); J Dewar (Sir Charles Gairdner Hospital, Nedlands, Australia); M Dowsett (Royal Marsden Hospital, London, UK); H Earl (Addenbrooke’s Hospital, Cambridge, UK); D Eccles (Princess Anne Hospital, Southampton, UK); R Edwards (Cancer Research UK, London, UK; Trial Statistician); G Evans (Christie Hospital NHS Trust, Manchester, UK); L Fallowfield (University of Sussex, Brighton, UK); I Fentiman (Guy’s Hospital, London, UK); J Forbes (ANZ BCTG, University of Newcastle, Newcastle Mater Hospital, Australia; Clinical Adviser); M Friedlander (Prince of Wales Hospital, Sydney, Australia); C Furnival (Wesley Medical Centre, Brisbane, Australia); J Garcia (Hospiat Universitario “Principe de Asturias”, Madrid, Spain); W George (Western Infirmary, Glasgow, UK); F J Gilbert (University of Aberdeen, Aberdeen, UK); A Goldhirsch (Oncology Institute of Southern Switzerland, Lugano, Switzerland); H Hamed (Guy’s Hospital, London, UK; Clinical Adviser); A Hanby (St James’s University Hospital, Leeds, UK; Trial Pathologist); S Hart (Monash Medical Centre, Clayton, Australia); J Hearne (NCRN, London, UK; Observer); A Henry (Hopital Jolimont, St Paul, Belgium); C Hirst (Wesley Breast Clinic, Brisbane, Australia); C Holcombe (Royal Liverpool University Hospital, Liverpool, UK); K Holli (Tampere University, and University Hospital, Finland); A Howell (Christie Hospital NHS Trust, Manchester, UK; Clinical Adviser); I Jackson (Astra Zeneca Pharmaceuticals, Macclesfield, UK; Observer); J Kirk (Westmead Hospital, Sydney, Australia); M Lansdown (St James’s University Hospital, Leeds, UK); K Law (Cancer Research UK, London, UK; Observer); M Lee (City Hospital NHS Trust, Birmingham, UK); T Lennard (University of Newcastle upon Tyne, Newcastle Upon Tyne, UK); F MacNeil (Essex County Hospital, Colchester, UK); P Maddox (Royal United Hospital, Bath, UK); R Mansel (University Hospital of Wales, Cardiff, UK); P McAleese (Monaghan General Hospital, Monaghan, Ireland); J MacKay (Addenbrooke’s NHS Trust, Cambridge, UK); K MacMichael (Huddersfield Royal Infirmary, Huddersfield, UK); C Mitine (Hopital Jolimont, St Paul, Belgium); C Normand (London School of Hygiene and Tropical Medicine, London, UK); W Odling-Smee (Belfast City Hospital, Belfast, UK); T Oivanen (Pirkanmaa Cancer Society, Tampere, Finland); C O’Neill (Cancer Research UK, London, UK; Senior Trial Coordinator); O Pagani (Ospedale Regionale della Beata Vergine, Mendrisio, Switzerland); T Powles (Royal Marsden Hospital, London, UK; Clinical Adviser); Z Raytor (Bristol Royal Infirmary, Bristol, UK); B Richmond (St Mary’s Hospital, London, UK); J Robertson
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
(Nottingham City Hospital, Nottingham, UK); R Sainsbury (University College Hospital, London, UK); T Sahmoud (Astra Zeneca Pharmaceuticals, Macclesfield, UK; Observer); P Sauven (Broomfield NHS Trust, Chelmsford, UK); R J Simes (University of Sydney, Sydney, Australia); R Snyder (St Vincent’s Hospital, Melbourne, Australia); R Stewart (Kettering General Hospital, Kettering, UK); A Stotter (Glenfield Hospital, Leicester, UK); A Thompson (Ninewells Hospital and Medical School, Dundee, UK); J Toy (Cancer Research UK, London, UK; Observer); P Twentyman (UKCCCR, London, UK; Observer); C Underhill (Border Medical Oncology, Wodonga, Australia); R Ward (St Vincent’s Hospital, Sydney, Australia); A Wilkinson (Belfast City Hospital, Belfast, UK); S Wilkinson (Royal Hobart Hospital, Hobart, Australia); J Williamson (City Hospital NHS Trust, Birmingham, UK); C Wynne (Christchurch Hospital, Christchurch, New Zealand). IBIS Coordinating Centre—Cancer Research UK, London, UK: C Duggan, R Edwards, S Falk, J Hickman, R Kealy, C O’Neill, E Pinney, S Tilling. IBIS–ANZ BCTG Operations/Statistical Office—H Dhillon, V Gebski, D Lindsay, A Melmeth, A Newton, L Paksec, M Seccombe, R Thornton. Study Coordinators—J Affen (Christie Hospital NHS Trust, Manchester, UK); S Bailey (University Hospital Wales, Cardiff, UK); S Bensted (Geelong Hospital, Geelong, Australia); J Bickerton (Monash Medical Centre, Clayton, Australia); K Brimson (Wesley Medical Centre, Brisbane, Australia); R Buder (Flinders Medical Centre, Adelaide, Australia); M Busteed (Canberra Hospital, Canberra, Australia); A Carnall (Prince of Wales Hospital, Sydney, Australia); O Claber (Northern Genetics Services, Newcastle upon Tyne, UK); S Clark (Bendigo Hospital, Bendigo, Australia); M Cross (Royal South Hants Hospital, Southampton, UK); S Daly (Flinders Medical Centre, Adelaide, Australia); A Deluca (Royal Melbourne Hospital, Melbourne, Australia); A Dowd (Geelong Hospital, Geelong, Australia); S Drummond (Breast Screening Unit, Edinburgh, UK); S Durell (Oxford Radcliffe Hospital, Oxford, UK); E Etherington (Huddersfield Royal Infirmary, Huddersfield, UK); W Fowler (Royal Hobart Hospital, Hobart, Australia); C Gano (Westmead Hospital, Sydney, Australia); D Garratt (Border Medical Oncology, Wodonga, Australia); C Gradige (Royal South Hants Hospital, Southampton, UK); E Granger (Guy’s Hospital, London, UK); J Gray (Belfast City Hospital, Belfast, UK); R Greenhalgh (Christie Hospital NHS Trust, Manchester, UK); L Gunn (Breast Screening Centre, Aberdeen, UK); A Hayes (Austin and Repatriation Medical Centre, West Heidelburg, Australia); J Hepper (St James’s Hospital, Leeds, UK); S Holcombe (Royal Liverpool University Hospital, Liverpool, UK); G Hoogeveen (Royal Melbourne Hospital, Melbourne, Australia); L Hughes (St Vincent’s Hospital, Melbourne, Australia); N Humphreys (Bendigo Hospital, Bendigo, Australia); S Ingelido (Wesley Breast Clinic, Brisbane, Australia); J Innes-Rowe (Sir Charles Gairdner Hospital, Nedlands, Australia); L Jolly (St Vincent’s Hospital, Sydney, Australia); A Jonson (Addenbrooke’s NHS Trust, Cambridge, UK); C Kershaw (Kettering General Hospital, Kettering, UK); T Klau (Flinders Medical Centre, Adelaide, Australia); M Lloyd (Prince of Wales Hospital, Sydney, Australia); J Magnay (Prince of Wales Hospital, Sydney, Australia); K Makinson (Royal Liverpool University Hospital, Liverpool, UK); A Miller (Flinders Medical Centre, Adelaide, Australia); B Murray (Royal Prince Alfred Hospital, Sydney, Australia); S Napier (Bendigo Hospital, Bendigo, Australia); S Pearce (Frenchay Hospital, Bristol, UK); A Pickersgill (University Hospital Wales, Cardiff, UK); N Ranieri (St Vincent’s Hospital, Melbourne, Australia); J Rice (Flinders Medical Centre, Adelaide, Australia); D Ridley (Guy’s Hospital, London, UK); F Richardson (Austin and Repatriation Medical Centre, West Heidelburg, Australia); J Roberts (Ninewells Hospital, Dundee, UK); K Rooke (Chelmsford and Essex Centre, Chelmsford, UK); J Scarlet (Waikato Hospital, Hamilton, New Zealand); N Scott (City Hospital, Nottingham UK); H Shirley (Sir Charles Gairdner Hospital, Nedlands, Australia); E Singleton (Breast Screening Centre, Aberdeen UK); A Smith (Christchurch Hospital, Christchurch, New Zealand); J Stein (Elizabeth Garrett Anderson Hospital, London, UK); B Thompson (Frenchay Hospital, Bristol, UK); R Toivanen (Pirkanmaa Cancer Society, Tampere, Finland); L Walsh (Newcastle Mater Misericordiae Hospital, Newcastle, Australia); C Watt (Western Infirmary, Glasgow, UK); F Way (Newcastle Mater Misericordiae Hospital, Newcastle, Australia); R Winter (Westmead Hospital, Sydney, Australia); A Woollett (Geelong Hospital, Geelong, Australia); M Young (Royal Hobart Hospital, Hobart, Australia).
Conflict of interest statement M Baum, J Cuzick, J Forbes, and A Howell have served as occasional consultants and advisory board members to AstraZeneca. J Forbes, M Baum, and J Cuzick are principal investigators for trials for which their institutions receive funding fron AstraZeneca. T Powles is a member of the advisory board to Eli Lilly and Pfizer and J Forbes has occasionally been a consultant to advisory boards for Pharmacia Oncology.
823
For personal use. Only reproduce with permission from The Lancet Publishing Group.
ARTICLES
Acknowledgments We express our gratitude and appreciation to the thousands of women volunteers who have taken part in this study during the past 10 years. We also thank the nurses and clinicians in the local centres for their continuing support and Jane Armitage for her review and coding of the deaths and cardiac and vascular events. The IBIS Trial was supported in the UK by Cancer Research UK. In Australia it was supported by the National Health and Medical Research Council via project grant numbers 209811, 980381, 950319, 920876, awarded to the ANZ Breast Cancer Trials Group, University of Newcastle.
9
10
11
12
References 1
2 3
4
5 6 7
8
Beatson GW. On the treatment of inoperable cases of carcinoma of the mamma: suggestions for a new method of treatment with illustrative cases. Lancet 1896; 2: 104–07. Lacassagne A. Hormonal pathogenesis of adenocarcinoma of the breast. Am J Cancer 1936; 14: 217–25. Pike MC, Krailo MD, Henderson BE, et al. Hormonal risk factors, breast tissue age and the age-incidence of breast cancer. Nature 1983; 303: 767–70. Thomas HV, Key TJ, Allen DS, et al. A prospective study of endogenous serum hormone concentrations and breast cancer risk in post-menopausal women on the island of Guernsey. Br J Cancer 1997; 76: 401–05. Cuzick J, Baum M. Tamoxifen and contralateral breast cancer. Lancet 1985; 1: 282. Cuzick J, Wang DY, Bulbrook RD. The prevention of breast cancer. Lancet 1986; 2: 83–86. Early Breast Cancer Trialists’ Collaborative Group. Tamoxifen for early breast cancer: an overview of the randomised trials. Lancet 1998; 351: 1451–67. Powles TJ, Tillyer CR, Jones AL, et al. Prevention of breast cancer with tamoxifen: an update on the Royal Marsden Hospital pilot programme. Eur J Cancer 1990; 26: 680–84.
824
13
14
15 16
17
18
Powles TJ, Eeles R, Ashley S, et al. Interim analysis of the incidence of breast cancer in the Royal Marsden Hospital tamoxifen randomised chemoprevention trial. Lancet 1998; 352: 98–101. Fisher B, Costantino JP, Wickerham DL, et al. Tamoxifen for prevention of breast cancer: Report of the National Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 1998; 90: 1371–87. Veronesi U, Maisonneuve P, Costa A, et al. Prevention of breast cancer with tamoxifen: preliminary findings from the Italian randomised trial among hysterectomised women. Lancet 1998; 352: 93–97. Cuzick J. Chemoprevention of breast cancer with tamoxifen. In: Hakama M, Beral V, Cuiatti E, Faivre J, Parkin DM, eds. Chemoprevention in cancer control. IARC Scientific Publication number 136. Lyon: IARC, 1996: 95–109. Cuzick J. A brief review of the current breast cancer prevention trials and proposals for future trials. Eur J Cancer 2000; 36: 1298–302. Cauley JA, Norton L, Lippman ME, et al. Continued breast cancer risk reduction in postmenopausal women treated with raloxifene: 4-year results from the MORE trial. Breast Cancer Res Treat 2001; 65: 125–34. National Health Service. Clinical terms (the Read codes). London: Stationery Office, 2001. Grady D, Wenger NK, Herrington D, et al. Postmenopausal hormone therapy increases risk for venous thromboembolic disease: the Heart and Estrogen/progestin Replacement Study. Ann Intern Med 2000; 132: 689–96. Mosca L, Collins P, Herrington DM, et al. Hormone replacement therapy and cardiovascular disease: a statement for healthcare professionals from the American Heart Association. Circulation 2001; 104: 499–503. The ATAC (Arimidex, Tamoxifen Alone or in Combination) Trialists’ Group. Anastrozole alone or in combination with tamoxifen versus tamoxifen alone for adjuvant treatment of postmenopausal women with early breast cancer: first results of the ATAC randomised trial. Lancet 2002; 359: 2131–39.
THE LANCET • Vol 360 • September 14, 2002 • www.thelancet.com
For personal use. Only reproduce with permission from The Lancet Publishing Group.