tCopright, 1997, by thc sahutt, b1cdcalSoccty MAY 1, 1997
G
PHYSICAL ACTIVITY AND THE RJSK OF BREAST CANCER NtJMBER 18
INGER THUNE, MD.,ToRMoD BRENN, M.Sc., EILIv LuNo,MD., PH.D., AND MARIA GAARD, MD.
ABsTRAcT
Background Because physical activity may affect hormonal concentrations and energy balance, we decided to investigate whether everyday exercise is related to the risk of breast cancer.
Metbods During 1974 to 1978 and 1977 to 1983, a total of 25,624 women, 20 to 54 years of age at entry, enrolled in health surveys and answered question naires about leisure-time and work activity.
Results During a median follow-up of 13.7 years, we identified 351 cases of invasive breast cancer among the 25,624 women in the cohort. Greater lei sure-time activity was associated with a reduced risk of breast cancer, after adjustments for age, body mass index (the weight in kilograms divided by the square of the height in meters), height, parity, and county of residence )re)ative risk, 0.63; 95 percent confidence interval, 0.42 to 0.95), among women who exercised regularly, as compared with sedentary women IP for trend=0.04). In regularly exercising women, the reduction in risk was greater in pre menopausal women than in postmenopausal wom en, and greater in younger women (<45 years at study entry) than in older women (45 years) Irela tive risk, 0.38; 95 percent confidence interval, 0.19 to 0.79). In stratified analyses the risk of breast cancer was Iowest in lean women (body-mass index, <22.8) who exercised at least four hours per week (relative risk, 0.28; 95 percent confidence interval, 0.11 to 0.70). The risk was also reduced with higher levels of activity at work, and again there was a more pro nounced effect among premenopausal than post menopausal women.
Conclusions Physical activity during leisure time and at work is associated with a reduced risk of breast cancer. (N Eng) J Med 1997;336:1269-75.)
©1997,Massachusetts Medical Society.
V
IGOROUS physical training’t and even moderate exercise9 can interrupt the menstrual cvcle, perhaps by suppressing the pulsatile release of gonadotropin releasing hormone.’°’’ This effect ofphysical activity maylOVCra voman’s cumulative exposure to estrogen and progesterone, thereby inhibiting carcinogen esis in the breast)22’ Energy balance might also in fluence the risk ofbreast cancer. Caloric restriction in rodents reduces the proliferative activity ofthe mam mary glands23 and inhibits carcinogenesis.24.25 How evcr, thc effect of energy balance, as indicated by energy intake, body-mass mdcx (the weight in kilo grams divided by the square ofthe height in meters), and energy expenditure, on the risk of breast cancer has not been examined thoroughly in humans.
In this studywe evaluated the influence ofphysi calactivity, both at work and during leisure time, on the risk of breast cancer in a cohort of 2 5,624 pre menopausal and postmenopausal women. Data on paritv, dietary factors, and body-mass indexallowed adjustment for potentially confounding factors, and reassessment of physical activity after thrce to five years gave an indication of the effect of sustained physical activityon the risk of breast cancer.
Study Population
METHODS
From 1974 to 1978, the National Health Scrcening Service in vited people in three counties in Norway (Oppland, Sogn og Fjordane, and Finnrnark) to participate in isurvcy of risk factors for cardiovascular disease. All svomen who ssere 35 to 49 vears of age and a random sample of 10 percent of those svho ss’cre 20 to 34 years of age were invited. In four municipaliries in Finnmark all women ss’ho seere 20 to 34 years ofage sverc invited. A com prehensive decription of these populations has been published previously.
m A total of 31,556 women seere invited to participate, and 28,621 (91 pcrcent) actually did.
All svomen in this survey as stell as a random sample ofwomcn who were 20 to 39 ycars of age wcre invited to participate in a second survey three to live ycars later (1977 to 1983). Of these 34,378 women, 31,209 (91 pcrcent) participated.2’ Tids second survey svas used as the hase line, because no inforniation on parity and dietary factors svas collectcd during tIte first survey.
Each wornan received a svritten invitation to participate, to
From thc Insticute ol Community Mcdicrnc, University ni Tromsø, Tromsø (l.T, T.B., EL), msd the Caneer Regotryol Norsvay,Oslo (MO
—hcnhinNorway. Address rcprintrcqucsts toDr.Tisuncarthe tnstirurr ofC,,mmunity Medicine,University olTromsø, N9037 Trc,ms,,, Norway.
Volume 336 Numher 18 ‘ 1269
getber with a one-page questionnaire. The participants seere asked to anssvcr the qucstionnaire aud bringitto the clinical examina tion. At scrcening, trained nurses checked the questionnaire for inconsistencies regarding phvsical activity and mcnopausal status, rncasurcd ss’eight and height, and collected blood samples.
During scrcening in the second survcy, the participants were askcd to III out a food-li-equcncy qtlestionflaire, to be returned be mail. Aller one reminder, 25,892 (83 percent) returned the qucstionnaire. The enes-fly aud fat intakes for cach woman were dcrivcd from thc sunt of all food consumed. Thc semiquantitatis’c food-frequency questionnaire that sve usedhas been described in detail and ealidatcd.’
Assessment of Physical Activity
ScIf-reported categories ofphysical activitv during leisure hours in the ecar preceding each survey seere assessed when the svonsen entercd the srudv and graded from i to 4 according to the par ticipant’s usual levd ofphysical aceivity. A grade of i ss’as assigned to thosc whose leisure time ss’as spent reading, watching teles’i sion, ur engaging in other sedentare actis’ities; a grade of 2 to those who spent at least four hours a weck walking, bicycling, or engaging in other tvpes of physical activity; a grade of 3 to those svho spent at least four hours a sveek exercising to keep lit and participating in recreational athletics; and a grade of 4 to those ss’ho engaged in rcgular, sigorous rraining or participating in com petitis-e sports ses-eral times a sveek. The seif-reported lesel of physicai actis’ity during work hours in the preceding year was also graded on a four-point scale. A grade of 1 svas assigned to those nshose ssork ss’as mostly sedentary; a grade of 2 to those whose job involvcd a lot of ss’alking; a grade of 3 to those svhose job required a lot of lifting and ss’alking; and a grade of4to those engagedin heavy manual labor.
Tsvo identical assessments of leisure-time activity seere made at an interval of three to live years, and the resultawerecombined for all groups. Vomen who reported moderate (grade 2) ur reg ular (grade 3 ur4)exercise during leisure time in thefirst surs’ey aud regular exercise (grade 3 ur 4) in the second survey seere characterized an being consistently physically aceive. Women who ssere sedentary (grade 1) during Icisure time in both surveys seere characterized as being consistently sedentary. The women ss-ho ss’ere ucither consistently sedentary for consistently active during leisure time seere characterized an being modcratcly active.
FoIIow-up and dentification of Cases of Breast Cancer Wc followed a total of 25,707 women ss-ho bad not been given diagnosis ofcancer before our bane-linesurvey(1977 to 1983).
We used tI-se participants’ national i 1-digit personal identification numbcrs to identify esery incident case of brcast canccr reported to thc Cancer Registry of Norsvay aud Statistics Norway through the end of follow-up (December 31, 1994). A total of98 percent ofthe cases were verified histologically. Women in whom cancer des-eloped (n=72) or who died (n=11) ss’ithin the first vear of thc stud’ seere excludcd from the analyses to account for thepos sibilite that undiagnosed canccr ur severe iliness might influence the les-el of physical activity. Through a linkage to the Central Population Register at Statistics Norss’ay, sse obtained informa tion concerning the rcproductive history ofeach woman, includ
ingthe date of birth of each liveborn child through December 31, 1992, and dcaths in the cohort through December 31, 1994.
The ultimate stud1’ cohort consisted of 25,624 ss’omen ss-ho participated in hoth surs’eys (age range, 20 to 69 ecars( during 359,930 person-vearsoffollosv.up.
Statistical Analysis
Bane-line variahies seere adjusted for age aud compared be anal esis of covariancc. Cox proportional-hazards regression analsis svas carried out to investigate the simultaneous efFect of physical activity and cos’ariates on the incidcnce of breast cancer. To cal culate the risk of breast cancer, wornen seere obsereed for the
de-velopnsent of breast cancer from entry into the stud’ to the date ofdiagnosis ofany cancer, the time ofdeath, or the end ofibliose up, whichever es’ent came first. In the analesis, grades 3 and 4 of leisure-time actis’ity were merged hecause of the small numbers of seomen with a grade of 4 in both surs’eys (48 svonienin the first surs’e aud 57 in the second surs-ee). An a reference group wc used ss’omen seho ss’ere sedentary at ss’ork or during Icisure time.
In the analvses, sne adjusted for age at entry (a continuous sas-iable(, county of residence, number of children, age at birth of first child, intake of total fat aud energv, aud hody-mass mdcx.
Women ss-ho reported that they seere premenopausal at hase line seere treated as premcnopausal until thev reached the age of 50 during folloss’-up, at sshich time they seere cousidered posttneuo pausal. Women ss’ho reported that they ss’erc postrnenopausal at base line seere treated as postmenopausal.
Because there ss’ere fess’ women with breast caucer ss-ho ss’ere sedentary both at work aud during leisure time, thc effcct ofthis combination on the risk of breast cancer could not be analyzed.
All significance tests were two-tailcd, and the les’el ofsiguiflcancc ss’as set at 5 pereent. The analyses seere perfomied svith the SAS statistical package s’ersion 6.11.
RESULTS
There were 351 incident cases of breast canccr (100 among premenopausal wonsen aud 251 among postmenopausal women) among 25,624 women. The mean length of follow-up was 14.0 years (median, 13.7), and the median age at diagnosis was 54.7 years (range, 36.3 to 68.0).
Table i gives the base-line characteristics of the participants. Two thirds of the women reported moderate activity during leisure time, whereas 15 percent exercised regularly. Only 14 percent report ed being sedentary at work, whereas 20 percent re ported lifting and 5 percent reported doing heavy manual labor. Women who reported regularly exer cising during leisuretimedid not differ from women were inactivc during their leisure timewith re spect to age at entry or number ofchildren, but they tended to be taller and to have a lower body-mass mdcx, a relatively low ratio of total cholcstcrol to high-density lipoprotein (HDL) cholesterol in se mm, lower serum triglyceride lcvels, and higher HDL cholesterol levels. Women whose work involved lift ing or heavy manual labor bad a higher body-mass index and more children than those engaged in sed entary work. Energy intake was positively rclatcd to physical activity, but the association was more pro nounced with work activity than xvith leisure-tirne
activity.
We analyzed other possible age-adjusted risk fac tors for breast cancer at base line and found a 28 percent increase in risk for cach additional 6 cm of height and a 13 percent reduction in risk for each child. An older maternal age at the birth of a first child xvas associated with a borderline incrcasc in risk, whereas body-mass mdcx (In the group as a whole or in the subgroups of premenopausal aud postmcnopausal women), energy intake, aud total fat intake did not infiuencc the overall risk of breast canccr (data not shovn).
Table 2 shows the relation bctwecn the levd of
- I
1270 - May 1, 1997
TABLE 1. BA%E-t.INE CRAwscrERIsTtcs OF TRE W0MEN ACC0ROINO To TRE LEvEL OE PRISaAL AUTIVITY IN TRE 1977—1983 SuRvEy.*
TABLE 2. ADIVITED RELATIVE R.ISK OF IloiiAsT CANCER AccoRotNo TO TRE LEVEI, OF PyoslciL AcTlvtry OVIUNO LEIsUO.E TIME ANO AT W0KK IN THE1977—1983SIJIeVEY.*
LEVELOF CASESOF RELATNERistt CASESOf REUTtEE RISK
PsysicAl. Acrtvtry BREAST CANcER 95% CIlt BnEasT CAsIcER 95%CIII Daring Irisurr time
Srdrntarr 66 1.00 60 1.00
Moderate 249 0.98 (0.75—1.20) 245 0.93 (0.71—1.22)
Regular rtrrcise 36 0.67 (0.44—1.00) 36 0.63 (0.42—0.95)
P for trend 0.08 0.04
At tror5
Sedentan 62 1.00 61 I 00
SValking 212 0.76 10.87—1.01) 210 004 10.63—1.12)
Lifting 64 066 (0 47—0.94) 63 074 (0.02—1 06)
Hearr manual lahor 12 0.46 (0.25—0 861 11 045 (0.25—0.921
P 6w trend 0.004 0.02
*Tltr trdrntart group is thr rrfrrrncr group. Cldrttotrs confldrnrrtntrrtal. Sahjrcts for ttltttm inftnmation concrming crrtain rariahirs n at mtsstng arr not tncludrd.
t’hrtahlrt nrrr adjattrd for age at rntn’.
)titrtahlrsnrrr adjustrd fttr age at rntrr, hod -mao tndrn, hrtgltt, ronnrrttfrcstdrttcr, and num her of tltildren.
leisure-tirne or w’ork activity and the overall risk of breast cancer. After adjustmenr for age and with the sedenrary groop as tlie reference groop, rhe relarive risk of breast caocer was redoced among women whose jobs involved walking, lifting, or heavy man ual labor. Adjusrments for orher factors (body-mass mdcx, counrv ofresideoce, numher ofchildrcn, aod height)inaddirion to age changed the nok estimates oniv slightly. Furrhcr adjusrmenrs for age at first
birth or dietary factors (energ inrake, total far in take, and fiber inrake) did not infiococe oor esti
matesofrelarive nok and w’ere omirted from rhc final model. A 52pcrcenr reducrion in nokwasobserved
amongrhe women who reporred doing heavy man ual labor (relative nok, 0.48; 95 percenr confideoce inrerval, 0.25 to 0.92). The overall adjusred nok of breast caocer decreased in a dose—response mannen w’irh increasing activity levd during leisore time
LEVEt nr AcrtvtryATW0RK
,krr at rtttry)tr) Rodt ntatt ittdrx Hright (ent) Trtglrtrrtdrt )ntg/dllt HDL cltatlrttrn,l lntg/dI%
Tittal cltolrttrrol:
I4DL cht,lrstrrol Paritt’
Nat. rtfclttldrrn Olotltrr’s agr at fot
barth (yr) Daily rnrrgv tntakr çkJ) Total dailr fat intakr (g) Datly sutoktug (5)
LEVEL nr Acnnanv 50555 10505K TtME
HE tVY
nrt,t.Ltu tttrt ta.
sEttEr rtay ttttttatxt EsEuttsE SEOENTAHY tt tLktrt. tItt St, Ltttttk (N=44l0) (s= 17,4011 (%=3719) l=3034l 5=10,385) 1N5240) (5=1300)
451 450 434 441 40.0 45.4 460
25.5 24.8 24.5 24.3 240 20.2 25.6
161.9 162.7 1633 1634 162.1 162.4 162.8
139.9 127.5 1240 1250 130.2 130.2 122.2
045 561 87.3 05.7 557 565 08.0
4.71 4.60 4.34 4.47 4.55 4.01 4.27
2.7 2.6 2.7 2.1 2.7 2.0 29
24.2 244 24.0 24.6 244 24.1 246
5725 5761 5797 5561 5716 5054 6434
554 54.0 54.7 524 54.6 55.9 61.5
40.1 34.5 31.6 38.0 38.3 36.4 26.1
‘All tariahlrs escrpt agrHerrad;ustrdforagr. All talurs rscrpt tltotr for datlr smokattg arr ntrans. Suh1rctt for aaItrtnt ittfoottatittn concrrning rrrtaintariablrs taas missing arr tot tnrlatdrd.
tToLonarra valurs Str triglycerides to millimolrs per liter, mnltiply by 0.01129.
flo ronrrrt ralursforHOLrholr,trrol to millimolrs per liter, mnlaiplr by 0 02586.
‘oIumr 336 Nombrr 18 1211
TABLE3. Anjus’rsoRsIATIvt RISK OS BluAs’rCsxct.RAccoRnlNG TO 1\ILNOPAUSAK STATUS ANE) THE Lsvst.os Pt-IvstcAL Acrivt’ryINil-Is 1977—I983 Sitrvo.*
LEVEL OF
PRYsIcAL ActIvnV PSEMENOPAUSAL WOMEN POSTMENOPAUSAL W0MEN
CASES (ti- (555%(IF
ICRE*.ST IttLs’I’(VF 05K (PlAST RF(ATT5’F (55K
I 51> l’R (95%Cl) (A1Ct’ER (98%CI)
liuring IcisurcInne
Slidcntarv 20 1.00 45 I 00
Moderate 65 0.77 (0.46—1.27) 177 1.00 (0.72—1.39)
lteguItresereise 10 0.53 (0.25—1.14) 26 0.67)0 41—1.10)
PSirrrirnd 0.10 0.15
At It’ork
Sedcntarv 22 1.00 39 1.00
SVaIkng 62 082 (0.50—1.34) 148 0,87 (0.61—1.24)
LIfrIngorheas’t ntannal 14 0.48 (0.24—0.95) 60 0.78 (0.82—1.18)
laht,r
5fnrtre,td 0.03 0.24
*Thescdeittary group isthc refcrcnce group. ‘artahTcs \s-crc adjusted fbr age at enrry, hody-mass mdcx,heighr, counrv of resrdencc, and nuniher ofchildren. CI denotes cr,nfidence interval. Suhjccts for svhom mnformation concerning cenain variablesstatmissing arc not includcd.
TABLE4.ADJUSTED RELSTIVERosos Bp.EAsT CANcERAcc0RJMNG TO BODY-MASS Ixnex ANE)TI-IS LEvEL05PHY5IcAL AcTIvtTy DUI1JNGLEtsuETIME IN THE
1977—1983 SURvEY.*
LEvEL05 PHY5IcAL
Acilvirn Booy.MAss INnEn, <22.8 Bony-Muss INnEn, 22.8—25.7 BoDy-MA5S INDEX, >25.7
(.555505 CASES 05 (TAES(IF
RRJ*AST RFLSTIx’E EIER IIKF.AS1’ RELATT’F P555 tRETT RElATIVt 1(15K
(.51> ER (95% CI) cs’stny (95% CI) csxcER (95% CI)
Sedcntary 21 1.00 14 1.00 30 1.00
Moderate 104 1.12 (0.70—1.79) 73 1.09(0 61—1.93) 68 0.70 (0.46—1.08) Regniar 6 0.28 (0.11—0.70) 14 0.96 (0.45—2.01) 16 0,53 (0.45—1.53)
5 for trcnd 0 02 0.90 0.36
T1,escdcntarv gronp isthe refcrencr group. Variahies wereadjusted for age at crttrv,licrght. coun
ty ofresidence,and nuntber ofchi)drcn. CI dcnotcs confidence inrers’aI. Suhjccts for sshonsinfor nianonconcernrngcerramn varrahles Rasmissing are notinduded.
(P for trend 0.04). Wornen vhoexercised at kast four hours a sveek during Icisure time bad a 37 perccnt reduction in the risk of breast canccr (relative risk, 0.63; 95 percent confidence interval, 0.42 to 0.95).
When thc group w’as divided according to meno pausal status (Table 3), a consistentiv inverse asso ciation was observed between the levet of leisure
time activitv and the prcmenopausal risk of breast cancer; tbe adjusted relative risk declined to 0.77 (95 percent confldence interval, 0.46 to 1.27) and further to 0.53 (95 percent confidencc interval, 0.25 to 1.14) as the levet of activity increased (P for trend=0.10). A weaker association was observed be
twcentbelevet of Icisure-time activity and tbe post menopausal risk of breast cancer. Tbe inverse associ ation between the levet ofactivity at work and the risk of breast caucer was also pronounced arnong
premenopausal women; among premenopausalSs’orn en whose jobs involved lifting or heavy manual la bor, rhe relative risk s’as 0.48 (95 percent confi dence interval, 0.24 to 0.95).
‘We also divided thc cohort into women who were younger than 45 years ofage at entry and those ‘ho were 45 or older. Among those s’ounger than 45 years at entrv for whorn data were complete (of ss’horn breast cancer developed in 138; mean age at diag nosis, 48.3 3’ears), the adjusted relative risk declined to 0.80 (95 percent confidence interval, 0.52 to 1.22) and further to 0.38 (95 percent confidcnce in terval, 0.19 to 0.79) as the levet of activity during leisure time incrcased (P for trend 0.01). The re spective adjusted relative risks werc 1.03 (95 percent confidence interval, 0.72 to 1.48) and 0.84 (95 per cent confidcnce interval,0.51 to1.39) (P for trend=
1272 Mat 1, 1997
TArni 5.AI)JusTEII RkLTIVsRISKOF BREAST C.xcER AcEORDING TOBony-MAS5lo)ExAND OvsLL LEvEL (SF PHVSICAL
AciiviryDUIUNGLElsulusTIME IN TRE1974—1978AND1977—1983 SuK”Eys.
0.54) among those for ‘hom data were complete were 45 years ofage or older at entry (ofwhom breast cancer developed in 208; mean age at diagno sis, 58.2 years). These values indicate that physical activiry had a protective cffect, particularly with re spect to the rjsk of breast cancer before and soon af ter menopause.
We examined models stratified according to body mass mdcx (Table 4). Among lean (body-mass mdcx,
<22.8), regularly exercising women, the risk of breast cancer was reduced by 72 percent (relative risk, 0.28; 95 percent confidence inrerval, 0.11 to 0.70). No such association was observed in the mid die or upper thirds ofbody-mass mdcx among regu lariv exercising svornen. In models stratified accord ing to both bodv-mass mdcx and menopausal status, this association was seen among both premenopausal and postrnenopausal leansvomen (data not shown).
In the second survev 61.2 pcrcent of the partici pants reported the same leve! ofleisure-time activirv as in the first survev, 23.5 percent reported an in creased leve!, and 15.3 percent reported a reduced leve!. By combining thcse two asscssments ofleisure time activitv, wc observed that the relative risk declined to 0.23 (95 percent confidence interval, 0.09 to 0.60) as thc leve! of sustained activirv in crcased in lean (bodv-mass index, <22.8) women (P for trend=0.002) (Table 5). Tbis protective ef fect across increasing levcls of sustained leisure-timc activirv svas observed in both lean premenopausal svomen (relative risk, 0.23; 95 pcrcent condence
Interva!,0.06 to 0.88; P for linear trend=0.02) and lean posrmenopausa! wornen (relative risk, 0.24; 95 percent confidcnce interval, 0.06 to 0.96; P for lin car trcnd=0.03).
DISCUSSJON
Our results support the idea that physical activirv protects against breast cancer, particularlv among premenopausal and vounger postmenopausal wom en. Activity during both leisure time and work re duced the overall risk. There was a significant inverse dose—response relation between leisure-time activir and the risk of breast cancer. The protective effeut was evident arnong lean premenopausa! and post mcnopausal women, and repeated assessment cm pbasized the preventive effect ofphvsical activitv.
The overal! reduction in the risk of breast cancer among active women is consistent with findings in other cohort’5-’7 and case—control’922studies, but at variancc svith the findings ofa few others.503’ Inone of these discrepant studies,3’ most of the svomen were older than in the present study and breast can
The overal! reduction in the risk of breast cancer among active women is consistent with findings in other cohort’5-’7 and case—control’922studies, but at variancc svith the findings ofa few others.503’ Inone of these discrepant studies,3’ most of the svomen were older than in the present study and breast can