Maternal Intakes and Nutritional Status During Lactation and the Implications for Maternal and Infant Health
Evidence on the complex interplay between maternal nutrition, health, exposures, and lactation and the implications for maternal and infant health has grown in recent years. There have been improvements in the techniques for assessing the nutrient content of breast milk, which could inform updates of estimates of infant nutrient needs. With the growing prevalence of obesity and diabetes in the population, studies have explored the relationship between these conditions and lactogenesis and breast milk composition. A growing body of evidence also suggests that breastfeeding has long-term metabolic consequences for the mother, lowering her risk of cardiometabolic disease. Session 6 of the workshop, moderated by Deborah O’Connor of the University of Toronto, provided an overview of these topics. Highlights from the session presentations are presented in Box 7-1.
Lindsay H. Allen, director of the U.S. Department of Agriculture’s (USDA’s) Agricultural Research Service’s Western Human Nutrition Research Center acknowledged that human milk is an infant’s best source of nutrition. “I think the fear of upsetting this concept has in a way [inhibited] research on what is in milk as far as micronutrient composition,” she said. In her review of current evidence on the relationship between maternal nutrient status, maternal intake, and human milk composition, Allen discussed the analysis of human milk micronutrient composition, updated
estimates for selected nutrients, and the Mothers, Infants and Lactation Quality (MILQ) Study.
Analysis of Human Milk Micronutrient Composition
Describing the state of the evidence on human milk micronutrient composition in the United States as poor, Allen reviewed several challenges that exist. A primary challenge is determining the appropriate approach to measure each of the 40 or more nutrients of interest. The human milk matrix is different from plasma, requiring different analytical techniques. Human milk collection is also an issue, as the best approach is not yet
known. Additionally, there are various maternal factors (e.g., supplement use) that need to be considered in analyzing and interpreting data.
Some, but not all, nutrients in human milk are dependent on maternal status and to a lesser extent on maternal intake. For instance, human milk concentrations of folate, calcium, iron, copper, and zinc do not change with maternal supplementation. By contrast, human milk concentrations of the fat-soluble vitamins, thiamin, riboflavin, vitamin B6, vitamin B12, choline, iodine, and selenium are affected by maternal status and are the nutrients that have the most implications for public health interventions, noted Allen.
The Dietary Reference Intakes (DRIs) include estimates of nutrient needs. For infants, Adequate Intakes (AIs) are established using breast milk composition data available at the time of review. Reiterating that the evidence on breast milk nutrient composition is poor, Allen showed the basis of select infant AIs and noted that some estimates have been derived from single studies with few participants (see Table 7-1). She characterized the
state of the evidence as “an appalling sparsity of data on which to set the recommendations for infants and for lactation.”
TABLE 7-1 Summary of Data Available for Setting Adequate Intakes for Infants and Lactation for Select Nutrients
NOTE: μg/d = micrograms per day; μg/L = micrograms per liter; d = days; IOM = Institute of Medicine; mg/L = milligrams per liter; mo = months; RDA = Recommended Dietary Allowance; wk = weeks.
SOURCES: Presented by Lindsey Allen. Adapted from Allen et al., 2018.
Allen’s lab focuses on developing efficient analytical methods for analyzing breast milk composition, using techniques such as inductively coupled plasma mass spectrometry and high-performance liquid chromatography. She indicated that mass spectrometry has been transformative and can be used to measure most of the B vitamins. Allen’s team is also starting to investigate metabolomics. Recent breast milk estimates globally show median concentrations below the values that were used for the existing infant AIs for several micronutrients.
Updated Estimates for Select Nutrients
Allen showed examples of where new evidence has emerged on breast milk micronutrient concentrations. Her remarks included comments on thiamin, vitamin B6, iodine, vitamin B12, and vitamin D.
The breast milk concentration used to establish the infant AI for thiamin was 210 μg/L and was derived from a single study. New evidence from a collection of high-income countries revealed a median thiamin concentration in breast milk of 125 μg/L. A supplementation trial conducted in Cambodia was able to increase breast milk concentrations from approximately this level to the concentrations used to establish the infant AI for thiamin. However, the new estimate of median concentrations is likely closer to normal and suggests the infant AI for thiamin is higher than infant needs, said Allen.
The prevalence of vitamin B6 deficiency is not well characterized nationally or internationally. Based on 2014 National Health and Nutrition Examination Survey (NHANES) data, 24 percent of U.S. women who do not use supplements have low serum pyridoxal phosphate; serum and breast milk concentrations are correlated, added Allen. In several countries, particularly low-income countries, breast milk vitamin B6 concentrations are below 0.13 mg/L, the value used to establish the infant AIs. One small study, conducted in lactating women from Davis, California, who were not taking supplements during lactation, found an average vitamin B6 concentrations of 0.3 mg/L. Allen suggested that this higher concentration could possibly be attributable to high vitamin B6 content of some prenatal vitamins.
“I would say of all the nutrients, iodine is the most sensitive to maternal intake,” stated Allen. She noted that despite salt fortification programs, low concentrations of iodine in breast milk are common. Across longitudinal studies iodine concentrations in breast milk are variable, but generally there is a marked decrease in iodine content during the first month of lactation. The only longitudinal study assessing the iodine content of breast milk in the United States was conducted in 31 Mexican American women and found concentrations below the values used to establish the iodine AIs for infants. Breast milk iodine concentrations mirror maternal urinary iodine excretion. In a systematic review of 57 studies, the concentration of iodine in breast milk was 13–18 μg/L among women with goiter, whereas women in areas with effective salt iodization had breast milk iodine concentrations exceeding 90 μg/L.
Different groups have considered improving the iodine status of lactating women. For instance, the American Pediatric Association and American Thyroid Association recommend that lactating women supplement with 150 μg/day, and the World Health Organization (WHO) recommends that lactating women in areas with moderate or severe iodine deficiency supplement their diets with 250 μg/day. Iodine is preferentially lost in breast milk over urine when status is low during lactation. To that end, it has been suggested that the iodine content of breast milk may be a better indicator of iodine status among lactating women (Dold et al., 2017).
Allen noted that older analytical techniques for measuring iodine concentrations in breast milk are no longer considered valid. The breast milk iodine concentration used in the derivation of the iodine AIs for infants and lactating women came from a single study from 1984. New evidence suggests infant needs are lower than previously estimated, meaning the iodine AI for lactating women is likely too high, said Allen. She continued with the caveat that infants have limited opportunity for iodine exposure. Weaning infants are at risk of iodine deficiency, particularly when breast milk iodine content is low, as complementary foods tend to be low in iodine.
A recent systematic review has assessed the state of the evidence on the vitamin B12 content of breast milk (Dror and Allen, 2018). Allen reported that 7 of the 26 identified studies used invalid analytical techniques. Vitamin B12 concentrations vary widely, but they appear to decrease during early lactation. The vitamin B12 content of breast milk correlated with maternal intake, along with maternal and infant status.
There is no consensus on whether lactating women in high-income countries should be supplementing with vitamin B12. Danish investigators who support supplementation have reported that breast milk vitamin B12 concentrations dramatically decrease during the middle of lactation, which is accompanied by increases in infant methylmalonic acid concentrations. Allen’s recent data do not support a drop in vitamin B12 concentrations during the middle of lactation.
Allen’s lab found that the vitamin B12 concentrations of breast milk in Canada, Denmark, Ghana, and the United States are close to the value used to establish the infant AIs for vitamin B12. Concentrations were markedly lower in countries with lower intakes of animal sources of food. This evidence indicates vitamin B12 concentrations in breast milk are heavily influenced by maternal intake. Despite this, there is virtually no data available on vegetarian and vegan lactating women in the United States, indicated Allen.
Allen explained that three studies conducted in Guatemala have characterized the continuum of maternal–child vitamin B12 depletion. Depletion during pregnancy can lead to low vitamin B12 stores in infants at birth and in breast milk. Infant stores will be depleted by 3 months of age and will remain so for 7 to 12 months, despite supplementation or more frequent breastfeeding. Cow milk intake can improve infant status, as its vitamin B12 concentration is higher than breast milk. Vitamin B12 depletion of infants after about 3 months of age can lead to growth and motor development deficits.
Evidence from intervention studies underscores the importance of improving maternal vitamin B12 status during pregnancy, noted Allen. Maternal serum vitamin B12 concentrations in early pregnancy are associated with infant status at 4–6 months of age. Allen suggested the dose matters, and smaller repeated exposures during lactation, rather than a single one-a-day supplement appear to be more effective. Despite the link with maternal dietary patterns, Allen stated that current intake is not the main influence on breast milk vitamin B12 concentrations.
Breast milk is a poor source of vitamin D. As such, the American Academy of Pediatrics recommends 400 IU/day of vitamin D be given to breastfed infants. A recent study has demonstrated that lactating women who take a 6,400 IU/day vitamin D supplement can supply infants with sufficient vitamin D through breast milk (Hollis et al., 2015). Pointing out that the supplement dose exceeds the vitamin D Tolerable Upper Intake Level of 4,000 IU/day, Allen indicated that such an intervention may not be
practical, but demonstrates the level of supplementation needed to increase vitamin D concentrations in breast milk.
Allen stated that there is no breast milk composition data for several micronutrients. As an example, she explained that survey data from Canada, Europe, Ireland, and the United Kingdom found riboflavin depletion in 20–60 percent of adults and commented that there are no estimates for the United States. Severe riboflavin deficiency has been associated with poor infant growth and development. Evidence from low-income countries indicate that breast milk concentrations respond to maternal intake to a greater degree than many of the other nutrients. Maternal supplementation during lactation, however, may not be an effective strategy for some nutrients. A study of exclusively breastfed infants in Malawi found that 2–5 percent of nutrients in a supplement given to lactating women appears in their breast milk.
Allen closed her presentation by providing a brief overview of the MILQ Study. The goal of the project is to create human milk reference values for nutrients during the first 9 months of lactation. The study aims to recruit 1,000 well-nourished, nonsupplemented women and their infants from four countries.
Ellen Demerath, professor of epidemiology and community health at the University of Minnesota School of Public Health, opened her remarks by emphasizing that the weight and metabolic status of the population has significantly changed over the past 30 years. When Nutrition During Lactation was published in the early 1990s, the prevalence of obesity among adults was 11.4 percent and the prevalence of gestational diabetes was 2 percent (IOM, 1991). Current estimates are markedly higher: 37 percent of U.S. women 20–39 years of age are estimated to have obesity (Flegal et al., 2016); 19.3 percent of U.S. adults have prediabetes (Menke et al., 2018); and anywhere from 5–6 percent to 15–20 percent of women have gestational diabetes (NIH Consensus Development Panel, 2013). Given this context, Demerath went on to discuss the implications of these conditions for maternal health and infant programming, breastfeeding outcomes, and breast milk composition.
Implications for Maternal Health and Infant Programming
Maternal obesity has been linked to a host of negative pregnancy, maternal, and childhood outcomes. Odds of gestational diabetes, preeclampsia, stillbirth, Cesarean section, and being large for gestational age increase with maternal obesity. Excessive gestational weight gain has been associated with postpartum weight retention, and maternal obesity has been associated with maternal risk for type 2 diabetes. The offspring of mothers who have obesity are at increased odds of having obesity during childhood and developing type 2 diabetes (Catalano and Shankar, 2017; Kulie et al., 2011; Lahti-Pulkkinen et al., 2019).
The Hyperglycemia and Adverse Pregnancy Outcome (HAPO) Study and HAPO Follow-Up Study provided critical insight into the influence of gestational diabetes. Compared to women who did not develop gestational diabetes, those who did had higher rates of Cesarean section, offspring who were large for gestational age, preterm infants, preeclampsia, maternal prediabetes, maternal type 2 diabetes, and offspring diabetes (Hod et al., 2019).
Demerath noted that, in the past 30 years, the fields of epidemiology and public health have embraced the concept that chronic disease risk is set in early life. “We talk about two-generational policies and programs now,” she said. Breastfeeding helps to set the infant’s metabolic expectations for future growth and may help to reset maternal insulin resistance that occurred during pregnancy. The concept of lactational programming suggests that maternal nutrition, health, and exposures can affect the composition of her breast milk, having implications for programmed offspring outcomes (see Figure 7-1).
Using a rodent model, Vogt et al. (2014) demonstrated that lactation serves as a critical period for programming offspring appetite. A high-fat diet, provided exclusively during lactation, led to increases in milk glucose and insulin concentrations, reductions in neuronal fiber density in regions of the brain associated with appetite regulation, and increases in offspring bodyweight, adiposity, and glucose tolerance. The exposure to milk from a maternal high-fat diet during lactation led to lifelong effects in the offspring’s metabolic function and weight status, even when the offspring consumed a low-fat diet after weaning.
Breastfeeding appears to modify metabolic risk transmission, indicated Demerath. The Study of Women, Infant Feeding and Type 2 Diabetes after GDM1 Pregnancy (SWIFT) found that infants of mothers with gestational diabetes had lower weight status in the first year if they were intensively breastfed (Gunderson et al., 2018). Other studies have reported that breast-
1 GDM is gestational diabetes mellitus.
feeding exclusivity and duration mitigate the effects of maternal gestational diabetes and obesity on childhood outcomes, including adiposity and childhood obesity (Crume et al., 2011; Li et al., 2005; Mayer-Davis et al., 2006). Demerath emphasized that breastfeeding has important maternal and child health benefits, as compared to formula feeding, and that changes to maternal metabolic status as a result of lactation have biological underpinnings. She also underscored that the composition of breast milk is superior to infant formula, but it can be affected by maternal obesity and diabetes.
Implications for Breastfeeding Outcomes
The Infant Feeding Practices Study (IFPS) found that one in eight women did not achieve their breastfeeding goals (Steube et al., 2014). One of the identified barriers to success was depression. “What is psychological for the mother can become physiological and nutritional for the offspring,” commented Demerath. The study also found obesity to be among the key barriers.
Other studies have provided evidence that obesity affects breastfeeding outcomes. For instance, an analysis of 2004–2011 data from the Preg-
nancy Risk Assessment Monitoring System (PRAMS) reported lower rates of breastfeeding continuation among mothers with higher levels of prepregnancy obesity (Kachoria et al., 2015). Maternal obesity has also been associated with delayed prolactin release and a lower prolactin response to suckling, and it appears to lower milk transfer, noted Demerath. She added that women who have obesity have higher rates of Cesarean section, a stressor that could have an effect on lactation. Issues of positioning, latching, body confidence, and comfort breastfeeding in public are other factors that shorten breastfeeding duration among women with obesity. Demerath explained there are disparities across different race and ethnicity groups, which are confounded by socioeconomic status. Among 81,669 participants in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) in Minnesota, for example, there was a graded increase in the risk of breastfeeding cessation by 6 months among non-Hispanic white and Hispanic mothers with increasing prepregnancy body mass index (BMI) status, but the relationship was not see among African American mothers.
Evidence on the relationship between gestational diabetes and breastfeeding outcomes is less clear. In SWIFT, approximately one-third of women with gestational diabetes had delayed lactation, which was exacerbated by insulin treatment, noted Demerath. In contrast, IFPS II did not find shortened breastfeeding durations among women who self-reported they had gestational diabetes (Wallenborn et al., 2017).
A more recent line of investigation has explored hormonal determinants of lactation. The hormone leptin is elevated in individuals with obesity. In vitro studies suggest leptin may inhibit oxytocin from promoting a muscle-contracting effect, which could possibly affect breast milk ejection (Moynihan et al., 2006). Elevated leptin may also promote prolactin resistance (Buonfiglio et al., 2016). Insulin appears to promote expression of genes related to milk protein synthesis, whereas insulin resistance may have an opposing effect, reducing expression of a key milk production protein (Lemay et al., 2013). Less is known about cortisol, but elevated levels may counter the effects of insulin and prolactin, said Demerath.
Relationships between lactation and both maternal obesity and diabetes appear to be bidirectional, but they may be confounded by metabolic status prior to and during pregnancy, Demerath said. To improve breastfeeding rates, she suggested that trials are needed to better understand causal relationships with breastfeeding outcomes among women with obesity and diabetes.
Implications for Breast Milk Composition
Breast milk is composed of more than 1,000 compounds, including macronutrients, fatty acids, hormones, immune factors, antimicrobials,
oligosaccharides, microbiomes, and environmental pollutants. A host of maternal, physiological, behavioral, and infant factors contributes to breast milk composition (Fields et al., 2016). Breast milk often mirrors maternal serum concentrations but not for all factors. Some compounds are actively transported or synthesized in the lactocytes, leading to differences in concentrations. For instance, concentrations of adiponectin are higher, and leptin, insulin, and fatty acid are lower in breast milk than in maternal serum. Breast milk composition also changes as a woman progresses from producing colostrum to late mature milk. Demerath identified some limitations in the literature, including a lack of evidence on breast milk composition after 6 months and studies being primarily observational in design with small sample sizes.
Relationships between maternal metabolic status and breast milk macronutrient composition have been explored. The Davis Area Research on Lactation, Infant Nutrition and Growth (DARLING) Study reported that mothers with higher relative weight had higher lipid concentrations in their breast milk (Nommsen et al., 1991). Women with obesity have also been found to have higher omega-6 and lower omega-3 long-chain polyunsaturated fatty acid composition in their breast milk, as compared to lean women (García-Ravelo et al., 2018; Panagos et al., 2016). Other studies of total breast milk lipid, protein, and lactose concentrations, however, reported only small or no differences in composition across prepregnancy BMI or gestational weight gain groups (Chang et al., 2015; Michaelsen et al., 1994; Quinn et al., 2012). Demerath suggested that additional research is needed to disaggregate the roles of maternal diet and obesity on breast milk composition.
Evidence is less robust linking maternal metabolic status to breast milk micronutrient composition. Carotenoid and vitamin D concentrations have been reported to be lower in the breast milk of women with obesity (Panagos et al., 2016). Characterizing the evidence as sparse, Demerath thought that more evidence on the role of maternal metabolic status on the vitamin, mineral, and polyphenol content of breast milk is warranted.
Demerath next discussed bioactive compounds in breast milk. Leptin, insulin, and ghrelin are three hormones that play a role in appetite, satiety, and metabolism, and concentrations are altered in the serum of adults with obesity. Because they can survive digestion, these compounds have been found to enter infant circulation intact and retain their bioactive properties. Maternal prepregnancy BMI is positively associated with breast milk leptin and insulin content (Fields and Demerath, 2012; Sadr Dadres et al., 2019; Whitaker et al., 2017; Young et al., 2017). Gestational weight gain, particularly among women with a normal prepregnancy BMI, also appears to influence bioactive concentrations in breast milk, indicated Demerath. Emerging data suggest that insulin may have a stronger effect on gene
expression and phenotype in female infants as compared to male infants (Fields et al., 2017). Other factors may not have relationships with maternal weight status. Concentrations of interleukin-6 in breast milk at 1 and 3 months postpartum were not associated with maternal BMI or gestational weight gain (Whitaker et al., 2017), although Collado et al. (2012) reported higher concentrations among women with obesity. With the studies varying in design and timing of sample collection, evidence of relationships between concentrations of adiponectin in breast milk and prepregnancy BMI is mixed (Chan et al., 2018; Ley et al., 2012; Sadr Dadres et al., 2019). Few studies have explored breast milk immunological function by maternal weight status (Erliana and Fly, 2019).
Diabetes may also alter breast milk composition. Higher breast milk concentrations of insulin and glucose have been found in women with type 1 and type 2 diabetes (Jovanovic-Peterson et al., 1989; Schaefer-Graf et al., 2006; Whitmore et al., 2012). Concentrations of insulin and some cytokines appear to be higher in the breast milk of women with gestational diabetes, noted Demerath. Concentrations of ghrelin in breast milk appear to be lower among women with gestational diabetes or preexisting diabetes, but this relationship is not seen in mature breast milk (Aydin et al., 2007).
Preliminary data have explored breast milk metabolomic differences by maternal BMI (Isganaitis et al., 2019). At 1 month postpartum, concentrations of nine breast milk metabolites were significantly different between women who had overweight or obesity compared to women with normal BMIs. Three of the differing metabolites were oligosaccharides, including 2′-fucosyllactose, said Demerath. Maternal BMI and infant body fatness were both positively associated with breast milk concentrations of adenine and 5-methylioadenoisine. Evidence on maternal metabolism, human milk oligosaccharides (HMOs), and the human milk microbiome is rapidly emerging and evolving, Demerath said. In the largest study to date, prepregnancy BMI and HMO concentrations in breast milk were not found to be associated (Azad et al., 2018).
Demerath pointed out that little is known about the breast milk components that influence infant growth, although evidence is emerging. Whereas maternal obesity has been associated with higher circulating leptin concentration, infant weight status and adiposity appear to be negatively associated with leptin concentrations in breast milk. Demerath thought that the elevated concentrations of leptin may be serving their functional role in infants and downregulating appetite.
To close her presentation, Demerath highlighted a few implications from the evidence she presented. She suggested that the issues leading women to not achieve their breastfeeding goals underscores the importance of lactation support. Estimates of breast milk composition and its relationship with maternal nutrition need to be updated, Demerath said. She also
noted that the comparative analysis of the existing body of evidence is difficult because of differences in study design and confounding. She said that women of lower socioeconomic status are less likely to be reflected in the data. Finally, she explained that while donor milk banks standardize for macronutrients, they do not standardize for other breast milk factors, and she thought further work was needed to preserve the content donor milk.
Erica P. Gunderson, epidemiologist and senior research scientist at the Kaiser Permanente Northern California Division of Research focused her remarks on evidence of the lasting effects of lactation on metabolic and cardiovascular health outcomes in women. Gunderson provided an overview of the continuum of maternal health from preconception through postdelivery; she also reviewed the limitations of the evidence base on lactation and the development of type 2 diabetes and cardiovascular diseases in women. In addition, Gunderson discussed the study designs and key findings from the Coronary Artery Risk Development in Young Adults (CARDIA) Study and SWIFT, and offered her reflections on the importance of this emerging evidence base.
The Continuum of Maternal Health
The relationships between pregnancy, lactation, and long-term maternal health outcomes has only recently become an active area of research. In 2010, an Agency for Healthcare Research and Quality review found that women who did not breastfeed had higher risk of breast cancer and ovarian cancer, but it reported that there were insufficient data to assess the risk of type 2 diabetes and cardiovascular disease.
“Pregnancy and lactation form a single continuum within the reproductive cycle where alterations of physiologic demands may have lasting consequence for future disease outcomes in women,” said Gunderson. Although this continuum extends before, during, and after pregnancy, most studies have only focused on pregnancy rather than the preconceptual or postpartum periods. During pregnancy, there are metabolic changes, such as increased insulin resistance, altered glycemic control, fat mass deposition, and inflammation. Blood volume also expands 40–50 percent during pregnancy, increasing cardiac output and decreasing blood pressure. “Each of these different adaptations, if they happened in any other physiological state, would be considered pathologic,” suggested Gunderson. She further noted that prepregnancy obesity may exacerbate these metabolic aberrations.
Women who had an uncomplicated pregnancy experience an increase in chronic disease risk with age, typically occurring after age 60. Pregnancy complications, however, increase the risk of women experiencing chronic disease earlier in life. Women who had gestational diabetes are at increased risk for type 2 diabetes, with a substantial portion of conversion occurring in the first few years postpartum (Bellamy et al., 2009). As a risk factor for cardiovascular disease, type 2 diabetes serves as an early predictor. Even if a woman does not go on to develop type 2 diabetes, she is still at a 30 percent higher risk for cardiovascular disease (Retnakaran and Shah, 2017; Tobias et al., 2017). Lactation can “reset maternal metabolism,” Gunderson said, including lowering triglycerides, low-density lipoprotein-cholesterol, blood glucose concentrations, and fasting insulin levels, and preserving higher high-density lipoprotein-cholesterol concentrations.
Limitations of Previous Studies
Studies assessing the relationships between lactation and lasting effects on chronic disease risk have suffered from issues of study design, asserted Gunderson. Most studies have not been designed to address the potential for reverse causality. Randomized trials to investigate this topic are not feasible, as randomization of individual women to breastfeed or not is not ethical and randomizing clusters to breastfeeding support is expensive and has low statistical power. Observational studies often rely on self-reported data. Retrospective studies are limited by the lack antecedent biochemical data (i.e., metabolic status prior to lactation). Some prospective observational studies of older women can also suffer from issues of retrospective reproductive exposure data and lack of antecedent biochemical measures.
Pooled data from observational studies, many of which were cross sectional or case–control design, found an approximate 7 percent reduction in hypertension among women who breastfed, but there was no graded association based on reported duration of breastfeeding (Qu et al., 2018). These studies did not assess pregnancy complications or antecedent risk factors, and some included older women. A recently published 6-year prospective study reported slightly stronger protective effect of lactation on cardiovascular disease hospitalization, but it did not find a dose–response relationship with breastfeeding duration (Nguyen et al., 2019). The protective effect was stronger for cardiovascular disease mortality, particularly among older women. Gunderson showed that several prospective epidemiological studies have found relatively weak relationships between breastfeeding and cardiovascular disease, and noted that many of the studies were conducted in older women. The European Prospective Investigation into Cancer and Nutrition (EPIC) Study, which included young women, found breastfeed-
These previous epidemiological studies have suffered from issues of internal validity, said Gunderson. None of the studies had metabolic or cardiovascular disease risk factors measured prior to women initiating breastfeeding, nor did they include assessments of pregnancy complications. Measures of disease outcomes are also limited, as they are typically self-reported. As women younger than 45 years of age are not routinely tested for cardiometabolic risk factors or type 2 diabetes, Gunderson suggested that this could lead to the selection bias of study participants. Additionally, despite an overall relationship, the studies failed to find a dose–response relationship.
CARDIA and SWIFT Studies: Longitudinal Biochemical Evidence
CARDIA and SWIFT are two robust longitudinal studies that were designed to overcome the limitations of previous assessments of the relationships between lactation and long-term metabolic disease risk based on longitudinal biochemical data. Whereas previous studies have enrolled postmenopausal women and asked women to recall their lactation history (retrospective design), CARDIA and SWIFT enrolled participants before and during pregnancy, respectively, and followed them prospectively. Both studies also obtained biochemical measurements both prior to and after lactation, across time. The prospective and longitudinal measures of the antecedent metabolic risk factors are crucial for any causal inferences, because it is important to establish that women who breastfed or did not breastfeed, or who differed in lactation duration, had equivalency of maternal metabolism or obesity before lactation to address potential confounding, explained Gunderson.
Gunderson highlighted some of the key study design elements that set CARDIA apart from other epidemiological studies that have explored the relationship between lactation and long-term chronic disease risk. The study enrolled 2,787 women 18–30 years of age (mean age: 24 years) who were followed for 30 years. Half of the cohort was black and half was white. Biochemical data were collected longitudinally beginning from the preconception period, and assessments were performed years later to collect measures of diabetes and metabolic disease. The study’s data collection schedule allowed for a greater understanding of the cardiometabolic risk factors present prior to pregnancy, the changes in the metabolic risk factors, and other factors that may affect the lactation–disease relationship, includ-
ing adverse pregnancy outcomes. The prospective cohort naturally included a group of women who never breastfed.
A 2010 analysis of CARDIA data found positive relationships between increasing lactation duration and favorable changes in cardiometabolic outcomes (Gunderson et al., 2010). There was a strong inverse association between lactation duration and the incidence of metabolic syndrome in both women with and without gestational diabetes. Although women with gestational diabetes were at increased risk of developing metabolic syndrome, longer breastfeeding duration conferred to them a similar lower risk of metabolic syndrome, for whom the risk of the metabolic syndrome was similar among women with and without gestational diabetes with breastfeeding duration of 9 months or longer.
Overall, the CARDIA women were a relatively healthy cohort at baseline. Those who never breastfed had slightly higher BMIs and modestly higher insulin resistance than those with longer breastfeeding duration, although Gunderson questioned the clinical significance of the differences within normal levels. History of gestational diabetes and fasting glucose results, however, did not differ across the lactation duration groups. Given this context, Gunderson et al. (2018) found a strong, graded protective association among lactation duration groups and the risk of developing type 2 diabetes. Women who had gestational diabetes were at higher risk of developing diabetes during the 30 years, but they showed a similar decrease in relative risk with longer breastfeeding duration. Based on multivariate models that accounted for numerous covariates, including preconception metabolic risk factors,2 there remained strong inverse, graded associations between lactation duration and diabetes risk. Compared to women who never breastfed, those who breastfed for longer than 6 months had approximately a 50 percent relative reduction in the risk of developing diabetes. Weight change did not explain the association, indicating that lactation may exert other cardiometabolic effects.
SWIFT, also a prospective cohort study design, enrolled 1,035 women who had gestational diabetes diagnosed by Carpenter and Coustan criteria (Gunderson et al., 2012). All women underwent a research 2-hour, 75 g oral glucose tolerance test at each of three research visits from study baseline (6 to 9 weeks postpartum) and annually for up to 2 years later. A fourth in-person research visit will be conducted in 2021 to 2023 to reclassify glucose
2 Gunderson reported the models controlled for fasting glucose, insulin resistance, family history of diabetes, physical activity, diet quality, perinatal outcomes, and follow-up changes in weight.
tolerance in the cohort approximately 10 years postbaseline. Each research visit also gathered research measurements of anthropometry, and surveys of lactation, infant diet, lifestyle behaviors, sleep, and sociodemographic factors. Electronic health records continued to follow up after 2 years postbaseline for new diagnoses of diabetes. Gunderson described the systematic biochemical testing for diabetes as objective outcomes collected under research protocols. This study design allowed for the assessment of changes in metabolic risk factors, factors that may affect the lactation–disease relationships, and perinatal outcomes. The cohort also included a group of women who did not breastfeed. Gunderson noted there was high participant retention and the cohort was racially and ethnically diverse (75 percent minority).
SWIFT observed that a substantial portion of progression to diabetes occurred during the first years after a gestational diabetes pregnancy. Approximately 12 percent of SWIFT participants progressed to diabetes within the 2 years postbaseline, which increased to 20 percent by 8 to 9 years postbaseline. Higher breastfeeding intensity at 6–9 weeks postpartum was associated with a 50 percent lower relative risk of incident diabetes (Gunderson et al., 2015). The lactation intensity groups did not differ with respect to the sum of the z-scores for the prenatal 3-hour, 100 g oral glucose tolerance test, a measure of the severity of gestational diabetes, or gestational diabetes treatment (e.g., medication, insulin), indicating similar antecedent metabolic risk status. Multivariate models accounting for potential confounders showed a 50 percent lower relative risk of incident diabetes among women who breastfeed for 5 or more months, as compared to those who breastfed for 0–2 months. The lactation–incident diabetes risk was not mediated through weight loss, reported Gunderson.
Importance of Findings from CARDIA and SWIFT
Both CARDIA and SWIFT provide robust evidence for breastfeeding reducing the risk of developing diabetes in women across the childbearing years. There were strong, graded inverse relationships for lactation measures and the 2-year and 30-year incidence of diabetes in women who did and did not have gestational diabetes. This inverse relationship was seen after controlling for antecedent biochemical and clinical risk factors prior to lactation, psychosocial factors, sociodemographic risk factors, and lifestyle behaviors.
Gunderson acknowledged that the biological mechanisms by which lactation may protect against cardiometabolic disease have not been elucidated and would be necessary for causal inferences. She suggested that prolactin could exert an effect on pancreatic beta cells or that there could be underlying differences in mobilization of fat depots (e.g., visceral fat)
(Gunderson et al., 2014), but she surmised that other mechanisms could be driving the observed differences.
To contextualize the importance of the findings from CARDIA and SWIFT, Gunderson explained that the Diabetes Prevention Program had found that a 5 kilogram weight loss through a diet and physical activity intervention reduced the risk of incident diabetes in adults with prediabetes by 58 percent. Given the similar strength of associations for lactation intensity and duration, she conjectured, “This is a specific lifestyle behavior in a critical period, the postpartum period, that may actually have very strong metabolic consequences.” Gunderson cited evidence attributing an annual excess of deaths for women later in life to suboptimal breastfeeding. As heart disease is the leading cause of death among U.S. women, and type 2 diabetes is a risk factor, she concluded that the impact of lactation on women’s health has been underappreciated.
Allen, Demerath, and Gunderson responded to audience questions. In the discussion, moderated by O’Connor, questions were raised related to lactational programming, studying breast milk composition, and factors affecting lactation success.
Prefacing that her question related to the application of the concepts of lactational programming, Leanne Redman from the Pennington Biomedical Research Center asked if there were existing recommendations related to nutrition before, during, and after pregnancy to optimize breast milk composition. Allen explained that there are no WHO recommendations on supplementation during lactation. She thought that, until functional deficits in infants are identified, there will remain inadequate interest and investment in the topic.
A webcast audience member, who referenced a possible U-shaped relationship between maternal leptin concentrations and offspring obesity, wanted to know if there was an effect modification by BMI status. Demerath responded that more data are needed to better characterize the relationships. She noted that most studies do adjust for BMI, but effect modification could still be a factor.
Studying Breast Milk Composition
Redman asked the panelists for their ideas on how to advance the evidence on the relationships between maternal diet, maternal health, breast
milk composition, and infant outcomes, given that some randomized controlled trials may not be ethical to conduct. Demerath acknowledged that only certain types of trials would be considered unethical, such as removing an essential nutrient from the diet. She suggested that trials promoting healthy dietary patterns coupled with pregnancy weight management would be useful. Allen added that she has been involved in several micronutrient trials in developing countries and said the challenge is identifying and measuring functional outcomes in infants. When asked by a webcast audience member if it is known whether breast milk that is pumped, as opposed to consumed directly from the breast, confers the same health benefits for mother and child, Gunderson responded that SWIFT collected such data, but it has yet to be analyzed.
Referring to Allen’s presentation on the content of vitamin B12 in breast milk among vegetarian and vegans, Johanna Dwyer of the National Institutes of Health’s Office of Dietary Supplements pointed out that there were observational studies conducted in the 1970s on mothers in Boston following macrobiotic diets. Allen, who was involved in the study Dwyer referenced, noted that detrimental effects were found in the infants, but she said the analytical technique to measure vitamin B12 concentrations used in those studies is no longer considered valid.
Factors Affecting Lactation Success
Amelia Foley of Community of Hope asked Demerath and Gunderson whether their research had accounted for weight stigma and psychological stress. From a clinical perspective, Demerath agreed that there is too much focus on weight during pregnancy and postpartum, but she noted that weight gain outside of the current guidelines could have deleterious effects on birth outcomes and child health. She suggested that each individual’s context should be considered. Although stress is not a topic her group has investigated, Demerath thought it would be important to explore. Gunderson added that, despite associations between weight status and diabetes risk, lactation appears to have a physiological effect independent of weight change. Agreeing that weight should not be the primary focus during the postpartum period, she suggested that attention should be given to preventing stress, preventing and addressing depression, helping women get sufficient sleep, and helping women eat a healthy diet. Gunderson also emphasized the importance of paid family leave to help women recover from the physiological demands of pregnancy and lactation.
An unidentified audience member wanted to know if social factors that affect a woman’s ability to initiate and continue breastfeeding could be confounding results, and if there was evidence from countries with more generous parental leave policies. Gunderson responded that it is difficult to
compare studies from Scandinavian countries and China to results from the United States, as their leave policies can last up to 1 year. She explained that SWIFT did not find relationships between perinatal outcomes and breastfeeding intensity. Agreeing that social factors like employment and social support systems play a role, Gunderson mentioned that CARDIA and SWIFT showed that black women were 50 percent less likely to breastfeed, compared to their white, Asian, and Hispanic counterparts. She suggested that more research into the factors that lead to suboptimal breastfeeding is needed. Demerath added that a Canadian study reported differences with respect to the microbiome, but echoed Gunderson’s sentiment that this is an area that is understudied.