Aromatase inhibitors versus tamoxifen in early breast cancer: patient-level meta-analysis of the randomised trials
Abstract
BACKGROUND: The optimal ways of using aromatase inhibitors or tamoxifen as endocrine treatment for early breast cancer remains uncertain. METHODS: We undertook meta-analyses of individual data on 31,920 postmenopausal women with oestrogen-receptor-positive early breast cancer in the randomised trials of 5 years of aromatase inhibitor versus 5 years of tamoxifen; of 5 years of aromatase inhibitor versus 2-3 years of tamoxifen then aromatase inhibitor to year 5; and of 2-3 years of tamoxifen then aromatase inhibitor to year 5 versus 5 years of tamoxifen. Primary outcomes were any recurrence of breast cancer, breast cancer mortality, death without recurrence, and all-cause mortality. Intention-to-treat log-rank analyses, stratified by age, nodal status, and trial, yielded aromatase inhibitor versus tamoxifen first-event rate ratios (RRs). FINDINGS: In the comparison of 5 years of aromatase inhibitor versus 5 years of tamoxifen, recurrence RRs favoured aromatase inhibitors significantly during years 0-1 (RR 0·64, 95% CI 0·52-0·78) and 2-4 (RR 0·80, 0·68-0·93), and non-significantly thereafter. 10-year breast cancer mortality was lower with aromatase inhibitors than tamoxifen (12·1% vs 14·2%; RR 0·85, 0·75-0·96; 2p=0·009). In the comparison of 5 years of aromatase inhibitor versus 2-3 years of tamoxifen then aromatase inhibitor to year 5, recurrence RRs favoured aromatase inhibitors significantly during years 0-1 (RR 0·74, 0·62-0·89) but not while both groups received aromatase inhibitors during years 2-4, or thereafter; overall in these trials, there were fewer recurrences with 5 years of aromatase inhibitors than with tamoxifen then aromatase inhibitors (RR 0·90, 0·81-0·99; 2p=0·045), though the breast cancer mortality reduction was not significant (RR 0·89, 0·78-1·03; 2p=0·11). In the comparison of 2-3 years of tamoxifen then aromatase inhibitor to year 5 versus 5 years of tamoxifen, recurrence RRs favoured aromatase inhibitors significantly during years 2-4 (RR 0·56, 0·46-0·67) but not subsequently, and 10-year breast cancer mortality was lower with switching to aromatase inhibitors than with remaining on tamoxifen (8·7% vs 10·1%; 2p=0·015). Aggregating all three types of comparison, recurrence RRs favoured aromatase inhibitors during periods when treatments differed (RR 0·70, 0·64-0·77), but not significantly thereafter (RR 0·93, 0·86-1·01; 2p=0·08). Breast cancer mortality was reduced both while treatments differed (RR 0·79, 0·67-0·92), and subsequently (RR 0·89, 0·81-0·99), and for all periods combined (RR 0·86, 0·80-0·94; 2p=0·0005). All-cause mortality was also reduced (RR 0·88, 0·82-0·94; 2p=0·0003). RRs differed little by age, body-mass index, stage, grade, progesterone receptor status, or HER2 status. There were fewer endometrial cancers with aromatase inhibitors than tamoxifen (10-year incidence 0·4% vs 1·2%; RR 0·33, 0·21-0·51) but more bone fractures (5-year risk 8·2% vs 5·5%; RR 1·42, 1·28-1·57); non-breast-cancer mortality was similar. INTERPRETATION: Aromatase inhibitors reduce recurrence rates by about 30% (proportionately) compared with tamoxifen while treatments differ, but not thereafter. 5 years of an aromatase inhibitor reduces 10-year breast cancer mortality rates by about 15% compared with 5 years of tamoxifen, hence by about 40% (proportionately) compared with no endocrine treatment.
Full text
A icles
www. helance .com Vol 386 Oc obe 3, 2015
1341
A oma ase inhibi o s e sus amoxi en in ea ly b eas
cance : pa ien -le el me a-analysis o he andomised ials
Ea ly B eas Cance T ialis s’ Collabo a i e G oup (EBCTCG)*
Summa y
Backg ound The op imal ways o using a oma ase inhibi o s o amoxi en as endoc ine ea men o ea ly b eas
cance emains unce ain.
Me hods We unde ook me a-analyses o indi idual da a on 31 920 pos menopausal women wi h oes ogen- ecep o -
posi i e ea ly b eas cance in he andomised ials o 5 yea s o a oma ase inhibi o e sus 5 yea s o amoxi en; o
5 yea s o a oma ase inhibi o e sus 2–3 yea s o amoxi en hen a oma ase inhibi o o yea 5; and o 2–3 yea s
o amoxi en hen a oma ase inhibi o o yea 5 e sus 5 yea s o amoxi en. P ima y ou comes we e any ecu ence
o b eas cance , b eas cance mo ali y, dea h wi hou ecu ence, and all-cause mo ali y. In en ion- o- ea
log- ank analyses, s a ifi ed by age, nodal s a us, and ial, yielded a oma ase inhibi o e sus amoxi en fi s -e en
a e a ios (RRs).
Findings In he compa ison o 5 yea s o a oma ase inhibi o e sus 5 yea s o amoxi en, ecu ence RRs a ou ed
a oma ase inhibi o s signifi can ly du ing yea s 0–1 (RR 0·64, 95% CI 0·52–0·78) and 2–4 (RR 0·80, 0·68–0·93), and
non-signifi can ly he ea e . 10-yea b eas cance mo ali y was lowe wi h a oma ase inhibi o s han amoxi en
(12·1% s 14·2%; RR 0·85, 0·75–0·96; 2p=0·009). In he compa ison o 5 yea s o a oma ase inhibi o e sus
2–3 yea s o amoxi en hen a oma ase inhibi o o yea 5, ecu ence RRs a ou ed a oma ase inhibi o s signifi can ly
du ing yea s 0–1 (RR 0·74, 0·62–0·89) bu no while bo h g oups ecei ed a oma ase inhibi o s du ing yea s 2–4, o
he ea e ; o e all in hese ials, he e we e ewe ecu ences wi h 5 yea s o a oma ase inhibi o s han wi h
amoxi en hen a oma ase inhibi o s (RR 0·90, 0·81–0·99; 2p=0·045), hough he b eas cance mo ali y educ ion
was no signifi
can (RR 0·89, 0·78–1·03; 2p=0·11). In he compa ison o 2–3 yea s o amoxi en hen a oma ase
inhibi o o yea 5 e sus 5 yea s o amoxi en, ecu ence RRs a ou ed a oma ase inhibi o s signifi can ly du ing
yea s 2–4 (RR 0·56, 0·46–0·67) bu no subsequen ly, and 10-yea b eas cance mo ali y was lowe wi h swi ching o
a oma ase inhibi o s han wi h emaining on amoxi en (8·7% s 10·1%; 2p=0·015). Agg ega ing all h ee ypes o
compa ison, ecu ence RRs a ou ed a oma ase inhibi o s du ing pe iods when ea men s diff e ed (RR 0·70,
0·64–0·77), bu no signifi can ly he ea e (RR 0·93, 0·86–1·01; 2p=0·08). B eas cance mo ali y was educed bo h
while ea men s diff e ed (RR 0·79, 0·67–0·92), and subsequen ly (RR 0·89, 0·81–0·99), and o all pe iods combined
(RR 0·86, 0·80–0·94; 2p=0·0005). All-cause mo ali y was also educed (RR 0·88, 0·82–0·94; 2p=0·0003). RRs
diff e ed li le by age, body-mass index, s age, g ade, p oges e one ecep o s a us, o HER2 s a us. The e we e ewe
endome ial cance s wi h a oma ase inhibi o s han amoxi en (10-yea incidence 0·4% s 1·2%; RR 0·33, 0·21–0·51)
bu mo e bone ac u es (5-yea isk 8·2% s 5·5%; RR 1·42, 1·28–1·57); non-b eas -cance mo ali y was simila .
In e p e a ion A oma ase inhibi o s educe ecu ence a es by abou 30% (p opo iona ely) compa ed wi h amoxi en
while ea men s diff e , bu no he ea e . 5 yea s o an a oma ase inhibi o educes 10-yea b eas cance mo ali y
a es by abou 15% compa ed wi h 5 yea s o amoxi en, hence by abou 40% (p opo iona ely) compa ed wi h no
endoc ine ea men .
Funding Cance Resea ch UK, Medical Resea ch Council.
Copy igh © Ea ly B eas Cance T ialis s’ Collabo a i e G oup (EBCTCG). Open Access a icle dis ibu ed unde he
e ms o CC BY.
In oduc ion
T ea men o 5 yea s wi h he selec i e oes ogen ecep o
(ER) modula o amoxi en educes ecu ence a es in
ER-posi i e ea ly b eas cance by abou hal du ing
ea men and abou one- hi d in he subsequen 5 yea s,
and educes b eas cance mo ali y by almos one- hi d
h oughou he fi s 15 yea s.1 Fu he educ ions in b eas
cance mo ali y du ing yea s 10–14 a e achie ed by
ex ending amoxi en ea men o 10 yea s.2,3 In
pos menopausal women only, a oma ase inhibi o s can
g ea ly educe oes ogen concen a ions, hence a oiding
s imula ion o ER-posi i e b eas cance cells. A oma ase
inhibi o s, gi en ei he o 5 yea s o o 2–3 yea s a e
2–3 yea s o amoxi en, p oduce g ea e educ ions in
ecu ence han 5 yea s o amoxi en alone,4 bu he eff ec
on b eas cance mo ali y, and he op imal way o
schedule a oma ase inhibi o s and amoxi en in he
ea men o ea ly b eas cance , emain unce ain.
Lance 2015; 386: 1341–52
Published Online
July 24, 2015
h p://dx.doi.o g/10.1016/
S0140-6736(15)61074-1
See Commen page 1317
*Full lis o membe s a ailable a
h p://www.c su.ox.ac.uk/
esea ch/me a- ials/ebc cg/
ebc cg-page
Co espondence o:
EBCTCG Sec e a ia , Clinical T ial
Se ice Uni , Nuffi eld
Depa men o Popula ion
Heal h, Richa d Doll Building,
Ox o d OX3 7LF, UK
[email p o ec ed]
A icles
1342
www. helance .com Vol 386 Oc obe 3, 2015
Ame ican Socie y o Clinical Oncology (ASCO) clinical
p ac ice guidelines efl ec his, ecom mending ha
pos menopausal women wi h ea ly ER-posi i e b eas
cance be off e ed ei he amoxi en o 10 yea s, an
a oma ase inhibi o o 5 yea s, amoxi en ini ially o
5 yea s ollowed by an a oma ase inhibi o o up o a
u he 5 yea s, o amoxi en o 2–3 yea s ollowed by an
a oma ase inhibi o o up o a u he 5 yea s.5 To help
cla i y he ela i e benefi s o a oma ase inhibi o s and
amoxi en and he eff ec o diff e en scheduling du ing
5 yea s o endoc ine he apy, we unde ook collabo a i e
me a-analyses o indi idual pa ien da a om he ials o
a oma ase inhibi o s e sus amoxi en.
Me hods
Iden ifi ca ion o s udies and collec ion o da a
T ial iden ifi ca ion, da a checking, analysis, and
in ol emen o ialis s a e as desc ibed in p e ious Ea ly
B eas Cance T ialis s’ Collabo a i e G oup (EBCTCG)
epo s.1,6,7 Eligible ials began by 2005 and andomised
pos menopausal women wi h ER-posi i e ea ly b eas
cance be ween 5 yea s o an a oma ase inhibi o e sus
5 yea s o amoxi en (compa ison A); 5 yea s o a oma ase
inhibi o e sus 2–3 yea s o amoxi en, hen a oma ase
inhibi o o yea 5 (compa ison B); 2–3 yea s o amoxi en,
hen a oma ase inhibi o o yea 5 e sus 5 yea s o
amoxi en (compa ison C); 5 yea s o a oma ase inhibi o
e sus 2 yea s o a oma ase inhibi o , hen amoxi en o
yea 5 (compa ison D); o 2 yea s o a oma ase inhibi o ,
hen amoxi en o yea 5 e sus 5 yea s o amoxi en
(compa ison E). Sepa a e analyses a e p o ided o each
o hese compa isons (A–E), hen esul s om some o
hem a e combined.
In o ma ion was sough du ing 2012–14 o each
indi idual pa ien on andomisa ion da e, alloca ed
ea men , age, menopausal s a us, body-mass index
(BMI), umou diame e , g ade, sp ead o loco egional
lymph nodes, ER, p oges e one ecep o (PR), and HER2
ecep o s a us, and da es o any loco egional,
con ala e al, o dis an b eas cance ecu ence, o he
second p ima y cance , bone ac u es, dea h, and cause
o dea h.
Ou comes
The p ima y ou comes we e any ecu ence o b eas
cance (dis an , loco egional, o new p ima y in he
con ala e al b eas ); b eas cance mo ali y; dea h
wi hou ecu ence; and all-cause mo ali y. Seconda y
ou comes we e incidence and si e o second p ima y
cance s, and bone ac u e. P especifi ed p ima y subg oup
in es iga ions we e o si e o ecu ence, age, nodal s a us,
PR s a us, his ological g ade, and ollow-up pe iod.
S a is ical analyses
S a is ical me hods (s a ifi ed log- ank s a is ics, Kaplan-
Meie g aphs) a e desc ibed elsewhe e.1,6,7 Time- o-e en
analyses we e s a ifi ed by age, nodal s a us, and ial.
Wi hin each s a um, hey compa ed all hose alloca ed
a oma ase inhibi o e sus all hose alloca ed amoxi en,
ega dless o ea men compliance (yielding in en ion- o-
ea analyses). Log- ank s a is ics we e used o assess he
eff ec s (a oma ase inhibi o s amoxi en) on a ious
ou comes, and, o each, o es ima e fi s -e en - a e a ios
(RRs) and hei CIs. I a log- ank s a is ic (o – e) has
a iance , hen, defi ning z=(o – e)/√ and b=(o – e)/ , b has
a iance 1/ and he e en RR (newe ea men s con ol)
is es ima ed as exp(b) wi h SE=(RR – 1)/z. CIs o RR a e
de i ed om hose o b (by no mal app oxima ions). 2p
indica es wo-sided signifi cance. The b eas cance
mo ali y a e in each yea is he o e all mo ali y a e
among all women minus ha among women o simila
age wi hou ecu ence. B eas cance mo ali y RRs a e
es ima ed om he co esponding log- ank analyses o
mo ali y wi h ecu ence (ob ained by sub ac ing
log- ank analyses o mo ali y wi hou ecu ence [ie,
censo ed a ecu ence] om hose o o e all mo ali y).
Analyses used EBCTCG Fo an p og ams. The policy on
da a sha ing om his s udy is a ailable online.
Role o he unding sou ce
The unde s o he s udy had no ole in s udy design,
da a collec ion, da a analysis, da a in e p e a ion, o
w i ing o he epo . The sec e a ia had ull access o all
da a and he w i ing commi ee had fi nal esponsibili y
o he decision o submi o publica ion.
Resul s
Indi idual pa ien da ase s we e p o ided o nine ials,8–16
including 35 129 (98%) o he 35 718 women andomised
be ween a oma ase inhibi o and amoxi en as pa o
abou 5 yea s o adju an endoc ine ea men (appendix).
This epo is es ic ed o he 31 920 (91%) wi h ER-posi i e
umou s o hese 35 129 pa ien s. All we e andomised
e enly be ween a oma ase inhibi o and amoxi en, hough
one ial (BIG 1-988) included a ou -way andomisa ion
ha con ibu es da a o all fi e compa isons (A–E); he
agg ega ed analyses a oid double coun ing i s esul s.
When epo s eme ged ha pa ien s on amoxi en had
hei ecu ence isk educed by swi ching a e 2–3 yea s
o an a oma ase inhibi o , c osso e o an a oma ase
inhibi o om he amoxi en-only g oup was sys ema ically
off e ed in wo ials (BIG 1-98,8 25% [619/2459] c osso e ;
ABCSG-8,9 18% [341/1949] c osso e ). In eigh ials,
compliance was simila in bo h g oups, bu in TEAM10 56%
(2698/4814) o hose alloca ed amoxi en hen a oma ase
inhibi o e sus 30% (1438/4852) o hose alloca ed only
a oma ase inhibi o discon inued ea men p ema u ely.
In compa ison A (5 yea s o a oma ase inhibi o s
5 yea s o amoxi en: wo ials, n=9885), ecu ence and
mo ali y we e bo h signifi can ly educed (fi gu e 1). The
numbe s wi h ecu ence we e 827 in he a oma ase
inhibi o g oup e sus 964 in he amoxi en g oup
(p<0·00001), wi h sepa a ely signifi can educ ions
du ing yea s 0–1 a e su ge y (RR 0·64, 95% CI
Fo he CTSU policy on da a
sha ing see h p://www.c su.ox.
ac.uk/ esea ch/da a-access-
policies/da a-access-and-
sha ing-policy/ iew
See Online o appendix
A icles
www. helance .com Vol 386 Oc obe 3, 2015
1343
0·52–0·78) and du ing yea s 2–4 (RR 0·80, 0·68–0·93),
bu no signifi can u he eff ec a e he scheduled
ea men pe iod, and li le ollow-up beyond yea 10.
The 10-yea ecu ence isk was 19·1% in he a oma ase
inhibi o g oup e sus 22·7% in he amoxi en g oup
(diff e ence 3·6%, 95% CI 1·7–5·4). Dis an ecu ence
(RR 0·86, 95% CI 0·77–0·96; 2p=0·007), local ecu ence
(RR 0·74, 0·58–0·95; 2p=0·020), and con ala e al
ecu ence (RR 0·62, 0·48–0·80; 2p=0·0003) we e all
educed (appendix). B eas cance mo ali y was also
educed (RR 0·85, 95% CI 0·75–0·96; 2p=0·009), as was
all-cause mo ali y (936 s 1000 dea hs; RR 0·89,
0·81–0·97; 2p=0·010), e en hough hal he dea hs we e
om non-b eas cance causes ha a e li le aff ec ed by
ea men .
In compa ison B (5 yea s o a oma ase inhibi o s
2–3 yea s o amoxi en hen a oma ase inhibi o o yea 5:
h ee ials, n=12 779), ecu ence was signifi can ly
educed only du ing yea s 0–1 (RR 0·74, 95% CI
0·62–0·89; 2p=0·002), ie, when he ea men s diff e ed,
and was simila du ing yea s 2–4 (RR 0·99, 0·85–1·15),
when bo h g oups we e ecei ing an a oma ase inhibi o
(fi gu e 2). The e was no signifi can u he eff ec a e
yea 5, bu li le ollow-up beyond yea 7. Pe haps because
he pe iod du ing which he ea men s diff e ed las ed
only hal as long as in compa ison A, he absolu e
educ ions in ecu ence and mo ali y appea ed smalle .
The o al numbe s wi h ecu ence we e 705 in he
a oma ase inhibi o g oup e sus 765 in he amoxi en
g oup (2p=0·045). Al hough b eas cance mo ali y
appea ed somewha educed (RR 0·89, 95% CI
0·78–1·03; 2p=0·11), his was no signifi can , and no
we e he eff ec s on o he mo ali y o all-cause mo ali y.
In compa ison C (2–3 yea s o amoxi en hen a oma ase
inhibi o o yea 5 s 5 yea s o amoxi en: six ials,
n=11 798), ecu ence and mo ali y we e bo h signifi can ly
educed (fi gu e 3). Fou ials did no andomise un il a e
2 yea s o amoxi en, bu wo andomised a yea 0; o
compa abili y wi h he o he ou , only pa ien s who
comple ed 2 yea s o amoxi en wi hou ecu ence o a
second p ima y a e included, bu sensi i i y analyses
(appendix) show his exclusion made li le diff e ence.
S a ing om when ea men s di e ged, he numbe s
wi h ecu ence we e 753 in he a oma ase inhibi o g oup
e sus 863 in he amoxi en g oup (2p=0·0001). Alloca ion
o an a oma ase inhibi o educed he ecu ence a e
du ing yea s 2–4 (RR 0·56, 95% CI 0·46–0·67; p<0·0001),
wi h no signifi can u he eff ec on ecu ence a e he
ea men pe iod, and li le ollow-up beyond yea 10. The
10-yea ecu ence isk was 17·0% in he a oma ase
inhibi o g oup e sus 19·0% in he amoxi en g oup
(diff e ence 2·0, 95% CI 0·2–3·8). Dis an ecu ence
(RR 0·86, 95% CI 0·77–0·97; 2p=0·02), and con ala e al
ecu ence (RR 0·67, 0·51–0·87; 2p=0·002) we e bo h
educed (appendix). B eas cance mo ali y was also
educed (RR 0·84, 95% CI 0·72–0·96; 2p=0·015), as was
all-cause mo ali y (639 s 764 dea hs; RR 0·82, 0·73–0·91;
2p=0·0002), helped by wha migh ha e been a chance
educ ion in non-b eas cance mo ali y.
The ecu ence esul s al eady desc ibed o
compa isons A–C a e summa ised in he appendix, using
black squa es o pe iods when he ea men s diff e ed
(a oma ase inhibi o in one g oup s amoxi en in he
o he ) and open squa es o pe iods when hey did no . I
also gi es he compa isons D and E, which bo h de i e
om BIG 1-98.8 Compa ison D was es ic ed o he
2558 women who we e ecu ence ee and s ill on
ea men a e 2 yea s o a oma ase inhibi o . Al hough
hey sugges no appa en gain om con inuing o ake an
a oma ase inhibi o a he han swi ching o amoxi en
a e 2 yea s, he CIs we e wide. Compa ison E included
3060 women; he p opo ional ecu ence educ ion
du ing yea s 0–1 (when he ea men s diff e ed) was
simila o ha in ea lie compa isons, and he appa en
fl uc ua ions in he ecu ence RR du ing he pe iod when
he ea men s no longe diff e ed could well be chance.
In each o compa isons A–C he e was signifi can
benefi only when ea men s diff e ed and no when hey
we e he same in bo h g oups. This pa e n is e en
clea e when esul s om all fi e compa isons a e
agg ega ed by ime pe iod (fi gu e 4). Recu ence RRs
a ou ed a oma ase inhibi o s du ing pe iods when
ea men s diff e ed (RR 0·70, 95% CI 0·64–0·77), bu no
signifi can ly he ea e (RR 0·93, 0·86–1·01; 2p=0·08).
The ecu ence a e was abou 30% lowe wi h an
a oma ase inhibi o han wi h amoxi en in yea s
0–1 (RR 0·70, 95% CI 0·61–0·80; 2p<0·0001), and in
yea s 2–4 (RR 0·71, 0·62–0·80; 2p<0·0001). Combining
ials whe e ea men s diff e ed only du ing yea s 0–1
and no du ing yea s 2–4, he e was no educ ion in
ecu ence du ing yea s 2–4 (RR 1·03, 95% CI
0·87–1·22). The e was li le u he eff ec du ing yea s
5–9 when no u he ea men was scheduled (RR 0·92,
95% CI 0·83–1·01), and li le ollow-up beyond yea 10.
B eas cance mo ali y was educed bo h while
ea men s diff e ed (RR 0·79, 95% CI 0·67–0·92), and
subsequen ly (RR 0·89, 0·81–0·99), and o all pe iods
combined (RR 0·86, 0·80–0·94; p=0·0005; appendix).
All-cause mo ali y was likewise educed (RR 0·90, 95%
CI 0·84–0·95; 2p=0·0005).
To enhance s a is ical powe , he main subg oup analyses
o ecu ence a e es ic ed o he pe iods when a oma ase
inhibi o was di ec ly compa ed wi h amoxi en (fi gu e 5).
The fi s such analyses compa e he six componen s
om p e ious fi gu es ha con ibu e o his: he ecu ence
RRs du ing yea s 0–1 we e, as expec ed, simila in
compa isons A, B, and E, bu he ecu ence RRs du ing
yea s 2–4 appea ed somewha mo e ex eme a e 2–3 yea s
o p e ious amoxi en (RR 0·56, 95% CI 0·46–0·67) han
a e 2–3 yea s o a oma ase inhibi o e sus amoxi en
(RR 0·83, 0·69–1·00), o a e 2 yea s o a oma ase
inhibi o (RR 1·08, 0·70–1·68).
Figu e 5 subdi ides he agg ega ed esul om he
pe iods when ea men s diff e ed by a oma ase
A icles
1344
www. helance .com Vol 386 Oc obe 3, 2015
inhibi o d ug, si e o fi s ecu ence, en y age, BMI,
and umou cha ac e is ics: PR s a us, nodal s a us,
umou diame e , umou g ade, and HER2 s a us
(a ailable o only one- hi d o pa ien s). The ecu ence
RRs we e simila wi h diff e en a oma ase inhibi o s
(each p<0·0001), wi h local ecu ence, con ala e al
0 5 10
0
10
20
30
40
50
A
Recu ence (%)
Recu ence a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 0–1
1·62 (157/9691)
2·41 (230/9542)
0·64 (0·52–0·78)
–41·1/92·8
Yea s 2–4
2·14 (285/13
336)
2·62 (338/12
906)
0·80 (0·68–0·93)
–34·1/149·0
Yea s 5–9
2·33 (365/15
648)
2·48 (372/14
985)
0·92 (0·79−1·06)
−15·5/177·2
Yea 10+
3·23 (20/619)
4·54 (24/529)
0·72 (0·39−1·30)
−3·6/10·7
9885 women, 1791 e en s
RR=0·80 (95% CI 0·73–0·88)
0 5 10
0
10
20
30
40
50
B
B eas cance mo ali y (%)
Dea h a es (%/yea : o al a e minus a e in women wi hou ecu ence) and log- ank s a is ics
Alloca ion
AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 0–1
0·52 (0·39−0·66)
0·51 (0·39−0·67)
0·98 (0·66−1·46)
−0·4/24·3
Yea s 2–4
1·23 (1·05−1·41)
1·60 (1·38−1·83)
0·74 (0·60−0·91)
−27·2/90·7
Yea s 5–9
1·66 (1·46−1·86)
1·81 (1·60−2·02)
0·90 (0·76−1·07)
−13·6/129·4
Yea 10+
1·93 (0·88−2·99)
1·88 (0·77−2·99)
1·01 (0·45−2·33)
0·1/5·6
9885 women, 1066 dea hs
RR=0·85 (95% CI 0·75–0·96)
0 5 10
Yea s
0
10
20
30
40
50
C
Dea h wi hou ecu ence (%)
Dea h-wi hou - ecu ence a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 0–1
0·54 (52/9691)
0·60 (57/9542)
0·89 (0·61−1·30)
−3·2/27·0
Yea s 2–4
0·99 (132/13
336)
0·95 (122/12
906)
1·03 (0·81−1·32)
2·0/62·5
Yea s 5–9
1·46 (229/15
648)
1·63 (244/14
985)
0·88 (0·73−1·05)
−14·9/114·4
Yea 10+
3·07 (19/619)
2·84 (15/529)
1·28 (0·64−2·56)
2·0/8·1
9885 women, 870 dea hs
RR=0·94 (95% CI 0·82–1·07)
0 5 10
Yea s
0
10
20
30
40
50
D
Dea h om any cause (%)
Dea h a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 0–1
1·05 (103/9775)
1·10 (106/9661)
0·93 (0·71−1·23)
−3·6/51·3
Yea s 2–4
2·18 (301/13
777)
2·50 (338/13
506)
0·85 (0·72−0·99)
−25·2/153·2
Yea s 5–9
3·06 (500/16
333)
3·35 (530/15
805)
0·89 (0·79−1·01)
−28·4/243·9
Yea 10+
4·76 (32/672)
4·44 (26/586)
1·16 (0·69−1·99)
2·1/13·6
9885 women, 1936 dea hs
RR=0·89 (95% CI 0·8–0·97)
10-yea gain 3·6% (95% CI 1·7 o 5·4)
Log- ank 2p<0·00001
Tamoxi en
22·7%
AI
19·1%
12·1%
9·0%
10-yea gain 2·1% (95% CI 0·5 o 3·7)
Log- ank 2p=0·009
Tamoxi en
14·2%
AI
12·1%
5·8%
4·5%
10-yea gain 1·1% (95% CI –0·4 o 2·6)
Log- ank 2p=0·34
Tamoxi en
11·9%
AI
10·8%
4·0%
4·0%
10-yea gain 2·7% (95% CI 0·1 o 4·7)
Log- ank 2p=0·01
Tamoxi en
24·0%
AI
21·3%
9·4%
8·2%
Figu e 1: 5 yea s o a oma ase inhibi o e sus 5 yea s o amoxi en
(A) Recu ence, (B) b eas cance mo ali y, (C) dea h wi hou ecu ence, and (D) dea h om any cause. RR= a e a io (wi h 95% CI). AI=a oma ase inhibi o . O–E=obse ed minus expec ed. V= a iance
o O–E.
A icles
www. helance .com Vol 386 Oc obe 3, 2015
1345
b eas cance , and dis an ecu ence all subs an ially
educed by a oma ase inhibi o compa ed wi h
amoxi en. In he agg ega ed da a, he RRs while
ea men s diff e ed appea ed simila in e e y subg oup,
sugges ing ha age, BMI, and umou cha ac e is ics
canno use ully p edic he RR.
Figu e 2: 5 yea s o a oma ase inhibi o e sus amoxi en o yea s 2–3 hen a oma ase inhibi o o yea 5
(A) Recu ence, (B) b eas cance mo ali y, (C) dea h wi hou ecu ence, and (D) dea h om any cause. RR= a e a io. AI=a oma ase inhibi o . O–E=obse ed minus expec ed. V= a iance o O–E.
0 1 2 3 4 5 6 7
1 2 3 4 5 6 7
0
10
20
30
40
50
A
Recu ence (%)
Recu ence a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
AI
Tamoxi en hen AI
Ra e a io (95% CI)
om (O-E)/V
Yea s 0–1
1·64 (204/12
435)
2·22 (273/12290)
0·74 (0·62−0·89)
−34·0/115·0
Yea s 2–4
2·31 (360/15
589)
2·29 (348/15
183)
0·99 (0·86−1·15)
−1·2/170·6
Yea 5+
2·43 (141/5811)
2·52 (144/5715)
0·96 (0·76−1·22)
−2·6/69·2
Yea s 0–1
0·46 (0·35−0·58)
0·55 (0·42−0·68)
0·84 (0·59−1·20)
−5·3/30·7
Yea s 2–4
1·41 (1·23−1·57)
1·49 (1·30−1·69)
0·93 (0·78−1·13)
−7·5/110·9
Yea 5+
1·77 (1·44−2·10)
2·08 (1·72−2·44)
0·84 (0·65−1·09)
−9·6/55·5
12
799 women, 1470 e en s
RR=0·90 (95% CI 0·81–0·99)
0
0
10
20
30
40
50
B
B eas cance mo ali y (%)
Dea h a es (%/yea : o al a e minus a e in women wi hou ecu ence) and log- ank s a is ics
Alloca ion
AI
Tamoxi en hen AI
Ra e a io (95% CI)
om (O-E)/V
Dea h-wi hou - ecu ence a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
AI
Tamoxi en hen AI
Ra e a io (95% CI)
om (O-E)/V
Yea s 0–1
0·52 (65/12
435)
0·50 (61/12290)
1·04 (0·73−1·48)
1·2/31·1
Yea s 2–4
0·94 (146/15
589)
0·82 (124/15
183)
1·11 (0·88−1·42)
7·1/66·2
Yea 5+
1·14 (66/5811)
1·07 (61/5715)
1·02 (0·72−1·46)
0·8/30·8
Yea s 0–1
0·98 (123/12
576)
1·04 (130/12
463)
0·94 (0·73−1·20)
−4·1/61·8
Yea s 2–4
2·31 (374/16
200)
2·27 (361/15
885)
1·00 (0·86−1·16)
−0·4/177·1
Yea 5+
2·84 (175/6152)
3·08 (187/6064)
0·90 (0·73−1·12)
−8·8/86·3
Dea h a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
AI
Tamoxi en hen AI
Ra e a io (95% CI)
om (O-E)/V
12
799 women, 827 dea hs
RR=0·89 (95% CI 0·78–1·03)
1 2 3 4 5 6 7
1 2 3 4 5 6 7
0
Yea s
0
10
20
30
40
50
C
Dea h wi hou ecu ence (%)
12
799 women, 523 dea hs
RR=1·07 (95% CI 0·90–1·28)
0
Yea s
0
10
20
30
40
50
D
Dea h om any cause (%)
12
799 women, 1350 dea hs
RR=0·96 (95% CI 0·86–1·07)
Tamoxi en hen AI
14·5%
AI
13·8%
10·7%
9·6%
7-yea gain 0·7% (95% CI –0·9 o 2·2)
Log- ank 2p=0·045
AI
6·1%
Tamoxi en hen AI
5·9%
3·8%
3·5%
7-yea gain 0·2% (95% CI –1·0 o 1·3)
Log- ank 2p=0·42
Tamoxi en hen AI
9·3%
AI
8·2%
5·5%
5·1%
7-yea gain 1·1% (95% CI –0·2 o 2·5)
Log- ank 2p=0·11
Tamoxi en hen AI
14·5%
AI
13·6%
8·7%
8·6%
7-yea gain 0·9% (95% CI –0·7 o 2·5)
Log- ank 2p=0·46
A icles
1346
www. helance .com Vol 386 Oc obe 3, 2015
Tumou cha ac e is ics we e, howe e , impo an ly
p edic i e o he absolu e isk o ecu ence, and hence
o he absolu e eff ec on b eas cance ou comes o
gi ing an a oma ase inhibi o a he han amoxi en
(appendix). Fo example, in he agg ega e o he ials
ha con ibu e o he black squa es in fi gu e 4, he
0 2 3 4 5 6 7 8 9 10 2 3 4 5 6 7 8 9 10
2 3 4 5 6 7 8 9 10 2 3 4 5 6 7 8 9 10
0
10
20
30
40
50
A
Recu ence (%)
Recu ence a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
Tamoxi en hen AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 2–4
1·48 (170/11
515)
2·64 (300/11
360)
0·56 (0·46–0·67)
−65·3/111·5
Yea s 5–9
2·48 (495/19
920)
2·51 (479/19
101)
0·97 (0·86−1·11)
−5·9/234·0
Yea 10+
3·26 (88/2696)
3·35 (84/2505)
0·92 (0·68−1·25)
−3·3/40·8
11
798 women, 1616 e en s
RR=0·82 (95% CI 0·75–0·91)
0
0
10
20
30
40
50
B
B eas cance mo ali y (%)
Dea h a es (%/yea : o al a e minus a e in women wi hou ecu ence) and log- ank s a is ics
Alloca ion
Tamoxi en hen AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 2–4
0·37 (0·25–0·48)
0·56 (0·43–0·70)
0·65 (0·44–0·96)
−11·0/25·8
Yea s 5–9
1·28 (1·12–1·44)
1·40 (1·26–1·56)
0·91 (0·77–1·08)
−11·9/132·0
Yea 10+
1·68 (1·63–1·72)
2·54 (2·45–2·59)
0·69 (0·48–1·00)
−10·6/28·9
11
798 women, 789 dea hs
RR=0·84 (95% CI 0·72–0·96)
0
Time since alloca ed ea men s diffe (yea s)
0
10
20
30
40
50
C
Dea h wi hou ecu ence (%)
Dea h-wi hou - ecu ence a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
Tamoxi en hen AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 2–4
0·51 (59/11
515)
0·55 (63/11
360)
0·91 (0·64–1·30)
−2·8/30·0
Yea s 5–9
0·85 (169/19
920)
1·15 (220/19
101)
0·73 (0·60–0·89)
−30·3/95·3
Yea 10+
1·85 (50/2696)
2·11 (53/2505)
0·95 (0·63–1·42)
−1·3/23·8
11
798 women, 614 dea hs
RR=0·79 (95% CI 0·67–0·93)
0
Time since alloca ed ea men s diffe (yea s)
0
10
20
30
40
50
D
Dea h om any cause (%)
Dea h a e/yea (%), e en s/woman-yea s and log- ank s a is ics
Alloca ion
Tamoxi en hen AI
Tamoxi en
Ra e a io (95% CI)
om (O-E)/V
Yea s 2–4
0·88 (102/11
644)
1·10 (128/11
618)
0·78 (0·60–1·01)
−13·9/55·9
Yea s 5–9
2·09 (437/20
949)
2·47 (509/20
593)
0·83 (0·73–0·95)
−42·3/227·3
Yea 10+
1·86 (50/2696)
2·11 (53/2505)
0·95 (0·63–1·42)
−1·3/23·8
11
798 women, 1403 dea hs
RR=0·82 (95% CI 0·73–0·91)
Tamoxi en
19·0%
Tamoxi en hen AI
17·0%
12·1%
9·5%
10-yea gain 2·0% (95% CI 0·2 o 3·8)
Log- ank 2p=0·0001
Tamoxi en
8·8%
Tamoxi en hen AI
7·0%
4·2%
3·2%
10-yea gain 1·7% (95% CI 0·3 o 3·2)
Log- ank 2p=0·005
Tamoxi en
17·5%
Tamoxi en hen AI
14·6%
8·8%
7·1%
10-yea gain 2·9% (95% CI 1·1 o 4·7)
Log- ank 2p=0·0002
Tamoxi en
10·01%
Tamoxi en hen AI
8·7%
5·0%
4·2%
10-yea gain 1·5% (95% CI 0·1 o 2·9)
Log- ank 2p=0·01
Figu e 3: Tamoxi en o yea s 2–3 hen a oma ase inhibi o o yea 5 e sus 5 yea s o amoxi en: e en s in women ali e and ee o ecu ence when ea men s di e ged
(A) Recu ence, (B) b eas cance mo ali y, (C) dea h wi hou ecu ence, and (D) dea h om any cause. RR= a e a io. AI=a oma ase inhibi o . O–E=obse ed minus expec ed. V= a iance o O–E.
A icles
www. helance .com Vol 386 Oc obe 3, 2015
1347
o e all Kaplan-Meie es ima e o he 5-yea ecu ence
isk was educed by 2·5% (7·3% s 9·8%, appendix).
Bu , in his same da a se , he 5-yea ecu ence isks o
women wi h N0, N1–3, and N4+ disease we e educed by
1·2%, 3·7%, and 6·4%, espec i ely.
Simila se s o subg oup analyses o each sepa a e
ca ego y o compa isons A–E a e in he appendix, bu
wi h so many subg oup analyses he appa en fi ndings
should be in e p e ed cau iously, as s iking alse-posi i e
and alse-nega i e esul s can easily a ise jus by chance.
Fo example, he hypo hesis om ATAC15 o a mo e
ex eme ecu ence RR in ER-posi i e PR-nega i e han
in ER-posi i e PR-posi i e disease is no suppo ed by
e idence om o he ials (fi gu e 5, appendix). Likewise,
he hypo hesis om compa ison C o equi alen effi cacy
o a oma ase inhibi o s and amoxi en in node-nega i e
disease is no suppo ed by e idence om he o he
compa isons. Such pa e ns migh be due mainly
o chance.
Resul s o cause-specifi c mo ali y, second cance
incidence, and bone ac u e be o e any b eas cance
ecu ence a e in he appendix. The e was a signifi can
educ ion in mo ali y wi hou ecu ence in
compa ison C ( amoxi en hen a oma ase inhibi o s
amoxi en alone) ha was no explained by any
pa icula cause and is unlikely o be due o
misclassifi ed b eas cance dea hs (pa ly because he
non-b eas -cance mo ali y a es we e sha ply age-
ela ed whe eas b eas cance mo ali y a es we e
simila in all age g oups).
The e we e ewe u e ine cance s and mo e bone
ac u es wi h a oma ase inhibi o s han wi h amoxi en.
Agg ega ing he fi e compa isons, he 10-yea incidence
o endome ial cance (defi ned as any u e ine cance
excep ce ix cance ) was 0·4% in he a oma ase inhibi o
g oup e sus 1·2% in he amoxi en g oup (absolu e
diff e ence 0·8%, 95% CI 0·6–1·0; p<0·0001), including
fi e e sus nine dea hs. The p opo ional dec ease in
endome ial cance incidence wi h a oma ase inhibi o s
(RR 0·33, 0·21–0·51) was app oxima ely independen o
age and pe sis ed o some yea s a e ea men ended.
As endome ial cance inc eases wi h age, he absolu e
excess wi h amoxi en was 0·7% (95% CI 0·5–0·9) a
ages 55–69 and 1·4% (95% CI 0·5–2·4) a olde ages
(appendix). The e was no signifi can eff ec on any o he
ype o cance (excep o con ala e al b eas cance ).
The incidence o bone ac u es was inc eased among
a oma ase-inhibi o -alloca ed pa ien s du ing yea s 0–4
(RR 1·42, 95% CI 1·28–1·57; p<0·0001), and emained
signifi can ly highe h ough yea s 5–9 (RR 1·29,
1·09–1·53; 2p=0·003) despi e ac u es being moni o ed
less eliably a e he 5-yea ea men pe iod. The 5-yea
ac u e isk was 8·2% in he a oma ase inhibi o g oup
e sus 5·5% in he amoxi en g oup (absolu e excess
2·7%, 95% CI 1·7–3·7). Again, he p opo ional inc ease
appea ed app oxima ely independen o age and he
absolu e incidence inc eased wi h age. Hence, among
women o age younge han 55, 55–69, and olde han
70 yea s a andomisa ion, he absolu e excess isks
(a oma ase inhibi o s amoxi en) o ha ing a ac u e
Figu e 4: Recu ence educ ions by ime since su ge y, combining da a om diff e en compa isons o a oma ase inhibi o (AI) e sus amoxi en ea men as
pa o 5 yea s o endoc ine he apy
Black squa es show pe iods when he p o ocol specifi ed ha one g oup should ecei e an a oma ase inhibi o and he o he should ecei e amoxi en; open squa es
show pe iods when he ea men s should ha e been he same in bo h g oups. *Agg ega ed o als a e adjus ed o a oid double coun ing o e en s in he ou -way
andomisa ion in BIG 1-98. AI=a oma ase inhibi o . O–E=obse ed minus expec ed.
E en s/women–yea s (%/yea )
Alloca ed
AI
Alloca ed
amoxi en
(a) AI s amoxi en
Yea s 0–1 AI s amoxi en
Yea s 2–4 AI s amoxi en
Sub o al
361/22
068 (1·6)
425/23
324 (1·8)
786/45
398 (1·7%/yea )
502/21
786 (2·3)
581/22
803 (2·5)
1083/44
589 (2·4%/yea )
0·70 (0·58–0·84)
0·71 (0·60–0·83)
0·70 (0·64–0·77)
2p<0·00001
–74·3
–83·7
–158·1
207·6
240·2
447·7
(b) Same o no ea men
Yea s 2–4 same ea men
Yea s 5–9 no ea men
Yea s 10+ no ea men
Sub o al
To al (a+b)
99% o 95% CIs
Diffe ence be ween ea men effec s in wo sub o als χ2
1=21·1; 2p<0·00001
He e ogenei y wi hin sub o als χ2
3=2·0; 2p=0·6
He e ogenei y be ween fi e compa isons χ2
4=23·1; 2p=0·0001
287/11
340 (2·5)
906/38
062 (2·4)
109/3384 (3·2)
1302/52
786 (2·5%/yea )
2088/98
184 (2·1%/yea )
266/10
954 (2·4)
940/36
499 (2·6)
114/3111 (3·7)
1320/50
564 (2·6%/yea )
2403/95
153 (2·5%/yea )
1·03 (0·83–1·29)
0·92 (0·81–1·03)
0·85 (0·60–1·21)
0·93 (0·86–1·01)
2p=0·08
0·829 (0·781–0·880)
2p<0·00001
4·2
–39·0
–8·6
–43·5
–201·5
133·1
443·7
53·1
629·9
1077·7
AI e en s Ra io o annual e en a es
AI: amoxi en
Ra e a io (CI)
Log- ank
O–E
Va iance
o O–E
AI be e Tamoxi en be e
T ea men effec 2p<0·00001
0 0·5 1·0 1·5 2·0
A icles
1348
www. helance .com Vol 386 Oc obe 3, 2015
AI be e Tamoxi en be e
T ea men effec 2p<0·00001
0 0·5 1·0 1·5 2·0
E en s/women-yea s (%/yea )
Alloca ed
AI
Alloca ed
amoxi en
(a) T ea men compa ison (χ2
5=13·3; p=0·02)
Yea s 0–1 compa ison
Yea s 0–1 compa ison
Yea s 0–1 compa ison
Yea s 2–4 compa ison
Yea s 2–4 compa ison
Yea s 2–4 compa ison
(b) AI agen (χ2
2=0·5; 2p=0·8)
Anas ozole
Exemes ane
Le ozole
(c) Si e o ecu ence (χ2
2=1·5; 2p=0·5)
Dis an
Isola ed local
Con ala e al
(d) Age ( end χ2
1=2·2; 2p=0·14)
<45 yea s
45–54 yea s
55–69 yea s
70+ yea s
(e) BMI ( end χ2
1=0·3; 2p=0·6)
<20
20–24
25–29
30+
Unknown
( ) PR s a us (χ2
1=5·7; 2p=0·02)
ER posi i e PR nega i e
ER posi i e PR unknown
ER posi i e PR posi i e
(g) Nodal s a us ( end χ2
1=0·0; 2p=1·0)
N0/N–
N1–3
N4+
N o he /unknown
(h) T s age ( end χ2
1=2·8; 2p=0·09)
1–20 mm (T1)
21–50 mm (T2)
>50 mm (T3/T4)
O he /unknown
(i) Tumou g ade ( end χ2
1=0·0; 2p=1·0)
Well diffe en ia ed
Mode a ely
Poo ly
G ade unknown
(j) HER2 s a us (χ2
2=0·0; 2p=1·0)
HER2 posi i e
HER2 nega i e
Unknown
To al
A 157/9676 (1·6)
B 204/12
418 (1·6)
E 35/2995 (1·2)
A 285/13
313 (2·1)
C 170/11
498 (1·5)
D 43/2505 (1·7)
291/16
842 (1·7)
233/13
164 (1·8)
262/15
392 (1·7)
617/45
398 (1·4)
101/45
398 (0·2)
68/45
398 (0·1)
4/215 (1·9)
108/6636 (1·6)
472/27
362 (1·7)
202/11
185 (1·8)
21/1506 (1·4)
194/11
001 (1·8)
202/11
397 (1·8)
166/6814 (2·4)
203/14
195 (1·4)
171/6928 (2·5)
49/2259 (2·2)
566/36
243 (1·6)
243/26
174 (0·9)
280/13
746 (2·0)
236/4067 (5·8)
27/1411 (1·9)
324/28
854 (1·1)
401/14
714 (2·7)
42/1116 (3·8)
19/708 (2·7)
60/8857 (0·7)
320/21
782 (1·5)
238/7557 (3·1)
168/7193 (2·3)
79/3264 (2·4)
112/8854 (1·3)
595/33
280 (1·8)
786/45
398 (1·7%/yea )
230/9526 (2·4)
273/12
273 (2·2)
45/3008 (1·5)
338/12
886 (2·6)
300/11
333 (2·6)
41/2491 (1·6)
387/16
415 (2·4)
341/12
990 (2·6)
355/15
184 (2·3)
819/44
590 (1·8)
158/44
590 (0·4)
106/44
590 (0·2)
9/183 (4·9)
144/6430 (2·2)
616/27
152 (2·3)
314/10
824 (2·9)
26/1328 (2·0)
272/10
695 (2·5)
264/11
016 (2·4)
203/7477 (2·7)
318/14
073 (2·3)
284/6687 (4·2)
57/2385 (2·4)
742/35
556 (2·1)
338/26
097 (1·3)
390/13
285 (2·9)
308/3850 (8·0)
47/1357 (3·5)
409/28
134 (1·5)
576/14
738 (3·9)
74/1008 (7·3)
24/710 (3·4)
84/8839 (1·0)
452/21
382 (2·1)
318/7287 (4·4)
229/7081 (3·2)
108/3074 (3·5)
166/8733 (1·9)
809/32
782 (2·5)
1083/44
589 (2·4%/yea )
0·64 (0·49–0·84)
0·74 (0·59–0·95)
0·78 (0·43–1·40)
0·80 (0·64–0·98)
0·56 (0·44–0·71)
1·08 (0·61–1·92)
0·71 (0·58–0·87)
0·67 (0·54–0·84)
0·73 (0·59–0·91)
0·73 (0·63–0·84)
0·64 (0·46–0·88)
0·64 (0·43–0·94)
0·73 (0·52–1·01)
0·75 (0·64–0·88)
0·60 (0·48–0·76)
0·79 (0·36–1·74)
0·71 (0·55–0·90)
0·72 (0·56–0·92)
0·79 (0·60–1·04)
0·63 (0·51–0·79)
0·57 (0·45–0·73)
0·91 (0·55–1·51)
0·74 (0·64–0·86)
0·72 (0·58–0·89)
0·68 (0·56–0·84)
0·73 (0·58–0·92)
0·59 (0·32–1·09)
0·76 (0·63–0·92)
0·70 (0·59–0·82)
0·51 (0·30–0·85)
0·71 (0·46–1·09)
0·68 (0·57–0·83)
0·70 (0·56–0·87)
0·71 (0·54–0·92)
0·66 (0·45–0·99)
0·67 (0·49–0·92)
0·71 (0·62–0·82)
0·702 (0·640–0·771)
2p<0·00001
−41·1
−34·0
−5·2
34·1
−65·3
1·6
−55·8
−54·4
−46·3
−109·4
−29·1
−19·7
−0·8
−19·2
−75·9
−62·2
−2·5
−37·9
−36·4
−20·8
−61·0
−61·3
−2·4
−94·8
−48·0
−61·6
−39·1
−9·4
−49·6
−84·4
−17·1
−2·6
−12·2
−71·3
−47·2
−33·1
−17·5
−26·1
−114·4
−158·1
92·8
115·0
19·5
149·0
111·5
20·1
163·2
137·9
149·6
344·5
64·3
43·4
1·4
60·4
262·2
123·7
10·7
109·1
111·1
86·8
133·7
109·2
25·8
320·1
143·4
162·4
124·4
17·5
179·2
233·8
25·3
9·7
35·1
188·0
131·3
95·5
43·0
65·4
339·3
447·7
AI e en s Ra io o annual e en a es
AI: amoxi en
Ra e a io (CI)
Log- ank
O–E
Va iance
o O–E
99% o 95% CIs
Figu e 5: Subg oup analyses
o ecu ence isk educ ions
combining da a om
fi e compa isons o
a oma ase inhibi o s e sus
amoxi en including only
da a du ing pe iods when
ea men s diff e ed
G ey squa es show unknown
s a us wi hin he subg oup.
Resul s a e plo ed as black
squa es wi h ho izon al lines
ha deno e 99% a he han
95% CIs o allow o mul iple
hypo hesis es ing. To al is
plo ed as a whi e diamond
ha deno es 95% CI.
AI=a oma ase inhibi o .
O–E=obse ed minus
expec ed.
Compa ison A=5 yea s o
a oma ase inhibi o e sus
5 yea s o amoxi en.
Compa ison B=5 yea s o
a oma ase inhibi o s e sus
2–3 yea s o amoxi en, hen
a oma ase inhibi o o yea 5.
Compa ison C=2–3 yea s o
amoxi en, hen a oma ase
inhibi o o yea 5 e sus
5 yea s o amoxi en.
Compa ison D=5 yea s o
a oma ase inhibi o e sus
2 yea s o a oma ase inhibi o ,
hen amoxi en o yea 5.
Compa ison E=2 yea s o
a oma ase inhibi o , hen
amoxi en o yea 5 e sus
5 yea s o amoxi en.
ER=oes ogen ecep o .
PR=p oges e one ecep o .
A icles
www. helance .com Vol 386 Oc obe 3, 2015
1349
wi hin 5 yea s we e, espec i ely, abou 1%, 2%, and 4%
(appendix). Diff e ences in ascula mo ali y, a oma ase
inhibi o e sus amoxi en, we e no signifi can :
h omboembolic, 14 e sus 19 dea hs; ce eb o ascula ,
44 e sus 52 dea hs; and ca diac, 137 e sus 128 dea hs.
Discussion
Indi idual ials ha e al eady shown educed ecu ence
a es wi h a oma ase inhibi o compa ed wi h amoxi en
bu none has shown in in en ion- o- ea analyses ha
b eas cance mo ali y is educed, no did p e ious
me a-analyses.4 Now, wi h longe ollow-up, he p esen
me a-analyses es ablish ha b eas cance mo ali y and
all-cause mo ali y a e also educed, be e cha ac e ise
ime-dependen eff ec s on ecu ence, and allow
in o ma i e in es iga ion o diff e en ial effi cacy wi hin
subg oups and o uncommon ad e se e en s.
The e was a ai ly consis en pa e n o subs an ial
ecu ence educ ions du ing pe iods when one g oup
was ecei ing an a oma ase inhibi o and he o he
amoxi en, bu li le u he educ ion du ing subsequen
pe iods when bo h g oups we e ecei ing he same
endoc ine ea men o a e scheduled endoc ine
ea men had ended in bo h g oups. Howe e , his
fi nding should no be in e p e ed as a oma ase inhibi o s
no ha ing he ca y-o e benefi s o amoxi en,1 a he
ha 5 yea s o endoc ine he apy ha includes an
a oma ase inhibi o educes ecu ence by abou
one- hi d du ing yea s 5–9, as does 5 yea s o amoxi en.
The mos ex eme ecu ence educ ion appea ed o be
in compa ison C in which, a e 2 yea s o amoxi en, an
a oma ase inhibi o was compa ed wi h amoxi en du ing
yea s 2–4. This esul is no explained by diff e ences
in effi cacy be ween diff e en a oma ase inhibi o s, as
indi ec compa isons in fi gu e 5, and di ec andomised
compa isons,16 show li le diff e ence be ween d ugs.
I has been hypo hesised ha he supe io i y o a oma ase
inhibi o s o e amoxi en is g ea e a e p e ious
exposu e o amoxi en,17 and he la ge ecu ence
educ ions epo ed in yea s 5–9 in ials o a oma ase
inhibi o e sus no u he ea men 18–20 a e 5 yea s o
amoxi en han in ials o 10 e sus 5 yea s o amoxi en2,3
p o ide some suppo o his. Howe e , he di ec ly
andomised fi ndings in compa ison B do no show any
eff ec o he ype o endoc ine he apy du ing yea s 0–1 on
he effi cacy o ea men du ing yea s 2–4, so he appa en
he e ogenei y o benefi om indi ec compa isons could
be la gely chance.
In compa ison E, a e an ini ial 2–3 yea s o an
a oma ase inhibi o he e appea ed o be no benefi
om con inuing an a oma ase inhibi o o 5 yea s
a he han swi ching o amoxi en, bu his esul was
based on one ial wi h ew e en s. Hence, i emains
unce ain whe he , a e 2–3 yea s o an a oma ase
inhibi o , any loss o benefi occu s om swi ching o
amoxi en— eassu ingly o women who do no
ole a e a oma ase inhibi o s. Resul s o ongoing ials
compa ing diff e en du a ions o a oma ase inhibi o
ea men will de e mine whe he , as wi h amoxi en,
longe is be e .2,3,21
The educ ion in b eas cance mo ali y wi h
a oma ase inhibi o compa ed wi h amoxi en is only
sligh , as expec ed in an al eady ela i ely good-p ognosis
popula ion, bu pe sis s du ing yea s 0–4 and 5–9,
signifi can ly educing 10-yea b eas cance mo ali y.
O e all 10-yea mo ali y was also signifi can ly educed,
e en hough abou hal he dea hs we e no due o b eas
cance . Non-b eas cance dea h a es we e simila wi h
a oma ase inhibi o and amoxi en excep ha , a e
2–3 yea s o amoxi en, he e appea ed o be ewe such
dea hs wi h an a oma ase inhibi o han wi h con inuing
amoxi en. This fi nding was unexpec ed, no explained
by any one cause, and no eplica ed in he o he
compa isons. Though likely o be a chance fi nding, i is
eassu ing o he sa e y o a oma ase inhibi o s.
Bone ac u es a e a conce n wi h a oma ase inhibi o s,
hough he absolu e excess o abou 0·5% pe yea migh
be pa ly explained by a bone-p o ec i e eff ec o
5 yea s o amoxi en s none: EBCTCG
p e ious me a-analysis1 (n=10 645)
5 yea s o a oma ase inhibi o s 5 yea s
o amoxi en: p esen me a-analyses*
(n=34 882)
5 yea s o a oma ase inhibi o s none:
es ima ed eff ec s (p oduc o wo RRs†)
RR (95% CI) p alue RR (95% CI) p alue RR (95% CI) p alue
B eas cance ecu ence
Du ing yea s 0–4 0·53 (0·48–0·57) 2p<0·0001 0·70 (0·64–0·77) 2p<0·0001 0·37 (0·33–0·42) 2p<0·0001
Du ing yea s 5–9 0·68 (0·60–0·78) 2p<0·0001 0·92 (0·83–1·01) 2p=0·082 0·63 (0·53–0·74) 2p<0·0001
B eas cance mo ali y
Du ing yea s 0–4 0·71 (0·62–0·80) 2p<0·0001 0·79 (0·67–0·92) 2p=0·002 0·56 (0·46–0·68) 2p<0·0001
Du ing yea s 5–9 0·66 (0·58–0·75) 2p=0·0001 0·91 (0·80–1·02) 2p=0·12 0·60 (0·50–0·72) 2p<0·0001
EBCTCG=Ea ly B eas Cance T ialis s’ Collabo a i e G oup. RR= a e a io. *Es ima ed om he agg ega ed da a (appendix). †Es ima ed a e a io o 5 yea s o a oma ase
inhibi o s none (RR3) is ob ained by di ec mul iplica ion o he a e a io o 5 yea s o amoxi en s none (RR1) by he a e a io o 5 yea s o a oma ase inhibi o s 5 yea s o
amoxi en (RR2) es ima ed om he agg ega ed da a; 95% confi dence limi s o RR3 a e exp[(o – e)1/ 1 + (o – e)2/ 2) – 1·96 √(1/ 1 + 1/ 2)] and exp[(o – e)1/ 1 + (o – e)2/ 2) + 1·96 √(1/
1 + 1/ 2)], espec i ely, whe e (o – e) and a e he obse ed minus expec ed s a is ics and hei a iances o he compa isons o 5 yea s o amoxi en s none and 5 yea s o
a oma ase inhibi o s 5 yea s o amoxi en (es ima ed om agg ega ed da a om ials con ibu ing o sub o al (a) in fi gu e 4).
Table: Es ima ion o he eff ec o 5 yea s o an a oma ase inhibi o e sus no endoc ine ea men