Eur J Epidemiol (2015) 30:485–492 DOI 10.1007/s10654-015-0045-2
CANCER
Size at birth and risk of breast cancer: update from a prospective population-based study Marie Søfteland Sandvei1,2 • Pagona Lagiou3,4 • Pa˚l Richard Romundstad1 Dimitrios Trichopoulos3,4 • Lars Johan Vatten1,4
•
Received: 24 April 2015 / Accepted: 16 May 2015 / Published online: 31 May 2015 Ó Springer Science+Business Media Dordrecht 2015
Abstract Birth size variables (birth weight, birth length and head circumference) have been reported to be positively associated with adult breast cancer risk, whereas a possible association of placental weight has not been adequately studied. It has also been suggested that maternal height may modify the association of birth size with adult breast cancer risk, but this has not been studied in detail. We updated a long-term follow-up of 22,931 Norwegian women (average of 51 years of follow up during which 870 women were diagnosed with breast cancer) and assessed placental weight in relation to breast cancer risk, in addition to providing updated analyses on breast cancer risk in relation to birth weight, birth length and head circumference. Placental weight was not associated with risk for breast cancer in adulthood, but there was a positive association of breast cancer risk with birth length (HR 1.13, 95 % CI 1.05–1.21, per 2 cm increment), though not with birth weight (HR 1.02, 95 % CI 0.95–1.10 per 0.5 kg increment). For birth length, the graded increase in risk was particularly strong among women whose mothers were relatively tall (p for trend, 0.001), compared to the trend among women whose mothers were relatively short (p for trend, 0.221). The results showed a robust and positive & Lars Johan Vatten [email protected] 1
Department of Public Health and General Practice, Norwegian University of Science and Technology, Trondheim, Norway
2
Nordland Hospital Bodø, Bodø, Norway
3
Department of Hygiene and Medical Statistics, School of Medicine, University of Athens, Athens, Greece
4
Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
association of birth length with breast cancer risk, and may be especially strong in women whose mothers were relatively tall. We found no association of placental weight with risk for breast cancer. Keywords Breast cancer Birth size Birth length Prospective Population-based
Introduction The hypothesis that breast cancer may originate in utero was proposed by Trichopoulos in 1990 [1]. Since then, many studies have addressed the question, including a collaborative cohort study between Trichopoulos and ourselves. In the latest incidence update (2005) of that cohort, we had found that characteristics of birth size, notably birth weight, birth length and head circumference, were positively associated with risk for breast cancer, and the association was particularly strong for birth length [2]. Since the mammary gland is not fully differentiated at birth [3], it is plausible that intrauterine and perinatal processes may influence later risk for breast cancer. Thus, high intrauterine hormone levels may increase the mammary gland-specific stem cell pool, and increase the likelihood for cancer later in life [4]. The hypothesis is supported by four types of research: (a) women exposed to a synthetic estrogen, diethylstilbestrol (DES) in utero, were at increased risk of breast cancer several decades later [5, 6]; (b) ecological contrasts of cord blood hormone levels between women in the United States, who are at high risk of breast cancer, and Chinese women, who are at low risk, indicate substantially higher levels of sex hormone binding globulin (SHBG; and inferentially lower levels of bioavailable estrogens and testosterone), as well as lower
123
486
levels of insulin-like growth factor 1 (IGF-1) in Chinese women [7]; (c) many studies indicate a positive association of birth size with breast cancer risk in adulthood [8–10]; and (d) studies have indicated a link of the stem cell pool with birth weight as well as with relevant cord blood hormones (IGF, estrogens, SHBG) [11–13]. Thus, a sequence may be envisaged from production of cord blood hormones (mostly from the placenta), to bigger size of the mammary stem cell pool, which is correlated with higher birth size, and to higher breast cancer risk in adult life. What is missing in this sequence of events is a link between the size of the placenta and breast cancer risk, since placental weight may reflect the production of relevant pregnancy hormones [14]. There is currently little data available to show any association of placental weight with breast cancer risk [15–17]. Also, the positive association of birth length needs to be verified, and studied in more detail, especially with reference to a potentially modifying effect of maternal height on the offspring’s risk of breast cancer [2, 18, 19]. Therefore, we have updated our cohort [2] with 11 additional years of follow up, and assessed whether placental weight is associated with breast cancer risk, in addition to providing updated analyses of birth weight, birth length and head circumference.
Materials and methods Study population The study has been described in detail previously [2]. Briefly, we abstracted information from birth records of all births that took place at St. Olav’s University Hospital in Trondheim, Norway, from 1920 to 1966. After excluding 524 twins, 11 triplets, and 32 with missing plurality, we identified a total of 22,943 female singletons as eligible for breast cancer follow-up. The regional committee for ethics in medical research approved the study. In 1960, every Norwegian citizen was assigned a unique 11-digit identification number, and each citizen’s record is continuously updated on vital status, as well as residential and childbearing history through the national Central Person Registry. By combining the mother’s name and the daughter’s unique 11-digit number, we could identify women who were alive in 1960 and were born at St. Olav’s Hospital between 1920 and 1960. However, for women whose mothers had died before being assigned the unique identification number, we could not be absolutely certain about the identity of the daughter. Also, most daughters changed their last name when they married, and if their mother had died before 1960, their identity could not always be verified. For women born after 1960, identification
123
M. S. Sandvei et al.
created no problem, because they were assigned the 11 digit person number at birth. Thus, among a total of 22,943 female singletons, we reliably identified 22,931 women who could be followed up for breast cancer. The reporting of cancer to the Norwegian Cancer Registry is mandatory, and breast cancer was registered according to the international classification of diseases (ICD7, code 170). For women born before 1941, follow up time was calculated from 1 January 1961, and women born later were followed up for breast cancer from their 20th birthday. Follow up ended when a cancer (at any site) was diagnosed, at emigration, at death, or on December 31, 2012, whichever occurred first. From birth records of the eligible women, we abstracted information on birth weight (g), birth length (cm), head circumference (cm), placental weight (measured and recorded since 1934), duration of gestation (reported in weeks or months), and birth order. In addition, we abstracted information on maternal age, height, marital status, and socioeconomic status (occupation) of the mother at childbearing. We also collected information from the Central Person Registry on childbearing history in adulthood for cohort members born in 1930 or later who were still residing in Trondheim (87 %). Thus, age at first birth and parity, two established factors associated with breast cancer risk, were included in this subset of the population. A large proportion of the women in the cohort underwent a compulsory mass examination for tuberculosis conducted between 1961 and 1975, which included measurements of height and weight. Thus, for around 50 % of the cohort, we also have information about adult height and weight. Statistical methods We categorized most continuous birth size variables (birth length, head circumference, and placental weight) in five approximately equal categories (exact quintiles were not possible for birth length and head circumference due to the discrete nature of these variables). Birth weight was categorized in four categories (\3000, 3000–3499, 3500–3999 and [4000 g). We also examined the birth size measures according to increments of approximately one standard deviation (SD), i.e., per 0.5 kg differences in birth weight, per 2 cm differences in birth length, per 1.5 cm differences in head circumference at birth, and per 150 g differences in placental weight. We used Cox regression analysis to estimate hazard ratios (HRs) for breast cancer associated with categories, and increments of each of the birth size characteristics. Precision was estimated using 95 % confidence intervals (CI), and p values for linear trend across categories were calculated
Size at birth and risk of breast cancer: update from a prospective population-based study
by treating the categories of birth size as continuous variables in the regression model. Departure from the proportional hazards assumption was evaluated by Scho¨nfelds residuals and by inspection of the log–log plots. We also studied breast cancer risk separately for women diagnosed before 50 years, as well as at 50 years of age or older, and used this cut-off as an approximate distinction between premenopausal and postmenopausal breast cancer. Attained age was used as the time scale in all analyses. In the multivariable analyses, we adjusted for length of gestation, birth order (1, 2, C3), maternal age at childbearing (continuous), and maternal socioeconomic status (low or high). In addition, we adjusted for gestational age in weeks; for participants with information on gestational months only (n = 4280), gestational week was set to the middle week for the given month (i.e. 30.5 weeks for 7 months, 34.5 weeks for 8 months, 38.5 weeks for 9 months and 42.5 weeks for 10 months). In a subset of women (72 %), we could also adjust for maternal height, and for the adult risk factors age at first birth and parity, and in another subset (50 %), we could also adjust for adult height and body mass index. In a stratified analysis, we assessed birth size and breast cancer risk in women whose mothers were relatively tall (at or above the median of 163 cm) or relatively short (below the median). In supplementary analysis, we assessed the association between maternal height (in five approximately equal categories) and daughters’ risk for breast cancer, and in a separate analysis, we studied whether the association of maternal height on the offspring’s risk could be mediated by birth size variables. The statistical analyses were conducted in Stata (version 13.1).
Results We followed 22,931 women born between 1920 and 1966 for breast cancer incidence from 1961 until 2012. During an average of 51 years of follow up, 870 women were diagnosed with breast cancer: 318 were diagnosed before 50 years of age and 552 were diagnosed at the age of 50 years or older. Median age at diagnosis was 54 years (range 25–86 years), illustrating that despite a very long follow-up, the cohort as a whole is still relatively young. The results show a positive and graded increase in breast cancer risk associated with birth length (p for trend 0.001, Table 1). Women in the highest category of birth length (C52 cm) were at 30 % higher risk (HR 1.3, 95 % CI 1.0–1.7) compared to women in the reference category (\49 cm). The risk increase per SD increment (2 cm) in
487
birth length was 13 % (HR 1.13, 95 % CI 1.05–1.21). For birth weight, we found no clear association with breast cancer risk (HR 1.02, 95 % CI 0.95–1.10 per 0.5 kg increase). Regarding head circumference, women in the highest category (C37 cm) were at slightly higher risk than women in the lowest category (HR 1.2, 95 % CI 0.9–1.5), but there was no clear trend (HR 1.01, 95 % CI 0.94–1.08 per 1.5 cm increase). For placental weight, we found no association with risk for breast cancer in adulthood. The results were essentially unchanged after adjustment for maternal height and the adult risk factors age at first birth and parity. Also, in the subset of women with information about adult height and BMI, adjustment of these factors did not substantially influence the estimated associations (data not shown). However, it should be noted that among women with information on age at first birth, only 34 % (6174 out of 17,907) had their first birth after 25 years of age. In a separate analysis, we distinguished between breast cancer diagnosed before 50 years, as well as at 50 years of age or older (Table 2). In these age-specific analyses, the results related to birth length, birth weight, head circumference, and placental weight did no substantially differ according to age at diagnosis. In supplementary analyses, we studied whether the associations of birth size variables with breast cancer risk were modified by maternal height (Table 3). Regarding birth weight, birth length and placental weight, the estimates did not substantially differ between women whose mothers were 163 cm (population median) or taller and those whose mothers were shorter than 163 cm. However, among women whose mothers were relatively tall, 2 cm greater birth length was associated with 20 % higher risk (HR 1.20, 95 % CI 1.07–1.34), as compared to 10 % among women whose mothers were relatively short (HR 1.10, 95 % CI 0.98–1.23). Nonetheless, formal testing showed no interaction on a multiplicative scale between birth length and strata of maternal height (p = 0.736). We found that maternal height at childbearing was positively associated with the daughters’ breast cancer risk in adulthood (Table 4). Thus, daughters whose mothers were 168 cm or taller were at 40 % higher risk (HR 1.4, 95 % CI 1.1–1.8) compared to daughters whose mothers were shorter than 158 cm. The risk increase per SD increment (5 cm) in maternal height was 10 % (HR 1.10, 95 % CI 1.03–1.17). The association was slightly attenuated after adjustment for birth size variables (birth weight, birth length, head circumference and placental weight), and also further attenuated after adjustment for the daughter’s own adult height (data not shown), as would be expected from its correlation with maternal height (r = 0.48, p \ 0.001).
123
488
M. S. Sandvei et al.
Table 1 Birth size characteristics in relation to breast cancer among 22,931 Norwegian women followed from 1961 to 2012 Number of cases
Number of women
Person years
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trendb
Birth weight (g) \3000
133
3587
36,800
3000– 3499
308
8516
137,572
1.1
1.1 (0.9–1.3)
1
1
3500–3999
306
7568
414,856
1.2
1.2 (1.0–1.4)
C4000
100
2881
510,245
1.0
1.0 (0.8–1.3)
0.639
Per 0.5 kg increment
847
22,552
1,099,472
1.03
1.02 (0.95–1.10)
0.576
Birth length (cm) \49
107
3092
148,856
1
1
49
100
2985
144,485
0.9
0.9 (0.7–1.2)
50
195
5202
251,749
1.0
1.0 (0.8–1.3)
51
159
4259
208,705
1.2
1.1 (0.9–1.5)
C52
287
7015
345,691
1.3
1.3 (1.0–1.7)
0.001
Per 2 cm increment
847
22,533
1,099,485
1.12
1.13 (1.05–1.21)
0.001
Head circumference (cm) \34
141
3772
183,842
1
1
34
203
5297
257,102
1.0
1.0 (0.8–1.3)
35
237
6297
307,347
1.0
1.0 (0.8–1.2)
36
156
4588
224,381
1.0
0.9 (0.7–1.2)
C37
107
2523
123,169
1.2
1.2 (0.9–1.5)
0.559
Per 1.5 cm increment
843
22,477
1,095,839
1.01
1.01 (0.94–1.08)
0.855
\560 560–619
142 107
4201 3342
207,595 163,821
1 0.9
1 0.9 (0.7–1.1)
620–699
181
5064
249,766
1.0
1.0 (0.8–1.2)
700–779
166
4384
215,371
1.0
1.0 (0.8–1.3)
C780
149
3936
192,823
1.0
1.0 (0.8–1.3)
0.710
Per 150 g increment
744
20,927
1,029,377
1.01
1.00 (0.93–1.08)
0.994
Placental weight (g)
a Adjusted for age, length of gestation, socioeconomic status (low, high), maternal age (\20, 20–24, 25–29, 30–34, C35), birth order (1st, 2nd, C3rd) b
Two-sided p values for trend in Cox regression
Discussion In this prospective cohort with long-term follow-up of 22,931 Norwegian women, we could confirm a robust positive association of birth length with breast cancer risk, but found no association of placental weight with risk for breast cancer. The effect modification of birth length by maternal height that was indicated previously [2] could not be confirmed by formal testing of interaction at the multiplicative scale, however, greater birth length was associated with a stronger risk increase among women whose mothers were relatively tall for women whose mothers were relatively tall, greater birth length was associated with a higher increase in risk (HR 1.20, 95 % CI 1.07–1.34 per 2 cm difference in birth length) than for women whose mothers were relatively short (HR 1.10, 95 % CI 0.98–1.23
123
per 2 cm difference in birth length). Also, the trend test across categories of birth length was very strong (p \ 0.01) in the former, and weak in the latter group (p = 0.221). The major strengths of this study include the collection of accurate measurements of birth size from birth records, the complete breast cancer follow-up through the Norwegian Cancer Registry, and the ability to control for potential confounding by length of gestation, birth order, maternal age at childbearing and maternal socioeconomic status, as well as age at first birth, parity, maternal height and the women’s adult height and BMI, for a subset of the women. The study is limited by imprecise information about gestational age in the birth records, however, this is a limitation in most studies of perinatal factors that use data from the time before ultrasound technology was introduced.
Size at birth and risk of breast cancer: update from a prospective population-based study
489
Table 2 Birth size characteristics in relation to breast cancer diagnosed before and after the age of 50 years \50 years (premenopausal) No cases
Age-adjusted HR
C50 years (postmenopausal) Multivariable HR (95 % CI)a
p for trendb
No cases
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trendb
Birth weight (g) \3000 3000–3499 3500–3999 C4000 Per 0.5 kg increment
42
0.9
0.9 (0.6–1.4)
91
1.2
1.2 (0.9–1.5)
117 115
1 1.1
1 1.1 (0.9–1.4)
191 191
1 1.2
1 1.2 (1.0–1.5)
39
1.0
1.0 (0.7–1.4)
0.536
61
1.0
1.0 (0.8–1.4)
0.948
313
1.05
1.03 (0.91–1.16)
0.666
534
1.01
1.02 (0.93–1.11)
0.738
0.9
0.9 (0.6–1.2)
Birth length (cm) \49
37
1.0
1
70
49
39
1.1
1.0 (0.–1.6)
61
50
64
1.0
0.9 (0.6–1.4)
131
1.0
1.0 (0.8–1.4)
51
60
1.2
1.1 (0.7–1.6)
99
1.2
1.2 (0.9–1.7)
1
C52
114
1.3
1.2 (0.8–1.8)
0.166
173
1.3
1.4 (1.0–1.9)
0.001
Per 2 cm increment
314
1.17
1.15 (1.02–1.30)
0.021
534
1.09
1.11 (1.01–1.22)
0.026
Head circumference (cm) \34
52
1.0
1
34
68
0.9
0.9 (0.6–1.3)
135
35
90
1.0
1.0 (0.7–1.4)
147
1.0
1.0 (0.8–1.3)
36 C37
61 41
1.0 1.2
0.9 (0.6–1.3) 1.1 (0.7–1.7)
0.730
95 66
1.0 1.2
1.0 (0.7–1.3) 1.2 (0.9–1.7)
0.654
312
1.04
1.01 (0.90–1.13)
0.897
532
1.00
1.01 (0.92–1.10)
0.908
Per 1.5 cm increment
89
1
1
1.1
1.1 (0.8–1.5)
Placental weight (g) \560
57
1.0
1
85
1
1
560–619
46
1.0
1.0 (0.7–1.5)
61
0.8
0.8 (0.6–1.2)
620–699
69
1.0
1.0 (0.7–1.4)
112
1.0
1.0 (0.7–1.3)
700–779
69
1.2
1.1 (0.8–1.6)
97
0.9
0.9 (0.7–1.2)
58
1.1
1.1 (0.7–1.5)
0.539
91
1.0
1.0 (0.7–1.3)
0.965
299
1.00
0.99 (0.89–1.12)
0.925
446
1.01
1.00 (0.91–1.10)
0.941
C780 Per 150 g increment a
Adjusted for age, length of gestation, socioeconomic status (low, high), maternal age, birth order (1st, 2nd, C3rd)
b
Two-sided p values for trend in Cox regression
Our results regarding birth length correspond to the results of the meta-analysis by dos Santos Silva et al. [9]. In studies of singletons with perinatal information extracted from birth records, that meta-analysis showed that 2 cm increase in birth length was associated with 10 % higher risk, our corresponding estimate was 13 %. The metaanalysis reported a positive association of head circumference with breast cancer risk in adulthood, and we found a slightly higher risk in women in the highest compared to the lowest quintile of head circumference. The association of placental weight was not assessed in the meta-analysis [9]. At best, placental weight is a crude
indicator of placental function, and its weight may not adequately reflect its production of pregnancy hormones. However, placental size is correlated with birth size, and it has been suggested that the ratio between placental weight and birth weight could be a useful indicator for placental efficiency [20, 21]. In a study from Sweden the results suggested that placental weight may be positively associated with high risk mammographic parenchymal patterns in the offspring [22], and mammographic patterns are correlated with breast cancer risk [23]. In relation to breast cancer risk, however, one Swedish study suggested a positive association of placental weight [16] that has not
123
490
M. S. Sandvei et al.
Table 3 Birth size characteristics in relation to breast cancer diagnosed stratified by maternal height Maternal height \ 163 cm No cases
Maternal height C 163 cm
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trend
No cases
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trend
Birth weight (g) \3000 3000–3499 3500–3999 C4000 Per 0.5 kg increment
62
1.0
1.0 (08–1.4)
46
1.1
1.1 (0.8–1.6)
151 122
1 1.1
1 1.1 (0.9–1.4)
123 157
1 1.2
1 1.2 (1.0–1.6)
35
1.0
0.9 (0.6–1.3)
0.974
59
1.1
1.1 (0.8–1.5)
0.406
370
1.05
1.02 (0.91–1.14)
0.791
385
1.06
1.07 (0.96–1.19)
0.242
Birth length (cm) \49
53
1
1
30
1
1
49
56
1.1
1.0 (0.7–1.5)
32
0.9
0.9 (0.6–1.6)
50
95
1.1
1.0 (0.7–1.4)
70
1.0
1.0 (0.6–1.5)
51
64
1.1
1.0 (0.7–1.5)
81
1.3
1.3 (0.8–2.0)
C52
102
1.3
1.2 (0.9–1.7)
0.221
172
1.5
1.5 (1.0–2.3)
0.001
Per 2 cm increment
370
1.13
1.10 (0.98–1.23)
0.101
385
1.18
1.20 (1.07–1.34)
0.001
Head circumference (cm) \34
58
1
1
60
1
1
34
96
1.3
1.3 (0.9–1.8)
88
0.9
0.9 (0.6–1.2)
35
102
1.2
1.2 (0.8–1.6)
103
0.8
0.8 (0.6–1.1)
64 50
1.2 1.7
1.1 (0.8–1.6) 1.6 (1.1–2.4)
0.101
77 53
0.8 0.9
0.8 (0.5–1.1) 0.9 (0.6–1.3)
0.385
370
1.11
1.09 (0.98–1.21)
0.130
381
0.97
0.97 (0.87–1.08)
0.527
\560
67
1.0
1
56
1
1
560–619
50
1.0
0.9 (0.6–1.3)
49
1.0
1.0 (0.7–1.4)
620–699
77
1.0
1.0 (0.7–1.4)
93
1.2
1.1 (0.8–1.6)
700–779
77
1.1
1.1 (0.8–1.5)
79
1.1
1.1 (0.8–1.5)
C780
54
0.9
0.9 (0.6–1.3)
0.916
86
1.2
1.2 (0.9–1.7)
0.230
325
0.99
0.97 (0.86–1.09)
0.640
363
1.04
1.04 (0.95–1.14)
0.373
36 C37 Per 1.5 cm increment Placental weight (g)
Per 150 g increment a
Adjusted for age, length of gestation, socioeconomic status (low, high), maternal age, birth order (1st, 2nd, C3rd)
been confirmed by two other Nordic studies [15, 17]. On the other hand, there is evidence that maternal preeclampsia may be associated with reduced breast cancer risk in the offspring [16, 17, 24, 25], and placentas in pregnancies with early onset preeclampsia are smaller than placentas in normotensive pregnancies [26, 27]. Therefore, we examined the lower part of the distribution in detail, but found no indication that offspring of pregnancies with low weight placentas were at lower breast cancer risk than others. It is also possible that a relatively high placental weight may reflect a compensatory response to suboptimal fetal growth [28, 29]. To take this possibility into account, we examined the ratio of placental weight and birth weight,
123
but found no indication that high placental weight relative to birth weight was associated with higher risk (data not shown). Placental weight is prone to measurement error which may have contributed to the lack of any association in our study. The variability of placental weight may be attributed to many factors, including mode of delivery, trimming of the membranes, length of pregnancy, length of the cord, and time to cord clamping [30]. For instance, during the first 2 min after birth there is a transfusion of about 100 ml of blood from the placenta to the baby [31], and the timing of the clamping of the cord may vary between institutions [32]. Thus, non-differential misclassification of placental
Size at birth and risk of breast cancer: update from a prospective population-based study Table 4 The relation of maternal height to daughter’s risk for breast cancer n
HR (95 % CI)a
HR (95 % CI)b
\158 cm
129
1
1
158–161 cm
201
1.3 (1.0–1.6)
1.3 (1.0–1.6)
162–163 cm
95
1.1 (0.8–1.4)
1.0 (0.8–1.3)
Maternal height (cm)
164–167 cm
195
1.3 (1.1–1.7)
1.3 (1.0–1.6)
C168 cm
151
1.4 (1.1–1.8)
1.3 (1.0–1.6)
p for trend Per 5 cm increment
771
0.013 1.10 (1.03–1.17)
0.087 1.05 (0.99–1.10)
a
Crude
b
Adjusted for birth length
weight measurements may have biased the results for placental weight towards the null, and could have masked a true association. As stated in the meta-analyses of dos Santos Silva et al. [9], we also found that birth length is more strongly associated with breast cancer risk than other perinatal factors. This was evident despite the fact that varying muscle tonus and the extent to which the child is stretched during length measurement make birth length a less reliable measure than birth weight and head circumference [33]. It has been suggested that maternal height may play an important role for the intrauterine origin of breast cancer [18, 34]. The hypothesis is supported by the positive association of the cord blood IGF-1 with birth size, which appears to be restricted to daughters of relatively tall mothers [18, 35, 36]. In contrast to mothers of shorter stature, taller mothers may not impose mechanical constraints on fetal growth, and growth enhancing pregnancy hormones, including IGFs, may therefore be allowed to exercise their effects, and result in a positive association with birth size [34] and subsequently, with breast cancer risk [8, 9] in daughters of taller mothers. When we examined maternal height as a risk factor for breast cancer, the risk increase was 10 % per 5 cm (HR 1.10, 95 % CI 1.03–1.17). The association was slightly attenuated after adjustment for birth size variables. Although many studies have examined the intrauterine origin of breast cancer, the underlying mechanisms are still largely unknown. The etiological model proposes that high intrauterine hormone levels increase the mammary glandspecific stem cell pool, that may increase the likelihood for cancer later in life [4]. The model also proposes that growth enhancing and mammotropic hormones (such as estrogens, prolactin, and IGF) are important in the etiology of breast cancer, but the timing of effects and possible interactions, either between them or with other factors,
491
remain to be understood [4]. If the birth size associations with breast cancer risk reflect underlying effects of pregnancy hormones, it is obvious that using birth size measures as indicators of pregnancy hormones will result in serious exposure misclassification and to attenuation of any effect of the pregnancy hormones. But if a complex interplay of pregnancy hormones and other intrauterine factors is related to adult breast cancer risk, birth size may constitute a summary measure that could be more relevant than each individual hormone [9]. Recent studies suggest that established breast cancer risk factors may be differently related to molecular breast cancer subtypes [37–40]. Therefore, it may also be useful to assess whether birth size measures could be differentially associated with the risk of molecular breast cancer subtypes, because distinctly different associations may advance our understanding of how perinatal factors influence breast cancer risk later in life. Acknowledgments The St. Olav Birth Cohort study was established by a Grant from the National Institutes of Health (RO1 CA 78761), and subsequently supported by the Norwegian Cancer Society and the Norwegian Research Council. MSS is a postdoctoral fellow financed by the Norwegian Cancer Society. Conflict of interest The authors have no conflict of interest to declare.
References 1. Trichopoulos D. Hypothesis: does breast cancer originate in utero? Lancet. 1990;335:939–40. 2. Vatten LJ, Nilsen TI, Tretli S, Trichopoulos D, Romundstad PR. Size at birth and risk of breast cancer: prospective populationbased study. Int J Cancer. 2005;114:461–4. 3. Russo J, Rivera R, Russo IH. Influence of age and parity on the development of the human breast. Breast Cancer Res Treat. 1992;23:211–8. 4. Trichopoulos D, Adami HO, Ekbom A, Hsieh CC, Lagiou P. Early life events and conditions and breast cancer risk: from epidemiology to etiology. Int J Cancer. 2008;122:481–5. 5. Hoover RN, Hyer M, Pfeiffer RM, Adam E, Bond B, Cheville AL, et al. Adverse health outcomes in women exposed in utero to diethylstilbestrol. N Engl J Med. 2011;365:1304–14. 6. Adami HO, Lagiou P, Trichopoulos D. Breast cancer following diethylstilbestrol exposure in utero: insights from a tragedy. Eur J Epidemiol. 2012;27:1–3. 7. Lagiou P, Samoli E, Okulicz W, Xu B, Lagiou A, Lipworth L, et al. Maternal and cord blood hormone levels in the United States and China and the intrauterine origin of breast cancer. Ann Oncol. 2011;22:1102–8. 8. Michels KB, Xue F. Role of birthweight in the etiology of breast cancer. Int J Cancer. 2006;119:2007–25. 9. dos Santos Silva I, De Stavola B, McCormack V. Birth size and breast cancer risk: re-analysis of individual participant data from 32 studies. PLoS Med. 2008;5:e193. 10. Park SK, Kang D, McGlynn KA, Garcia-Closas M, Kim Y, Yoo KY, et al. Intrauterine environments and breast cancer risk: metaanalysis and systematic review. Breast Cancer Res. 2008;10:R8.
123
492 11. Savarese TM, Strohsnitter WC, Low HP, Liu Q, Baik I, Okulicz W, et al. Correlation of umbilical cord blood hormones and growth factors with stem cell potential: implications for the prenatal origin of breast cancer hypothesis. Breast Cancer Res. 2007;9:R29. 12. Strohsnitter WC, Savarese TM, Low HP, Chelmow DP, Lagiou P, Lambe M, et al. Correlation of umbilical cord blood haematopoietic stem and progenitor cell levels with birth weight: implications for a prenatal influence on cancer risk. Br J Cancer. 2008;98:660–3. 13. Qiu L, Low HP, Chang CI, Strohsnitter WC, Anderson M, Edmiston K, et al. Novel measurements of mammary stem cells in human umbilical cord blood as prospective predictors of breast cancer susceptibility in later life. Ann Oncol. 2012;23:245–50. 14. Lagiou P, Hsieh CC, Samoli E, Lagiou A, Xu B, Yu GP, et al. Associations of placental weight with maternal and cord blood hormones. Ann Epidemiol. 2013;23:669–73. 15. Vatten LJ, Maehle BO, Lund Nilsen TI, Tretli S, Hsieh CC, Trichopoulos D, et al. Birth weight as a predictor of breast cancer: a case-control study in Norway. Br J Cancer. 2002;86:89–91. 16. Ekbom A, Trichopoulos D, Adami HO, Hsieh CC, Lan SJ. Evidence of prenatal influences on breast cancer risk. Lancet. 1992; 340:1015–8. 17. Ekbom A, Hsieh CC, Lipworth L, Adami HQ, Trichopoulos D. Intrauterine environment and breast cancer risk in women: a population-based study. J Natl Cancer Inst. 1997;89:71–6. 18. Lagiou P, Samoli E, Lagiou A, Hsieh CC, Adami HO, Trichopoulos D. Maternal height, pregnancy estriol and birth weight in reference to breast cancer risk in Boston and Shanghai. Int J Cancer. 2005;117:494–8. 19. Lagiou P, Trichopoulos D, Hsieh CC. Is maternal height a risk factor for breast cancer? Eur J Cancer Prev. 2013;22:389–90. 20. Heinonen S, Taipale P, Saarikoski S. Weights of placentae from small-for-gestational age infants revisited. Placenta. 2001;22: 399–404. 21. Thame M, Osmond C, Bennett F, Wilks R, Forrester T. Fetal growth is directly related to maternal anthropometry and placental volume. Eur J Clin Nutr. 2004;58:894–900. 22. Ekbom A, Thurfjell E, Hsieh CC, Trichopoulos D, Adami HO. Perinatal characteristics and adult mammographic patterns. Int J Cancer. 1995;61:177–80. 23. McCormack VA, dos Santos Silva I. Breast density and parenchymal patterns as markers of breast cancer risk: a metaanalysis. Cancer Epidemiol Biomark Prev. 2006;15:1159–69. 24. Xue F, Michels KB. Intrauterine factors and risk of breast cancer: a systematic review and meta-analysis of current evidence. Lancet Oncol. 2007;8:1088–100. 25. Potischman N, Troisi R. In-utero and early life exposures in relation to risk of breast cancer. Cancer Causes Control. 1999;10: 561–73. 26. Boyd PA, Scott A. Quantitative structural studies on human placentas associated with pre-eclampsia, essential hypertension
123
M. S. Sandvei et al.
27.
28. 29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
and intrauterine growth retardation. Br J Obstet Gynaecol. 1985;92:714–21. Akhlaq M, Nagi AH, Yousaf AW. Placental morphology in preeclampsia and eclampsia and the likely role of NK cells. Indian J Pathol Microbiol. 2012;55:17–21. Teasdale F. Histomorphometry of the human placenta in maternal preeclampsia. Am J Obstet Gynecol. 1985;152:25–31. Risnes KR, Romundstad PR, Nilsen TI, Eskild A, Vatten LJ. Placental weight relative to birth weight and long-term cardiovascular mortality: findings from a cohort of 31,307 men and women. Am J Epidemiol. 2009;170:622–31. Burkhardt T, Schaffer L, Schneider C, Zimmermann R, Kurmanavicius J. Reference values for the weight of freshly delivered term placentas and for placental weight-birth weight ratios. Eur J Obstet Gynecol Reprod Biol. 2006;128:248–52. Farrar D, Airey R, Law GR, Tuffnell D, Cattle B, Duley L. Measuring placental transfusion for term births: weighing babies with cord intact. BJOG Int J Obstet Gynaecol. 2011;118:70–5. Lundberg C, Oian P, Klingenberg C. Umbilical cord clamping at birth–practice in Norwegian maternity wards. Tidsskr Nor Laegeforen. 2013;133:2369–73. Johnson TS, Engstrom JL, Gelhar DK. Intra- and interexaminer reliability of anthropometric measurements of term infants. J Pediatr Gastroenterol Nutr. 1997;24:497–505. Lagiou P, Hsieh CC, Trichopoulos D, Xu B, Wuu J, Mucci L, et al. Birthweight differences between USA and China and their relevance to breast cancer aetiology. Int J Epidemiol. 2003;32: 193–8. Lagiou P, Hsieh CC, Lipworth L, Samoli E, Okulicz W, Troisi R, et al. Insulin-like growth factor levels in cord blood, birth weight and breast cancer risk. Br J Cancer. 2009;100:1794–8. Lagiou P, Samoli E, Hsieh CC, Lagiou A, Xu B, Yu GP, et al. Maternal and cord blood hormones in relation to birth size. Eur J Epidemiol. 2014;29:343–51. Horn J, Alsaker MD, Opdahl S, Engstrom MJ, Tretli S, Haugen OA, et al. Anthropometric factors and risk of molecular breast cancer subtypes among postmenopausal Norwegian women. Int J Cancer. 2014;135:2678–86. Horn J, Opdahl S, Engstrom MJ, Romundstad PR, Tretli S, Haugen OA, et al. Reproductive history and the risk of molecular breast cancer subtypes in a prospective study of Norwegian women. Cancer Causes Control. 2014;25:881–9. Tamimi RM, Colditz GA, Hazra A, Baer HJ, Hankinson SE, Rosner B, et al. Traditional breast cancer risk factors in relation to molecular subtypes of breast cancer. Breast Cancer Res Treat. 2012;131:159–67. Yang XR, Chang-Claude J, Goode EL, Couch FJ, Nevanlinna H, Milne RL, et al. Associations of breast cancer risk factors with tumor subtypes: a pooled analysis from the Breast Cancer Association Consortium studies. J Natl Cancer Inst. 2011; 103:250–63.
CANCER
Size at birth and risk of breast cancer: update from a prospective population-based study Marie Søfteland Sandvei1,2 • Pagona Lagiou3,4 • Pa˚l Richard Romundstad1 Dimitrios Trichopoulos3,4 • Lars Johan Vatten1,4
•
Received: 24 April 2015 / Accepted: 16 May 2015 / Published online: 31 May 2015 Ó Springer Science+Business Media Dordrecht 2015
Abstract Birth size variables (birth weight, birth length and head circumference) have been reported to be positively associated with adult breast cancer risk, whereas a possible association of placental weight has not been adequately studied. It has also been suggested that maternal height may modify the association of birth size with adult breast cancer risk, but this has not been studied in detail. We updated a long-term follow-up of 22,931 Norwegian women (average of 51 years of follow up during which 870 women were diagnosed with breast cancer) and assessed placental weight in relation to breast cancer risk, in addition to providing updated analyses on breast cancer risk in relation to birth weight, birth length and head circumference. Placental weight was not associated with risk for breast cancer in adulthood, but there was a positive association of breast cancer risk with birth length (HR 1.13, 95 % CI 1.05–1.21, per 2 cm increment), though not with birth weight (HR 1.02, 95 % CI 0.95–1.10 per 0.5 kg increment). For birth length, the graded increase in risk was particularly strong among women whose mothers were relatively tall (p for trend, 0.001), compared to the trend among women whose mothers were relatively short (p for trend, 0.221). The results showed a robust and positive & Lars Johan Vatten [email protected] 1
Department of Public Health and General Practice, Norwegian University of Science and Technology, Trondheim, Norway
2
Nordland Hospital Bodø, Bodø, Norway
3
Department of Hygiene and Medical Statistics, School of Medicine, University of Athens, Athens, Greece
4
Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
association of birth length with breast cancer risk, and may be especially strong in women whose mothers were relatively tall. We found no association of placental weight with risk for breast cancer. Keywords Breast cancer Birth size Birth length Prospective Population-based
Introduction The hypothesis that breast cancer may originate in utero was proposed by Trichopoulos in 1990 [1]. Since then, many studies have addressed the question, including a collaborative cohort study between Trichopoulos and ourselves. In the latest incidence update (2005) of that cohort, we had found that characteristics of birth size, notably birth weight, birth length and head circumference, were positively associated with risk for breast cancer, and the association was particularly strong for birth length [2]. Since the mammary gland is not fully differentiated at birth [3], it is plausible that intrauterine and perinatal processes may influence later risk for breast cancer. Thus, high intrauterine hormone levels may increase the mammary gland-specific stem cell pool, and increase the likelihood for cancer later in life [4]. The hypothesis is supported by four types of research: (a) women exposed to a synthetic estrogen, diethylstilbestrol (DES) in utero, were at increased risk of breast cancer several decades later [5, 6]; (b) ecological contrasts of cord blood hormone levels between women in the United States, who are at high risk of breast cancer, and Chinese women, who are at low risk, indicate substantially higher levels of sex hormone binding globulin (SHBG; and inferentially lower levels of bioavailable estrogens and testosterone), as well as lower
123
486
levels of insulin-like growth factor 1 (IGF-1) in Chinese women [7]; (c) many studies indicate a positive association of birth size with breast cancer risk in adulthood [8–10]; and (d) studies have indicated a link of the stem cell pool with birth weight as well as with relevant cord blood hormones (IGF, estrogens, SHBG) [11–13]. Thus, a sequence may be envisaged from production of cord blood hormones (mostly from the placenta), to bigger size of the mammary stem cell pool, which is correlated with higher birth size, and to higher breast cancer risk in adult life. What is missing in this sequence of events is a link between the size of the placenta and breast cancer risk, since placental weight may reflect the production of relevant pregnancy hormones [14]. There is currently little data available to show any association of placental weight with breast cancer risk [15–17]. Also, the positive association of birth length needs to be verified, and studied in more detail, especially with reference to a potentially modifying effect of maternal height on the offspring’s risk of breast cancer [2, 18, 19]. Therefore, we have updated our cohort [2] with 11 additional years of follow up, and assessed whether placental weight is associated with breast cancer risk, in addition to providing updated analyses of birth weight, birth length and head circumference.
Materials and methods Study population The study has been described in detail previously [2]. Briefly, we abstracted information from birth records of all births that took place at St. Olav’s University Hospital in Trondheim, Norway, from 1920 to 1966. After excluding 524 twins, 11 triplets, and 32 with missing plurality, we identified a total of 22,943 female singletons as eligible for breast cancer follow-up. The regional committee for ethics in medical research approved the study. In 1960, every Norwegian citizen was assigned a unique 11-digit identification number, and each citizen’s record is continuously updated on vital status, as well as residential and childbearing history through the national Central Person Registry. By combining the mother’s name and the daughter’s unique 11-digit number, we could identify women who were alive in 1960 and were born at St. Olav’s Hospital between 1920 and 1960. However, for women whose mothers had died before being assigned the unique identification number, we could not be absolutely certain about the identity of the daughter. Also, most daughters changed their last name when they married, and if their mother had died before 1960, their identity could not always be verified. For women born after 1960, identification
123
M. S. Sandvei et al.
created no problem, because they were assigned the 11 digit person number at birth. Thus, among a total of 22,943 female singletons, we reliably identified 22,931 women who could be followed up for breast cancer. The reporting of cancer to the Norwegian Cancer Registry is mandatory, and breast cancer was registered according to the international classification of diseases (ICD7, code 170). For women born before 1941, follow up time was calculated from 1 January 1961, and women born later were followed up for breast cancer from their 20th birthday. Follow up ended when a cancer (at any site) was diagnosed, at emigration, at death, or on December 31, 2012, whichever occurred first. From birth records of the eligible women, we abstracted information on birth weight (g), birth length (cm), head circumference (cm), placental weight (measured and recorded since 1934), duration of gestation (reported in weeks or months), and birth order. In addition, we abstracted information on maternal age, height, marital status, and socioeconomic status (occupation) of the mother at childbearing. We also collected information from the Central Person Registry on childbearing history in adulthood for cohort members born in 1930 or later who were still residing in Trondheim (87 %). Thus, age at first birth and parity, two established factors associated with breast cancer risk, were included in this subset of the population. A large proportion of the women in the cohort underwent a compulsory mass examination for tuberculosis conducted between 1961 and 1975, which included measurements of height and weight. Thus, for around 50 % of the cohort, we also have information about adult height and weight. Statistical methods We categorized most continuous birth size variables (birth length, head circumference, and placental weight) in five approximately equal categories (exact quintiles were not possible for birth length and head circumference due to the discrete nature of these variables). Birth weight was categorized in four categories (\3000, 3000–3499, 3500–3999 and [4000 g). We also examined the birth size measures according to increments of approximately one standard deviation (SD), i.e., per 0.5 kg differences in birth weight, per 2 cm differences in birth length, per 1.5 cm differences in head circumference at birth, and per 150 g differences in placental weight. We used Cox regression analysis to estimate hazard ratios (HRs) for breast cancer associated with categories, and increments of each of the birth size characteristics. Precision was estimated using 95 % confidence intervals (CI), and p values for linear trend across categories were calculated
Size at birth and risk of breast cancer: update from a prospective population-based study
by treating the categories of birth size as continuous variables in the regression model. Departure from the proportional hazards assumption was evaluated by Scho¨nfelds residuals and by inspection of the log–log plots. We also studied breast cancer risk separately for women diagnosed before 50 years, as well as at 50 years of age or older, and used this cut-off as an approximate distinction between premenopausal and postmenopausal breast cancer. Attained age was used as the time scale in all analyses. In the multivariable analyses, we adjusted for length of gestation, birth order (1, 2, C3), maternal age at childbearing (continuous), and maternal socioeconomic status (low or high). In addition, we adjusted for gestational age in weeks; for participants with information on gestational months only (n = 4280), gestational week was set to the middle week for the given month (i.e. 30.5 weeks for 7 months, 34.5 weeks for 8 months, 38.5 weeks for 9 months and 42.5 weeks for 10 months). In a subset of women (72 %), we could also adjust for maternal height, and for the adult risk factors age at first birth and parity, and in another subset (50 %), we could also adjust for adult height and body mass index. In a stratified analysis, we assessed birth size and breast cancer risk in women whose mothers were relatively tall (at or above the median of 163 cm) or relatively short (below the median). In supplementary analysis, we assessed the association between maternal height (in five approximately equal categories) and daughters’ risk for breast cancer, and in a separate analysis, we studied whether the association of maternal height on the offspring’s risk could be mediated by birth size variables. The statistical analyses were conducted in Stata (version 13.1).
Results We followed 22,931 women born between 1920 and 1966 for breast cancer incidence from 1961 until 2012. During an average of 51 years of follow up, 870 women were diagnosed with breast cancer: 318 were diagnosed before 50 years of age and 552 were diagnosed at the age of 50 years or older. Median age at diagnosis was 54 years (range 25–86 years), illustrating that despite a very long follow-up, the cohort as a whole is still relatively young. The results show a positive and graded increase in breast cancer risk associated with birth length (p for trend 0.001, Table 1). Women in the highest category of birth length (C52 cm) were at 30 % higher risk (HR 1.3, 95 % CI 1.0–1.7) compared to women in the reference category (\49 cm). The risk increase per SD increment (2 cm) in
487
birth length was 13 % (HR 1.13, 95 % CI 1.05–1.21). For birth weight, we found no clear association with breast cancer risk (HR 1.02, 95 % CI 0.95–1.10 per 0.5 kg increase). Regarding head circumference, women in the highest category (C37 cm) were at slightly higher risk than women in the lowest category (HR 1.2, 95 % CI 0.9–1.5), but there was no clear trend (HR 1.01, 95 % CI 0.94–1.08 per 1.5 cm increase). For placental weight, we found no association with risk for breast cancer in adulthood. The results were essentially unchanged after adjustment for maternal height and the adult risk factors age at first birth and parity. Also, in the subset of women with information about adult height and BMI, adjustment of these factors did not substantially influence the estimated associations (data not shown). However, it should be noted that among women with information on age at first birth, only 34 % (6174 out of 17,907) had their first birth after 25 years of age. In a separate analysis, we distinguished between breast cancer diagnosed before 50 years, as well as at 50 years of age or older (Table 2). In these age-specific analyses, the results related to birth length, birth weight, head circumference, and placental weight did no substantially differ according to age at diagnosis. In supplementary analyses, we studied whether the associations of birth size variables with breast cancer risk were modified by maternal height (Table 3). Regarding birth weight, birth length and placental weight, the estimates did not substantially differ between women whose mothers were 163 cm (population median) or taller and those whose mothers were shorter than 163 cm. However, among women whose mothers were relatively tall, 2 cm greater birth length was associated with 20 % higher risk (HR 1.20, 95 % CI 1.07–1.34), as compared to 10 % among women whose mothers were relatively short (HR 1.10, 95 % CI 0.98–1.23). Nonetheless, formal testing showed no interaction on a multiplicative scale between birth length and strata of maternal height (p = 0.736). We found that maternal height at childbearing was positively associated with the daughters’ breast cancer risk in adulthood (Table 4). Thus, daughters whose mothers were 168 cm or taller were at 40 % higher risk (HR 1.4, 95 % CI 1.1–1.8) compared to daughters whose mothers were shorter than 158 cm. The risk increase per SD increment (5 cm) in maternal height was 10 % (HR 1.10, 95 % CI 1.03–1.17). The association was slightly attenuated after adjustment for birth size variables (birth weight, birth length, head circumference and placental weight), and also further attenuated after adjustment for the daughter’s own adult height (data not shown), as would be expected from its correlation with maternal height (r = 0.48, p \ 0.001).
123
488
M. S. Sandvei et al.
Table 1 Birth size characteristics in relation to breast cancer among 22,931 Norwegian women followed from 1961 to 2012 Number of cases
Number of women
Person years
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trendb
Birth weight (g) \3000
133
3587
36,800
3000– 3499
308
8516
137,572
1.1
1.1 (0.9–1.3)
1
1
3500–3999
306
7568
414,856
1.2
1.2 (1.0–1.4)
C4000
100
2881
510,245
1.0
1.0 (0.8–1.3)
0.639
Per 0.5 kg increment
847
22,552
1,099,472
1.03
1.02 (0.95–1.10)
0.576
Birth length (cm) \49
107
3092
148,856
1
1
49
100
2985
144,485
0.9
0.9 (0.7–1.2)
50
195
5202
251,749
1.0
1.0 (0.8–1.3)
51
159
4259
208,705
1.2
1.1 (0.9–1.5)
C52
287
7015
345,691
1.3
1.3 (1.0–1.7)
0.001
Per 2 cm increment
847
22,533
1,099,485
1.12
1.13 (1.05–1.21)
0.001
Head circumference (cm) \34
141
3772
183,842
1
1
34
203
5297
257,102
1.0
1.0 (0.8–1.3)
35
237
6297
307,347
1.0
1.0 (0.8–1.2)
36
156
4588
224,381
1.0
0.9 (0.7–1.2)
C37
107
2523
123,169
1.2
1.2 (0.9–1.5)
0.559
Per 1.5 cm increment
843
22,477
1,095,839
1.01
1.01 (0.94–1.08)
0.855
\560 560–619
142 107
4201 3342
207,595 163,821
1 0.9
1 0.9 (0.7–1.1)
620–699
181
5064
249,766
1.0
1.0 (0.8–1.2)
700–779
166
4384
215,371
1.0
1.0 (0.8–1.3)
C780
149
3936
192,823
1.0
1.0 (0.8–1.3)
0.710
Per 150 g increment
744
20,927
1,029,377
1.01
1.00 (0.93–1.08)
0.994
Placental weight (g)
a Adjusted for age, length of gestation, socioeconomic status (low, high), maternal age (\20, 20–24, 25–29, 30–34, C35), birth order (1st, 2nd, C3rd) b
Two-sided p values for trend in Cox regression
Discussion In this prospective cohort with long-term follow-up of 22,931 Norwegian women, we could confirm a robust positive association of birth length with breast cancer risk, but found no association of placental weight with risk for breast cancer. The effect modification of birth length by maternal height that was indicated previously [2] could not be confirmed by formal testing of interaction at the multiplicative scale, however, greater birth length was associated with a stronger risk increase among women whose mothers were relatively tall for women whose mothers were relatively tall, greater birth length was associated with a higher increase in risk (HR 1.20, 95 % CI 1.07–1.34 per 2 cm difference in birth length) than for women whose mothers were relatively short (HR 1.10, 95 % CI 0.98–1.23
123
per 2 cm difference in birth length). Also, the trend test across categories of birth length was very strong (p \ 0.01) in the former, and weak in the latter group (p = 0.221). The major strengths of this study include the collection of accurate measurements of birth size from birth records, the complete breast cancer follow-up through the Norwegian Cancer Registry, and the ability to control for potential confounding by length of gestation, birth order, maternal age at childbearing and maternal socioeconomic status, as well as age at first birth, parity, maternal height and the women’s adult height and BMI, for a subset of the women. The study is limited by imprecise information about gestational age in the birth records, however, this is a limitation in most studies of perinatal factors that use data from the time before ultrasound technology was introduced.
Size at birth and risk of breast cancer: update from a prospective population-based study
489
Table 2 Birth size characteristics in relation to breast cancer diagnosed before and after the age of 50 years \50 years (premenopausal) No cases
Age-adjusted HR
C50 years (postmenopausal) Multivariable HR (95 % CI)a
p for trendb
No cases
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trendb
Birth weight (g) \3000 3000–3499 3500–3999 C4000 Per 0.5 kg increment
42
0.9
0.9 (0.6–1.4)
91
1.2
1.2 (0.9–1.5)
117 115
1 1.1
1 1.1 (0.9–1.4)
191 191
1 1.2
1 1.2 (1.0–1.5)
39
1.0
1.0 (0.7–1.4)
0.536
61
1.0
1.0 (0.8–1.4)
0.948
313
1.05
1.03 (0.91–1.16)
0.666
534
1.01
1.02 (0.93–1.11)
0.738
0.9
0.9 (0.6–1.2)
Birth length (cm) \49
37
1.0
1
70
49
39
1.1
1.0 (0.–1.6)
61
50
64
1.0
0.9 (0.6–1.4)
131
1.0
1.0 (0.8–1.4)
51
60
1.2
1.1 (0.7–1.6)
99
1.2
1.2 (0.9–1.7)
1
C52
114
1.3
1.2 (0.8–1.8)
0.166
173
1.3
1.4 (1.0–1.9)
0.001
Per 2 cm increment
314
1.17
1.15 (1.02–1.30)
0.021
534
1.09
1.11 (1.01–1.22)
0.026
Head circumference (cm) \34
52
1.0
1
34
68
0.9
0.9 (0.6–1.3)
135
35
90
1.0
1.0 (0.7–1.4)
147
1.0
1.0 (0.8–1.3)
36 C37
61 41
1.0 1.2
0.9 (0.6–1.3) 1.1 (0.7–1.7)
0.730
95 66
1.0 1.2
1.0 (0.7–1.3) 1.2 (0.9–1.7)
0.654
312
1.04
1.01 (0.90–1.13)
0.897
532
1.00
1.01 (0.92–1.10)
0.908
Per 1.5 cm increment
89
1
1
1.1
1.1 (0.8–1.5)
Placental weight (g) \560
57
1.0
1
85
1
1
560–619
46
1.0
1.0 (0.7–1.5)
61
0.8
0.8 (0.6–1.2)
620–699
69
1.0
1.0 (0.7–1.4)
112
1.0
1.0 (0.7–1.3)
700–779
69
1.2
1.1 (0.8–1.6)
97
0.9
0.9 (0.7–1.2)
58
1.1
1.1 (0.7–1.5)
0.539
91
1.0
1.0 (0.7–1.3)
0.965
299
1.00
0.99 (0.89–1.12)
0.925
446
1.01
1.00 (0.91–1.10)
0.941
C780 Per 150 g increment a
Adjusted for age, length of gestation, socioeconomic status (low, high), maternal age, birth order (1st, 2nd, C3rd)
b
Two-sided p values for trend in Cox regression
Our results regarding birth length correspond to the results of the meta-analysis by dos Santos Silva et al. [9]. In studies of singletons with perinatal information extracted from birth records, that meta-analysis showed that 2 cm increase in birth length was associated with 10 % higher risk, our corresponding estimate was 13 %. The metaanalysis reported a positive association of head circumference with breast cancer risk in adulthood, and we found a slightly higher risk in women in the highest compared to the lowest quintile of head circumference. The association of placental weight was not assessed in the meta-analysis [9]. At best, placental weight is a crude
indicator of placental function, and its weight may not adequately reflect its production of pregnancy hormones. However, placental size is correlated with birth size, and it has been suggested that the ratio between placental weight and birth weight could be a useful indicator for placental efficiency [20, 21]. In a study from Sweden the results suggested that placental weight may be positively associated with high risk mammographic parenchymal patterns in the offspring [22], and mammographic patterns are correlated with breast cancer risk [23]. In relation to breast cancer risk, however, one Swedish study suggested a positive association of placental weight [16] that has not
123
490
M. S. Sandvei et al.
Table 3 Birth size characteristics in relation to breast cancer diagnosed stratified by maternal height Maternal height \ 163 cm No cases
Maternal height C 163 cm
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trend
No cases
Age-adjusted HR
Multivariable HR (95 % CI)a
p for trend
Birth weight (g) \3000 3000–3499 3500–3999 C4000 Per 0.5 kg increment
62
1.0
1.0 (08–1.4)
46
1.1
1.1 (0.8–1.6)
151 122
1 1.1
1 1.1 (0.9–1.4)
123 157
1 1.2
1 1.2 (1.0–1.6)
35
1.0
0.9 (0.6–1.3)
0.974
59
1.1
1.1 (0.8–1.5)
0.406
370
1.05
1.02 (0.91–1.14)
0.791
385
1.06
1.07 (0.96–1.19)
0.242
Birth length (cm) \49
53
1
1
30
1
1
49
56
1.1
1.0 (0.7–1.5)
32
0.9
0.9 (0.6–1.6)
50
95
1.1
1.0 (0.7–1.4)
70
1.0
1.0 (0.6–1.5)
51
64
1.1
1.0 (0.7–1.5)
81
1.3
1.3 (0.8–2.0)
C52
102
1.3
1.2 (0.9–1.7)
0.221
172
1.5
1.5 (1.0–2.3)
0.001
Per 2 cm increment
370
1.13
1.10 (0.98–1.23)
0.101
385
1.18
1.20 (1.07–1.34)
0.001
Head circumference (cm) \34
58
1
1
60
1
1
34
96
1.3
1.3 (0.9–1.8)
88
0.9
0.9 (0.6–1.2)
35
102
1.2
1.2 (0.8–1.6)
103
0.8
0.8 (0.6–1.1)
64 50
1.2 1.7
1.1 (0.8–1.6) 1.6 (1.1–2.4)
0.101
77 53
0.8 0.9
0.8 (0.5–1.1) 0.9 (0.6–1.3)
0.385
370
1.11
1.09 (0.98–1.21)
0.130
381
0.97
0.97 (0.87–1.08)
0.527
\560
67
1.0
1
56
1
1
560–619
50
1.0
0.9 (0.6–1.3)
49
1.0
1.0 (0.7–1.4)
620–699
77
1.0
1.0 (0.7–1.4)
93
1.2
1.1 (0.8–1.6)
700–779
77
1.1
1.1 (0.8–1.5)
79
1.1
1.1 (0.8–1.5)
C780
54
0.9
0.9 (0.6–1.3)
0.916
86
1.2
1.2 (0.9–1.7)
0.230
325
0.99
0.97 (0.86–1.09)
0.640
363
1.04
1.04 (0.95–1.14)
0.373
36 C37 Per 1.5 cm increment Placental weight (g)
Per 150 g increment a
Adjusted for age, length of gestation, socioeconomic status (low, high), maternal age, birth order (1st, 2nd, C3rd)
been confirmed by two other Nordic studies [15, 17]. On the other hand, there is evidence that maternal preeclampsia may be associated with reduced breast cancer risk in the offspring [16, 17, 24, 25], and placentas in pregnancies with early onset preeclampsia are smaller than placentas in normotensive pregnancies [26, 27]. Therefore, we examined the lower part of the distribution in detail, but found no indication that offspring of pregnancies with low weight placentas were at lower breast cancer risk than others. It is also possible that a relatively high placental weight may reflect a compensatory response to suboptimal fetal growth [28, 29]. To take this possibility into account, we examined the ratio of placental weight and birth weight,
123
but found no indication that high placental weight relative to birth weight was associated with higher risk (data not shown). Placental weight is prone to measurement error which may have contributed to the lack of any association in our study. The variability of placental weight may be attributed to many factors, including mode of delivery, trimming of the membranes, length of pregnancy, length of the cord, and time to cord clamping [30]. For instance, during the first 2 min after birth there is a transfusion of about 100 ml of blood from the placenta to the baby [31], and the timing of the clamping of the cord may vary between institutions [32]. Thus, non-differential misclassification of placental
Size at birth and risk of breast cancer: update from a prospective population-based study Table 4 The relation of maternal height to daughter’s risk for breast cancer n
HR (95 % CI)a
HR (95 % CI)b
\158 cm
129
1
1
158–161 cm
201
1.3 (1.0–1.6)
1.3 (1.0–1.6)
162–163 cm
95
1.1 (0.8–1.4)
1.0 (0.8–1.3)
Maternal height (cm)
164–167 cm
195
1.3 (1.1–1.7)
1.3 (1.0–1.6)
C168 cm
151
1.4 (1.1–1.8)
1.3 (1.0–1.6)
p for trend Per 5 cm increment
771
0.013 1.10 (1.03–1.17)
0.087 1.05 (0.99–1.10)
a
Crude
b
Adjusted for birth length
weight measurements may have biased the results for placental weight towards the null, and could have masked a true association. As stated in the meta-analyses of dos Santos Silva et al. [9], we also found that birth length is more strongly associated with breast cancer risk than other perinatal factors. This was evident despite the fact that varying muscle tonus and the extent to which the child is stretched during length measurement make birth length a less reliable measure than birth weight and head circumference [33]. It has been suggested that maternal height may play an important role for the intrauterine origin of breast cancer [18, 34]. The hypothesis is supported by the positive association of the cord blood IGF-1 with birth size, which appears to be restricted to daughters of relatively tall mothers [18, 35, 36]. In contrast to mothers of shorter stature, taller mothers may not impose mechanical constraints on fetal growth, and growth enhancing pregnancy hormones, including IGFs, may therefore be allowed to exercise their effects, and result in a positive association with birth size [34] and subsequently, with breast cancer risk [8, 9] in daughters of taller mothers. When we examined maternal height as a risk factor for breast cancer, the risk increase was 10 % per 5 cm (HR 1.10, 95 % CI 1.03–1.17). The association was slightly attenuated after adjustment for birth size variables. Although many studies have examined the intrauterine origin of breast cancer, the underlying mechanisms are still largely unknown. The etiological model proposes that high intrauterine hormone levels increase the mammary glandspecific stem cell pool, that may increase the likelihood for cancer later in life [4]. The model also proposes that growth enhancing and mammotropic hormones (such as estrogens, prolactin, and IGF) are important in the etiology of breast cancer, but the timing of effects and possible interactions, either between them or with other factors,
491
remain to be understood [4]. If the birth size associations with breast cancer risk reflect underlying effects of pregnancy hormones, it is obvious that using birth size measures as indicators of pregnancy hormones will result in serious exposure misclassification and to attenuation of any effect of the pregnancy hormones. But if a complex interplay of pregnancy hormones and other intrauterine factors is related to adult breast cancer risk, birth size may constitute a summary measure that could be more relevant than each individual hormone [9]. Recent studies suggest that established breast cancer risk factors may be differently related to molecular breast cancer subtypes [37–40]. Therefore, it may also be useful to assess whether birth size measures could be differentially associated with the risk of molecular breast cancer subtypes, because distinctly different associations may advance our understanding of how perinatal factors influence breast cancer risk later in life. Acknowledgments The St. Olav Birth Cohort study was established by a Grant from the National Institutes of Health (RO1 CA 78761), and subsequently supported by the Norwegian Cancer Society and the Norwegian Research Council. MSS is a postdoctoral fellow financed by the Norwegian Cancer Society. Conflict of interest The authors have no conflict of interest to declare.
References 1. Trichopoulos D. Hypothesis: does breast cancer originate in utero? Lancet. 1990;335:939–40. 2. Vatten LJ, Nilsen TI, Tretli S, Trichopoulos D, Romundstad PR. Size at birth and risk of breast cancer: prospective populationbased study. Int J Cancer. 2005;114:461–4. 3. Russo J, Rivera R, Russo IH. Influence of age and parity on the development of the human breast. Breast Cancer Res Treat. 1992;23:211–8. 4. Trichopoulos D, Adami HO, Ekbom A, Hsieh CC, Lagiou P. Early life events and conditions and breast cancer risk: from epidemiology to etiology. Int J Cancer. 2008;122:481–5. 5. Hoover RN, Hyer M, Pfeiffer RM, Adam E, Bond B, Cheville AL, et al. Adverse health outcomes in women exposed in utero to diethylstilbestrol. N Engl J Med. 2011;365:1304–14. 6. Adami HO, Lagiou P, Trichopoulos D. Breast cancer following diethylstilbestrol exposure in utero: insights from a tragedy. Eur J Epidemiol. 2012;27:1–3. 7. Lagiou P, Samoli E, Okulicz W, Xu B, Lagiou A, Lipworth L, et al. Maternal and cord blood hormone levels in the United States and China and the intrauterine origin of breast cancer. Ann Oncol. 2011;22:1102–8. 8. Michels KB, Xue F. Role of birthweight in the etiology of breast cancer. Int J Cancer. 2006;119:2007–25. 9. dos Santos Silva I, De Stavola B, McCormack V. Birth size and breast cancer risk: re-analysis of individual participant data from 32 studies. PLoS Med. 2008;5:e193. 10. Park SK, Kang D, McGlynn KA, Garcia-Closas M, Kim Y, Yoo KY, et al. Intrauterine environments and breast cancer risk: metaanalysis and systematic review. Breast Cancer Res. 2008;10:R8.
123
492 11. Savarese TM, Strohsnitter WC, Low HP, Liu Q, Baik I, Okulicz W, et al. Correlation of umbilical cord blood hormones and growth factors with stem cell potential: implications for the prenatal origin of breast cancer hypothesis. Breast Cancer Res. 2007;9:R29. 12. Strohsnitter WC, Savarese TM, Low HP, Chelmow DP, Lagiou P, Lambe M, et al. Correlation of umbilical cord blood haematopoietic stem and progenitor cell levels with birth weight: implications for a prenatal influence on cancer risk. Br J Cancer. 2008;98:660–3. 13. Qiu L, Low HP, Chang CI, Strohsnitter WC, Anderson M, Edmiston K, et al. Novel measurements of mammary stem cells in human umbilical cord blood as prospective predictors of breast cancer susceptibility in later life. Ann Oncol. 2012;23:245–50. 14. Lagiou P, Hsieh CC, Samoli E, Lagiou A, Xu B, Yu GP, et al. Associations of placental weight with maternal and cord blood hormones. Ann Epidemiol. 2013;23:669–73. 15. Vatten LJ, Maehle BO, Lund Nilsen TI, Tretli S, Hsieh CC, Trichopoulos D, et al. Birth weight as a predictor of breast cancer: a case-control study in Norway. Br J Cancer. 2002;86:89–91. 16. Ekbom A, Trichopoulos D, Adami HO, Hsieh CC, Lan SJ. Evidence of prenatal influences on breast cancer risk. Lancet. 1992; 340:1015–8. 17. Ekbom A, Hsieh CC, Lipworth L, Adami HQ, Trichopoulos D. Intrauterine environment and breast cancer risk in women: a population-based study. J Natl Cancer Inst. 1997;89:71–6. 18. Lagiou P, Samoli E, Lagiou A, Hsieh CC, Adami HO, Trichopoulos D. Maternal height, pregnancy estriol and birth weight in reference to breast cancer risk in Boston and Shanghai. Int J Cancer. 2005;117:494–8. 19. Lagiou P, Trichopoulos D, Hsieh CC. Is maternal height a risk factor for breast cancer? Eur J Cancer Prev. 2013;22:389–90. 20. Heinonen S, Taipale P, Saarikoski S. Weights of placentae from small-for-gestational age infants revisited. Placenta. 2001;22: 399–404. 21. Thame M, Osmond C, Bennett F, Wilks R, Forrester T. Fetal growth is directly related to maternal anthropometry and placental volume. Eur J Clin Nutr. 2004;58:894–900. 22. Ekbom A, Thurfjell E, Hsieh CC, Trichopoulos D, Adami HO. Perinatal characteristics and adult mammographic patterns. Int J Cancer. 1995;61:177–80. 23. McCormack VA, dos Santos Silva I. Breast density and parenchymal patterns as markers of breast cancer risk: a metaanalysis. Cancer Epidemiol Biomark Prev. 2006;15:1159–69. 24. Xue F, Michels KB. Intrauterine factors and risk of breast cancer: a systematic review and meta-analysis of current evidence. Lancet Oncol. 2007;8:1088–100. 25. Potischman N, Troisi R. In-utero and early life exposures in relation to risk of breast cancer. Cancer Causes Control. 1999;10: 561–73. 26. Boyd PA, Scott A. Quantitative structural studies on human placentas associated with pre-eclampsia, essential hypertension
123
M. S. Sandvei et al.
27.
28. 29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
and intrauterine growth retardation. Br J Obstet Gynaecol. 1985;92:714–21. Akhlaq M, Nagi AH, Yousaf AW. Placental morphology in preeclampsia and eclampsia and the likely role of NK cells. Indian J Pathol Microbiol. 2012;55:17–21. Teasdale F. Histomorphometry of the human placenta in maternal preeclampsia. Am J Obstet Gynecol. 1985;152:25–31. Risnes KR, Romundstad PR, Nilsen TI, Eskild A, Vatten LJ. Placental weight relative to birth weight and long-term cardiovascular mortality: findings from a cohort of 31,307 men and women. Am J Epidemiol. 2009;170:622–31. Burkhardt T, Schaffer L, Schneider C, Zimmermann R, Kurmanavicius J. Reference values for the weight of freshly delivered term placentas and for placental weight-birth weight ratios. Eur J Obstet Gynecol Reprod Biol. 2006;128:248–52. Farrar D, Airey R, Law GR, Tuffnell D, Cattle B, Duley L. Measuring placental transfusion for term births: weighing babies with cord intact. BJOG Int J Obstet Gynaecol. 2011;118:70–5. Lundberg C, Oian P, Klingenberg C. Umbilical cord clamping at birth–practice in Norwegian maternity wards. Tidsskr Nor Laegeforen. 2013;133:2369–73. Johnson TS, Engstrom JL, Gelhar DK. Intra- and interexaminer reliability of anthropometric measurements of term infants. J Pediatr Gastroenterol Nutr. 1997;24:497–505. Lagiou P, Hsieh CC, Trichopoulos D, Xu B, Wuu J, Mucci L, et al. Birthweight differences between USA and China and their relevance to breast cancer aetiology. Int J Epidemiol. 2003;32: 193–8. Lagiou P, Hsieh CC, Lipworth L, Samoli E, Okulicz W, Troisi R, et al. Insulin-like growth factor levels in cord blood, birth weight and breast cancer risk. Br J Cancer. 2009;100:1794–8. Lagiou P, Samoli E, Hsieh CC, Lagiou A, Xu B, Yu GP, et al. Maternal and cord blood hormones in relation to birth size. Eur J Epidemiol. 2014;29:343–51. Horn J, Alsaker MD, Opdahl S, Engstrom MJ, Tretli S, Haugen OA, et al. Anthropometric factors and risk of molecular breast cancer subtypes among postmenopausal Norwegian women. Int J Cancer. 2014;135:2678–86. Horn J, Opdahl S, Engstrom MJ, Romundstad PR, Tretli S, Haugen OA, et al. Reproductive history and the risk of molecular breast cancer subtypes in a prospective study of Norwegian women. Cancer Causes Control. 2014;25:881–9. Tamimi RM, Colditz GA, Hazra A, Baer HJ, Hankinson SE, Rosner B, et al. Traditional breast cancer risk factors in relation to molecular subtypes of breast cancer. Breast Cancer Res Treat. 2012;131:159–67. Yang XR, Chang-Claude J, Goode EL, Couch FJ, Nevanlinna H, Milne RL, et al. Associations of breast cancer risk factors with tumor subtypes: a pooled analysis from the Breast Cancer Association Consortium studies. J Natl Cancer Inst. 2011; 103:250–63.
