点击显示 收起
1 From Cincinnati Children's Hospital Medical Center (LJM, JGW, SRG, MA, BSD, WB, LMD, and ALM) and the Department of Pediatrics, University of Cincinnati College of Medicine (LJM, JGW, SRG, MA, LMD, and ALM), Cincinnati, OH, and the National Institute of Medical Sciences and Nutrition, Mexico City, Mexico (GMR-P).
2 Supported by grants from the NIH (HD13021), the ADA (7-03-CD-06), Cincinnati Children's Hospital Medical Center Lactation Grant (anonymous donor), and the Cincinnati Hospital Research Foundation Translational Research Initiative. 3 Address reprint requests to AL Morrow, Center for Epidemiology and Biostatistics, Cincinnati Children's Hospital Medical Center, 3333 Burnet Avenue, MLC 5041, Cincinnati, OH 45229. E-mail: ardythe.morrow{at}cchmc.org.
ABSTRACT
Background: Previous studies have shown that human milk has a role in the gastrointestinal, neural, and immune development of neonates. If present in milk, adiponectin would be a promising candidate for influencing infant development, given its metabolic functions.
Objectives: Our objectives were to determine whether adiponectin is present in human milk and to characterize maternal factors associated with potential variation in milk adiponectin concentrations.
Design: We quantified adiponectin concentrations in human milk samples from donors to the Cincinnati Children's Research Human Milk Bank and randomly selected participants in a cohort study in Mexico City funded by the National Institutes of Health. Using cross-sectional and longitudinal data, we examined milk adiponectin concentrations in relation to lactation duration, maternal body mass index (BMI; in kg/m2), and ethnicity.
Results: Adiponectin was detected in human skim milk (range: 4.2–87.9 ng/mL). In cross-sectional and longitudinal analyses, duration of lactation was negatively associated with milk adiponectin concentrations (ß = –0.059 ± 0.024 and –0.059 ± 0.007, respectively; P < 0.02 for both). Maternal postpregnancy BMI was positively associated with milk adiponectin concentrations (ß = 0.08 ± 0.02, P < 0.0001; longitudinal analysis). Mexican mothers had lower median milk adiponectin concentrations at 1 mo than did the non-Hispanic white subjects from Cincinnati (11.5 and 19.8 ng/mL; P = 0.003).
Conclusions: Adiponectin is present in human milk and its concentrations are associated with duration of lactation, maternal adiposity, and ethnicity. Given the importance of adiponectin in inflammation, insulin sensitivity, and fatty acid metabolism, future studies should examine milk adiponectin's role in infant metabolic development.
Key Words: Adiponectin • human milk • lactation • maternal adiposity
INTRODUCTION
Breastfed children tend to be healthier, have a lower incidence of allergy and infectious disease, and tend to be leaner than formula-fed children (1). Although the reasons for these protective effects are not completely understood, components in milk clearly play a role. In addition, previous studies have shown that human milk influences gastrointestinal, neural, and immunologic development in breastfeeding infants (2–6).
Adiponectin, a protein produced primarily in adipose tissue, influences several physiologic processes that may affect human development. High concentrations of circulating adiponectin have positive health effects through the reduction of pro-inflammatory cytokines (7, 8), improvement of insulin sensitivity (9), and increase in fatty acid metabolism (10). Given the biological properties of adiponectin and the expression of adiponectin receptor 1 in the small intestine of neonatal mice (11), dietary adiponectin may influence infant development. Because adipose tissue comprises a large proportion of the human breast and is the primary source of adiponectin, we hypothesized that adiponectin would be present in human milk.
Human milk varies in composition within and between lactating women. Intraindividual variation in milk proteins is likely due, in part, to changes in milk composition throughout lactation (12–14). Interindividual variability in milk protein concentrations has been attributed to genetic variation (15, 16) and maternal adiposity (17), among other factors. Therefore, we hypothesized that duration of lactation, ethnicity (as a proxy for genetic variation), and maternal adiposity are associated with variations in milk adiponectin concentrations.
To test these hypotheses we quantified the concentration of adiponectin in cross-sectional and longitudinal human milk samples donated by lactating women residing in the United States and Mexico. Furthermore, we examined the associations between maternal factors and milk adiponectin concentrations.
SUBJECTS AND METHODS
Subjects
We analyzed human milk samples from 2 distinct populations of women: 1) donors to the Cincinnati Children's Research Human Milk Bank (RHMB) in Cincinnati, OH, and 2) participants in a cohort study funded by the National Institute of Child Health and Human Development entitled "The Role of Human Milk in Infant Nutrition and Health," which was conducted in Mexico City. The Institutional Review Board at Cincinnati Children's Hospital Medical Center approved the protocols and consent forms for both of these studies, and the Institutional Review Board at the National Institute of Medical Sciences and Nutrition in Mexico City also approved the protocol for the Mexican cohort study.
Cincinnati Children's Research Human Milk Bank
The RHMB is a repository to which any lactating woman can voluntarily donate breast milk as either a one-time donation (RHMB–ad hoc cohort) or regularly throughout the course of lactation (RHMB-longitudinal cohort) (18).
Mothers in the RHMB–ad hoc cohort donate milk that they have expressed. The following information is collected at the time of donation: gestational age of the infant at delivery, day of lactation when the milk was expressed, volume donated, and whether the milk was brought for donation to the bank fresh or frozen. However, because the mothers provide the expressed milk, the collection is considered to be nonstandardized. From the RHMB–ad hoc cohort, a single milk sample was randomly selected from 30 donor mothers.
For women who wish to be part of the RHMB-longitudinal cohort, a more rigorous process is involved. The mothers must provide consent within 1 wk after delivery and must have delivered a full-term (37 wk gestation) singleton infant with no congenital or medical complications. Mothers must commit to breastfeeding, at least partially, for 6 mo, speak English, and live within a 25-mile (40-km) radius of Cincinnati Children's Hospital Medical Center. Women of all races and ethnic groups were eligible. Milk collection by the RHMB-longitudinal cohort follows a standardized procedure during home visits (see below). To be selected for this analysis from the RHMB-longitudinal cohort, mothers had to donate 7 samples by 7 mo of lactation. A total of 199 milk samples from 22 mothers were included in this study.
At the time of the first visit, the research nurse visited the mother in her home and conducted an extensive questionnaire-based interview. Data collected during this interview included demographics, reproductive history, previous breastfeeding experience, general health status of the mother and infant since birth, and medication and environmental exposure. During subsequent visits, an abbreviated questionnaire was administered to obtain updates on medications, health status, and other factors. Maternal anthropometric measurements were taken during each home visit with portable scales that were calibrated regularly.
Mexican Human Milk Study cohort
The Mexican Human Milk Study was conducted between 1998 and 2003 as a collaboration between the National Institute of Medical Sciences and Nutrition in Mexico City and Cincinnati Children's Hospital Medical Center. Mothers were included in the study if they had a healthy full-term infant born without congenital malformations and intended to breastfeed. Mothers received 3 visits from a peer counselor to support exclusive breastfeeding. During the home visits, the mothers provided milk samples weekly for the first month and then monthly for the duration of lactation. No data were collected on maternal height or weight or infant gestational age at birth. From the Mexican cohort, 37 mothers were randomly selected from subjects who had donated milk samples at approximately 1 mo of lactation. Characteristics of all selected participants in the 3 cohorts are provided in Table 1.
View this table:
TABLE 1. Characteristics of the study cohorts1
Milk collection
For the RHMB–ad hoc cohort, milk collection involved the mother bringing her milk sample to the RHMB at Cincinnati Children's Hospital Medical Center. For the RHMB-longitudinal and Mexico breastfeeding cohorts, milk samples were collected by the study nurse during home visits for the duration of lactation. The mothers were asked to not feed their infants from the breast from which they intended to donate for 2 h before their appointment. Milk collection involved draining an entire breast with the use of a standard electric pump. For each cohort, a single study nurse collected milk between 1000 am and 1300. Collected milk was stored on ice for transportation to the local institution, where it was portioned and frozen at –80°C.
Maternal anthropometric measures
In the RHMB-longitudinal cohort only, a single trained research nurse (BSD) performed the anthropometric measurements. Maternal weight was measured to the nearest 0.1 kg at each monthly visit with the use of an E-Z Carry Portable Digital Scale (Hopkins Medical Products, Baltimore, MD). Women were measured while wearing street clothing and no shoes. Height was measured at the initial visit while the subject was in a standing position, was wearing socks, and had their heels together, toes apart at a 45° angle, and head in the Frankfort horizontal plane. Height was marked on the wall and the vertical distance to the floor was measured to the nearest 0.1 cm. Body mass index (BMI) was calculated as weight (in kg) divided by the square of height (in m).
Assay of adiponectin and leptin
Because lipids interfere with radioimmunoassays (RIAs), skim milk was used, and the same person (WB) assayed all samples from the 3 cohorts. Milk samples were thawed and mixed by vortex. Skim milk (aqueous phase) was prepared by centrifugation (1500 x g, 20 min, 4°C), after which the fat layer was removed. Immunoreactive adiponectin was measured in duplicate by using a commercial RIA kit (Linco Research, St Charles, MO) with the use of 100 µL of a 1:3 dilution of skim milk (33.3-µL equivalent). The inter- and intraassay CVs were 8.5% and 3.9%, respectively. In the RHMB–ad hoc samples, leptin was also assayed in duplicate in skim milk by using a commercial RIA kit (Linco Research) following the protocol of Houseknecht et al (19). The inter- and intraassay CVs were 4.5% and 5.0%, respectively, and had a limit of detection of 0.3 ng/mL.
The assay methods for adiponectin were validated by using standards, and skim milk samples were spiked with the 5-, 20-, and 100-ng/mL human adiponectin standards to determine the recovery of added mass.
Statistical analyses
Statistical analyses were conducted by using SAS software (version 9.1.3; SAS Institute Inc, Cary, NC). Descriptive statistics are reported as medians and ranges because of the nonnormality of the data (Table 2). To improve the normality of milk adiponectin for use in statistical models, the data were natural log (ln) transformed. Milk leptin concentrations were evaluated in their original units. Eight milk leptin concentrations were below the limit of detection (0.3 ng/mL); for the purposes of the analysis, these values were set at 0.2 ng/mL. Analyses of leptin were also conducted with the values below the detection limit set at 0; however, the results did not change substantially and thus are not reported.
View this table:
TABLE 2. Human milk adiponectin and leptin concentrations in the study cohorts1
To examine the effect of lactation duration on ln(milk adiponectin) concentrations, we used RHMB–ad hoc and -longitudinal data. In cross-sectional analyses, linear regression models were constructed by using ln(milk adiponectin) as the dependent variable. In the longitudinal analyses, mixed models with repeated measures were constructed for which the intercepts for each donor were treated as random. The best correlation structure, in this case equal correlation among ln(milk adiponectin) obtained across different time points, was selected by using the Bayesian Information Criterion derived based on the data. The dependent variable was ln(milk adiponectin) and the independent variable was day of lactation.
To determine the effect of ethnicity on milk adiponectin concentrations, median milk adiponectin at 1 mo of lactation from non-Hispanic whites in the RHMB-longitudinal cohort (range: 26–42 d) and Hispanics in the Mexican cohort (range: 31–40 d) were compared by using a Wilcoxon's rank-sum test.
To determine the effect of maternal adiposity on milk adiponectin concentrations, multiple data points from each mother in the RHMB-longitudinal cohort were analyzed cross-sectionally (at each collection time point) to assess potential interaction with time postpartum. Given the lack of time effects, we summarized the overall effect using a longitudinal repeated-measures model of ln(milk adiponectin) and maternal BMI. This analysis involved application of mixed-model procedures.
RESULTS
Validation of adiponectin assay for human milk
Adiponectin concentrations of serial dilutions (10–40 µL) of skim milk were parallel to the standard curve. Skim milk samples were spiked with the 5-, 20-, and 100- ng/mL human adiponectin standard, and recovery of added mass averaged 109% with an SD of 6.7%.
Adiponectin and leptin in human milk
Immunoreactive adiponectin was detected in skim milk in all samples (Table 2). In the RHMB–ad hoc samples, both leptin and adiponectin concentrations were assayed in milk (Figure 1). Median adiponectin concentrations were >40 times those of leptin concentrations (16.6 compared with 0.4 ng/mL, respectively; P < 0.0001 by Wilcoxon's rank-sum test).
FIGURE 1.. Adiponectin and leptin concentrations in 30 cross-sectional milk samples from the ad hoc cohort of the Cincinnati Children's Research Human Milk Bank. Inset: detail of milk leptin distribution. The box represents the interquartile range, the horizontal line represents the median, and the whiskers represent the 95th and 5th percentiles of distribution. Difference between medians: P < 0.0001 (Wilcoxon's rank-sum test).
Adiponectin decreases through lactation
In the RHMB–ad hoc samples (n = 30), month of lactation was negatively associated with ln(milk adiponectin): ß = –0.059 ± 0.024, P = 0.02. These results were verified in RHMB-longitudinal samples, in which the month of lactation was also negatively associated with ln(milk adiponectin), ß = –0.059 ± 0.007, P < 0.0001 (Figure 2). On the basis of these results, milk adiponectin is predicted to be 6.9 ng/mL lower by 7 mo of lactation than at 1 wk of lactation. This is equivalent to a 5.72% decrease in milk adiponectin concentration with each month of lactation.
FIGURE 2.. Adiponectin concentrations by duration of lactation in milk samples from 199 participants in the longitudinal cohort of the Cincinnati Children's Research Human Milk Bank. The solid line represents the predicted regression line determined from the repeated-measures analysis of month of lactation and natural log(milk adiponectin): ß± SE: –0.059 ± 0.007. Three data points from 2 subjects with milk adiponectin concentrations between 45 and 60 ng/mL are not presented.
Adiponectin concentrations in Hispanic mothers
Hispanic mothers from the Mexico cohort (n = 37) had significantly lower median milk adiponectin concentration at 1 mo than did non-Hispanic whites from the RHMB-longitudinal sample (11.7 compared with 19.8 ng/mL; P = 0.0005 by Wilcoxon's rank-sum test; n = 19). After the outliers were removed (n = 2 from each group), this difference remained highly significant (11.5 compared with 18.6 ng/mL; P = 0.0006).
Human milk adiponectin and postpregnancy maternal weight
In longitudinal analyses accounting for repeated measures (Figure 3), maternal postpregnancy BMI was significantly associated with ln(milk adiponectin): ß = 0.10 ± 0.02, P < 0.0001. There was no interaction between month of lactation and maternal BMI. When outlier milk adiponectin concentrations >50 ng/mL were excluded (14 longitudinal values from 2 subjects), this relation remained highly significant (ß = 0.08 ± 0.02, P < 0.0001). This equates to an 8.33% increase in milk adiponectin concentration with each unit increase in maternal BMI unit.
FIGURE 3.. Adiponectin concentrations by maternal BMI in milk samples from 199 participants in the longitudinal cohort of the Cincinnati Children's Research Human Milk Bank. The solid line represents the predicted regression line determined from the repeated-measures analysis of maternal BMI and natural log(milk adiponectin); data for 2 women (n = 14 longitudinal samples) with milk adiponectin concentrations >50 ng/mL (ß± SE: 0.08 ± 0.02) were excluded. The dashed line represents the predicted regression line including these 2 women (ß± SE: 0.10 ± 0.02). Three data points from 2 subjects with milk adiponectin concentrations between 45 and 60 ng/mL are not presented.
DISCUSSION
To our knowledge, this is the first study to report the presence of adiponectin in human milk. The concentration of adiponectin in milk is much lower than in serum. In neonates, serum adiponectin concentrations have been reported to range from 20 to 60 µg/mL (20, 21), whereas we found milk adiponectin concentrations to range from 4.2 to 87.9 ng/mL. However, milk adiponectin concentrations are consistent with concentrations of expressed adiponectin messenger RNA from mesenteric adipose tissue (22), which suggests possible local effects in the gut.
Leptin has previously been reported in human milk (19, 23–25). In the RHMB–ad hoc samples in this study, milk leptin concentrations were 40-fold lower than adiponectin. The leptin concentrations in milk measured in this study were slightly lower than those reported for leptin in other studies (17, 19, 23). The lower concentrations may have been due to the extended lactation period in this study.
Whether adiponectin in human milk has biological significance for breastfeeding infants is not clear; however, several lines of research suggest that it might. Previous studies have shown that milk components are not often degraded in the stomach, in part because the composition of human milk forms a protective environment for proteins (26) and in part because of the reduced acidity of the infant stomach (27) and limited gastric proteolysis (28). Indeed, oral insulin is not degraded and thus can stimulate gut maturation (26, 29, 30). Second, adiponectin's physiologic actions could be important in developing infants. Because adiponectin has been shown to increase insulin sensitivity (31, 32), it may also augment insulin's action in the gut of infants. Adiponectin may also have direct effects on the gut of infants, because previous studies have documented that adiponectin receptor 1 is expressed in fetal small intestine (11).
The current study also reports maternal factors associated with milk adiponectin concentrations. First, milk adiponectin concentrations decline throughout lactation. These results were significant in both the cross-sectional and longitudinal cohorts, which suggests a robust phenomenon, even in the case of nonuniform milk collection in the cross-sectional cohort. This finding agrees with that of previous studies that documented a decrease in many milk proteins throughout lactation (12–14).
We also examined the effect of ethnicity on adiponectin concentrations. Hispanic mothers of Mexican descent had significantly lower concentrations of milk adiponectin than did the non-Hispanic white donors at 1 mo of lactation. Milk samples were collected according to the same protocol, during the same period of lactation, and were assayed by the same person, thereby minimizing sampling and assay error. Nevertheless, there are several possible causes for these differences. First, differences in sample storage time may have been responsible. All samples were stored at –80°C; however, the Mexico samples had been frozen for 5–8 y, whereas the RHMB samples had been frozen for <2 y. Although most proteins are stable at –80°C, it is possible that degradation occurred. Second, these differences may be attributed to differences in BMI in the Cincinnati and Mexico mothers. However, national surveys support similar rates of overweight in both populations (33, 34). Lastly, the difference could be physiologic because the ethnic difference in adiponectin concentrations in milk parallel reported differences in adiponectin concentrations in serum. Several studies have shown that whites have higher serum adiponectin concentrations than do Asians (35, 36), Amerindians (37), and African Americans (38). Although there has been no direct comparison of serum adiponectin concentrations between Non-Hispanic whites and Hispanics, the Mexican population is considered to be a combination of Amerindian and white populations (39).
In contrast with concentrations in serum, where adiposity is negatively correlated with adiponectin (40), a positive association was found between adiponectin concentrations in milk and maternal adiposity. One potential explanation for this finding is the relation between adiponectin, prolactin, and adiposity. Adiponectin is negatively regulated by prolactin (41, 42), a major determinant of mammary gland development in lactating women. Prolactin secretion is dampened in obesity (43, 44). Thus, we speculate that if adiponectin is produced by breast adipose tissue, reduced negative regulation by prolactin in heavier mothers may effectively increase concentrations of adiponectin produced locally in the breast tissue and secreted into human milk. Because this association is reported in only 22 women, it will be important to confirm this relation in other studies.
In summary, this study reports that the adipocyte protein adiponectin is present in human milk at concentrations significantly higher than those of milk leptin. In addition, maternal factors—including duration of lactation, ethnicity, and adiposity—are associated with the concentration of adiponectin in human milk. Given the importance of adiponectin in inflammation, insulin sensitivity, and fatty acid metabolism, milk adiponectin may influence infant development. Future studies are required to clarify milk adiponectin's physiologic role.
ACKNOWLEDGMENTS
We thank all of the women who generously provided milk samples and Shannon Hatfield for assistance with manuscript preparation.
LJM and ALM were responsible for the study concept and LMD and JGW provided significant advice on this concept. SRG, ALM, BSD, and GMR-P were involved in the study design and data collection. WB developed the protocol for the milk adiponectin assay. LJM, JGW, and MA were involved in the data analysis. LJM, JGW, SRG, LMD, GMR-P, and ALM were involved in drafting the manuscript. All authors read and approved the final version of the manuscript. None of the authors had a conflict of interest to disclose.
REFERENCES