Effects of ovarian ablation or suppression on breast cancer recurrence and

# Effects of ovarian ablation or suppression on breast cancer recurrence and survival: patient-level meta-analysis of 15 000 women in 23 randomised trials *The Lancet 2026* Articles Effects of ovarian ablation or suppression on breast cancer recurrence and survival: patient-level meta-analysis of 15 000 women in 23 randomised trials Early Breast Cancer Trialists’ Collaborative Group (EBCTCG)* Summary Background For premenopausal women with oestrogen receptor (ER)-positive early breast cancer, the additional Lancet 2026; 407: 1699–711 protective effect of ovarian function suppression (OFS, by ablation or drugs) may depend on menopausal status after *Members and writing any chemotherapy, and tamoxifen usage. We assess the effects of OFS on breast cancer outcomes among committee listed at end of the premenopausal women and how they vary by patient or tumour characteristics and receipt of other treatments. Article Correspondence to: Early Breast Cancer Trialists’ Methods We conducted a meta-analysis of individual participant data from the randomised trials comparing OFS Collaborative Group Secretariat, versus no OFS, in women with ER-positive or ER-unknown early breast cancer who were premenopausal at Clinical Trial Service Unit, randomisation and younger than 55 years. Trials were categorised by whether premenopausal status was or was not Nuffield Department of confirmed after chemotherapy (if given), and by allocation to tamoxifen. Primary outcomes were invasive breast Population Health, Richard Doll Building, Old Road Campus, cancer recurrence, breast cancer mortality, other mortality, and all-cause mortality. ER-weighted log-rank methods Oxford OX3 7LF, UK estimated event rate ratios (RRs) for ER-positive disease. bc.overview@ndph.ox.ac.uk Findings Datasets were provided for 23 of 25 identified eligible trials, comprising 18 851 (98·9%) of 19 053 randomly assigned women. Among 15 075 premenopausal women with ER-positive or ER-unknown tumours, allocation to OFS significantly reduced recurrence rates (RR 0·82, 95% CI 0·77–0·87; p<0·00001), with larger reductions in women who were confirmed premenopausal after chemotherapy (or who did not receive chemotherapy) than in those with unconfirmed premenopausal status after chemotherapy; heterogeneity p=0·0004. Among confirmed premenopausal women, recurrence reductions were larger in older trials without tamoxifen (RR 0·61, 0·52–0·71; p<0·0001) than in more recent trials of OFS plus tamoxifen versus tamoxifen (RR 0·79, 0·70–0·91; p=0·0008). In these more recent trials, the additional recurrence reduction with OFS appeared larger in women younger than 45 years than in women aged 45–54 years (RR 0·73, 0·63–0·86 vs RR 0·95, 0·75–1·21; p=0·072); in those younger than 45 years, breast cancer mortality was similarly improved (RR 0·74, 0·58–0·94; p=0·012). There was no increase in deaths without recurrence. Findings did not differ significantly by OFS method or other recorded patient or tumour characteristics. Interpretation For premenopausal women with ER-positive early breast cancer, even if chemotherapy or tamoxifen are given, OFS significantly reduces the 15-year risk of recurrence and death. Funding Nuffield Department of Population Health, University of Oxford; Cancer Research UK; the Breast Cancer Research Foundation; and the UK Medical Research Council. Copyright © 2026 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Introduction tamoxifen, and about the extent to which suppression of In women with oestrogen receptor (ER)-positive early ovarian function might or might not further reduce this breast cancer, following surgery and, where indicated, risk. radiotherapy and chemotherapy, adjuvant endocrine Early trials examined the effect of irreversible ovarian therapy can substantially improve survival.1 In ablation (with surgery or ovarian irradiation), whereas postmenopausal women, suppression of oestrogen more recent trials have generally used a gonadotrophin- production with aromatase inhibitors is more effective releasing hormone agonist (GnRHa) to achieve than treatment with the selective oestrogen receptor potentially reversible pharmacological suppression of modulator tamoxifen.2 In premenopausal women, ovarian oestrogen production.3–7 Previous meta-analyses however, ovarian oestrogen production means that by the Early Breast Cancer Trialists’ Collaborative Group although tamoxifen is effective, aromatase inhibitors (EBCTCG) of individual participant data, which were alone are not effective. Some uncertainty remains about restricted to older trials, reported fewer breast cancer the effect of circulating oestrogens on recurrence risk in recurrences and deaths with ovarian ablation or premenopausal women, particularly those taking suppression, but only in the absence of other systemic Articles Research in context Evidence before this study unbiased estimates of the risks and benefits of OFS among A previous Early Breast Cancer Trialists’ Collaborative Group premenopausal women, including those who remained (EBCTCG) meta-analysis showed that, for women with early premenopausal after any chemotherapy. For women with breast cancer, ablation of ovarian function significantly reduces confirmed premenopausal status, the results show that OFS breast cancer recurrence and death in the absence of other greatly reduces the 15-year risk of breast cancer recurrence and systemic treatments. Subsequent trials and systematic reviews death among women who were not scheduled to receive any on the effects of pharmacological ovarian suppression using a other endocrine therapy. Among women who were allocated gonadotrophin-releasing hormone agonist, and of ablation by tamoxifen (for about 4 years), the improvements were still irradiation, reported no clear benefit, so questions remain as to definite, and for women younger than 45 years they were whether ovarian function suppression (OFS) by ablation or substantial, with recurrence, distant recurrence, breast cancer drugs can materially improve long-term outcomes beyond the mortality and all-cause mortality rates during the first decade effects of tamoxifen or chemotherapy. The EBCTCG’s ongoing all reduced by about a quarter. There were few deaths from systematic searches of bibliographic databases, including causes other than breast cancer, and no increase with OFS. MEDLINE, Embase, the Cochrane Library, and meeting Numbers were insufficient for reliable subgroup analyses, but abstracts, up to Sept 1, 2025, sought all randomised trials of there was no evidence that the proportional risk reductions OFS versus no OFS, with other treatments the same in both differed by prognostic factors or method of OFS. arms. Implications of all the available evidence Added value of this study For premenopausal women with early-stage, ER-positive breast This collaborative meta-analysis collated, checked, and cancer, including those who are still premenopausal after any analysed individual patient-level data from chemotherapy, the additional long-term benefits of OFS over 15 075 premenopausal women with oestrogen receptor and above the established benefits of tamoxifen and of (ER)-positive or ER-unknown disease in 23 trials. Long-term chemotherapy can reliably inform clinical guidelines as well as follow-up of all the available randomised evidence provides individual clinician and patient discussions. treatments.1,8 More recent reviews that focused on the collected up to Nov 1, 2024, with no additional trials separate effects of ovarian suppression by GnRHa7,9 and identified by a further search done on Sept 1, 2025. For by ovarian irradiation10 reported unclear benefit. the main analyses presented here, information was Hence, questions remain about the comparative sought on all women with early breast cancer who effectiveness of ovarian ablation and pharmacological entered a randomised trial of OFS versus no OFS that ovarian suppression, and about whether long-term started recruitment before Jan 1, 2010, and had recurrence benefits outweigh any potential adverse effects on quality or mortality as a main outcome. Trials were not eligible if of life11,12 and non-cancer mortality.13 The effect of ovarian they included additional therapies (other than ablation and suppression (hereafter ovarian function prednisone) in one group but not the other (eg, trials that suppression, or OFS) when added to chemotherapy and compared OFS vs chemotherapy)20 or if they had only tamoxifen is also uncertain.14 This report addresses these non-oncological outcomes (eg, trials testing short-term questions, providing updated and extended meta- ovarian suppression during chemotherapy to try to analyses of the effects of OFS, whether by surgery, prevent chemotherapy-induced menopause).21 In some radiotherapy, or drug therapy, for premenopausal women eligible trials of OFS versus no OFS, no other adjuvant with ER-positive breast cancer or tumours with unknown therapy was to be given, whereas in others all participants ER status. were to be offered the same additional therapy, usually chemotherapy or 2–5 years of tamoxifen, or both. Methods Information was requested from study investigators Study design and participants about each woman in all eligible trials, including This meta-analysis was conducted in accordance with the individual patient data on randomisation date, allocated Preferred Reporting Items for a Systematic Review and treatment, age, tumour diameter, grade, histology, Meta-analysis of Individual Participant Data statement,15 axillary-lymph-node involvement, ER, progesterone For more on EBCTCG see https:// with additional information available on the EBCTCG receptor, and HER2 status, follow-up duration, dates of www.ctsu.ox.ac.uk/research/ website. Trial identification procedures have been any breast cancer recurrence (locoregional, contralateral, ebctcg described previously1,16–19 and include extensive searching or distant), dates and sites of any other second primary of bibliographic databases, journals and conference cancer; and dates and causes of death. The most recent proceedings, and communications with breast cancer follow-up data request was supplemented for one UK researchers to identify and include published and trial6 by mortality information from National Health unpublished trials. The current paper uses data that were Service Digital and its predecessors. Datasets were 1700 Articles checked for consistency, with queries resolved, if therefore divided trials into four categories: by whether possible, through correspondence with investigators. premenopausal status was certain because no Protocol-specified primary outcomes were recurrence of chemotherapy was given or premenopausal status was invasive breast cancer (locoregional, distant, or new assessed after chemotherapy (confirmed premenopausal) contralateral), breast cancer mortality, other mortality, and or uncertain when not assessed after chemotherapy all-cause mortality. Deaths without recorded recurrence (unconfirmed premenopausal), and also by whether or from an unknown cause were ascribed to breast cancer, as not tamoxifen was given. other causes are uncommon at these ages. Role of the funding source Statistical analysis The funders of the study had no role in study design, Statistical methods have been described in earlier data collection, data analysis, data interpretation, or EBCTCG reports1,16–19 and the Statistical Analysis Plan writing of the report. (appendix pp 38–45). Forest plots and Kaplan–Meier See Online for appendix graphs describe the separate trials and their combined Results results, and subgroup analyses help explore whether the We identified 27 unconfounded trials of OFS versus not proportional risk reductions produced by treatment that recruited women between 1948 and 2014: trial depend strongly on patient or tumour characteristics. designs and patient characteristics are shown online Time-to-first-event analyses, stratified by age, nodal (appendix pp 3–4). Individual patient datasets were status, year of follow-up, and trial, give the log-rank provided from 25 of them, comprising 19 235 (99·0%) of observed minus expected (O – E) statistic and its variance 19 439 women randomly assigned. Two of the 25 trials (V). These statistics yield the significance test and the that provided data (n=384) were ineligible as they first-event rate ratio, RR, and its CI (using the one-step included only postmenopausal women. Women who estimate logRR=[O – E]/V with variance 1/V). The were aged 55 years or older, postmenopausal, or had e contribution of tumours with unknown ER status to each ER-negative disease were excluded. Analyses presented of these (O – E) values is multiplied by 0·75 (yielding here focus on the remaining 23 trials and the ER-weighted RRs that assume the effect of OFS in 15 075 women who were considered to be premenopausal ER-unknown disease is only 75% of that in ER-positive at the time of randomisation and had ER-positive or disease).18 For overall results, 95% CIs are given. Results unknown ER status tumours (figure 1). Sensitivity for subgroups and individual trials are given with analyses on the groups excluded from the main analyses 99% CIs. χ² tests for heterogeneity or, where appropriate, found no benefit from OFS (appendix pp 6–8). trend compare RRs in different subgroups. Median 15 of the trials included women who were confirmed (IQR) follow-up duration (among women who would premenopausal prior to allocation to OFS or not, 12 have been survivors) was from Kaplan–Meier graphs of (n=4012) because participants did not receive any time to follow-up cessation. All p values are two-sided. chemotherapy after randomisation, and three (n=3430) In-house FORTRAN programs were used for analyses. because premenopausal status was confirmed following Breast cancer mortality RRs are estimated by chemotherapy by measurement of serum oestradiol25 or subtracting the log-rank statistics (O – E and V) for mortality without recurrence from those for overall 19 235 women randomised in 25 trials to OFS vs no OFS mortality (log-rank subtraction).20 This avoids having to determine which deaths after recurrence were from breast cancer without inappropriately assuming all were. The main analyses include only women younger than 9608 allocated OFS 9627 allocated no OFS 55 years who were premenopausal at randomisation and had ER-positive or ER-unknown tumours. Where 2084 ineligible 2076 ineligible individual information on menopausal status was 650 postmenopausal 657 postmenopausal unavailable, we assumed women were premenopausal if 73 premenopausal, age 85 premenopausal, age trial eligibility required this criterion, and only included ≥55 years ≥55 years 1361 premenopausal, age 1334 premenopausal, age those younger than 50 years if trials did not restrict <55 years, ER-negative <55 years, ER-negative eligibility by premenopausal status. One trial had a tumour tumour three-way randomisation phase;22 for balance, the 78 control patients in this phase are counted twice in 7524 eligible 7551 eligible totals (but not in calculation of log-rank statistics). 3734 confirmed premenopausal 3708 confirmed premenopausal Women who were premenopausal at randomisation but 3790 unconfirmed menopausal 3843 unconfirmed menopausal status status then received chemotherapy might have developed chemotherapy-induced menopause before commencing Figure 1: Flow chart for the trials of OFS versus no OFS in early breast cancer OFS.23,24 Premenopausal status was not confirmed after OFS=ovarian function suppression. To make all comparisons evenly (1:1) balanced, 78 controls in the only chemotherapy in some trials so the main analyses three-way comparison are counted twice. Articles measurement of follicle stimulating hormone levels and was 42 years (IQR 38–46). 13 trials (n=7633) recruited assessment of vaginal bleeding history.26,27 Median women whose premenopausal status was not follow-up was 11·0 years (IQR 8·2–14·9) and median age reconfirmed following post-randomisation chemotherapy Treatment Events/women OFS events Ratio of annual event rate (95% CI or 99% CI) Allocated Allocated Log-rank Variance OFS vs control OFS control O − E* of O − E* A No chemotherapy or premenopausal status confirmed after chemotherapy; no tamoxifen 1948 Christie A ± ovarian irradiation 239/352 252/356 −21·4 54·3 1957 Norwegian RH ± ovarian irradiation 17/76 35/93 −6·1 7·0 1961 NSABP B−03 ± oophorectomy 62/129 30/55 −3·2 8·9 1964 Saskatchewan CF ± oophorectomy 67/169 81/144 −13·7 16·8 1965 PMH Toronto† ±(ovarian irradiation [± prednisone]) 126/239 135/218 −15·1 31·1 1974 Bradford RI ± oophorectomy 7/25 11/25 −2·0 1·7 1987 ZIPP ± goserelin 2+ years 60/176 81/187 −9·9 27·8 1990 IBCSG VIII ± goserelin 2 years 16/42 17/35 −2·5 7·5 1991 CAMS China ± ovarian ablation 6/60 14/59 −3·6 3·2 1992 ICCG C/9/91 UK 8FECq3w; ± GnRHa 3 years 22/63 33/90 −6·0 9·9 Subtotal 622/1331 689/1262 −83·5 168·2 0·61 (0·52–0·71) (46·7%) (54·6%) reduction p<0·00001 B No chemotherapy or premenopausal status confirmed after chemotherapy; tamoxifen 1987 ZIPP Tamoxifen 2 years ± goserelin 2+ years 98/406 133/435 −16·1 46·5 1992 ICR−CTSU/NCRI BCSG Tamoxifen 5 years ± ovarian ablation 45/176 45/182 2·7 13·9 1994 EST3193/INT0142 Tamoxifen 5 years ± goserelin/leuprorelin/ 18/170 18/165 −0·7 8·7 oophorectomy/ovarian irradiation 1994 JATBCZ ZXBC−1002 Tamoxifen 2 years ± goserelin 2 years 6/20 2/18 1·4 1·6 2003 SOFT/IBCSG 24−02 ± chemotherapy; tamoxifen 5 years ± triptorelin 181/997 212/1000 −19·4 94·3 5 years/oophorectomy/ovarian irradiation 2008 Zhejiang China Tamoxifen 5 years ± goserelin 1·5 years (110 patients) (no data) 2009 ASTRRA Korea Neoadjuvant/adjuvant chemotherapy + 86/634 121/646 −17·4 50·4 tamoxifen 5 years ± goserelin 2+ years Subtotal with data‡ 434/2403 531/2446 −49·5 215·4 0·79 (0·70–0·91) (18·1%) (21·7%) reduction p=0·0008 C Premenopausal status not confirmed after chemotherapy; no tamoxifen 1974 Bradford RI MTt ± oophorectomy 5/23 8/19 −1·2 1·6 1978 Toronto-Edmonton CMF ± BCG ± (ovarian ablation + prednisone) 55/95 60/98 −3·5 20·4 1978 BCCA Vancouver 6CMFq3w ± (ovarian irradiation + prednisone) 46/63 37/58 0·6 17·1 1978 IBCSG/Ludwig II (± oophorectomy); (CMFq4w + prednisone) 1 year 86/123 99/129 −8·7 29·5 1979 SWOG 7827 B (± oophorectomy); (CMFV + prednisone) 1 year 81/150 68/134 1·6 33·6 1987 ZIPP (6CMFq4w) ± goserelin 2+ years 66/144 83/145 −9·4 23·9 1987 Pretoria RSA 6CMFq4w ± buserelin 5 years 18/60 24/65 −4·6 5·4 1989 FNCLCC France 6FA/ECq3w or q4w; ± triptorelin 3 years/ 147/377 147/371 −3·8 57·2 ovarian irradiation/oophorectomy 1989 ECOG EST5188 6CAFq4w; ± goserelin 5 years 198/436 199/443 −2·7 91·8 1990 IBCSG VIII 6CMFq4w; ± goserelin 18 months 56/271 85/281 −14·9 33·9 1991 CAMS China CMF ± ovarian ablation 172/989 195/1060 −2·0 66·4 1993 GABG 4 Germany (±4ECq3w; 3CMFq4w); ± goserelin 2 years 36/163 39/146 −3·5 17·0 Unk IT Naples 4Eq3w; 4CMFq4w; ± goserelin 2 years (92 patients) (no data) Subtotal with data‡ 966/2894 1044/2949 −52·1 397·8 0·88 (0·80–0·97) (33·4%) (35·4%) reduction p=0·0090 D Premenopausal status not confirmed after chemotherapy; tamoxifen 1987 ZIPP (6CMF + tamoxifen 2 years) 103/203 92/200 6·1 34·7 ± goserelin 2+ years 1992 ICR−CTSU/NCRI BCSG (Chemotherapy [73% CMF] + tamoxifen 5 years) 233/693 241/694 0·2 78·4 ± ovarian irradiation/suppression Subtotal 336/896 333/894 6·2 113·0 1·06 (0·88–1·27) (37·5%) (37·2%) increase p=0·56 Total (A + B + C + D) 2358/7524 2597/7551 −178·8 894·5 0·819 (0·767–0·874) (31·3%) (34·4%) reduction p<0·00001 Heterogeneity between four subtotals: χ2=24·2; p=0·00002 3 0 0·5 1·0 1·5 2·0 Heterogeneity within subtotals: χ2=25·6; p=0·49 99% CI Two-way heterogeneity tests: (A) versus (B) χ2 1 =7·0, p=0·0080; OFS better OFS worse 95% CI (C) versus (D) χ2=2·8, p=0·10; (A + B) versus (C + D) χ2=12·4, p=0·00042 1 1 1702 Articles (unconfirmed premenopausal). Most women were unconfirmed premenopausal status after chemot herapy treated with cyclophosphamide, methotrexate, and there was no heterogeneity by tamoxifen or not (χ²₁=9·8; fluorouracil, and a substantial proportion would have p=0·098). Consequently, further analyses combine all become postmenopausal following such chemotherapy.23 women with unconfirmed premenopausal status. Median follow-up for these comparisons was 11·2 years Where premenopausal status was confirmed at the start (IQR 8·5–14·3) and median age was 43 years (IQR 39–47). of OFS and women were not allocated tamoxifen, Five trials contributed women to both categories. recurrence rates were 39% lower with OFS compared with For the 23 eligible trials that provided data, results for control (RR 0·61, 95% CI 0·52–0·71; p<0·0001), with a any first invasive breast cancer recurrence (distant, loco- 15-year recurrence risk of 39·1% versus 56·5% regional or new contralateral disease) are shown in (figure 2A, 3A). Breast cancer mortality was similarly figure 2. Trials are categorised by confirmed (either no reduced (RR 0·65, 95% CI 0·56–0·76; p<0·0001; 15-year chemotherapy administered or premenopausal status breast cancer mortality of 35·9% vs 49·0%; figure 3A). In assessed after chemotherapy) or unconfirmed (not the second category, where confirmed premenopausal assessed after chemotherapy) premenopausal status at women received tamoxifen and were randomised to the the start of OFS and by the presence or absence of addition of OFS or not, there was a 21% lower recurrence tamoxifen in both comparator arms. For each trial, the rate with than without OFS (RR 0·79, 95% CI 0·70–0·91; information includes the year recruitment started, trial p=0·0008) and an absolute 5·9% lower 15-year recurrence name, method of OFS, use of chemotherapy, log-rank risk (24·8% vs 30·7%, figure 2B, 3B). Rates of breast cancer statistics, and the ratio of event rates. Similar plots for mortality were 16% (RR 0·84, 95% CI 0·69–1·02; p=0·080) distant recurrence at any time, locoregional, and lower but this did not reach statistical significance; contralateral recurrence as first event, breast cancer figure 3B. In trials in which premenopausal status was not mortality, death without recurrence (in the first year and confirmed after chemotherapy, OFS reduced the risk of overall), and all-cause mortality are shown in the recurrence (RR 0·91, 95% CI 0·84–1·00, p=0·042) but appendix (pp 10–16). Across all trials, women assigned to there was little apparent effect on breast cancer mortality OFS had an 18% lower rate of breast cancer recurrence (RR 0·96, 95% CI 0·88–1·05, p=0·38; figure 3C). (RR 0·82, 95% CI 0·77–0·87; p<0·00001) than did Figure 4 shows the effect of OFS (with or without women assigned to control; the 15-year absolute risks tamoxifen) on recurrence risk in confirmed were 36·5% versus 41·9% (appendix p 17). The rate of premenopausal women by age at randomisation. In breast cancer death was reduced by 14% (RR 0·86, those aged under 45 years, in the absence of tamoxifen, 0·80–0·93; p=0·0009), with 15-year absolute risk there was a 42% recurrence reduction (RR 0·58, 95% CI of 31·3% versus 34·4% (appendix pp 13, 17). 0·47–0·70; p<0·0001) and in those aged 45–54 years a As anticipated, there was significant heterogeneity 35% reduction (RR 0·65, 95% CI 0·52–0·82, p=0·0003). (χ²₃=22·2, p<0·0001) in the recurrence and breast cancer In the presence of tamoxifen, the reduction again mortality reductions between the four categories of trial appeared larger in women aged under 45 years than comparisons. Allowing for the different lengths of those aged 45–54 years; RR 0·73 (95% CI 0·63–0·86; follow-up in the trials, heterogeneity tests restricted to p=0·0002) versus RR 0·95 (0·75–1·21; p=0·67), although years 0–14 of follow-up showed a significant difference a test for interaction by age did not reach significance (χ²₁=12·4; p=0·0004) between trials where pre- (χ²₁=3·2; p=0·072). menopausal status was confirmed after chemotherapy In current practice, as in the more recent trials included (or no chemotherapy was administered) and those in in these meta-analyses, premenopausal women are more which it was not, and, within women with confirmed likely to be offered tamoxifen with consideration given to premenopausal status, between those allocated, and not the addition of OFS or not. Figure 5 shows subgroup allocated tamoxifen (χ²₁=7·0; p=0·0080). In women with analyses, for the trials of OFS versus not in confirmed premenopausal women who received tamoxifen, by site of recurrence, period of follow-up, age, ER and PR status, use of previous chemotherapy, method of ovarian Figure 2: Trial-specific recurrence rate ratios by allocation to OFS versus no suppression (surgery or irradiation, or pharmacological OFS in ER-positive or ER-unknown early breast cancer, estimated from ER- suppression), nodal status, T-stage, tumour grade and weighted analyses of four types of randomised comparison Recurrence rate ratios (local, distant, or new contralateral invasive disease) are HER2 status, with similar plots for trials in the absence for ER-positive disease*. A=doxorubicin. BCG=Bacillus Calmette–Guèrin. of tamoxifen given in the appendix (p 25). Distant, C=cyclophosphamide. E=epirubicin. F=fluorouracil. GnRHa=gonadotrophin- locoregional and contralateral recurrences were all releasing hormone agonist. M=methotrexate. O – E=observed minus expected. reduced by OFS (figure 5A; appendix p 19). Statistically OFS=ovarian function suppression. Tt=Triethylenephosphoramide. q4w=4-weekly. q3w=3-weekly. V=vincristine. Unk=unknown. *ER weighting: and clinically significant reductions in recurrence were the contribution to each O – E value from ER-unknown disease is 0∙75 of what seen in years 0–4, and these persisted undiminished the unweighted contribution would have been. †To make all comparisons through to years 5–9, and 10–14, with little follow-up after evenly (1:1) balanced, the 78 controls in the three-way phase of this one trial are year 15 (figure 5B). Proportional reductions did not differ counted twice. ‡The two trials with no data do not contribute to the subtotals or total. significantly by nodal status, tumour size or grade Articles 80 40 0 1704 )ES ,%( ytilatrom recnac tsaerB A 2593 women RR 0·65 (0·56−0·76); p<0·00001 41·7% Control 49·0% 30·1% OFS 35·9% 30·6% 21·5% Years 0−4 Years 5−9 Years 10−14 Year 15+ 5·61 (5·00 to 6·23) 2·83 (2·32 to 3·33) 1·76 (1·30 to 2·21) 2·08 (1·72 to 2·45) 6·50 (5·82 to 7·19) 3·47 (2·86 to 4·07) 2·57 (1·97 to 3·17) 2·57 (2·11 to 3·03) 0·68 (0·55 to 0·85) 0·67 (0·48 to 0·94) 0·59 (0·37 to 0·95) 0·60 (0·41 to 0·86) −31·2/80·8 −13·6/34·5 −9·0/17·3 −14·6/28·3 80 40 0 )ES ,%( ytilatrom recnac tsaerB B 4849 women RR 0·84 (0·69−1·02); p=0·080 10·0% Tamoxifen 15·4% 4·6% 2·8% 8·7% OFS + tamoxifen 13·9% Years 0−4 Years 5−9 Years 10−14 Year 15+ 0·56 (0·43 to 0·70) 1·28 (1·04 to 1·53) 1·16 (0·75 to 1·56) 0·0 0·91 (0·73 to 1·08) 1·17 (0·94 to 1·40) 1·23 (0·81 to 1·65) 1·90 (−0·25 to 4·05) 0·60 (0·43 to 0·83) 1·08 (0·81 to 1·44) 0·94 (0·56 to 1·58) 0·11 (0·01 to 1·13) −18·7/36·4 3·6/47·1 −0·8/14·4 −1·5/0·7 80 40 0 )ES ,%( ytilatrom recnac tsaerB 80 40 0 Event rate (% per year) Death rate per year and log-rank analyses (% [95% CI]: total rate – rate in women without recurrence) Years 0−4 Years 5−9 Years 10−14 Year 15+ and log-rank analyses OFS 8·20 (424/5172) 2·37 (90/3805) 1·32 (39/2965) 1·28 (69/5396) OFS + tamoxifen Control 10·57 (484/4580) 3·02 (94/3109) 2·39 (55/2301) 1·37 (56/4094) Tamoxifen RR (95% CI) 0·58 (0·48 to 0·70) 0·71 (0·49 to 1·04) 0·47 (0·28 to 0·80) 0·83 (0·50 to 1·38) RR (95% CI) (O − E)/V −61·3/112·7 −9·2/26·8 −10·1/13·4 −2·8/15·2 (O − E)/V Death rate per year (% [95% CI]: total rate – rate in women without recurrence) and log-rank analyses OFS + tamoxifen Tamoxifen RR (95% CI) (O − E)/V C 7633 women RR 0·96 (0·88−1·05); p=0·38 Control 40·3% 31·7% OFS 39·6% 16·7% 29·2% 15·7% 0 5 10 15 Time (years) Death rate per year (% [95% CI]: total rate – rate in women without recurrence) and log-rank analyses Years 0−4 Years 5−9 Years 10−14 Year 15+ OFS + tamoxifen 3·36 (3·08 to 3·64) 3·57 (3·22 to 3·92) 3·15 (2·65 to 3·65) 2·50 (1·81 to 3·19) Tamoxifen 3·54 (3·26 to 3·82) 3·90 (3·53 to 4·26) 2·72 (2·24 to 3·19) 2·26 (1·61 to 2·92) RR (95% CI) 0·93 (0·82 to 1·06) 0·91 (0·78 to 1·06) 1·17 (0·90 to 1·51) 1·22 (0·77 to 1·94) (O − E)/V −15·3/218·1 −16·0/166·2 8·7/56·8 3·7/18·2 )ES ,%( ecnerruceR 2593 women RR 0·61 (0·52−0·71); p<0·00001 50·9% Control 56·5% 42·7% OFS 39·1% 35·6% 28·2% 80 40 0 Event rate (% per year) and log-rank analyses Years 0−4 Years 5−9 Years 10−14 Year 15+ OFS + tamoxifen 2·50 (270/10815) 1·60 (120/7517) 1·73 (41/2371) 2·40 (3/125) Tamoxifen 3·00 (327/10897) 2·08 (154/7411) 2·18 (48/2201) 1·85 (2/108) RR (95% CI) 0·82 (0·69 to 0·97) 0·74 (0·58 to 0·95) 0·77 (0·49 to 1·20) 1·76 (0·27 to 11·28) (O − E)/V −26·7/133·2 −18·3/61·6 −5·1/19·5 0·6/1·1 )ES ,%( ecnerruceR 0 5 10 15 0 5 10 15 4849 women RR 0·79 (0·70−0·91); p=0·0008 22·6% Tamoxifen 30·7% 14·0% OFS + tamoxifen 24·8% 18·2% 11·7% 0 5 10 15 0 5 10 15 80 40 0 0 5 10 15 Event rate (% per year) and log-rank analyses Years 0−4 Years 5−9 Years 10−14 Year 15+ OFS 5·30 (836/15786) 3·37 (338/10040) 2·69 (107/3973) 1·20 (21/1747) Control 5·87 (928/15813) 3·48 (342/9836) 2·35 (91/3869) 0·96 (16/1675) RR (95% CI) 0·86 (0·77 to 0·96) 0·97 (0·82 to 1·15) 1·18 (0·86 to 1·61) 1·07 (0·53 to 2·18) (O − E)/V −49·1/330·6 −3·8/133·0 6·5/39·6 0·5/7·6 )ES ,%( ecnerruceR 7633 women RR 0·91 (0·84−1·00); p=0·042 37·6% Control 44·4% 25·9% OFS 43·2% 34·9% 23·0% Time (years) Figure 3: Outcome by allocation to OFS versus no OFS in women with ER-positive or ER-unknown early breast cancer, for each of three types of randomised comparison (A) No chemotherapy (or still premenopausal after chemotherapy) and tamoxifen not given. (B) No chemotherapy (or still premenopausal after chemotherapy), but tamoxifen given (mean 4 years). (C) chemotherapy, with unknown premenopausal status after. Plots are smoothed beyond year 10. RRs are from ER-weighted estimates of effect of OFS in women with ER-positive disease. O – E=observed minus expected. ER=oestrogen receptor. OFS=ovarian function suppression. RR=rate ratio. V=variance. Articles Age <45 years 80 40 0 (figure 5H–J). Few women had HER2 status measured, For confirmed premenopausal women aged under and trials were mostly conducted before the availability 45 years, figure 6 shows that adding OFS to tamoxifen of trastuzumab, but benefits appeared at least as large in compared with tamoxifen alone leads to a one quarter HER2-positive as in HER2-negative disease. In this reduction in distant recurrence (RR 0·77, 95% CI subgroup analysis there was limited power to assess the 0·64–0·93; p=0·0067) and breast cancer mortality relative benefits of ovarian ablation versus suppression, (RR 0·74, 95% CI 0·58–0·94; p=0·012; figure 6A, B). or duration of pharmacological OFS as most of the trials Rates of death from other causes were less than 0·1% per administered 2 to 5 years of GnRHa to suppress ovarian year, and not significantly affected by OFS (figure 6C), oestrogen production. leading to an overall survival benefit in these women )ES ,%( ecnerruceR A Age 45–54 years 1402 women 1191 women RR 0·58 (0·47–0·70); p<0·00001 RR 0·65 (0·52−0·82); p=0·0003 Control 55·2% 46·3% 38·6% Control 45·8% OFS 38·2% OFS 32·3% 31·9% 23·8% Recurrence rates (% per year) and log-rank analyses Years 0−4 Years 5−9 Year 10+ Years 0−4 Years 5−9 Year 10+ OFS 9·83 (259/2635) 2·29 (43/1879) 1·16 (49/4220) 6·50 (165/2537) 2·44 (47/1926) 1·42 (59/4141) Control 11·77 (284/2413) 3·61 (56/1550) 1·89 (61/3228) 9·23 (200/2167) 2·44 (38/1558) 1·58 (50/3167) RR (95% CI) 0·61 (0·48–0·77) 0·56 (0·34–0·93) 0·47 (0·28–0·79) 0·55 (0·41–0·73) 0·96 (0·54–1·69) 0·86 (0·51–1·44) (O − E)/V −33·0/65·8 −8·7/14·9 −10·7/14·3 −28·3/46·9 −0·5/11·9 −2·2/14·4 80 40 0 )ES ,%( ecnerruceR 0 5 10 0 5 10 B 3156 women 1693 women RR 0·73 (0·63−0·86); p=0·0002 RR 0·95 (0·75−1·21); p=0·67 Tamoxifen 25·4% 16·6% Tamoxifen 17·4% OFS + tamoxifen 11·8% 19·2% OFS + tamoxifen 11·8% 16·6% 8·9% 0 5 10 0 5 10 Time (years) Time (years) Recurrence rates (% per year) and log-rank analyses Years 0−4 Years 5−9 Year 10+ Years 0−4 Years 5−9 Year 10+ OFS + tamoxifen 2·51 (180/7168) 1·84 (88/4780) 1·80 (26/1441) 2·47 (90/3646) 1·17 (32/2737) 1·71 (18/1055) Tamoxifen 3·61 (248/6861) 2·18 (95/4365) 1·97 (24/1219) 1·96 (79/4036) 1·94 (59/3046) 2·38 (26/1091) RR (95% CI) 0·68 (0·56–0·83) 0·83 (0·61–1·12) 0·93 (0·52–1·67) 1·34 (0·97–1·86) 0·60 (0·39–0·92) 0·68 (0·36–1·28) (O − E)/V −37·4/96·9 −7·8/41·4 −0·8/11·2 10·7/36·2 −10·5/20·2 −3·7/9·5 Figure 4: Recurrence by allocation to OFS versus not OFS in confirmed premenopausal women with ER-positive or ER-unknown early breast cancer, estimated by age and by tamoxifen use (A) No chemotherapy (or still premenopausal after chemotherapy) and tamoxifen not given. (B) No chemotherapy (or still premenopausal after chemotherapy), but tamoxifen given (mean 4 years). RRs are from ER-weighted estimates of effect of OFS in women with ER-positive disease. ER=oestrogen receptor. O – E=observed minus expected. OFS=ovarian function suppression. RR=rate ratio. V=variance. Articles Events/woman-years (% per year) OFS + tamoxifen Ratio of annual event rate events (95% CI or 99% CI) Allocated Allocated Log-rank Variance OFS + tamoxifen vs tamoxifen OFS + tamoxifen tamoxifen O – E* of O – E* A Site of first recurrence (χ2=3·8; p=0·15) Distant 269/20 784 (1·3%) 302/20 577 (1·5%) –14·6 127·0 0·89 (0·71–1·12) Isolated local 120/20 781 (0·6%) 166/20 576 (0·8%) –25·1 65·4 0·68 (0·50–0·94) Contralateral 45/20 781 (0·2%) 63/20 562 (0·3%) –9·9 25·5 0·68 (0·41–1·13) B Follow-up period, years (trend χ2=0·1; p=0·79) 0–4 270/10 765 (2·5%) 327/10 846 (3·0%) –26·7 133·2 0·82 (0·65–1·02) 5–9 120/7489 (1·6%) 154/7376 (2·1%) –18·3 61·6 0·74 (0·54–1·03) 10–14 41/2343 (1·7%) 48/2173 (2·2%) –5·1 19·5 15+ 3/116 (2·6%) 2/96 (2·1%) 0·6 1·1 C Age at entry, years (trend χ2=2·7; p=0·10†) Age <35 62/2040 (3·0%) 77/1714 (4·5%) –13·8 30·3 0·63 (0·40–1·01) Age 35–39 99/3664 (2·7%) 127/3807 (3·3%) –13·9 49·8 0·76 (0·53–1·09) Age 40–44 133/7642 (1·7%) 163/6874 (2·4%) –21·1 66·8 0·73 (0·53–1·00) Age 45–49 114/5908 (1·9%) 127/6336 (2·0%) –1·8 53·0 Age 50–54 26/1509 (1·7%) 37/1811 (2·0%) –2·2 10·9 D ER status (χ2=0·1; p=0·77) ER-unknown 49/1561 (3·1%) 60/1586 (3·8%) –4·3 13·8 ER-positive 385/19 247 (2·0%) 471/19 009 (2·5%) –47·6 205·5 0·79 (0·66–0·95) E PR status (χ2=0·1; p=0·98) PR-positive 201/11 804 (1·7%) 240/11 664 (2·1%) –23·4 105·6 0·80 (0·62–1·03) PR-negative 16/653 (2·5%) 20/607 (3·3%) –1·6 7·1 PR-unknown 217/8320 (2·6%) 271/8291 (3·3%) –24·6 99·6 0·78 (0·60–1·01) F Previous chemotherapy use (χ2=0·7; p=0·39) Absence 220/11 274 (2·0%) 260/11 399 (2·3%) –16·5 99·0 0·85 (0·65–1·10) Presence 214/9510 (2·3%) 271/9178 (3·0%) –33·0 116·4 0·75 (0·59–0·96) G Ablation or suppression (χ2=0·5; p=0·47) Ablation 56/2273 (2·5%) 66/2227 (3·0%) –1·9 21·4 Suppression 360/17 051 (2·1%) 447/16 907 (2·6%) –46·9 185·3 0·78 (0·64–0·94) H Nodal status (trend χ2=0·7; p=0·41) N0 239/14 562 (1·6%) 300/14 312 (2·1%) –33·5 120·1 0·76 (0·60–0·96) N1–3 85/2920 (2·9%) 97/2964 (3·3%) –5·5 40·2 0·87 (0·58-1·31) N4+ 49/880 (5·6%) 58/812 (7·1%) –3·1 22·2 0·87 (0·50-1·50) N unknown 61/2422 (2·5%) 76/2489 (3·1%) –7·3 33·0 0·80 (0·51–1·25) I Tumour diameter (trend χ2=0·1; p=0·85) 1–20 mm (T1) 203/13 407 (1·5%) 256/13 587 (1·9%) –25·1 102·8 0·78 (0·61–1·01) 21–50 mm (T2) 187/6598 (2·8%) 237/6185 (3·8%) –28·3 93·8 0·74 (0·57–0·97) >50 mm (T3–T4) 24/348 (6·9%) 18/354 (5·1%) 1·0 7·2 Other or unknown 20/421 (4·8%) 20/430 (4·7%) –0·3 6·7 J Tumour grade (trend χ2=0·6; p=0·44) Well differentiated 60/4214 (1·4%) 67/4164 (1·6%) –3·5 28·3 0·88 (0·55-1·43) Moderately 173/8698 (2·0%) 206/8584 (2·4%) –15·9 85·0 0·83 (0·63–1·10) Poorly 100/3215 (3·1%) 130/3207 (4·1%) –14·6 49·2 0·74 (0·51–1·07) Unknown 101/4610 (2·2%) 128/4572 (2·8%) –14·0 42·8 0·72 (0·49–1·07) K HER2 status (χ2=2·5; p=0·11) Positive 34/1918 (1·8%) 49/1597 (3·1%) –10·7 19·2 0·57 (0·32–1·03) Negative 211/11 170 (1·9%) 245/11 189 (2·2%) –18·4 109·8 0·85 (0·66–1·08) Unknown 189/7685 (2·5%) 237/7783 (3·0%) –21·8 85·5 0·77 (0·59–1·02) Total 434/20 784 (2·1%) 531/20 577(2·6%) –49·5 215·4 0·795 (0·695–0·908) p=0·0008 99% CI 95% CI Other or unknown 0 0·5 1·0 1·5 2·0 OFS + tamoxifen better Tamoxifen better Figure 5: Subgroup analyses of recurrence by allocation to OFS plus tamoxifen versus tamoxifen in confirmed premenopausal women with ER-positive or ER-unknown early breast cancer RRs are from ER-weighted estimates of effect of OFS in women with ER-positive disease. EST3193 did not provide individual patient data for method used for ovarian suppression so has been excluded from (G): 55% had oophorectomy or radiation, and 36% had GnRHa. ER=oestrogen receptor. GnRHa=gonadotrophin-releasing hormone agonist. O – E=observed minus expected. OFS=ovarian function suppression. PR=progesterone receptor. RR=rate ratio. *ER weighting: the contribution to each O – E value from ER-unknown disease is 0∙75 of what the unweighted contribution would have been. †Heterogeneity between participants younger than 45 years and those aged 45 years and older, χ²1=3∙2; p=0∙072. 1706 Articles (RR 0·72, 95% CI 0·57–0·90; p=0·0045), and 10-year rates of 1·10 (0·86–1·41; p=0·45). Analyses of deaths mortality of 9·4% versus 12·2%. without recorded recurrence showed no relationship Across all trials there was no evidence that OFS caused between nodal status and mortality in the 426 (42·9%) of any increase or decrease in non-breast second primary 993 women with a specified cause of death, indicating cancers, or deaths from known causes other than breast that only a few of these are likely to be misclassified breast cancer (appendix pp 15, 28–30). There were comparatively cancer deaths (appendix p 31). However, the 567 (57·1%) few non-breast cancer deaths recorded among these of 993 deaths without recorded recurrence that were women, who were all aged under 55 years at reported as being from an unknown cause occurred at randomisation, with an RR of non-breast cancer death twice the rate in women with node-positive disease than 30 10 )ES ,%( ecnerrucer tnatsiD A 3156 women 100 RR 0·77 (0·64–0·93); p=0·0067 30 20 Tamoxifen 18·6% 11·8% OFS + tamoxifen 14·1% 8·9% Distant recurrence rates (% per year) and log-rank analyses Years 0−4 Years 5−9 Year 10+ OFS + tamoxifen 1·88 (137/7269) 1·22 (61/4987) 1·24 (19/1535) Tamoxifen 2·50 (176/7028) 1·59 (75/4718) 0·98 (14/1434) RR (95% CI) 0·73 (0·58–0·92) 0·77 (0·54–1·09) 1·34 (0·66–2·71) (O − E)/V −22·6/70·9 −8·2/31·6 2·2/7·7 )ES ,%( ytilatrom recnac tsaerB B 3156 women RR 0·74 (0·58–0·94); p=0·012 Tamoxifen 11·7% OFS + tamoxifen 9·1% 5·2% 2·3% Death rate per year (% [95% CI]: total rate – rate in women without recurrence) and log-rank analyses Years 0−4 Years 5−9 Year 10+ OFS + tamoxifen 0·49 (0·33/0·65) 1·45 (1·12/1·77) 1·17 (0·64/1·69) Tamoxifen 1·02 (0·79/1·25) 1·44 (1·11/1·77) 1·41 (0·82/2·00) RR (95% CI) 0·46 (0·31–0·68) 1·02 (0·73–1·42) 0·83 (0·44–1·55) (O − E)/V −19·6/25·1 0·6/34·4 −1·9/9·8 30 10 )ES ,%( htaed tsaerb−noN C 3156 women 100 RR 0·46 (0·18–1·16); p=0·10 30 10 0·4% Tamoxifen 0·7% 0·1% OFS + tamoxifen 0·4% 0 5 10 Time (years) Non-breast-cancer death rates (% per year) and log-rank analyses Years 0−4 Years 5−9 Year 10+ OFS + tamoxifen 0·01 (1/7168) 0·06 (3/4780) 0·21 (3/1441) Tamoxifen 0·07 (5/6861) 0·07 (3/4365) 0·41 (5/1219) RR (95% CI) 0·24 (0·05–1·29) 0·80 (0·15–4·25) 0·49 (0·11–2·19) (O − E)/V −2·0/1·4 −0·3/1·4 −1·2/1·7 )ES ,%( htaed ynA 0 5 10 0 5 10 D 3156 women RR 0·72 (0·57–0·90); p=0·0045 Tamoxifen 12·2% OFS + tamoxifen 9·4% 5·5% 2·4% 0 5 10 00 55 1100 Time (years) Death rates (% per year) and log-rank analyses Years 0−4 Years 5−9 Year 10+ OFS + tamoxifen 0·51 (38/7498) 1·51 (80/5315) 1·35 (22/1629) Tamoxifen 1·09 (80/7334) 1·50 (76/5075) 1·75 (27/1542) RR (95% CI) 0·44 (0·30–0·65) 1·01 (0·73–1·40) 0·76 (0·43–1·36) (O − E)/V −21·6/26·5 0·3/35·8 −3·1/11·5 Figure 6: Main outcomes by allocation to OFS plus tamoxifen versus tamoxifen for confirmed premenopausal women younger than 45 years with ER-positive or ER-unknown early breast cancer (A) Distant recurrence at any time. (B) Breast cancer mortality. (C) Non-breast cancer mortality. (D) All-cause mortality. RRs are from ER-weighted estimates of effect of OFS in women with ER-positive disease. ER=oestrogen receptor. O – E=observed minus expected. OFS=ovarian function suppression. RR=rate ratio. V=variance. Articles in node-negative disease, indicating that, in a young did not receive tamoxifen. Nonetheless, even in more population with few other expected causes of death, most recent trials of OFS plus tamoxifen versus tamoxifen of these deaths were from breast cancer. Sensitivity alone (tamoxifen allocated for a mean of about 4 years) a analyses which counted deaths from unknown causes significant benefit was observed, particularly among without recurrence as non-breast cancer deaths, gave very women younger than 45 years. It is unclear if the smaller similar results (RR 0·86 vs 0·85 for breast cancer effect in these trials can be at least partially explained by mortality; 1·10 vs 1·00 for non-breast cancer mortality; adherence to therapy (either pharmacological OFS or appendix pp 32–33). tamoxifen) because adherence data were not routinely Recording of toxicity varied considerably between collected. For premenopausal women who received trials, so patient-level data on non-fatal toxicity were not chemotherapy (and most chemotherapy in these trials sought. Instead, toxicities reported in trial publications involved cyclophosphamide, methotrexate, and are summarised in the appendix (pp 34–37). Toxicity was fluorouracil, which often causes menopause),23 if more consistently reported from the more recent trials menopausal status was uncertain after chemotherapy, of OFS plus tamoxifen versus tamoxifen. While adverse there appeared to be little benefit from OFS. event reporting varied between trials, the addition of Nowadays, many premenopausal women with OFS was consistently associated with more higher grade ER-positive disease are prescribed tamoxifen, and this hot flushes. For other adverse events, for example, meta-analysis reliably confirms that the use of OFS offers weight gain, musculoskeletal symptoms, fatigue, further benefit. In the presence of OFS, however, insomnia, and neuropsychiatric toxicity, there was no aromatase inhibitors are more effective than tamoxifen consistent increase with OFS versus control. Two trials in such women,19 indicating that, depending on their reported more glucose intolerance and more grade 3 absolute risk of recurrence and tolerability, the optimal hypertension with the addition of OFS but it is unclear endocrine treatment for premenopausal women might whether this represents selective reporting. Of the be a combination of aromatase inhibitors and either 23 trials, only four studies (SOFT, ZIPP, EST3193, and ovarian ablation or long-term adherence to ovarian ICR-CTSU) collected data on quality of life (QOL) suppression. The optimal duration of endocrine measurements and these are best reviewed in the treatment in these circumstances is not known, but in individual publications. postmenopausal women 10 years of an aromatase inhibitor appears somewhat more effective than just Discussion 5 years.29 Additionally, if a premenopausal woman cannot Ablation or suppression of premenopausal ovarian adhere long term to tamoxifen or an aromatase inhibitor, function is one of the earliest forms of endocrine therapy this meta-analysis confirms a substantial oncological for breast cancer, first used in the 19th century.3,4 It was benefit from OFS as the sole endocrine treatment, but the first cancer treatment assessed in a randomised trial,5 OFS alone can also be associated with intrusive endocrine initiated at the Christie Hospital in 1948. Further trials of symptoms. radiotherapy or surgery to ablate ovarian function were Previous chemotherapy did not appear to affect the undertaken over the next half-century. More recently, benefits of OFS for women who remained premenopausal these irreversible procedures have often been replaced by and with modern chemotherapy recovery of ovarian pharmacological suppression of ovarian function, which function is common, particularly for younger women.30 leaves open the possibility that, for some women, ovarian Despite the introduction of targeted agents, and oestrogen production might resume after the scheduled increasing use of genomic assays to help inform end of suppression. chemotherapy treatment decisions, it is probable that This updated meta-analysis summarises all available OFS remains relevant to premenopausal women with a unconfounded randomised evidence on OFS for current diagnosis of early breast cancer.31,32 For these premenopausal women with early breast cancer. The women, further research is needed to fully evaluate the ER-weighted analyses confirm that, for women with safety and efficacy of adding newer therapies such as ER-positive disease who are confirmed to be selective ER degraders or CDK 4/6 inhibitors, with or premenopausal (after any other treatments), OFS without previous chemotherapy, to OFS. Concerns about substantially reduces the risk of recurrence and death the risk of thrombosis33 or cardiac events34 with tamoxifen from breast cancer, with the breast cancer death rate and some CDK 4/6 inhibitors will limit consideration of reduced considerably not only during the first decade but this combination, but most premenopausal women also during the second decade. This is clinically relevant, receiving adjuvant CDK 4/6 inhibitors will be at sufficient as the risk of recurrence of ER-positive disease remains risk of recurrence to be recommended OFS with an appreciable for at least 20 years.28 aromatase inhibitor. Results are consistent with, but more definite than, Many trials did not report side-effects or QOL, although those in previous meta-analyses.7,8 The proportional OFS can cause more menopausal symptoms.12 There are reduction in breast cancer recurrence and death is limited data on late toxicities, making reliable assessment greater in the earlier trials where women in both groups of any long-term harms of OFS difficult. However, after 1708 Articles pharmacological OFS ends, QOL symptom scores can Declaration of interests recover.35 While OFS is known to cause osteoporosis,14,36 JBr reports financial interest by being clinical advisor for National Institute bone mineral density was not routinely measured; but for Health and Care Excellence and for Genesis Care. JBe reports research grants from Amgen, AstraZeneca, Bayer, Merck, Pfizer, Roche, and Sanofi- this risk can be ameliorated with bisphosphonates and Aventis to Karolinska Institutet or University Hospital for molecular monitoring of bone health. There was no evidence of marker studies or clinical studies; payment or honoraria from Roche, adverse effects on non-breast-cancer mortality, or cancer AstraZeneca, Novartis, and Stratipath; stocks in Stratipath; and financial incidence. interests from UpToDate. JBl reports grants from AstraZeneca, Merck Sharp & Dohme, Puma Biotecnology, Pfizer, Janssen-Cilag, Novartis, A limitation is that many trials took place before Eli Lilly, Breast Cancer Now, Breast International Group, Glaxo Smith the 1980s, when ER status was not routinely available, Kline, Kortuc, National Institute for Health and Care Research, and diagnosis of recurrence was less sensitive, and adjuvant Verastem; and participation on a Data Safety Monitoring Board for ATNEC therapy was not routinely used. However, even comparing (University of Warwick) and POSNOC (University of Nottingham). MMR reports support from Breast Cancer Research Foundation; grants OFS plus tamoxifen versus tamoxifen alone, this meta- from Pfizer, TerSera Therapeutics, Ispen, DebioPharm, and AstraZeneca; analysis shows that the addition of OFS reduces breast consulting fees from TerSera Therapeutics, Tolmar Pahrmaceuticals, and cancer recurrence and mortality rates by about a quarter AstraZeneca; and payments or honoraria from Bristol-Myers Squibb, among premenopausal women younger than 45 years. St Gallen Oncology Conferences and McGill University funded by Merck; participation on a Data Safety Monitoring Board for ABCSG; and Interestingly this benefit was seen in the trials where leadership or fiduciary role for ETOP IBCSG Partners Foundation. most premenopausal women receiving OFS plus PAF reports honorarium from Eli Lilly. SMS reports consulting fees from tamoxifen had pharmacological OFS rather than ovarian TERSERA and Molecular Templates; payment or honoraria from Chugai ablation, many receiving only about 2 years of such Pharmaceuticals, Merck and AstraZeneca; support for attending meetings from Roche, Chugai, SEAGEN, IMMUNOME, and NAPO; participation therapy, but the benefits appeared to be persistent. on Data Safety Monitoring Board for Napo Pharmaceuticals Scientific Further research is needed to assess the optimal duration Advisory board, Merck Advisory board, and AstraZeneca Advisory board; of pharmacological OFS. leadership or fiduciary role for SEAGEN BOD (November, 2022, to Proportional risk reductions appeared similar for December, 2023), Immunome BOD (April, 2024, to present), and NSABP BOD (unpaid); stock or stock options in SEAGEN and Immunome; and women regardless of their individual or tumour medical writing third party with AstraZeneca. characteristics, so the absolute benefits, unlike any Data sharing harms, are likely to increase with increasing recurrence All datasets provided to the EBCTCG remain the property of the trial risk. Since these studies were conducted, there has been groups sending them, to whom data sharing requests should be made. a steady improvement in breast cancer outcomes37 (and The EBCTCG data sharing policy is available at https://www.ctsu.ox. corresponding decreases in population-wide breast ac.uk/research/the-early-breast-cancer-trialists-collaborative- groupebctcg/data-policy-for-the-early-breast-cancer-trialists2019- cancer mortality rates in middle age) which complicates collaborativegroup-ebctc. prediction of the additional absolute benefit from OFS Acknowledgments for women diagnosed today. Nevertheless, this meta- The chief acknowledgment is to the women who took part in these trials analysis of trials in ER-positive early breast cancer of and the trialists who conducted the studies and shared their data. OFS, together with those trials of OFS with 5 years of Lifelong outcome data for the 1948 Christie trial was provided by NHS Digital or its predecessors. This study is supported by core funding to tamoxifen or of an aromatase inhibitor, demonstrate the Population Health Research Unit and Clinical Trial Service Unit, large reductions in distant recurrence or breast cancer Nuffield Department of Population Health, University of Oxford, from mortality rates throughout the first decade and, at least Cancer Research UK, the Breast Cancer Research Foundation, and the for OFS and tamoxifen (and perhaps for aromatase UK Medical Research Council. CT, DD, and PM received funding from Cancer Research UK (grants c8225/A21133 and PRCRPG-Nov21\100001). inhibitors),29 into the second decade. Collectively, these results remain relevant to contemporary practice and Writing committee Rosie Bradley, Jeremy Braybrooke, Mike Clarke, Robert K Hills, emphasise the importance of effective long-term Amanda Kerr, Richard Peto, David Dodwell, Paul McGale, prevention of ER stimulation by oestrogen from the Hongchao Pan, Carolyn Taylor, Rodrigo Arriagada, Judith Bliss, ovaries or from the aromatase enzymes, while Allan Hackshaw, Hyun-Ah Kim, Woo Chul Noh, John Yarnold, minimising side-effects from ER blockade or oestrogen Jonas Bergh, Sandra M Swain, Nancy E Davidson, Prudence A Francis, Meredith M Regan, Richard Gray (1949–2023). deprivation. EBCTCG steering committee Contributors Jonas Bergh, Sandra M Swain (Co-Chairs), David Cameron (Vice-Chair), RB, JBr, RG, and RKH designed and carried out the analyses, with all Kathy Albain, Stewart Anderson, Rodrigo Arriagada, Karla Ballman, members of the writing committee providing input to the analysis plan. John Bartlett, Elizabeth Bergsten-Nordström*, Judith Bliss, RB, RG and RKH accessed and verified the data. RB, JBr, RG, RKH and Rosie Bradley†, Etienne Brain, Jeremy Braybrooke†, Lisa Carey, RP drafted the report and all other writing committee members Almona Choudhury*, Mike Clarke†, Robert Coleman, Jack Cuzick, contributed to revising it. Interim analyses were presented and Nancy Davidson, Lucia Del Mastro, James Dignam, David Dodwell†, discussed at steering committee meetings. The EBCTCG secretariat Mitch Dowsett, Fran Duane†, Bent Ejlertsen, Theodoros Foukakis, (G Beake, R Berry, C Boddington, R Bradley, J Braybrooke, M Clarke, Prudence A Francis, Jose Angel Garcia-Saenz, Richard Gelber, C Davies, L Davies, D Dodwell, F Duane, V Evans, J Gay, L Gettins, Michael Gnant, Matthew P Goetz, Pamela Goodwin, Richard Gray†, J Godwin, R Gray, R K Hills, F Holt, S James, A Kerr, H Liu, Z Liu, Daniel Hayes, Catherine Hill, Robert K Hills†, Reshma Jagsi, E MacKinnon, G Mannu, P McGale, T McHugh, P Morris, M Nakahara, Wolfgang Janni, Sibylle Loibl, Elizabeth MacKinnon†, H Pan, R Peto, S Read, E Straiton, C Taylor, H Taylor) was responsible Eleftherios Mamounas, Gurdeep Mannu†, Paul McGale†, for maintaining collaboration, identifying trials, and obtaining and Stuart A McIntosh, Hirofumi Mukai, Valentina Nekljudova, checking datasets. All authors of the writing committee had Larry Norton, Hongchao Pan†, Richard Peto†, Martine Piccart, responsibility for the decision to submit for publication. Articles Philip Poortmans, Kathy I Pritchard, Vinod Raina, Daniel Rea, 7 Cuzick J, Ambroisine L, Davidson N, et al. Use of luteinising- Meredith M Regan, John Robertson, Shigerhira Saji, Roberto Salgado, hormone-releasing hormone agonists as adjuvant treatment in Dennis Slamon, Tanja Spanic*, Joseph Sparano, Patty Spears*, premenopausal patients with hormone-receptor-positive breast Guenther Steger, Gong Tang, Carolyn Taylor†, Sandra J Thompson, cancer: a meta-analysis of individual patient data from randomised adjuvant trials. Lancet 2007; 369: 1711–23. Masakazu Toi, Andrew Tutt, Giuseppe Viale, Xiang Wang, Tim Whelan, 8 Early Breast Cancer Trialists’ Collaborative Group. Ovarian ablation Nicholas Wilcken, Antonio Wolff, Norman Wolmark, Ke-Da Yu. *Patient in early breast cancer: overview of the randomised trials. Lancet representatives. †EBCTCG secretariat, Oxford. 1996; 348: 1189–96. Contributing trial groups (lead investigators) 9 Goel S, Sharma R, Hamilton A, Beith J. LHRH agonists for Bradford Royal Infirmary, Radford, UK—M B Masood, D Parker, J J Price; adjuvant therapy of early breast cancer in premenopausal women. British Columbia Cancer Agency, Vancouver, BC, Canada—C Lohrisch, Cochrane Database Syst Rev 2009; 2009: CD004562. A Nichol; Chinese Academy of Medical Sciences, Beijing, China 10 Asiri MA, Tunio MA, Abdulmoniem R. Is radiation-induced ovarian (in collaboration with the Oxford NDPH)—Y Shan, Y F Shao, X Wang, ablation in breast cancer an obsolete procedure? Results of a meta- B Xu, D B Zhao (NDPH: Z M Chen, H C Pan); Christie Hospital and analysis. Breast Cancer (Dove Med Press) 2016; 8: 109–16. Holt Radium Institute, Manchester, UK—A Howell, R Swindell; Eastern 11 Tevaarwerk AJ, Wang M, Zhao F, et al. Phase III comparison of Cooperative Oncology Group, Boston, MA, USA—R L Comis, tamoxifen versus tamoxifen plus ovarian function suppression in N E Davidson, R Gray, N Robert, G Sledge, L J Solin, J A Sparano, premenopausal women with node-negative, hormone receptor- positive breast cancer (E-3193, INT-0142): a trial of the Eastern D C Tormey, W Wood; Federation Nationale des Centres de Lutte Contre le Cooperative Oncology Group. J Clin Oncol 2014; 32: 3948–58. Cancer (FNCLCC) Breast Group, Paris, France—T Delozier, M Spielman, 12 Ribi K, Luo W, Bernhard J, et al. Adjuvant tamoxifen plus ovarian J Mace Lesec’h; German Adjuvant Breast Group (GABG), Frankfurt, function suppression versus tamoxifen alone in premenopausal Germany—W Eiermann, J Hilfrich, W Jonat, M Kaufmann, women with early breast cancer: patient-reported outcomes in the R Kreienberg, M Schumacher; Institute of Cancer Research Clinical Trials suppression of ovarian function trial. J Clin Oncol 2016; 34: 1601–10. and Statistics Unit (ICR-CTSU, NCRI), Lonodn, UK—R A’Hern, J Bliss, 13 Cusimano MC, Chiu M, Ferguson SE, et al. Association of bilateral P Ellis, L Kilburn, J R Yarnold; International Breast Cancer Study Group salpingo-oophorectomy with all cause and cause specific mortality: (IBCSG), Bern, Switzerland—M Castiglione, A Coates, M Colleoni, population based cohort study. BMJ 2021; 375: e067528. J Collins, J Forbes, P A Francis, R D Gelber, A Goldhirsch, J Lindtner, 14 Burstein HJ, Lacchetti C, Anderson H, et al. Adjuvant endocrine O Pagani, K N Price, M M Regan, C M Rudenstam, H J Senn, therapy for women with hormone receptor–positive breast cancer: B Thürliman; International Collaborative Cancer Group, London, ASCO clinical practice guideline update on ovarian suppression. UK—J M Bliss, C E D Chilvers, R C Coombes, M Espie, E Hall, J Clin Oncol 2016; 34: 1689–701. L Kilburn, M Marty; Japanese Association for Treatment of Breast Cancer 15 Stewart LA, Clarke M, Rovers M, et al. Preferred reporting items for with Zoladex, Toyoko, Japan—S Mitsuyama, Y Nomura, S Ohno; Korean a systematic review and meta-analysis of individual participant data: Breast Cancer Study Group (KBCSG), Seoul, Soth Korea—H A Kim, the PRISMA-IPD statement. JAMA 2015; 313: 1657–65. W C Noh; National Surgical Adjuvant Breast and Bowel Project (NSABP), 16 Early Breast Cancer Trialists’ Collaborative Group. Treatment of Pittsburgh, PA, USA—S Anderson, H Bandos, A Brown, J Bryant, early breast cancer. Volume 1: Worldwide evidence 1985–1990. R Cecchini, J Costantino, J Dignam, B Fisher, C Geyer, E P Mamounas, Oxford, UK; Oxford University Press, 1990. S Paik, C Redmond, E Romond, S Swain, G Tang, M A Torres, 17 Early Breast Cancer Trialists’ Collaborative Group. Comparisons L Wickerham, N Wolmark, G Yothers; Norwegian Radium Hospital, Oslo, between different polychemotherapy regimens for early breast Norway—B Erikstein, S Gundersen, M Hauer-Jensen, H Høst, cancer: meta-analyses of long-term outcome among 100 000 women A B Jacobsen, R Nissen-Meyer; Pretoria University, Pretoria, South in 123 randomised trials. Lancet 2012; 379: 432–44. Africa—C I Falkson; Saskatchewan Cancer Foundation, Regina, SK, 18 Early Breast Cancer Trialists’ Collaborative Group. Relevance of Canada—A J S Bryant, G H Ewing, L A Firth, J L Krushen-Kosloski; breast cancer hormone receptors and other factors to the efficacy of adjuvant tamoxifen: patient-level meta-analysis of randomised Southwest Oncology Group, San Antonio, TX, USA—K Albain, W Barlow, trials. Lancet 2011; 378: 771–84. G T Budd, J Gralow, D Hayes, G Hortobagyi, R Jagsi, S Martino, 19 Early Breast Cancer Trialists’ Collaborative Group. Aromatase L Pusztai, P Sharma, A Thompson; Toronto-Edmonton Breast Cancer inhibitors versus tamoxifen in premenopausal women with Study Group, Toronto, ON, Canada—A H G Paterson, K I Pritchard; oestrogen receptor-positive early-stage breast cancer treated with Toronto Princess Margaret Hospital, Toronto, ON, Canada—A Fyles, ovarian suppression: a patient-level meta-analysis of 7030 women J W Meakin, T Panzarella, K I Pritchard; ZIPP trial (Cancer Research from four randomised trials. Lancet Oncol 2022; 23: 382–92. UK; GIVIO, Italy; Southeast Sweden, and Stockholm)—M Baum, 20 Scottish Cancer Trials Breast Group. Adjuvant ovarian ablation T Fornander, A Hackshaw, A Nicolucci, B Nordenskjold, R Sainsbury. versus CMF chemotherapy in premenopausal women with The full list of EBCTCG collaborating trialists is available at https:// pathological stage II breast carcinoma: the Scottish trial. Scottish www.ctsu.ox.ac.uk/research/the-early-breast-cancer-trialists- Cancer Trials Breast Group and ICRF Breast Unit, Guy’s Hospital, collaborative-group-ebctcg. London. Lancet 1993; 341: 1293–98. 21 Lambertini M, Moore HCF, Leonard RCF, et al. Gonadotropin- References releasing hormone agonists during chemotherapy for 1 Early Breast Cancer Trialists’ Collaborative Group. Effects of preservation of ovarian function and fertility in premenopausal chemotherapy and hormonal therapy for early breast cancer on patients with early breast cancer: a systematic review and meta- recurrence and 15-year survival: an overview of the randomised analysis of individual patient-level data. J Clin Oncol 2018; trials. Lancet 2005; 365: 1687–717. 36: 1981–90. 2 Early Breast Cancer Trialists’ Collaborative Group. Aromatase 22 Meakin JW, Allt WE, Beale FA, et al. Ovarian irradiation and inhibitors versus tamoxifen in early breast cancer: patient-level prednisone therapy following surgery and radiotherapy for meta-analysis of the randomised trials. Lancet 2015; 386: 1341–52. carcinoma of the breast. Can Med Assoc J 1979; 120: 1221–29. 3 Beatson GT. On the treatment of inoperable cases of carcinoma of 23 Del Mastro L, Venturini M, Sertoli MR, Rosso R. Amenorrhea the mamma: suggestions for a new method of treatment with induced by adjuvant chemotherapy in early breast cancer patients: illustrative cases. Lancet 1896; 2: 104–07. prognostic role and clinical implications. Breast Cancer Res Treat 4 Clarke MJ. Ovarian ablation in breast cancer, 1896 to 1998: 1997; 43: 183–90. milestones along hierarchy of evidence from case report to 24 Bernhard J, Zahrieh D, Castiglione-Gertsch M, et al. Adjuvant Cochrane review. BMJ 1998; 317: 1246–48. chemotherapy followed by goserelin compared with either modality 5 Paterson R, Russel MH. Clinical trials in malignant disease. Part II- alone: the impact on amenorrhea, hot flashes, and quality of life in breast cancer: value of irradiation of the ovaries. J Fac Radiol 1959; premenopausal patients—the International Breast Cancer Study 10: 130–33. Group Trial VIII. J Clin Oncol 2007; 25: 263–70. 6 Goel S, Sharma R, Hamilton A, Beith J. LHRH agonists for adjuvant therapy of early breast cancer in premenopausal women. Cochrane Database Syst Rev 2009; 2009: CD004562. 1710 Articles 25 Francis PA, Pagani O, Fleming GF, et al, and the SOFT and TEXT 32 Slamon D, Lipatov O, Nowecki Z, et al. Ribociclib plus endocrine Investigators and the International Breast Cancer Study Group. therapy in early breast cancer. N Engl J Med 2024; 390: 1080–91. Tailoring adjuvant endocrine therapy for premenopausal breast 33 Rugo HS, O’Shaughnessy J, Boyle F, et al. Adjuvant abemaciclib cancer. N Engl J Med 2018; 379: 122–37. combined with endocrine therapy for high-risk early breast cancer: 26 Kim HA, Lee JW, Nam SJ, et al. Adding ovarian suppression to safety and patient-reported outcomes from the monarchE study. tamoxifen for premenopausal breast cancer: a randomized phase III Ann Oncol 2022; 33: 616–27. trial. J Clin Oncol 2020; 38: 434–43. 34 Tripathy D, Im SA, Colleoni M, et al. Ribociclib plus endocrine 27 Coombes RC, Kilburn LS, Tubiana-Mathieu N, et al. Epirubicin therapy for premenopausal women with hormone-receptor-positive, dose and sequential hormonal therapy—mature results of the advanced breast cancer (MONALEESA-7): a randomised phase 3 HMFEC randomised phase III trial in premenopausal patients with trial. Lancet Oncol 2018; 19: 904–15. node positive early breast cancer. Eur J Cancer 2016; 60: 146–53. 35 Tevaarwerk AJ, Wang M, Zhao F, et al. Phase III comparison of 28 Pan H, Gray R, Braybrooke J, et al. 20-year risks of breast cancer tamoxifen versus tamoxifen plus ovarian function suppression in recurrence after stopping endocrine therapy at 5 years. N Engl J Med premenopausal women with node-negative, hormone receptor- 2017; 377: 1836–46. positive breast cancer (E-3193, INT-0142): a trial of the Eastern 29 Early Breast Cancer Trialists’ Collaborative Group. Extending the Cooperative Oncology Group. J Clin Oncol 2014; 32: 3948–58. duration of endocrine treatment for early breast cancer: patient-level 36 Melton LJ 3rd, Crowson CS, Malkasian GD, O’Fallon WM. Fracture meta-analysis of 12 randomised trials of aromatase inhibitors in risk following bilateral oophorectomy. J Clin Epidemiol 1996; 22 031 postmenopausal women already treated with at least 5 years 49: 1111–15. of endocrine therapy. Lancet 2025; 406: 603–14. 37 Early Breast Cancer Trialists’ Collaborative Group. Reductions in 30 Furlanetto J, Marmé F, Seiler S, et al. Chemotherapy-induced recurrence in women with early breast cancer entering clinical trials ovarian failure in young women with early breast cancer: between 1990 and 2009: a pooled analysis of 155 746 women in Prospective analysis of four randomised neoadjuvant/adjuvant 151 trials. Lancet 2024; 404: 1407–18. breast cancer trials. Eur J Cancer 2021; 152: 193–203. 31 Johnston SRD, Toi M, O’Shaughnessy J, et al. Abemaciclib plus endocrine therapy for hormone receptor-positive, HER2-negative, node-positive, high-risk early breast cancer (monarchE): results from a preplanned interim analysis of a randomised, open-label, phase 3 trial. Lancet Oncol 2023; 24: 77–90. --- [PDF原文](https://sci-net.xyz/storage/7932541/af8b646a9341140c27105e3c148db8f9644e587eeb694f70b1e318ec0df906b8/Effects-of-ovarian-ablation-or-suppression-on-breast-cancer-recurrence-and-survival-patient.pdf) DOI: 10.1016/S0140-6736(26)00313-2