Int. J . Cancer: 45, 428430 (1990) 0 1990 Wiley-Liss, Inc.
Publication of the International Union Against Cancer Publication de I'Union Intarnationale Contra 18 Cancer
AGE AT FIRST AND SECOND BIRTHS AND BREAST CANCER RISK IN BIPAROUS WOMEN E. NEGRI',C. LA VECCHIA'S~, S.W. Durn3, P. BRUZZP,F. PARAZZINI'and N.E. DAY^ 'Istituto di Ricerche Famcologiche "Mario Negri" , Via Eritrea 62, 20157 Milan, Italy; 21nstitut universitaire de mkdecine sociale et prkventive, Bugnon 17, 1005 Lausanne, Switzerland; 3Medical Research Council Biostatistics Unit, Cambridge, CB2 23W,UK;and 41stitutoNationule per la Ricerca sul Cancro, 16132 Genoa,Italy. The role of age at first and at second birth on subsequent breast cancer risk was analyzed using pooled data from 2 hospital-based case-control studies conducted in Italy, for a total of 1,200 biparous cases and 987 controls. Compared with women who gave birth for the first time below a e 20, the relative risks were above unity for those with later Wrst birth, even after allowance for age at second birth, althou h the trend was inconsistent across subsequent strata. A sim lor direct trend in risk was observed in relation t o age at second birth: com ared with less than 25 years and after allowance for a e at f r s t birth, the point estimates were I.2, I.4 and I.4 for 2!-29. 3&34 and 3 3 5 (p for linear trend = 0.04). The results for age at first and at second birth were similar in the 2 studies pooled in this analysis. A significant interaction with age was observed in relation t o age at first and at second birth. In younger women (below age 50) a strong and direct association with age at first birth was found, while no apparent protection was conveyed by earlier second birth. Among older women (aged 50 or over), there was no apparent relationship with age at first birth after allowance for age at second, but the role of age at second birth was independent and statistically si nlficant. Thus, our study confirms an independent and signfficant role of age at second birth in biparous women, after allowance for age at f i r s t birth, and indicates that, after reciprocal allowance, the role of f i r s t and second birth was not apparently different. The relative risks for both variables were quantitatively moderate, and may be influenced by age or other temporal variables.
f
difficult to interpret since there are few data which, considering multiparous women only, made simultaneous allowance for age at first and subsequent births, and for parity. Rather than attempt to adjust for the effect of parity, a clearer picture of the effect of a second birth will be obtained by confining the analysis to women who have had exactly 2 births. We decided, therefore, to analyze the role of age at first and at second birth in women with 2 births, using pooled data from 2 case-control studies conducted in Italy. SUBJECTS AND METHODS
Two studies were included in the present overview; a detailed description of their design and materials is given in the original reports. Briefly, the first study (La Vecchia et al., 1987) was a hospital-based case-control investigation conducted since January 1983 in the greater Milan area, Northern Italy, on a total of 2,125 histologically confirmed breast cancer patients under 75 years of age, admitted before September 1988 to the National Cancer Institute and the Ospedale Maggiore, which includes the 4 largest teaching and general hospitals in Milan. The comparison group consisted of 2,160 women admitted to the same network of hospitals for acute conditions (3 I % traumas, 27% other orthopaedic conditions, 14% acute surgical diseases, 28% other miscellaneous disorders), and was comThere is considerable evidence that parity and age at first parable to the case series in terms of age and area of residence. birth are related to subsequent breast cancer risk. A review of The second study (Toti et al., 1986) was a cooperative case26 epidemiological studies published since 1970 has indicated control investigation conducted between 1980 and 1983 in 13 that, from an aetiological viewpoint, both parity and age at first breast cancer clinics from 10 Italian provinces: 8 provinces in birth have some independent effect on breast carcinogenesis. Northern Italy (75% of the cases), one in central Italy (20%) From a public health viewpoint, the importance of age at first and one in Southern Italy (5%). A total of 1,556 newly diagbirth appears greater, since the trend is linear across subsequent nosed patients were interviewed and compared with 1,505 conage levels, while the protection of parity appears to be quan- trols admitted for acute, orthopaedic, medical or surgical contitatively relevant only for women with 4 or 5 births or more ditions, unrelated to known or suspected risk factors for breast (La Vecchia et al., 1989). cancer. Although matching was not strictly respected, the disLess is known about the role of age at subsequent pregnancies. In a hospital-based case-control study, Wynder et al. (1978) found no association for age at last birth, when this was examined in separate strata of age at first birth. An analysis Received: October 5 , 1989 and in revised form November 23, 1989. based on the international study conducted by MacMahon et al. (1970) (Trichopoulos et al. 1983) suggested that age at subseTABLE I - DISTRIBUTION OF 1.2W BLPAROUS CASES OF BREAST CANCER quent births had some independent effect on breast cancer risk, AND 987 BIPAROUS CONTROL SUBJECTS ACCORDING TO AGE, CENTRE AND though smaller than the role of age at first birth. Hunt et al. MENOPAUSAL STATUS (1980) and Lubin et al. (1982) found significant direct associVariable Cases (8) Controls (%I ations with age at last birth. In the latter study, the significant association persisted after allowance for age at first birth. In a Ag:35 (2.6) study from Estonia (MacMahon et al., 1982), a significant 31 77 (7.8) 3 5 4 242 (20.2) 227 (23.0) reduction in risk was observed among biparous women whose 45-54 394 (32.8j 277 &.ij second birth occurred below age 25, as compared to uniparous 55-64 282 (23.5) 225 (22.8) ones. Further, a cohort study from Norway (Kvale and Heuch, 181 (18.3) a65 251 (20.9) 1987) found, among women with 2 or more births, no associCentre ation with age at first birth, but a direct, though moderate, Milan 811 (67.6) 643 (65.1) relation with age at last birth, after reciprocal allowance for the 344 (34.9) Other 389 (32.4) 2 variables. MenoDausal status 477 (39.8) 414 (41.9) Pre-menopausal These findings seem to suggest that age at subsequent births Post-menopausal 723 (60.3) 573 (58.1) has some independent effect on breast cancer risk, but are
429
PARITY AGE A N D BREAST CANCER RISK
TABLE 11 - DISTRIBUTION OF 1,200 BIPAROUS CASES OF BREAST CANCER AND 987 BIPAROUS CONTROL SUBJECTS ACCORDING TO AGE AT FIRST AND SECOND BIRTHS Age at second birth (cases:controls) 2M9
Publication of the International Union Against Cancer Publication de I'Union Intarnationale Contra 18 Cancer
AGE AT FIRST AND SECOND BIRTHS AND BREAST CANCER RISK IN BIPAROUS WOMEN E. NEGRI',C. LA VECCHIA'S~, S.W. Durn3, P. BRUZZP,F. PARAZZINI'and N.E. DAY^ 'Istituto di Ricerche Famcologiche "Mario Negri" , Via Eritrea 62, 20157 Milan, Italy; 21nstitut universitaire de mkdecine sociale et prkventive, Bugnon 17, 1005 Lausanne, Switzerland; 3Medical Research Council Biostatistics Unit, Cambridge, CB2 23W,UK;and 41stitutoNationule per la Ricerca sul Cancro, 16132 Genoa,Italy. The role of age at first and at second birth on subsequent breast cancer risk was analyzed using pooled data from 2 hospital-based case-control studies conducted in Italy, for a total of 1,200 biparous cases and 987 controls. Compared with women who gave birth for the first time below a e 20, the relative risks were above unity for those with later Wrst birth, even after allowance for age at second birth, althou h the trend was inconsistent across subsequent strata. A sim lor direct trend in risk was observed in relation t o age at second birth: com ared with less than 25 years and after allowance for a e at f r s t birth, the point estimates were I.2, I.4 and I.4 for 2!-29. 3&34 and 3 3 5 (p for linear trend = 0.04). The results for age at first and at second birth were similar in the 2 studies pooled in this analysis. A significant interaction with age was observed in relation t o age at first and at second birth. In younger women (below age 50) a strong and direct association with age at first birth was found, while no apparent protection was conveyed by earlier second birth. Among older women (aged 50 or over), there was no apparent relationship with age at first birth after allowance for age at second, but the role of age at second birth was independent and statistically si nlficant. Thus, our study confirms an independent and signfficant role of age at second birth in biparous women, after allowance for age at f i r s t birth, and indicates that, after reciprocal allowance, the role of f i r s t and second birth was not apparently different. The relative risks for both variables were quantitatively moderate, and may be influenced by age or other temporal variables.
f
difficult to interpret since there are few data which, considering multiparous women only, made simultaneous allowance for age at first and subsequent births, and for parity. Rather than attempt to adjust for the effect of parity, a clearer picture of the effect of a second birth will be obtained by confining the analysis to women who have had exactly 2 births. We decided, therefore, to analyze the role of age at first and at second birth in women with 2 births, using pooled data from 2 case-control studies conducted in Italy. SUBJECTS AND METHODS
Two studies were included in the present overview; a detailed description of their design and materials is given in the original reports. Briefly, the first study (La Vecchia et al., 1987) was a hospital-based case-control investigation conducted since January 1983 in the greater Milan area, Northern Italy, on a total of 2,125 histologically confirmed breast cancer patients under 75 years of age, admitted before September 1988 to the National Cancer Institute and the Ospedale Maggiore, which includes the 4 largest teaching and general hospitals in Milan. The comparison group consisted of 2,160 women admitted to the same network of hospitals for acute conditions (3 I % traumas, 27% other orthopaedic conditions, 14% acute surgical diseases, 28% other miscellaneous disorders), and was comThere is considerable evidence that parity and age at first parable to the case series in terms of age and area of residence. birth are related to subsequent breast cancer risk. A review of The second study (Toti et al., 1986) was a cooperative case26 epidemiological studies published since 1970 has indicated control investigation conducted between 1980 and 1983 in 13 that, from an aetiological viewpoint, both parity and age at first breast cancer clinics from 10 Italian provinces: 8 provinces in birth have some independent effect on breast carcinogenesis. Northern Italy (75% of the cases), one in central Italy (20%) From a public health viewpoint, the importance of age at first and one in Southern Italy (5%). A total of 1,556 newly diagbirth appears greater, since the trend is linear across subsequent nosed patients were interviewed and compared with 1,505 conage levels, while the protection of parity appears to be quan- trols admitted for acute, orthopaedic, medical or surgical contitatively relevant only for women with 4 or 5 births or more ditions, unrelated to known or suspected risk factors for breast (La Vecchia et al., 1989). cancer. Although matching was not strictly respected, the disLess is known about the role of age at subsequent pregnancies. In a hospital-based case-control study, Wynder et al. (1978) found no association for age at last birth, when this was examined in separate strata of age at first birth. An analysis Received: October 5 , 1989 and in revised form November 23, 1989. based on the international study conducted by MacMahon et al. (1970) (Trichopoulos et al. 1983) suggested that age at subseTABLE I - DISTRIBUTION OF 1.2W BLPAROUS CASES OF BREAST CANCER quent births had some independent effect on breast cancer risk, AND 987 BIPAROUS CONTROL SUBJECTS ACCORDING TO AGE, CENTRE AND though smaller than the role of age at first birth. Hunt et al. MENOPAUSAL STATUS (1980) and Lubin et al. (1982) found significant direct associVariable Cases (8) Controls (%I ations with age at last birth. In the latter study, the significant association persisted after allowance for age at first birth. In a Ag:35 (2.6) study from Estonia (MacMahon et al., 1982), a significant 31 77 (7.8) 3 5 4 242 (20.2) 227 (23.0) reduction in risk was observed among biparous women whose 45-54 394 (32.8j 277 &.ij second birth occurred below age 25, as compared to uniparous 55-64 282 (23.5) 225 (22.8) ones. Further, a cohort study from Norway (Kvale and Heuch, 181 (18.3) a65 251 (20.9) 1987) found, among women with 2 or more births, no associCentre ation with age at first birth, but a direct, though moderate, Milan 811 (67.6) 643 (65.1) relation with age at last birth, after reciprocal allowance for the 344 (34.9) Other 389 (32.4) 2 variables. MenoDausal status 477 (39.8) 414 (41.9) Pre-menopausal These findings seem to suggest that age at subsequent births Post-menopausal 723 (60.3) 573 (58.1) has some independent effect on breast cancer risk, but are
429
PARITY AGE A N D BREAST CANCER RISK
TABLE 11 - DISTRIBUTION OF 1,200 BIPAROUS CASES OF BREAST CANCER AND 987 BIPAROUS CONTROL SUBJECTS ACCORDING TO AGE AT FIRST AND SECOND BIRTHS Age at second birth (cases:controls) 2M9
