Ovarian ablation in early breast cancer: overview of the randomised trials

THE LANCET

Articles

Ovarian ablation in early breast cancer: overview of the randomised trials Early Breast Cancer Trialists’ Collaborative Group*

Summary Background Among women with early breast cancer, the effects of ovarian ablation on recurrence and death have been assessed by several randomised trials that now have long follow-up. In this report, the Early Breast Cancer Trialists’ Collaborative Group present their third 5-yearly systematic overview (meta-analysis), now with 15 years’ follow-up. Methods In 1995, information was sought on each patient in any randomised trial of ovarian ablation or suppression versus control that began before 1990. Data were obtained for 12 of the 13 studies that assessed ovarian ablation by irradiation or surgery, all of which began before 1980, but not for the four studies that assessed ovarian suppression by drugs, all of which began after 1985. Menopausal status was not consistently defined across trials; therefore, the main analyses are limited to women aged under 50 (rather than “premenopausal”) w hen randomised. Oestrogen receptors were measured only in the trials of ablation plus cytotox ic chemotherapy versus the same chemotherapy alone. Findings A mong 2102 w omen aged under 50 w hen randomised, most of w hom w ould have been premenopausal at diagnosis, 1130 deaths and an additional 153 recurrences were reported. 15-year survival was highly significantly improved among those allocated ovarian ablation (52·4 vs 46·1%, 6·3 [SD 2·3] fewer deaths per 100 women, logrank 2p=0·001), as was recurrence-free survival (45·0 vs 39·0%, 2p=0·0007). The numbers of events were too small for any subgroup analyses to be reliable. The benefit was, however, significant both for those with (“node positive”) and for those without (“node negative”) axillary spread when diagnosed. In the trials of ablation plus cytotox ic chemotherapy versus the same chemotherapy alone, the benefit appeared smaller (even for women with oestrogen receptors detected on the primary tumour) than in the trials of ablation in the absence of chemotherapy (where the observed survival improvements were about six per 100 node-negative women and 12 per 100 node-positive women). Among 1354 women aged 50 or over when randomised, most of w hom w ould have been perimenopausal or postmenopausal, there was only a nonsignificant improvement in survival and recurrence-free survival. Interpretation In women aged under 50 with early breast cancer, ablation of functioning ovaries significantly *List of participants at end of report Correspondence to: EBCTCG Secretariat, Clinical Trial Service Unit, Radcliffe Infirmary, Oxford OX2 6HE, UK

Vol 348 • November 2, 1996

improves long-term survival, at least in the absence of chemotherapy. Further randomised evidence is needed on the additional effects of ovarian ablation in the presence of other adjuvant treatments, and to assess the relevance of hormone-receptor measurements.

Lancet 1996; 348: 1189–96 See Commentary page 1184

Introduction Since 1985 the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) has, every 5 years, organised systematic overviews (meta-analyses) of data from individual patients from all randomised trials of the treatment of operable breast cancer.1–4 This report presents the information from the third cycle for trials of ovarian ablation, based on the data collected in 1995, which yields at least 15 years of follow-up for most women.

Methods Trial identification and data-checking procedures for the EBCTCG overviews have been described.1–3 The aim here was to review all properly randomised trials that began recruiting before 1990 and compared the ablation or suppression of ovarian function, sometimes with the addition of prednisone, versus no such adjuvant treatment. In practice, all the trials that can be reviewed here began before 1980, and all involved surgical or radiotherapeutic ablation (see below). To be “properly randomised”, trials had to have used a method of treatment allocation that precluded foreknowledge of the allocated treatment.1 Information was sought for each woman with apparently resectable breast cancer on allocated treatment, age, menopausal status, and whether or not involved axillary lymph nodes, oestrogen receptors (ER), or progesterone receptors had been found at surgery. Dates of randomisation, contralateral breast cancer, first local recurrence, first distant recurrence, and last known vital status were also sought (ideally to 1995). The cause of death was requested for women who had died without any record of distant recurrence. Data were checked for internal consistency and were amended through correspondence with the investigators. After the main analyses were presented, discussed, and revised at a 1995 meeting of all the collaborators, preliminary versions of this report were circulated for comment among those whose trial results it contains (and to the other members of the EBCTCG), and revised in the light of these comments by a writing committee.

Statistical methods Menopausal status was not consistently defined across the different trials, so instead, as before,1–3 the results were analysed separately for women under and over 50 years of age when randomised. The analyses were intended to be of all randomised women by allocated treatment (ie, to be intention-to-treat analyses). The main statistical methods for combining results from different trials have been described1–4 and involve calculating the logrank observed minus expected (O⫺E) for each trial. (For mortality analyses, O⫺E compares the number of deaths

1189

THE LANCET

Trial

Ovarian treatment

Common systemic therapy

Accrual period

Randomised (by 1995) <50 years

Data available

肁50 years

(a) Ovarian irradiation or surgery, without routine cytotoxic chemotherapy 450 rads None Christie (first part) 5 1000 rads None Norwegian RH6 7 Surgery None Boston 8 Surgery None NSABP B-03 9 Surgery None Saskatchewan CF 10 2000 rads† ±Pr PMH Toronto Ontario CTRF 1500 rads None CRFB Caen A 900/1400 rads None Bradford RI (stratum 1) Surgery None

1948–50 1957–63 1961–* 1961–67 1964–74 1965–72 1968–77 1971–76 1974–85

178 151 * 184 255 349 9 1 42

11 195 * 0 124 430 323 51 9

Yes Yes No Yes Yes Yes Yes Yes Yes

Subtotal (a), except Boston

1948–85

1169

1143

Yes

(b) Ovarian irradiation or surgery, with routine cytotoxic chemotherapy Bradford RI (stratum 2) Surgery MTt Toronto-Edmonton 1500 rads+Pr CMF±BCG 11 1600 rads+Pr CMF BCCA Vancouver 12 Surgery CMFPr IBCSG/Ludwig II SWOG 7827 B Surgery CMFVPr

1974–85 1978–88 1979–85 1978–81 1979–89

38 241 111 281 262

5 56 23 75 52

Yes Yes§ Yes§ Yes§ Yes§

Subtotal (b)

1974–89

933

211

Yes

1987–¶ 1989–¶ 1989–¶ 1989–94

972 746 191 1382

0 120 0 155

No No No No

1987–

3291

275

No

(c) Medical ovarian suppression CRC Under 50s FNCLCC France SE Sweden ECOG EST 5188

Goserelin Triptorelin or goserelin‡ Goserelin Goserelin

±Tam FAC or FEC ±Tam FAC±Tam

Subtotal (c)

*143 randomised, but no data on accrual period, age distribution, or outcome. †Stratum 1=control vs 2000 rads, stratum 2=control vs 2000 rads vs 2000 rads+prednisone. ‡Some hospitals used ovarian irradiation or surgery instead. §Hormone-receptor measurements available only from these trials. ¶Still randomising patients. A=adriamycin (doxorubicin), C=cyclophosphamide, E=epirubicin, F=fluorouracil, BCG=bacillus Calmette-Guerin, M=methotrexate, Pr=prednisone, Tam=tamoxifen, Tt=thiotepa, V=vincristine, and ±=randomisation of.

Table: Randomised trials of ovarian ablation or suppression versus control that began before 1990 observed among women allocated ovarian ablation in one particular trial with the number that would have been expected if those women had had the average prognosis of all the women, treated or control, in that trial.) Information from different trials can be combined by summing these logrank statistics, one per trial, and by summing their variances. From these two sums the significance level, or two-sided p value (2p), can be calculated, as can the overall proportional odds reduction and its standard deviation (SD). Crude overall estimates of the absolute differences in outcome are also given. When both are expressed as percentages, the odds reduction is larger than the absolute risk reduction: eg, an odds reduction of 18% might correspond to an absolute improvement in 15-year survival of only about 6% (ie, 50 vs 56%). In analyses of non-breast-cancer deaths, the statistical conventions used to avoid bias are as in the EBCTCG overview of radiotherapy trials, 4 and involve “logrank subtraction”.

Trials 17 trials of ovarian ablation or suppression that began before 1990 were identified (table). Four achieved suppression largely or wholly by the use of luteinising-hormone releasing-hormone (LH-RH) agonists such as goserelin acetate. These four trials all began during the 1980s, and data from them are not yet available for the EBCTCG overview. The other 13 trials all began before 1980, and all achieved ovarian ablation by irradiation or by surgery. Information is available from all but one of these, a small study (143 women) for which patients’ records seem irretrievable.7 The remaining 12 included a total of 3456 women, which is 96% of those known to have been randomised in such studies. In seven of these trials the ovarian ablation and control groups received no routine cytotoxic chemotherapy, in one there were randomisations both for cytotoxic therapy and for ovarian ablation in a “factorial” design, and in four both groups were scheduled to receive a common cytotoxic chemotherapy regimen (after ablation, in those allocated this treatment). Follow-up information to 1995 was not available for three of the 12 trials (Norway, Toronto, and Ontario), but these had already supplied the EBCTCG with follow-up to more than 15 years after randomisation.3 Two of the 12 trials had not been included in the previous EBCTCG overview: one is the National Surgical

1190

Adjuvant Breast Project (NSABP) B-03 trial,8 for which adequate data could not be retrieved in 1990, and the other is the Christie trial5 for which it has only recently been established that a proper randomisation method (sealed envelope) was used for the first part of the study. This allocation method was subsequently changed to a method of treatment allocation based on date of birth that did not preclude foreknowledge, and so patients from the latter part of the Christie trial continue to be excluded. Even when all 12 available trials are combined, the total number of randomised women is still small (2102 aged under 50 at randomisation and 1354 aged 50 or more, table). But, as all of these trials began more than 15 years ago, substantial numbers of deaths have occurred (1130 among women who were under 50 and 1018 among those who were older). Hence, an overview of the results should provide fairly reliable estimates of long-term overall and recurrence-free survival. Separate analyses are presented for women aged under 50 at randomisation (who are likely still to have been premenopausal when diagnosed, with substantial ovarian hormone production) and those aged 50 or more (who are likely to have been perimenopausal or postmenopausal, with much less ovarian function). Because cytotoxic chemotherapy can itself produce partial or complete ovarian suppression in premenopausal women,12,13 separate analyses are also presented of ovarian ablation in the absence and in the presence of chemotherapy. But, since the overall numbers randomised in ablation trials are not large, such subgroup analyses may yield unreliable results.2–4 Information on ER was available only for women in trials with routine cytotoxic chemotherapy.

Results Women aged under 50 at entry Recurrence-free survival—Among women aged under 50, by year 15 there were 6·0 (SD 2·3) fewer recurrences or deaths per 100 women allocated ovarian ablation (45·0 vs 39·0% alive and with no history of local or distant recurrence 15 years after randomisation, logrank 2p=0·0007, figure 1, recurrence-free survival). This highly significant difference was, in absolute terms, somewhat smaller than had been suggested by the previous cycle of

Vol 348 • November 2, 1996

THE LANCET 100

higher in the first 5 years than later, most of the absolute separation between the lines in figure 1 (recurrence-free survival) occurred during the first 5 years. The proportional reduction in recurrence or prior death appeared to be similar in trials of ablation by surgery (18% [8]) or by radiotherapy (19% [8]), but the numbers of patients were not sufficient for this apparent similarity to be reliably informative.

Recurrence-free survival

Recurrence-free survival (%)

80 62·8%

Age <50: Ovarian ablation

60 51·7% 56·0%

45·0% 45·3%

40

39·0%

Difference: 6·0 SD 2·3 events per 100 (logrank 2p = 0·0007

Control 20

0 5

Annual event rates Years 0 – 4 Ablation Control

100

10

Years 5 – 9 9·79% SD 0·47 3·92% SD 0·37 11·97% SD 0·59 4·50% SD 0·46

Years 15+ 3·07% SD 0·36 4·02% SD 0·51

Overall survival

74·2%

80

69·0%

Survival (%)

15+ years

Years 10 – 14 2·62% SD 0·36 2·91% SD 0·46

Age <50: Ovarian ablation

60·0%

60 52·4% 54·4% 46·1%

40

Difference: 6·3 SD 2·3 deaths per 100 (logrank 2p = 0·001)

Control

20

0 0 Annual death rates Years 0 – 4 Ablation Control

6·15% SD 0·35 7·18% SD 0·42

5

10 Years 5 – 9 4·39% SD 0·36 4·91% SD 0·43

15+ years

Years 10 – 14 2·90% SD 0·36 3·17% SD 0·44

Years 15+ 2·63% SD 0·31 3·90% SD 0·46

Figure 1: Absolute effects of ovarian ablation in all trials combined among women aged under 50 at entry Recurrence-free and overall survival for 2102 women aged under 50 when randomised between ovarian ablation (■) and control (●).

this overview.3 But, as it was based on a larger amount of evidence, especially in later years (see Discussion), the difference provides a more reliable estimate of the average benefit in the trials than was previously available. Proportionally, ovarian ablation reduced the event rate by about one-fifth (18·5% [SD 5·5], figure 2, recurrence-free survival), both in the first 5 years (ie, years 0–4) and in later years. But, since the absolute event rate was much

Vol 348 • November 2, 1996

Overall survival Although some recurrences are curable, most are not, so at 15 years the absolute difference was about as large for overall survival (52·4 vs 46·1% still alive 15 years after randomisation, logrank 2p=0.001, figure 1, overall survival) as for recurrence-free survival. This observation corresponds to an average of 6·3 (SD 2·3) fewer deaths per 100 women treated in all the trials. Again, although the proportional reduction in mortality was about the same in the earlier as in the later years, most of the absolute difference in overall survival appeared during the first 5 years. When these recurrence-free and survival analyses were further subdivided, no material differences between the benefits at ages under 40 or 40–49 were apparent (data not shown). Possible influence of cytotoxic chemotherapy —Because cytotoxic chemotherapy may itself suppress ovarian function, especially in women who are nearing the menopause,12,13 ovarian ablation may produce a much greater change in hormonal status in women who do not receive chemotherapy than in those who do. Analyses subdivided by whether or not women in both treatment groups of each trial were to receive routine cytotoxic chemotherapy are shown in figure 2. The proportional improvement in recurrence-free survival among women aged under 50 at entry was 25% (SD 7) in the absence of chemotherapy but only 10% (9) in its presence (figure 2, recurrence-free survival). If real, the difference between these two results could be important. But the amount of evidence available was small and the two results were not significantly different from each other, so we cannot determine just from this evidence whether chemotherapy modified the size of the effects of ovarian ablation. Similarly, the proportional improvement in survival was 24% (7) in the absence but only 8% (10) in the presence of chemotherapy (figure 2, overall survival). Again, the numbers of deaths were too small to assess any such modifications of the size of the survival effects reliably. In the trials of ovarian ablation in the absence of routine chemotherapy, only 20 (2%) of the 1169 women aged under 50 were classified as postmenopausal at randomisation. Hence, if the analyses are restricted to premenopausal women only (according to the definition used in each trial), the findings are virtually unchanged, with ovarian ablation producing a proportional improvement in recurrence-free survival of 27% (7) in the absence of chemotherapy. Possible influence of nodal status in trials without cytotox ic chemotherapy —The indirect comparisons between the size of benefit in the absence and in the presence of cytotoxic chemotherapy are strongly confounded by nodal status, and vice versa. Almost all of the node-negative women (473 of 502) aged under 50 were entered into ovarian ablation trials in the absence of 1191

THE LANCET

Events/patients Allocated Adjusted ablation control

Study name

Ablation events Var Obs of O-E –exp

Annual odds of event Reduction Ratio (and CIs), (% and SD) ablation: control

(a) Ovarian ablation in the absence of chemotherapy Christie A Norwegian RH NSABP B-03* Saskatchewan CF PMH Toronto* Ontario CTRF CRFB Caen A Bradford RI (stratum 1)

(a) Subtotal*

77/88 24/68 76/129 72/143 139/216 5/6 1/1 6/22

80/90 46/83 2(37/55) 68/112 148/204 3/3 0/0 7/20

–7·4 –11·4 –4·3 –8·8 –12·4 2·2 ·· –0·9

30·4 16·6 20·9 31·5 47·8 1·3 ·· 3·1

400/673 (59·4%)

426/622 (68·5%)

–43·1

151·6

25% SD 7 (2p = 0·0005)

(b) Ovarian ablation in the presence of chemotherapy Bradford RI (stratum 2) Toronto-Edmonton BCCA Vancouver IBCSG/Ludwig II SWOG 7827 B

(b) Subtotal Total (a+b)* 99% or

6/21 70/119 33/57 97/139 63/136

7/17 72/122 28/54 105/142 62/126

269/472 (57·0%)

274/461 (59·4%)

669/1145 700/1083 (58·4%) (64·6%)

–1·0 –1·4 0·9 –8·7 –2·6

3·1 31·3 13·9 44·0 28·9

–12·9

121·2

–55·9

272·7

10% SD 9 (2p > 0·1, NS)

18·5% SD 5·5 (2p = 0·0007)

95% interval

Test for heterogenity (12 trials): χ 211 = 11·1, p > 0·1, NS Test for interaction (a) vs (b): χ 21 = 2·1, p > 0·1, NS

0

* Adjustments for balance, control patients in 2:1 randomisations contribute twice. PMH contributes stratum 1: 49/68 vs 50/62 plus stratum 2: 90/148 vs 2(49/71)

Deaths/patients Allocated Adjusted ablation control

Study name

0·5

1·0

1·5

2·0

Ablation better Ablation worse Treatment effect 2p = 0·0007

Ablation deaths Obs Var –exp of O-E

Annual odds of death Reduction Ratio (and CIs), (% and SD) ablation: control

(a) Ovarian ablation in the absence of chemotherapy Christie A Norwegian RH NSABP B-03* Saskatchewan CF PMH Toronto* Ontario CTRF CRFB Caen A Bradford RI (stratum 1)

(a) Subtotal*

76/88 24/68 75/129 59/143 133/216 5/6 1/1 5/22

80/90 43/83 2(35/55) 62/112 143/204 3/3 0/0 7/20

–7·3 –9·1 –2·3 –11·6 –11·7 0·9 ·· –1·4

33·7 15·9 21·2 28·2 48·6 1·8 ·· 2·9

378/673 (56·2%)

408/622 (65·6%)

–42·5

152·3

24% SD 7 (2p = 0·0006)

(b) Ovarian ablation in the presence of chemotherapy Bradford RI (stratum 2) Toronto-Edmonton BCCA Vancouver IBCSG/Ludwig II SWOG 7827 B

(b) Subtotal Total (a+b)* 99% or

5/21 56/119 21/57 85/139 42/136

4/17 52/122 21/54 91/142 49/126

0·6 3·1 –1·4 –5·9 –5·0

1·9 25·1 10·0 39·8 21·7

209/472 (44·3%)

217/461 (47·1%)

–8·6

98·5

587/1145 625/1083

8% SD 10 (2p > 0·1, NS)

18·4% SD 5·7

–51·1 250·8 (51·3%) (57·7%)

(2p = 0·001)

95% interval

Test for heterogenity (12 trials): χ211 = 8·1, p > 0·1, NS Test for interaction (a) vs (b): χ21 = 2·2, p > 0·1, NS * Adjustments for balance, control patients in 2:1 randomisations contribute twice. PMH contributes stratum 1: 46/68 vs 49/62 plus stratum 2: 87/148 vs 2(47/71)

0

0·5

1·0

1·5

2·0

Ablation better Ablation worse Treatment effect 2p = 0·001

Figure 2: Proportional effects of ovarian ablation in each trial and overall, with subdivisions by absence or presence of chemotherapy, among women aged under 50 at entry Top=recurrence-free and bottom=overall survival. Each trial, or part of trial, is described by a single line showing numbers of events and patients and summary logrank statistics. For each subdivision, the ratio of the annual event rate in the ovarian ablation group to that in the control group (odds ◆. Solid vertical line=odds ratio of 1·0 (ie, no difference ratio) is plotted as ■ with 99% CI. For subtotals and total, 95% CIs are represented by ◆ between ovarian ablation and control), whereas broken vertical line=“typical odds ratio” in total of all these trials. For balance, control patients in 2:1 randomisations (ie, NSABP and part of PMH) are counted twice in adjusted control totals, but not in statistical calculations. Var=variance, NS=not significant.

1192

Vol 348 • November 2, 1996

THE LANCET

Recurrence-free survival

Overall survival

100

100

88·7% 84·0% 80·3%

Node-negative, age <50: Ovarian ablation 75·4%

80

73·6%

88·6% 80

Difference: 8·9 SD 4·2 events per 100 (logrank 2p = 0·01)

78·7% 61·3%

66°·5%

60

51·8% 44·1%

Control Node-positive, age <50: Ovarian ablation 37·4%

40

30·6%

48·3%

76·6% Difference: 5·6 SD 4·0 deaths per 100 (logrank 70·9% 2p = 0·01)

Node-positive, age <50: Ovarian ablation 41·7%

52·2%

Difference: 12·5 SD 3·9 deaths per 100 (logrank 2p = 0·0007)

40 38·3% 29·2%

Control

20

24·0%

Node-negative, age <50: Ovarian ablation

Control

60

Difference: 13·4 SD 3·8 events per 100 (logrank 2p = 0·0002)

38·6%

20

Survival (%)

Recurrence-free survival (%)

82·5%

82·8%

Control

0

0 0

5

10

15+ years

0

5

10

15+ years

Figure 3: Absolute effects of ovarian ablation in absence of routine chemotherapy in all trials combined among women aged under 50 at entry Recurrence-free and overall survival for 473 node-negative and 696 node-positive women who were aged under 50 when randomised between ovarian ablation (■) and control (●) in trials, or parts of trials, where cytotoxic therapy was not routinely used: Years of follow-up

Ablation

Control

Deaths

Person-years

Annual death rate (SD)

Deaths

Person-years

Annual death rate (SD)

Node-negative 0–4 5–9 10–14 15+

28 15 12 33

1170 1030 931 1580

2·4% 1·5% 1·3% 2·1%

25 21 15 43

1037 884 779 1309

2·4% (0·5) 2·4% (0·5) 1·9% (0·5) 3·3% (0·5)

Node-positive 0–4 5–9 10–14 15+

166 55 29 40

1620 1077 870 1151

10·3% (0·8) 5·1% (0·7) 3·3% (0·6) 3·5% (0·6)

134 37 23 28

997 577 426 491

13·4% (1·2) 6·4% (1·1) 5·4% (1·1) 5·7% (1·1)

(0·5) (0·4) (0·4) (0·4)

chemotherapy, whereas almost all of the women in the trials in the presence of chemotherapy (904 of 933) were node-positive. Hence, the relevance of nodal status can be assessed only in the ovarian ablation trials in which chemotherapy was not routinely given. As would be expected, the 473 women classified as node-negative (either by axillary clearance or just by axillary sampling) had a better prognosis than the remaining 696 women classified as node-positive (or of unknown nodal status). Because the number of women studied was not large, the separate effects of ovarian ablation in node-positive and node-negative women cannot be estimated reliably. But, whether or not the nodes were involved, ovarian ablation in the absence of chemotherapy was associated with significant improvements in recurrence-free survival and in overall survival. The proportional risk reductions for node-positive and for node-negative women were similar, but the absolute risk reduction was non-significantly greater for the node-positive women (figure 3). Most of the patients in these trials were randomised before 1970, and most survivors have been followed up to beyond 1990, so there is information beyond year 15. Even during this late period, the annual death rates, taking all women together, remained lower among those who had been allocated ablation (2·6% [SD 0·3]) than among the controls (3·9% [0·5]). This observation provides some

Vol 348 • November 2, 1996

reassurance against the later emergence of any substantial hazards.

ER status in trials with cytotoxic chemotherapy —ER measurements on the primary tumour were available for most women aged under 50 at entry in four of the five trials in which both randomised groups received chemotherapy, but not for any women in trials of ovarian ablation in the absence of chemotherapy. Among the 194 women with “ER poor” primary tumours, there was no apparent difference between ablation plus chemotherapy versus chemotherapy alone, either in recurrence-free survival (logrank O⫺E 0·3, variance 26·9, not significant) or in overall survival (logrank O⫺E 1·9, variance 24·0, not significant). Among the 550 women with “ER positive” primary tumours, however, ablation plus chemotherapy appeared to be more effective than chemotherapy alone, both for recurrence-free survival (logrank O⫺E ⫺9·5, variance 67·0, odds reduction 13% [11]) and for overall survival (logrank O⫺E ⫺9·2, variance 50·5, odds reduction 17% [13]), but these differences were not statistically significant. Cause-specific mortality —Most studies were able to supply some cause-specific mortality information. Among women aged under 50 at randomisation who died without a distant recurrence of their breast cancer being recorded, 1193

THE LANCET

116 were classified as having died of non-breast-cancer causes. After allowance had been made4 for those allocated ablation surviving longer, and therefore being at more prolonged risk of death from other causes, there was no significant difference between the treatment groups in vascular deaths (22/929 ablation vs 20/824 controls in trials with data, logrank O⫺E ⫺1.6, variance 9·5, not significant), in other non-breast-cancer deaths (44/929 vs 30/824, logrank O⫺E ⫺0·2, variance 16·5, not significant) or in all non-breast-cancer deaths.

Contralateral breast cancer—The suppression of ovarian function might be expected to reduce the incidence of contralateral breast cancer. But, even in the aggregate of all available trials, there was not enough information to confirm or refute this suggestion (30 contralateral breast cancers as first event among 712 women allocated ablation vs 32 among 679 controls in trials with data, logrank O⫺E ⫺2·8, variance 15·1, not significant). Women aged 50 or above at entry Because ovarian function is generally substantially less at older than at younger ages, any effects of ovarian ablation might be expected to be much less. Of the 1354 women aged 50 or above when randomised, 1018 were reported to have died and a further 48 to have had a distant or local recurrence. Despite these large numbers of events, there were no significant differences between the treatment groups in recurrence-free survival or in overall survival. By year 15 after randomisation there were 3·1 (SD 2·6) fewer recurrences or deaths per 100 women allocated ovarian ablation (32·0 vs 28·9% alive with no history of local or distant recurrence, logrank O⫺E ⫺17·6, variance 225·9, not significant) and 2·5 (SD 2·7) fewer deaths per 100 (36·9 vs 34·5% alive, logrank O⫺E ⫺8·9, variance 223·8, not significant).

Discussion For women aged under 50 years, most of whom were premenopausal when diagnosed, these updated analyses continue to show that ovarian ablation produced a highly significant improvement in recurrence-free and in overall survival. The relevance of nodal status could be assessed only in trials of ovarian ablation in the absence of cytotoxic chemotherapy, and in these there was significant benefit both for women with and for those without axillary spread. The risk reduction produced by ovarian ablation appeared to be smaller in the presence of routine cytotoxic chemotherapy (ie, in trials of ablation plus chemotherapy versus the same chemotherapy alone) and, in these trials, the benefits appeared to be greater in women with ER-positive primary tumours. All of these indirect comparisons of treatment effects in different circumstances were, however, based on numbers that were too small to be statistically reliable, and may be distorted by chance.14 The benefits of ovarian ablation might well be lessened by the presence of polychemotherapy, for such chemotherapy induces ovarian dysfunction in many premenopausal women.12,13 This may explain why the average effects of ovarian ablation look smaller in figure 1 than they did in the previous cycle of this overview,3 because the trials of ovarian ablation in the presence of routine cytotoxic chemotherapy started more recently than those in its absence, and now make a greater proportional contribution to the overall 15-year analyses. 1194

This worldwide collaboration of trialists is so extensive that no substantial amount of properly randomised evidence is likely to have been overlooked. The data have been extensively checked, but still the irregular denominators in some of the trials (figure 2) raise the possibility that some randomised patients may have been inappropriately excluded from follow-up. Nevertheless, the overall finding of some long-term benefit from ovarian ablation in women under 50 is secure, even though the exact size of that benefit for particular types of patients remains uncertain. By contrast, there was no significant benefit in older women, most of whom would have been perimenopausal or postmenopausal and might, therefore, have been expected to have much less to gain from ablation. Much additional information on ovarian ablation will be available for the next cycle of the EBCTCG (in 2000), both from longer follow-up of the trials included here and from the trials that began more recently. For example, a Chinese trial of ovarian ablation that began in 1991 has already randomised more than 3000 women aged under 50 between adjuvant cytotoxic chemotherapy alone and adjuvant chemotherapy plus ablation. In addition, there are more than 3000 women in the recent trials of ovarian suppression with LH-RH agonists (table). Further large randomised trials are, however, still needed not only to assess reliably the additional effects of ovarian ablation in the presence of cytotoxic chemotherapy (and vice versa), but also to assess these effects in the presence and absence of prolonged antioestrogen therapy, and to assess the relevance of hormone-receptor measurements. Such trials might involve three-way comparisons of ablation versus cytotoxic chemotherapy versus both: these comparisons would assess the additional value of each in the presence of the other, and would also add to the evidence15 on ablation versus chemotherapy. However, although the additional relevance of ablation in populations where other adjuvant treatments are being used requires further study, the value of ablation in populations where other adjuvants are not used routinely (which, worldwide, involve large numbers of women) is now known to be substantial and persistent, reducing both recurrences and deaths during the first 15 years. When deciding whether to use this treatment, however, it is also important to consider possible side-effects, including acute menopausal symptoms and any long-term effects of early menopause. 100 years ago Beatson published the first report of ovarian ablation for advanced breast cancer.16 Almost 50 years ago, when randomised trials were becoming important in medicine, Paterson began the first randomised trial of ovarian ablation for early breast cancer, which might also be the first randomised trial of any treatment for cancer.5 By 1980 another dozen randomised trials of adjuvant ovarian ablation had been undertaken but none was large, and our present report is the first definitive overview of findings on long-term survival. Much larger scale randomised evidence will not be available until the early years of the next century. In the meantime, the present analyses provide the best available randomised evidence of the long-term effects of ovarian ablation for women with early breast cancer. The analyses are already sufficient to show that ablation of functioning ovaries in early breast cancer improves long-term survival with (at least in the absence of other adjuvant treatments) significant benefit both in node-negative and in nodepositive premenopausal women.

Vol 348 • November 2, 1996

THE LANCET

EBCTCG Secretariat M Clarke, R Collins, C Davies, J Godwin, R Gray, and R Peto accept full responsibility for the overall content of this manuscript.

EBCTCG The trial groups and trialists who are members of the EBCTCG are listed below in alphabetical order of group, of institute, or of location. The groups whose data were included in the analyses presented here are marked *.

Trial groups and trialists ACETBC, Japan O Abe, R Abe, K Enomoto, K Kikuchi, H Koyama, Y Nomura, K Sakai, K Sugimachi, T Tominaga, J Uchino, M Yoshida. Amsterdam Integraal Kankercentrum, Netherlands AO van de Velde, JA van Dongen, JB Vermorken. Athens Metaxas Memorial Cancer Hospital, Greece A Arvelakis, G Giokas, B Lissaios. Auckland Breast Cancer Study Group, New Zealand VJ Harvey, TM Holdaway, RG Kay, BH Mason. Australian-New Zealand Breast Cancer Trials Group A Coates, JF Forbes. Belgian Adjuvant Breast Cancer Project, Belgium C Focan, JP Lobelle. Berlin-Buch Akademie der Wissenschaften, Germany U Peek. Birmingham General Hospital, UK GD Oates, J Powell. BMFT Freiberg, Germany G Bastert, H Rauschecker, R Sauer, W Sauerbrei, A Schauer, M Schumacher. Bordeaux Institut Bergonié, France M Durand, L Mauriac. Bordet Institute, Belgium S Bartholomeus, MJ Piccart. Boston Dana-Farber Cancer Institute, USA RS Gelman, IC Henderson, CL Shapiro. Bradford Royal Infirmary, UK* AK Hancock, MB Masood, D Parker, JJ Price. British Columbia Cancer Agency, Canada* S Jackson, J Ragaz. Caen Centre Regional François Baclesse, France* T Delozier, J Mace-Lesec’h. Cambridge Addenbrooke’s Hospital, UK JL Haybittle. Cancer & Leukemia Group B, USA C Cirrincione, IC Henderson, A Korzun, RB Weiss, WC Wood. Cancer Research Campaign, UK M Baum, J Houghton, D Riley. Cape Town Groote Schuur Hospital, South Africa DM Dent, CA Gudgeon, A Hacking. Cardiff Surgery Trialists, UK K Horgan, L Hughes, HJ Stewart. Case Western Reserve University, Ohio NH Gordon. Central Oncology Group, USA HL Davis. Centre Léon-Bérard, France P Romestaing, Y Lehingue. Cheltenham General Hospital, UK JR Owen. Chicago University, USA P Meier. Christie Hospital & Holt Radium Institute, UK* A Howell, GC Ribeiro, R Swindell. Coimbra Instituto de Oncologia, Portugal J Albano, CF de Oliveira, H Gervásio, J Gordilho. Copenhagen Danish Cancer Registry, Denmark B Carstensen, T Palshof. Copenhagen Radium Centre, Denmark H Johansen. Cracow Institute of Oncology, Poland S Korzeniowski, J Skolyszewski. Danish Breast Cancer Cooperative Group, Denmark KW Andersen, CK Axelsson, M Blichert-Toft, HT Mouridsen, M Overgaard, C Rose. Dublin St Luke’s Hospital, Ireland N Corcoran. Düsseldorf University, Germany HJ Trampisch. Eastern Cooperative Oncology Group, USA MD Abeloff, PC Carbone, J Glick, R Gray, DC Tormey. Elim Hospital, Germany J Rossbach. European Organization for Research and Treatment of Cancer, Belgium. Evanston Hospital, USA EF Scanlon, S Schurman. Ghent University Hospital, Belgium A de Schryver. Glasgow Beatson Oncology Centre, UK HMA Yosef. Glasgow Victoria Infirmary, UK CS McArdle, DC Smith. Granada University Hospital, Spain PC Lara. Gruppo Ricerca Ormono Chemio Terapia Adiuvante, Italy F Boccardo. Guadalajara Hospital de 20 Noviembre, Mexico A Erazo, JY Medina. Gunma University, Japan M Izuo, Y Morishita. Guy’s Hospital, UK A Bentley, Z Doran, IS Fentiman, JL Hayward, RD Rubens. Gynecological Adjuvant Breast Group, Germany M Kaufmann, W Jonat. Heidelberg University I, Germany H Scheurlen. Heidelberg University II, Germany D von Fournier, M Kaufmann. Hellenic Cooperative Oncology Group, Greece G Fountzilas. Helsinki Deaconess Medical Centre, Finland P Klefstrom. Helsinki University, Finland C Blomqvist. ICRF, UK J Cuzick. Innsbruck University, Austria R Margreiter. International Breast Cancer Study Group (Ludwig), Switzerland* M Castiglione, F Cavalli, J Collins, J Forbes, RD Gelber, A Goldhirsch, J Lindtner, KN Price, CM Rudenstam, HJ Senn. International Collaborative Cancer Group, London, UK JM Bliss, CED Chilvers, RC Coombes, M Marty. Israel NSABC, Israel R Borovik, G Brufman, H Hayat, E Robinson, N Wigler. Italian Cooperative ChemoRadio-Surgical Group, Italy F Pannuti. Japanese National Hospitals Group Breast Cancer Study Group, Japan S Takashima, T Yasutomi. Kawasaki Medical School, Japan H Sonoo. Kumamoto University Group, Japan J Yamashita, M Ogawa. Kyushu National Cancer Center, Japan Y Nomura. Limburg, Breast Cancer Study Group of the Comprehensive Cancer Centre, Netherlands PSGJ Hupperets. Louvain Academisch Ziekenhuis St Rafael, Belgium J Bonte. Lund University, Sweden I Tengrup, L Tennvall-Nittby. Marseille Laboratoire de Cancérologie Biologique APM, France P Martin, S Romain. Mayo Clinic, USA D Ahmann, DJ Schaid. MD Anderson Cancer Center, USA AU Buzdar, T Smith. Memorial Sloan-Kettering Cancer Center, USA T Hakes, L Norton, R Wittes. Mexican National Medical Centre, Mexico R de la Huerta, MG Sainz. Milan Istituto Nazionale per lo Studio e la Cura dei Tumori, Italy G Bonadonna, M del Vecchio, P Valagussa, U Veronesi. Montpellier Centre Paul Lamarque, France JB Dubois. Naples University, Italy AR Bianco. National Cancer Institute, USA ME Lippman, LJ Pierce, R Simon, SM Steinberg. National Surgical Adjuvant Project for Breast & Bowel Cancers, USA* A Brown, B Fisher, C Redmond, N Wolmark. Nolvadex Adjuvant Trial Organisation, UK M Baum, IM Jackson, MK Palmer. North Central Cancer Treatment Group, USA JN Ingle, VJ Suman. North Sweden Breast Cancer Group, Sweden NO Bengtsson, LG Larsson. North-Western British Surgeons, UK

Vol 348 • November 2, 1996

JP Lythgoe, R Swindell. Northwick Park Hospital, UK M Kissin. Norwegian Breast Cancer Group, Norway E Hannisdal, JE Varhaug; Norwegian Radium Hospital, Norway* R Nissen-Meyer. Nottingham City Hospital, UK RW Blamey, AK Mitchell, JFR Robertson. Oita Prefectural Hospital, Japan Y Nakamura. Oncofrance, France G Mathé, JL Misset. Ontario Cancer Treatment & Research Foundation, Canada* HT Abu-Zahra, EA Clarke, JR McLaughlin. Ontario Clinical Oncology Group, Canada RM Clark, M Levine. Ontario National Cancer Institute of Canada Clinical Trials Group, Canada JD Myles, JL Pater, KI Pritchard. Osaka City University, Japan K Morimoto. Osaka National Hospital, Japan K Sawa, Y Takatsuka. Oslo Radium Hospital, Norway S Gundersen, M Hauer-Jensen, H Høst. Oxford Churchill Hospital, UK E Crossley, K Durrant, A Harris. Oxford ICRF/MRC Clinical Trial Service Unit, UK A Beighton, M Clarke, R Collins, C Davies, V Evans, J Godwin, R Gray, E Greaves, C Harwood, S James, E Lau, G Mead, A Muldal, A Naughton, R Peto, A Tooth, K Wheatley. Paris Centre René Huguenin, France P Rambert. Paris Institut Curie, France B Asselain, RJ Salmon, JR Vilcoq. Paris Institut Gustave-Roussy, France R Arriagada, C Hill, A Laplanche, MG Lê, M Spielmann. Parma Hospital, Italy G Cocconi, B di Blasio. Philadelphia Fox Chase Cancer Centre, USA R Catalano, RH Creech. Piedmont Oncology Association, USA J Brockschmidt, MR Cooper. Prague Charles University, Czech Republic O Andrysek, J Barkmanova. Pretoria University, South Africa CI Falkson. Rosario, Instituto Cardiovascular de Rosario, Argentina M Abraham. Rotterdam Daniel den Hoed Cancer Center, Netherlands JGM Klijn, AD Treurniet-Donker, WLJ van Putten. Royal Marsden Hospital, Institute of Cancer Research, UK D Easton, TJ Powles. St George’s Hospital, UK JC Gazet. St Petersburg Petrov Research Institute of Oncology, Russia V Semiglazov. Sardinia Oncology Hospital A Businico, Sardinia N Deshpande, L di Martino. SASIB International Trialists, South Africa P Douglas, A Hacking, H Høst, A Lindtner, G Notter. Saskatchewan Cancer Foundation, Canada* AJS Bryant, GH Ewing, JL Krushen-Kosloski. Scandinavian Adjuvant Chemotherapy Study Group, Norway R Nissen-Meyer. Scottish Cancer Trials Office, UK APM Forrest, W Jack, C McDonald, HJ Stewart. South Swedish Breast Cancer Group, Sweden TR Möller, S Rydén. South-East Sweden Breast Cancer Group, Sweden J Carstensen, T Hatschek, M Söderberg. Southeastern Cancer Study Group & Alabama Breast Cancer Project, USA JT Carpenter. Southwest Oncology Group, USA* K Albain, J Crowley, S Green, S Martino, CK Osborne, PM Ravdin. Stockholm Breast Cancer Study Group, Sweden LE Rutqvist, A Wallgren. Stockholm Karolinska Hospital, Sweden LE Holm. Swiss Group for Clinical Cancer Research SAKK & OSAKO, Switzerland M Castiglione, A Goldhirsch, HJ Senn, B Thürlimann. Tel Aviv University, Israel H Brenner, A Hercbergs. Tokyo Cancer Institute Hospital, Japan M Yoshimoto. Toronto-Edmonton Breast Cancer Study Group, Canada G DeBoer, AHG Paterson, KI Pritchard. Toronto Princess Margaret Hospital, Canada* JW Meakin, T Panzarella, KI Pritchard. Toulouse Centre Claudius Regaud, France A Naja. Tunis Institut Salah Azaiz, Tunisia J Bahi. UK Multicentre Cancer Chemotherapy Study Group, UK M Reid, M Spittle. UK/Asia Collaborative Breast Cancer Group, UK F Senanayake. Uppsala-Örebro Cancer Study Group, Sweden J Bergh, L Holmberg. Vienna University Hospital 1st Department of Gynaecology, Austria P Sevelda, CC Zielinsky. Vienna University Hospital Department of Surgery, Austria R Jakesz, M Gnant. Wessex Radiotherapy Centre, UK RB Buchanan, M Cross. West Midlands Oncology Association, UK JA Dunn, WM Gillespie, K Kelly, JM Morrison. West of Scotland Breast Trial, UK A Litton. Western Cancer Study Group, USA RT Chlebowski. Witwatersrand University, South Africa WR Bezwoda. Würzburg University, Germany H Caffier.

Acknowledgments This continuing collaboration of breast cancer trialists is funded by a special grant from the Imperial Cancer Research Fund to the Clinical Trial Service Unit & Epidemiological Studies Unit in the Nuffield Department of Clinical Medicine, University of Oxford. The chief acknowledgement is to the thousands of women who took part in the trials reviewed here, and to the members of the relevant trial groups who supplied their data.

References 1

2

3

4

5

Early Breast Cancer Trialists’ Collaborative Group. Effects of adjuvant tamoxifen and of cytotoxic therapy on mortality in early breast cancer: an overview of 61 randomised trials among 28,896 women. N Engl J Med 1988; 319: 1681–92. Early Breast Cancer Trialists’ Collaborative Group. Treatment of early breast cancer, Vol 1: worldwide evidence 1985-1990. Oxford: Oxford University Press, 1990. Early Breast Cancer Trialists’ Collaborative Group. Systematic treatment of early breast cancer by hormonal, cytotoxic, or immune therapy: 133 randomised trials involving 31 000 recurrences and 24 000 deaths among 75 000 women. Lancet 1992; 339: 1–15, 71–85. Early Breast Cancer Trialists’ Collaborative Group. Effects of radiotherapy and surgery in early breast cancer: an overview of the randomised trials. N Engl J Med 1995; 333: 1444–55. Paterson R, Russell MH. Clinical trials in malignant disease: part II. Breast cancer: value of irradiation of the ovaries. J Fac Radiol 1959; 10:130–33.

1195

THE LANCET 6

Nissen-Meyer R. Primary breast cancer: the effect of primary ovarian irradiation. Ann Oncol 1991; 2: 343–46. 7 Nevinny HB, Nevinny D, Roscoff CB, Hall TC, Muench H. Prophylactic oophorectomy in breast cancer therapy. Am J Surg 1969; 117: 531–36. 8 Ravdin RG, Lewison EF, Slack NH, et al. Results of a clinical trial concerning the worth of prophylactic oophorectomy for breast carcinoma. Surg Gynecol Obstet 1970; 31: 1055–64. 9 Bryant AJS, Weir JA. Prophylactic oophorectomy in operable instances of carcinoma of the breast. Surg Gynecol Obstet 1981; 153: 660–64. 10 Meakin JW, Hayward JL, Panzarella T, et al. Ovarian irradiation and prednisone therapy following surgery and radiotherapy for carcinoma of the breast. Breast Cancer Res Treat 1996; 37: 11–19. 11 Ragaz J, Jackson S, Nilson K, et al. Randomized study of locoregional radiotherapy and ovarian ablation in premenopausal patients with breast cancer treated with adjuvant chemotherapy. Proc Am Soc Clin Oncol 1988; 7: 12 (abstr).

12 Ludwig Breast Cancer Study Group. Chemotherapy with or without oophorectomy in high-risk premenopausal patients with operable breast cancer. J Clin Oncol 1985; 3: 1059–67. 13 Richards MA, O’Reilly SM, Howell A, et al. Adjuvant cyclophosphamide, methotrexate and fluorouracil in patients with axillary node-positive breast cancer: an update of the Guy’s/Manchester Trial. J Clin Oncol 1990; 8: 2032–39. 14 Counsell CE, Clarke MJ, Slattery J, Sandercock PAG. The miracle of DICE therapy for acute stroke: fact or fictional product of subgroup analysis? BMJ 1994; 309: 1677–81. 15 Scottish Cancer Trials Breast Group. Adjuvant ovarian ablation versus CMF chemotherapy in premenopausal women with pathological stage II breast carcinoma: the Scottish trial. Lancet 1993; 341: 1293–98. 16 Beatson GT. On the treatment of inoperable cases of carcinoma of the mamma: suggestions for a new method of treatment, with illustrative cases. Lancet 1896; ii: 104–07.

Systematic review of amodiaquine treatment in uncomplicated malaria

P Olliaro, C Nevill, J LeBras, P Ringwald, P Mussano, P Garner, P Brasseur

Summary Background Opinion and policy over the use of amodiaquine for treating malaria vary. Amodiaquine is more palatable than chloroquine and may be more effective but serious adverse events have been reported in travellers taking it as prophylaxis. It is not recommended as first-line treatment. In the light of the global debate over the use of this drug, we conducted a systematic review of the effectiveness and tolerability of amodiaquine in the treatment of uncomplicated falciparum malaria. Methods This is a systematic review of published and unpublished randomised or pseudorandomised trials of amodiaquine. Observational reports w ere also systematically identified and reviewed to access evidence of serious adverse events. Findings 40 trials met the inclusion criteria. Symptomatic patients were enrolled in 24 studies in comparisons of amodiaquine ( n=1071) w ith chloroquine ( n=1097) . A modiaquine w as significantly more effective than chloroquine, with odds ratios and 99% confidence intervals (OR [99% CI]) of 4·29 (3·30–5·58) on day 7 and 6·00 (3·97–9·06) on day 14. Time to parasite clearance was significantly shorter with amodiaquine and fever clearance times were marginally faster. Eight studies compared amodiaquine w ith chloroquine in asymptomatic UNDP/World Bank/WHO Special Programme for Training and Research in Tropical Diseases (TDR), Geneva, Switzerland (P Olliaro MD); AMREF, Nairobi, Kenya (C Nevill MD); Institut de Médecine et d’Epidémiologie Tropicales, Hôpital Bichat-C Bernard, Paris, France (J LeBras PhD); ORSTOM/OCEAC, Yaoundé, Cameroon (P Ringwald MD); Cochrane Tropical Diseases Group, Liverpool, UK (P Mussano PhD, P Garner MD); and Laboratoire de Parasitologie, Hôpital C Nicolle, Rouen, France (P Brasseur MD) Correspondence to: Dr P Olliaro, CHEMAL, UNDP/World Bank/WHO Special Programme for Training and Research in Tropical Diseases (TDR), World Health Organization, 20 avenue Appia, 1211 Geneva 27, Switzerland

1196

parasitaemia, with effects on parasitological outcomes similar to those for symptomatic malaria. At twelve sites, 692 amodiaquine and 679 sulfadox ine/pyrimethamine (S/P) recipients were enrolled. The two drugs did not differ significantly on day 7 (OR 0·74 [0·48–1·15]) but the odds ratios favoured S/P on day 14 (OR 0·51 [0·28–0·93]) and on day 28 ( OR 0·30 [ 0·16–0·55] ) . The time to parasitological clearance was similar in the two groups; fever clearance times were significantly shorter with amodiaquine. Tolerability w as assessed for both comparative and non-comparative trials. The rates of adverse events in controlled trials were 10·7%, 8·8%, and 14·3% w ith amodiaquine, chloroquine, and S/ P, respectively. No life-threatening adverse events and no significant shifts in laboratory indices were reported. Interpretation This systematic review of published and unpublished trials supports the use of amodiaquine in the treatment of uncomplicated malaria. However, there is partial cross-resistance betw een chloroquine and amodiaquine, and monitoring of the effectiveness of this drug and surveillance for evidence of tox icity must continue.

Lancet 1996; 348: 1196–201 See Commentary page 1184

Introduction Amodiaquine is a 4-aminoquinoline used widely in the past to treat and prevent malaria. In the mid-1980s, reports of fatal adverse drug reactions (ADRs) were described in travellers using amodiaquine as prophylaxis.1,2 As a consequence one manufacturer (Parke-Davis) modified the labelling and withdrew prophylaxis as an indication, and in 1990 the World Health Organization (WHO) stopped using this drug in malaria control programmes.3 WHO’s 19th Expert Committee on Malaria modified this in 1993 to say that “amodiaquine could be used for treatment if the risk of infection outweighs the potential for [adverse drug reactions]”, but still do not

Vol 348 • November 2, 1996