Eur J Epidemiol (2014) 29:343–351 DOI 10.1007/s10654-014-9914-3
PERINATAL EPIDEMIOLOGY
Maternal and cord blood hormones in relation to birth size Pagona Lagiou • Evangelia Samoli • Chung-Cheng Hsieh • Areti Lagiou • Bio Xu • Guo-Pei Yu • Sagano Onoyama • Lucy Chie • Hans-Olov Adami Lars J. Vatten • Dimitrios Trichopoulos • Michelle A. Williams
•
Received: 12 February 2014 / Accepted: 13 May 2014 / Published online: 22 May 2014 Ó Springer Science+Business Media Dordrecht 2014
Abstract Birth size has been associated with adult life diseases, but the endocrine factors that are likely involved are not established. We evaluated the associations of maternal and cord blood hormones with birth size in normal pregnancies, and examined possible effect modification by maternal height, on the basis of prior suggestive evidence. In a prospective study of normal singleton pregnancies in Boston, USA and Shanghai, China, maternal hormone levels at the 27th gestational week were available for 225 pregnancies in Boston and 281 in Shanghai and cord blood measurements for 92 pregnancies in Boston and 110 in Shanghai. Pearson partial correlation coefficients of logtransformed hormone levels with birth weight and length were calculated. Overall, positive correlations with birth weight were found for maternal estriol (r = 0.19; p \ 0.001)
and progesterone (r = 0.15; p \ 0.001) and these associations were more evident among taller mothers. There was an inverse association of cord blood progesterone (r = -0.16; p \ 0.03) with birth weight. In Boston, cord blood IGF-1 was positively associated with birth weight (r = 0.22; p \ 0.04) and length (r = 0.25; p \ 0.02), particularly among taller mothers (r = 0.43 and 0.38, respectively; p \ 0.02), whereas among taller mothers in Shanghai the associations of IGF-2 with birth size appeared to be at least as strong as those of IGF-1. In conclusion, maternal estriol and progesterone, and cord blood IGF-1 were positively correlated with birth size. All correlations tended to be more pronounced among offspring of taller mothers. Among taller mothers in Shanghai, IGF-2 appeared to be at least as strongly associated with birth size as IGF-1.
P. Lagiou (&) C.-C. Hsieh H.-O. Adami D. Trichopoulos M. A. Williams Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA e-mail: [email protected]
G.-P. Yu Medical Informatics Center, Peking University, Beijing, China
P. Lagiou E. Samoli Department of Hygiene, Epidemiology and Medical Statistics, School of Medicine, University of Athens, Athens, Greece C.-C. Hsieh S. Onoyama Department of Cancer Biology, University of Massachusetts Medical School, Worcester, MA, USA A. Lagiou Department of Public Health and Community Health, Faculty of Health Professions, Athens Technological Educational Institute, Athens, Greece
L. Chie Department of Obstetrics and Gynecology, Beth Israel Deaconess Medical Center, Boston, MA, USA H.-O. Adami Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden L. J. Vatten Department of Community Medicine and General Practice, School of Medicine, Norwegian University of Science and Technology, Trondheim, Norway
B. Xu Department of Epidemiology, School of Public Health, Fudan University, Shanghai, China
123
344
Keywords Birth weight Birth length Insulin like growth factor Estriol Progesterone Pregnancy Maternal height Cord blood
Introduction The hypothesis that adult life chronic diseases may have intrauterine roots has been explored by several investigators during the last decades [1, 2]. Specifically, compromised fetal growth has been linked to several high risk factors for cardiovascular diseases [3, 4], whereas bigger birth size has been linked to breast cancer [5] and possibly other adult life malignancies [6]. However, notwithstanding important efforts in the field, it has not yet been possible to firmly document associations of birth weight and length with levels of maternal or cord blood hormones that may underlie associations between birth size and risk for adult disease. In particular, no attempt has been made to quantify the associations of maternal and cord blood hormones with birth size and, thus, gain an insight on possible intrauterine endocrine roots that may affect adult life chronic disease risk, which is known to differ between Caucasian and Asian populations. We have collected maternal and cord blood samples, as well as data on birth size parameters, from medically uncomplicated pregnancies in Boston, US and Shanghai, China [7]. Preliminary analyses concerning certain maternal blood hormones in relation to birth size, mainly in Boston, have been previously considered [8–10]. Regarding cord blood hormones in relation to birth weight, only the insulin-like growth factor (IGF) axis has been previously studied [11]. In this report, we examine the association of birth weight and length with the levels of steroid hormones (estradiol, estriol, testosterone, progesterone), as well as sex hormone binding globulin (SHBG), prolactin, adiponectin and hormones of the IGF system [IGF-I, IGF-II and the IGF-binding protein 3 (IGFBP-3)] in maternal sera collected at the 27th week of gestation and in the umbilical cord blood. The hormones measured were those most frequently studied in adults in relation to breast cancer and other malignancies. In this report, we also examine whether the associations of the measured hormones with birth weight and length differ by maternal height, as hypothesized [12] on the basis of empirical evidence [13, 14].
Subjects and methods The study relies on data collected in 1994–1995 from healthy singleton pregnancies in Boston, USA and
123
P. Lagiou et al.
Shanghai, China. Details on the study have been previously published [7, 15, 16]. Briefly, pregnant women were recruited during their first routine prenatal visit to the collaborating maternity clinics of the Beth Israel hospital in Boston, USA and the Shanghai Medical University (currently the Fudan University) in China. A total of 402 Caucasian women in Boston, USA and 424 Asian women in Shanghai, China were identified. The study was approved by the Institutional Review Boards of the Beth Israel Hospital, Shanghai Medical University and Harvard School of Public Health. Eligibility criteria included age less than 40 years, a maximum parity of two, absence of a prior diagnosis of diabetes mellitus or thyroid disease, no hormonal medication during the index pregnancy and no known fetal abnormality. Of the 402 eligible women in Boston, exclusions were made on account of refusals (response proportion 80.8 %), pregnancy interruptions, twin pregnancies, gestational age below 37 or above 42 weeks, pregnancy toxemia, missing covariates, or losses to follow up (only 10 women), leaving 241 women. Additional exclusions were necessary due to limited availability of biological samples for some pregnancies [15]. Eventually, associations of birth size with maternal serum hormone levels were examined for 225 women; for associations with cord blood hormone levels, samples were available for 92 women. Of the 424 eligible women in Shanghai, exclusions were made for the same reasons as for in women in Boston. The response rate was 96.5 % and there were only seven loses to follow up. Eventually, 295 Asian women were included in the study. Additional exclusions, however, were necessary due to limited availability of biological samples for some pregnancies [15]. Associations of birth size with maternal serum hormone levels were examined for 281 women; for associations with cord blood hormone levels, samples were available for 110 women. Hormone measurements During the visit to the maternity clinics at the 27th week of gestation, 10 ml of venous blood was drawn from every woman. At delivery, cord blood was also collected, as indicated. Details on blood draw, transportation and storage have been previously published [16]. Levels of estradiol (E2), estriol (E3), progesterone, sex hormone binding globulin (SHBG), testosterone, adiponectin, insulin-like growth factor I (IGF-I) and insulin-like growth factor binding protein 3 (IGFBP3) were measured in both maternal sera and cord blood. Prolactin was measured only in maternal sera, and insulin-like growth factor II (IGF-II) only in cord blood. Measurements were conducted in two time periods. Maternal levels of E2, E3, SHBG, progesterone, and prolactin were measured at the
Maternal and cord blood hormones
Department of Clinical Chemistry of the Uppsala University Hospital in Sweden in the late 1990s [7], whereas maternal levels of testosterone, adiponectin, IGF-I and IGFBP3, as well as all hormone levels in cord blood, were measured in 2006 at the ILAT Steroid RIA Laboratory of the University of Massachusetts Medical School [15]. Measurements for each hormone were conducted simultaneously for samples from Boston and Shanghai. Maternal estradiol-17b was measured with a timeresolved competitive solid-phase fluoroimmunoassay (AutoDELFIA estradiol kit; WallacOy, Turku, Finland). Maternal unconjugated E3 was measured with a similar time-resolved competitive solid-phase fluoroimmunoassay method (AutoDELFIA unconjugated oestriol kit; WallacOy). Maternal SHBG was measured with a time-resolved non-competitive solid-phase sandwich fluoroimmunoassay (AutoDELFIA SHBG kit; WallacOy). Maternal progesterone was measured with a time-resolved competitive solid-phase fluoroimmunoassay (AutoDELFIA Progesterone kit; WallacOy). Maternal prolactin was measured with a time-resolved non-competitive solid phase sandwich fluoroimmunoassay (AutoDELFIA prolactin kit; WallacOy). Maternal and cord blood testosterone was measured by radioimmunoassay kits from DPC (Diagnostic Products Corporation, Los Angeles, CA). Maternal and cord blood adiponectin was measured by radioimmunoassay (Linco Research, St. Charles, MO). Maternal and cord blood IGF-I and IGFBP-3, as well as cord blood IGF-II were measured by coated-tube immunoradiometric assay kits (Diagnostic System Laboratories, Inc., Webster, TX). There was no detectable cross-reactivity of the IGF-I assay with IGF-II according to the manufacturer’s specificity assessment. Cord blood E2 was measured by radioimmunoassay using kits from DPC (Diagnostic Products Corporation, Los Angeles, CA). Cord blood unconjugated E3, progesterone, and SHBG were measured using chemiluminescent immunoassay methodologies from DPC (Los Angeles, CA). Statistical analyses The steps to reach the study samples eventually analyzed in the Boston and the Shanghai centers are described in the Methods section. There were 45 women in Boston and 97 in Shanghai with missing information in one or more of the covariates (percentages 14.8 and 22.9 % of eligible women, respectively). We opted to analyze the clean dataset, as imputations would require assumptions that would be difficult to evaluate. We examined the distributions of Caucasian and Chinese women by maternal–offspring characteristics and estimated the quartiles of the endocrine compounds measured in maternal sera at the 27th gestational week and in cord blood. To investigate the
345
associations of birth weight and length with the log-transformed values of maternal/cord blood hormone levels, we used Pearson partial correlation coefficients. All coefficients were adjusted for maternal age, maternal height, body mass index (BMI) before pregnancy and maternal weight gain till the 27th gestational week, as well as parity, duration of gestation and gender of offspring. Analyses for the two centers combined were also controlled for center. Between-center heterogeneity was estimated by introducing an interaction term between birth weight or length and center in a multivariate model for the association between the indicated hormone and birth weight or length. Analyses were also conducted separately within each center, for taller and shorter women, with the cut-off for height set at 163 cm (median height in Boston, third quartile in Shanghai). Statistical significance level was set at p \ 0.05 for all analyses performed. The statistical analyses were conducted using the SPSS statistical package (Statistical Package for Social Sciences v. 21.0, Chicago, Illinois, USA).
Results Table 1 shows the distribution of women and their singleton offspring in the two centers by maternal, gestational and neonatal characteristics. Both birth weight and birth length are higher among newborns in Boston. Table 2 presents maternal serum levels at the 27th gestational week and umbilical cord blood levels of the studied hormones in Boston and Shanghai. In both centers, steroid hormone and adiponectin levels were substantially higher in cord blood compared to maternal blood, whereas the opposite was true with respect to IGF-I, IGFBP-3 and SHBG. Correlations of the maternal hormone levels with the cord blood hormone levels have been previously reported and they were generally poor [15]. In Table 3, Pearson partial correlation coefficients of log-transformed levels of maternal blood hormones at the 27th gestational week with birth weight (upper panel) and birth length (lower panel) are presented. Given the multitude of comparisons undertaken, strong evidence for heterogeneity between the two centers exists only with respect to estriol in relation to birth length (p for heterogeneity *0.006) and to a lesser extent for prolactin with respect to birth weight (p for heterogeneity *0.036). For both centers combined, there is evidence for a positive correlation of birth weight with estriol and progesterone and of birth length with progesterone alone. All these positive correlations appear to be driven by the results in Boston. Over both centers, the correlation of maternal IGF-I tends to be negative with birth length (r = -0.084, p * 0.066), but is essentially null with birth weight.
123
346
P. Lagiou et al.
Table 1 Characteristics of womena and their singleton offspring in Boston, USA and Shanghai, China (A) Age (years)
Boston (n = 225)
Shanghai (n = 281)
31.0 (0.3)
25.3 (0.2)
131 (58.2)
276 (98.2)
Parity 1 2 Duration of gestation (weeks)
94 (41.8)
5 (1.8)
40.0 (0.1)
40.0 (0.1)
Gestational week at blood draw
27.1 (0.1)
26.9 (0.1)
Maternal weight gain until the 27th gestational week (kg)
11.5 (0.3)
8.6 (0.3)
164.4 (0.4)
160.2 (0.3)
22.0 (0.2)
19.9 (0.1)
Maternal height (cm) Pre-pregnancy body mass index (BMI, kg/m2) Gender of offspring Male
119 (52.9)
169 (60.1)
Female
106 (47.1)
112 (39.9)
3,579.1 (32.3)
3,423.2 (26.0)
Birth weight (g) Birth length (cm) (B) Age (years)
50.6 (0.2) Boston (n = 92) 31.0 (0.3)
49.9 (0.2) Shanghai (n = 110) 25.1 (0.3)
Parity 1
48 (52.2)
2
44 (47.8)
Duration of gestation (weeks)
109 (99.1) 1 (0.9)
40.1 (0.1)
40.0 (0.1)
Gestational week at blood draw
27.3 (0.2)
26.9 (0.1)
Maternal weight gain until the 27th gestational week (kg)
11.5 (0.4)
8.9 (0.4)
164.1 (0.8) 22.0 (0.3)
160.2 (0.5) 19.9 (0.2)
Maternal height (cm) Pre-pregnancy body mass index (BMI, kg/m2) Gender of offspring Male
46 (50.0)
Female
46 (50.0)
46 (41.8)
3,557.7 (51.1)
3,492.6 (43.8)
Birth weight (g) Birth length (cm)
50.6 (0.3)
64 (58.2)
49.8 (0.3)
Upper panel (A): women with available maternal hormone measurements. Lower panel (B): women with available cord blood measurements For continuous variables mean (standard error); for categorical variables n (%). a Women with pregnancy duration 37 to 42 weeks, inclusive, no pregnancy toxemia and measured offspring birth weight and length
In Table 4, Pearson partial correlation coefficients of log-transformed levels of umbilical cord blood hormones with birth weight (upper panel) and birth length (lower panel) are presented. There is no statistically significant evidence for heterogeneity between the two centers. Cord blood progesterone is inversely correlated with birth weight, significantly so over both centers (r = -0.16, p * 0.031), but it is essentially uncorrelated with birth length. In Boston, but not in Shanghai, cord blood IGF-1 is significantly positively correlated with both birth weight (r = 0.22, p * 0.043) and birth length (r = 0.25, p * 0.020). In Table 5, Pearson partial correlation coefficients of log-transformed levels of maternal blood hormones at the 27th gestational week with birth weight (upper panel) and
123
birth length (lower panel) are presented, separately for mothers of shorter (163 cm or less) and higher (more than 163 cm) stature. The positive correlation of both maternal estriol and maternal progesterone with birth weight tended to be stronger among infants of taller mothers in both centers and statistically significant among taller women in Boston; p values for interaction, however, did not reach statistical significance at the 0.05 level. Maternal estriol tended to be positively correlated with birth length among taller women, albeit not significantly so, but there was also a positive correlation of estradiol with birth length among taller women in both Boston and Shanghai which, however, reached statistical significance in the latter center only. In Table 6, Pearson partial correlation coefficients of log-transformed levels of umbilical cord blood hormones
Maternal and cord blood hormones
with birth weight (upper panel) and birth length (lower panel) are presented, separately for mothers of shorter (163 cm or less) and taller (more than 163 cm) stature. With respect to birth weight, there was a strong and statistically significant positive correlation with cord blood IGF-1 among taller women in Boston (r = 0.43, p * 0.009)—the respective association was null among shorter women in Boston (p for interaction 0.246), as well as all women in Shanghai. For IGF-2, there was a strong [albeit not statistically significant (p * 0.066) in this fairly small group of 29 pregnancies] positive correlation with birth weight among taller women in Shanghai, whereas the respective correlation was essentially null among shorter women in Shanghai (p for interaction 0.532), as well as all women in Boston. Regarding birth length, there was again evidence that the positive correlations of IGF-1 (in both centers, r = 0.38, p * 0.021 in Boston and r = 0.53, p * 0.013 in Shanghai) and IGF-2 (in Shanghai only, r = 0.37, p * 0.095) were stronger among taller women; the p values for interaction of maternal height with IGF-1 in Boston and in Shanghai were, respectively, 0.033 and 0.057. Significant inverse associations in Boston only between birth length and SHBG among shorter women, and birth length and adiponectin among taller women, are likely due to chance, since there were no corroborating findings in other tabulations.
347
correlations with birth size were found for estriol and progesterone and their associations were more evident among taller mothers (Tables 3, 5). With respect to the measured cord blood hormones, there was an inverse association of progesterone with birth weight and the inverse association was more pronounced among taller mothers in Shanghai. Cord blood IGF-1 was positively associated with birth weight and length in Boston, particularly among taller mothers, whereas among taller mothers in Shanghai IGF-2 appears to play at least as important a role as IGF-1 on birth size, with correlation coefficients around 0.4 for both birth weight and birth length (p \ 0.10 in both instances, even though there were only 29 observations). Since infants with low and high birth weights are at increased risk of disease in later life [1–6], it would have been interesting to assess the associations of hormones among the low and the high birth weight infants, but our study had limited power for such an analysis.
Table 2 Maternal serum levels of the indicated hormones at the 27th gestational week in Boston, USA (n = 225) and Shanghai, China (n = 281) and cord blood levels in Boston, USA (n = 92) and Shanghai, China (n = 110) Maternal serum levels Median
IQR
a
Cord blood levels Median
IQRa
Boston Estradiol (ng/ml)
Discussion In sizeable cohorts of pregnant women in Boston, USA and Shanghai, China, we have measured maternal blood hormones at the 27th gestational week and cord blood hormones at delivery, and we have correlated the logtransformed values of the measured hormones with birth weight and birth length. The overall aim was to identify which of the measured hormones are linked to birth size. The focus on a Caucasian and a native Chinese cohort was motivated from the ecological contrast in chronic disease patterns in the two populations, notably with respect to breast cancer, and the possibility that these contrasts may be reflected also in perinatal endocrine differences. In interpreting the results, we took into account the magnitude of the absolute values of the respective correlation coefficients and the corresponding p values. We have assumed that the results concerning birth weight are somewhat more informative than those concerning birth length (birth weight is generally more accurately measured than birth length). In general, there was little heterogeneity of the results between the two cohorts; among the 36 calculated p values for heterogeneity (Tables 3, 4), two were ‘‘statistically significant’’, close to the 5 % expected by chance. With respect to the measured maternal hormones, positive
10.0
5.3
30.0
21.9
Estriol (ng/ml)
3.9
1.6
360.5
188.5
SHBG (nmol/l)
420.2
151.2
31.6
12.6
Prolactin (lg/l)
89.9
45.9
Progesterone (ng/ml)
78.8
29.8
878.0
750.2
Testosterone (ng/ml) Adiponectin (lg/ml)
0.6 14.0
0.4 9.0
2.6 52.5
1.3 29.1
231.5
142.8
IGF-I (ng/ml) IGF-II (ng/ml) IGFBP3 (ng/ml)
6,878.0
4,498.5
94.5
51.5
499.0
123.3
1,992.0
1,969.3
Shanghai Estradiol (ng/ml)
12.2
5.4
61.3
42.8
5.9
2.6
395.0
172.0
SHBG (nmol/l)
454.3
154.4
51.7
142.2
Prolactin (lg/l)
112.1
48.0
Progesterone (ng/ml)
74.1
23.7
875.5
712.3
Testosterone (ng/ml)
0.7
0.4
4.5
3.5
Adiponectin (lg/ml)
10.9
5.7
34.9
35.9
194.0
120.0
Estriol (ng/ml)
IGF-I (ng/ml) IGF-II (ng/ml) IGFBP3 (ng/ml)
6,182.0
4,495.0
73.0
51.7
587.0
140.7
2,532.0
2,303.0
a
IQR, interquartile range as the difference between the 75th percentile and the 25th percentile SHBG: sex hormone binding globulin; IGF-I, insulin-like growth factor 1; IGF-II, insulin-like growth factor 2; IGFBP3, insulin-like growth factor binding protein 3
123
348
P. Lagiou et al.
Table 3 Pearson partial correlation coefficients of maternal blood hormones at the 27th gestational week (log-transformed) with birth weight (upper panel) and birth length (lower panel) Boston (n = 225)
Shanghai (n = 281)
r
r
p-value
p for heterogeneity
p-value
Both centersa r
p-value
Estradiol
0.06
0.351
0.06
0.329
0.860
0.07
0.108
Estriol
0.27
PERINATAL EPIDEMIOLOGY
Maternal and cord blood hormones in relation to birth size Pagona Lagiou • Evangelia Samoli • Chung-Cheng Hsieh • Areti Lagiou • Bio Xu • Guo-Pei Yu • Sagano Onoyama • Lucy Chie • Hans-Olov Adami Lars J. Vatten • Dimitrios Trichopoulos • Michelle A. Williams
•
Received: 12 February 2014 / Accepted: 13 May 2014 / Published online: 22 May 2014 Ó Springer Science+Business Media Dordrecht 2014
Abstract Birth size has been associated with adult life diseases, but the endocrine factors that are likely involved are not established. We evaluated the associations of maternal and cord blood hormones with birth size in normal pregnancies, and examined possible effect modification by maternal height, on the basis of prior suggestive evidence. In a prospective study of normal singleton pregnancies in Boston, USA and Shanghai, China, maternal hormone levels at the 27th gestational week were available for 225 pregnancies in Boston and 281 in Shanghai and cord blood measurements for 92 pregnancies in Boston and 110 in Shanghai. Pearson partial correlation coefficients of logtransformed hormone levels with birth weight and length were calculated. Overall, positive correlations with birth weight were found for maternal estriol (r = 0.19; p \ 0.001)
and progesterone (r = 0.15; p \ 0.001) and these associations were more evident among taller mothers. There was an inverse association of cord blood progesterone (r = -0.16; p \ 0.03) with birth weight. In Boston, cord blood IGF-1 was positively associated with birth weight (r = 0.22; p \ 0.04) and length (r = 0.25; p \ 0.02), particularly among taller mothers (r = 0.43 and 0.38, respectively; p \ 0.02), whereas among taller mothers in Shanghai the associations of IGF-2 with birth size appeared to be at least as strong as those of IGF-1. In conclusion, maternal estriol and progesterone, and cord blood IGF-1 were positively correlated with birth size. All correlations tended to be more pronounced among offspring of taller mothers. Among taller mothers in Shanghai, IGF-2 appeared to be at least as strongly associated with birth size as IGF-1.
P. Lagiou (&) C.-C. Hsieh H.-O. Adami D. Trichopoulos M. A. Williams Department of Epidemiology, Harvard School of Public Health, 677 Huntington Avenue, Boston, MA 02115, USA e-mail: [email protected]
G.-P. Yu Medical Informatics Center, Peking University, Beijing, China
P. Lagiou E. Samoli Department of Hygiene, Epidemiology and Medical Statistics, School of Medicine, University of Athens, Athens, Greece C.-C. Hsieh S. Onoyama Department of Cancer Biology, University of Massachusetts Medical School, Worcester, MA, USA A. Lagiou Department of Public Health and Community Health, Faculty of Health Professions, Athens Technological Educational Institute, Athens, Greece
L. Chie Department of Obstetrics and Gynecology, Beth Israel Deaconess Medical Center, Boston, MA, USA H.-O. Adami Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden L. J. Vatten Department of Community Medicine and General Practice, School of Medicine, Norwegian University of Science and Technology, Trondheim, Norway
B. Xu Department of Epidemiology, School of Public Health, Fudan University, Shanghai, China
123
344
Keywords Birth weight Birth length Insulin like growth factor Estriol Progesterone Pregnancy Maternal height Cord blood
Introduction The hypothesis that adult life chronic diseases may have intrauterine roots has been explored by several investigators during the last decades [1, 2]. Specifically, compromised fetal growth has been linked to several high risk factors for cardiovascular diseases [3, 4], whereas bigger birth size has been linked to breast cancer [5] and possibly other adult life malignancies [6]. However, notwithstanding important efforts in the field, it has not yet been possible to firmly document associations of birth weight and length with levels of maternal or cord blood hormones that may underlie associations between birth size and risk for adult disease. In particular, no attempt has been made to quantify the associations of maternal and cord blood hormones with birth size and, thus, gain an insight on possible intrauterine endocrine roots that may affect adult life chronic disease risk, which is known to differ between Caucasian and Asian populations. We have collected maternal and cord blood samples, as well as data on birth size parameters, from medically uncomplicated pregnancies in Boston, US and Shanghai, China [7]. Preliminary analyses concerning certain maternal blood hormones in relation to birth size, mainly in Boston, have been previously considered [8–10]. Regarding cord blood hormones in relation to birth weight, only the insulin-like growth factor (IGF) axis has been previously studied [11]. In this report, we examine the association of birth weight and length with the levels of steroid hormones (estradiol, estriol, testosterone, progesterone), as well as sex hormone binding globulin (SHBG), prolactin, adiponectin and hormones of the IGF system [IGF-I, IGF-II and the IGF-binding protein 3 (IGFBP-3)] in maternal sera collected at the 27th week of gestation and in the umbilical cord blood. The hormones measured were those most frequently studied in adults in relation to breast cancer and other malignancies. In this report, we also examine whether the associations of the measured hormones with birth weight and length differ by maternal height, as hypothesized [12] on the basis of empirical evidence [13, 14].
Subjects and methods The study relies on data collected in 1994–1995 from healthy singleton pregnancies in Boston, USA and
123
P. Lagiou et al.
Shanghai, China. Details on the study have been previously published [7, 15, 16]. Briefly, pregnant women were recruited during their first routine prenatal visit to the collaborating maternity clinics of the Beth Israel hospital in Boston, USA and the Shanghai Medical University (currently the Fudan University) in China. A total of 402 Caucasian women in Boston, USA and 424 Asian women in Shanghai, China were identified. The study was approved by the Institutional Review Boards of the Beth Israel Hospital, Shanghai Medical University and Harvard School of Public Health. Eligibility criteria included age less than 40 years, a maximum parity of two, absence of a prior diagnosis of diabetes mellitus or thyroid disease, no hormonal medication during the index pregnancy and no known fetal abnormality. Of the 402 eligible women in Boston, exclusions were made on account of refusals (response proportion 80.8 %), pregnancy interruptions, twin pregnancies, gestational age below 37 or above 42 weeks, pregnancy toxemia, missing covariates, or losses to follow up (only 10 women), leaving 241 women. Additional exclusions were necessary due to limited availability of biological samples for some pregnancies [15]. Eventually, associations of birth size with maternal serum hormone levels were examined for 225 women; for associations with cord blood hormone levels, samples were available for 92 women. Of the 424 eligible women in Shanghai, exclusions were made for the same reasons as for in women in Boston. The response rate was 96.5 % and there were only seven loses to follow up. Eventually, 295 Asian women were included in the study. Additional exclusions, however, were necessary due to limited availability of biological samples for some pregnancies [15]. Associations of birth size with maternal serum hormone levels were examined for 281 women; for associations with cord blood hormone levels, samples were available for 110 women. Hormone measurements During the visit to the maternity clinics at the 27th week of gestation, 10 ml of venous blood was drawn from every woman. At delivery, cord blood was also collected, as indicated. Details on blood draw, transportation and storage have been previously published [16]. Levels of estradiol (E2), estriol (E3), progesterone, sex hormone binding globulin (SHBG), testosterone, adiponectin, insulin-like growth factor I (IGF-I) and insulin-like growth factor binding protein 3 (IGFBP3) were measured in both maternal sera and cord blood. Prolactin was measured only in maternal sera, and insulin-like growth factor II (IGF-II) only in cord blood. Measurements were conducted in two time periods. Maternal levels of E2, E3, SHBG, progesterone, and prolactin were measured at the
Maternal and cord blood hormones
Department of Clinical Chemistry of the Uppsala University Hospital in Sweden in the late 1990s [7], whereas maternal levels of testosterone, adiponectin, IGF-I and IGFBP3, as well as all hormone levels in cord blood, were measured in 2006 at the ILAT Steroid RIA Laboratory of the University of Massachusetts Medical School [15]. Measurements for each hormone were conducted simultaneously for samples from Boston and Shanghai. Maternal estradiol-17b was measured with a timeresolved competitive solid-phase fluoroimmunoassay (AutoDELFIA estradiol kit; WallacOy, Turku, Finland). Maternal unconjugated E3 was measured with a similar time-resolved competitive solid-phase fluoroimmunoassay method (AutoDELFIA unconjugated oestriol kit; WallacOy). Maternal SHBG was measured with a time-resolved non-competitive solid-phase sandwich fluoroimmunoassay (AutoDELFIA SHBG kit; WallacOy). Maternal progesterone was measured with a time-resolved competitive solid-phase fluoroimmunoassay (AutoDELFIA Progesterone kit; WallacOy). Maternal prolactin was measured with a time-resolved non-competitive solid phase sandwich fluoroimmunoassay (AutoDELFIA prolactin kit; WallacOy). Maternal and cord blood testosterone was measured by radioimmunoassay kits from DPC (Diagnostic Products Corporation, Los Angeles, CA). Maternal and cord blood adiponectin was measured by radioimmunoassay (Linco Research, St. Charles, MO). Maternal and cord blood IGF-I and IGFBP-3, as well as cord blood IGF-II were measured by coated-tube immunoradiometric assay kits (Diagnostic System Laboratories, Inc., Webster, TX). There was no detectable cross-reactivity of the IGF-I assay with IGF-II according to the manufacturer’s specificity assessment. Cord blood E2 was measured by radioimmunoassay using kits from DPC (Diagnostic Products Corporation, Los Angeles, CA). Cord blood unconjugated E3, progesterone, and SHBG were measured using chemiluminescent immunoassay methodologies from DPC (Los Angeles, CA). Statistical analyses The steps to reach the study samples eventually analyzed in the Boston and the Shanghai centers are described in the Methods section. There were 45 women in Boston and 97 in Shanghai with missing information in one or more of the covariates (percentages 14.8 and 22.9 % of eligible women, respectively). We opted to analyze the clean dataset, as imputations would require assumptions that would be difficult to evaluate. We examined the distributions of Caucasian and Chinese women by maternal–offspring characteristics and estimated the quartiles of the endocrine compounds measured in maternal sera at the 27th gestational week and in cord blood. To investigate the
345
associations of birth weight and length with the log-transformed values of maternal/cord blood hormone levels, we used Pearson partial correlation coefficients. All coefficients were adjusted for maternal age, maternal height, body mass index (BMI) before pregnancy and maternal weight gain till the 27th gestational week, as well as parity, duration of gestation and gender of offspring. Analyses for the two centers combined were also controlled for center. Between-center heterogeneity was estimated by introducing an interaction term between birth weight or length and center in a multivariate model for the association between the indicated hormone and birth weight or length. Analyses were also conducted separately within each center, for taller and shorter women, with the cut-off for height set at 163 cm (median height in Boston, third quartile in Shanghai). Statistical significance level was set at p \ 0.05 for all analyses performed. The statistical analyses were conducted using the SPSS statistical package (Statistical Package for Social Sciences v. 21.0, Chicago, Illinois, USA).
Results Table 1 shows the distribution of women and their singleton offspring in the two centers by maternal, gestational and neonatal characteristics. Both birth weight and birth length are higher among newborns in Boston. Table 2 presents maternal serum levels at the 27th gestational week and umbilical cord blood levels of the studied hormones in Boston and Shanghai. In both centers, steroid hormone and adiponectin levels were substantially higher in cord blood compared to maternal blood, whereas the opposite was true with respect to IGF-I, IGFBP-3 and SHBG. Correlations of the maternal hormone levels with the cord blood hormone levels have been previously reported and they were generally poor [15]. In Table 3, Pearson partial correlation coefficients of log-transformed levels of maternal blood hormones at the 27th gestational week with birth weight (upper panel) and birth length (lower panel) are presented. Given the multitude of comparisons undertaken, strong evidence for heterogeneity between the two centers exists only with respect to estriol in relation to birth length (p for heterogeneity *0.006) and to a lesser extent for prolactin with respect to birth weight (p for heterogeneity *0.036). For both centers combined, there is evidence for a positive correlation of birth weight with estriol and progesterone and of birth length with progesterone alone. All these positive correlations appear to be driven by the results in Boston. Over both centers, the correlation of maternal IGF-I tends to be negative with birth length (r = -0.084, p * 0.066), but is essentially null with birth weight.
123
346
P. Lagiou et al.
Table 1 Characteristics of womena and their singleton offspring in Boston, USA and Shanghai, China (A) Age (years)
Boston (n = 225)
Shanghai (n = 281)
31.0 (0.3)
25.3 (0.2)
131 (58.2)
276 (98.2)
Parity 1 2 Duration of gestation (weeks)
94 (41.8)
5 (1.8)
40.0 (0.1)
40.0 (0.1)
Gestational week at blood draw
27.1 (0.1)
26.9 (0.1)
Maternal weight gain until the 27th gestational week (kg)
11.5 (0.3)
8.6 (0.3)
164.4 (0.4)
160.2 (0.3)
22.0 (0.2)
19.9 (0.1)
Maternal height (cm) Pre-pregnancy body mass index (BMI, kg/m2) Gender of offspring Male
119 (52.9)
169 (60.1)
Female
106 (47.1)
112 (39.9)
3,579.1 (32.3)
3,423.2 (26.0)
Birth weight (g) Birth length (cm) (B) Age (years)
50.6 (0.2) Boston (n = 92) 31.0 (0.3)
49.9 (0.2) Shanghai (n = 110) 25.1 (0.3)
Parity 1
48 (52.2)
2
44 (47.8)
Duration of gestation (weeks)
109 (99.1) 1 (0.9)
40.1 (0.1)
40.0 (0.1)
Gestational week at blood draw
27.3 (0.2)
26.9 (0.1)
Maternal weight gain until the 27th gestational week (kg)
11.5 (0.4)
8.9 (0.4)
164.1 (0.8) 22.0 (0.3)
160.2 (0.5) 19.9 (0.2)
Maternal height (cm) Pre-pregnancy body mass index (BMI, kg/m2) Gender of offspring Male
46 (50.0)
Female
46 (50.0)
46 (41.8)
3,557.7 (51.1)
3,492.6 (43.8)
Birth weight (g) Birth length (cm)
50.6 (0.3)
64 (58.2)
49.8 (0.3)
Upper panel (A): women with available maternal hormone measurements. Lower panel (B): women with available cord blood measurements For continuous variables mean (standard error); for categorical variables n (%). a Women with pregnancy duration 37 to 42 weeks, inclusive, no pregnancy toxemia and measured offspring birth weight and length
In Table 4, Pearson partial correlation coefficients of log-transformed levels of umbilical cord blood hormones with birth weight (upper panel) and birth length (lower panel) are presented. There is no statistically significant evidence for heterogeneity between the two centers. Cord blood progesterone is inversely correlated with birth weight, significantly so over both centers (r = -0.16, p * 0.031), but it is essentially uncorrelated with birth length. In Boston, but not in Shanghai, cord blood IGF-1 is significantly positively correlated with both birth weight (r = 0.22, p * 0.043) and birth length (r = 0.25, p * 0.020). In Table 5, Pearson partial correlation coefficients of log-transformed levels of maternal blood hormones at the 27th gestational week with birth weight (upper panel) and
123
birth length (lower panel) are presented, separately for mothers of shorter (163 cm or less) and higher (more than 163 cm) stature. The positive correlation of both maternal estriol and maternal progesterone with birth weight tended to be stronger among infants of taller mothers in both centers and statistically significant among taller women in Boston; p values for interaction, however, did not reach statistical significance at the 0.05 level. Maternal estriol tended to be positively correlated with birth length among taller women, albeit not significantly so, but there was also a positive correlation of estradiol with birth length among taller women in both Boston and Shanghai which, however, reached statistical significance in the latter center only. In Table 6, Pearson partial correlation coefficients of log-transformed levels of umbilical cord blood hormones
Maternal and cord blood hormones
with birth weight (upper panel) and birth length (lower panel) are presented, separately for mothers of shorter (163 cm or less) and taller (more than 163 cm) stature. With respect to birth weight, there was a strong and statistically significant positive correlation with cord blood IGF-1 among taller women in Boston (r = 0.43, p * 0.009)—the respective association was null among shorter women in Boston (p for interaction 0.246), as well as all women in Shanghai. For IGF-2, there was a strong [albeit not statistically significant (p * 0.066) in this fairly small group of 29 pregnancies] positive correlation with birth weight among taller women in Shanghai, whereas the respective correlation was essentially null among shorter women in Shanghai (p for interaction 0.532), as well as all women in Boston. Regarding birth length, there was again evidence that the positive correlations of IGF-1 (in both centers, r = 0.38, p * 0.021 in Boston and r = 0.53, p * 0.013 in Shanghai) and IGF-2 (in Shanghai only, r = 0.37, p * 0.095) were stronger among taller women; the p values for interaction of maternal height with IGF-1 in Boston and in Shanghai were, respectively, 0.033 and 0.057. Significant inverse associations in Boston only between birth length and SHBG among shorter women, and birth length and adiponectin among taller women, are likely due to chance, since there were no corroborating findings in other tabulations.
347
correlations with birth size were found for estriol and progesterone and their associations were more evident among taller mothers (Tables 3, 5). With respect to the measured cord blood hormones, there was an inverse association of progesterone with birth weight and the inverse association was more pronounced among taller mothers in Shanghai. Cord blood IGF-1 was positively associated with birth weight and length in Boston, particularly among taller mothers, whereas among taller mothers in Shanghai IGF-2 appears to play at least as important a role as IGF-1 on birth size, with correlation coefficients around 0.4 for both birth weight and birth length (p \ 0.10 in both instances, even though there were only 29 observations). Since infants with low and high birth weights are at increased risk of disease in later life [1–6], it would have been interesting to assess the associations of hormones among the low and the high birth weight infants, but our study had limited power for such an analysis.
Table 2 Maternal serum levels of the indicated hormones at the 27th gestational week in Boston, USA (n = 225) and Shanghai, China (n = 281) and cord blood levels in Boston, USA (n = 92) and Shanghai, China (n = 110) Maternal serum levels Median
IQR
a
Cord blood levels Median
IQRa
Boston Estradiol (ng/ml)
Discussion In sizeable cohorts of pregnant women in Boston, USA and Shanghai, China, we have measured maternal blood hormones at the 27th gestational week and cord blood hormones at delivery, and we have correlated the logtransformed values of the measured hormones with birth weight and birth length. The overall aim was to identify which of the measured hormones are linked to birth size. The focus on a Caucasian and a native Chinese cohort was motivated from the ecological contrast in chronic disease patterns in the two populations, notably with respect to breast cancer, and the possibility that these contrasts may be reflected also in perinatal endocrine differences. In interpreting the results, we took into account the magnitude of the absolute values of the respective correlation coefficients and the corresponding p values. We have assumed that the results concerning birth weight are somewhat more informative than those concerning birth length (birth weight is generally more accurately measured than birth length). In general, there was little heterogeneity of the results between the two cohorts; among the 36 calculated p values for heterogeneity (Tables 3, 4), two were ‘‘statistically significant’’, close to the 5 % expected by chance. With respect to the measured maternal hormones, positive
10.0
5.3
30.0
21.9
Estriol (ng/ml)
3.9
1.6
360.5
188.5
SHBG (nmol/l)
420.2
151.2
31.6
12.6
Prolactin (lg/l)
89.9
45.9
Progesterone (ng/ml)
78.8
29.8
878.0
750.2
Testosterone (ng/ml) Adiponectin (lg/ml)
0.6 14.0
0.4 9.0
2.6 52.5
1.3 29.1
231.5
142.8
IGF-I (ng/ml) IGF-II (ng/ml) IGFBP3 (ng/ml)
6,878.0
4,498.5
94.5
51.5
499.0
123.3
1,992.0
1,969.3
Shanghai Estradiol (ng/ml)
12.2
5.4
61.3
42.8
5.9
2.6
395.0
172.0
SHBG (nmol/l)
454.3
154.4
51.7
142.2
Prolactin (lg/l)
112.1
48.0
Progesterone (ng/ml)
74.1
23.7
875.5
712.3
Testosterone (ng/ml)
0.7
0.4
4.5
3.5
Adiponectin (lg/ml)
10.9
5.7
34.9
35.9
194.0
120.0
Estriol (ng/ml)
IGF-I (ng/ml) IGF-II (ng/ml) IGFBP3 (ng/ml)
6,182.0
4,495.0
73.0
51.7
587.0
140.7
2,532.0
2,303.0
a
IQR, interquartile range as the difference between the 75th percentile and the 25th percentile SHBG: sex hormone binding globulin; IGF-I, insulin-like growth factor 1; IGF-II, insulin-like growth factor 2; IGFBP3, insulin-like growth factor binding protein 3
123
348
P. Lagiou et al.
Table 3 Pearson partial correlation coefficients of maternal blood hormones at the 27th gestational week (log-transformed) with birth weight (upper panel) and birth length (lower panel) Boston (n = 225)
Shanghai (n = 281)
r
r
p-value
p for heterogeneity
p-value
Both centersa r
p-value
Estradiol
0.06
0.351
0.06
0.329
0.860
0.07
0.108
Estriol
0.27
