Micronutrient supplementation of lactating Guatemalan women acutely increases infants’ intake of riboflavin, thiamin, pyridoxal, and cobalamin, but not niacin, in a randomized crossover trial

ABSTRACT

Background

Maternal supplementation during lactation could increase milk B-vitamin concentrations, but little is known about the kinetics of milk vitamin responses.

Objectives

We compared acute effects of maternal lipid-based nutrient supplement (LNS) consumption (n = 22 nutrients, 175%–212% of the RDA intake for the nutrients examined), as a single dose or at spaced intervals during 8 h, on milk concentrations and infant intake from milk of B-vitamins.

Methods

This randomized crossover trial in Quetzaltenango, Guatemala included 26 mother–infant dyads 4–6 mo postpartum who were randomly assigned to receive 3 treatments in a random order: bolus 30-g dose of LNS (Bolus); 3 × 10-g doses of LNS (Divided); and no LNS (Control), with control meals. Mothers attended three 8-h visits during which infant milk consumption was measured and milk samples were collected at every feed. Infant intake was assessed as over 8 h.

Results

Maternal supplementation with the Bolus or Divided dose increased least-squares mean (95% CI) milk and infant intakes of riboflavin [milk: Bolus: 154.4 (138.2, 172.5) μg · min⁻¹ · mL⁻¹; Control: 84.5 (75.8, 94.3) μg · min⁻¹ · mL⁻¹; infant: Bolus: 64.5 (56.1, 74.3) μg; Control: 34.5 (30.0, 39.6) μg], thiamin [milk: Bolus: 10.9 (10.1, 11.7) μg · min⁻¹ · mL⁻¹; Control: 7.7 (7.2, 8.3) μg · min⁻¹ · mL⁻¹; infant: Bolus: 5.1 (4.4, 6.0) μg; Control: 3.4 (2.9, 4.0) μg], and pyridoxal [milk: Bolus: 90.5 (82.8, 98.9) μg · min⁻¹ · mL⁻¹; Control: 60.8 (55.8, 66.3) μg · min⁻¹ · mL⁻¹; infant: Bolus: 39.4 (33.5, 46.4) μg; Control: 25.0 (21.4, 29.2) μg] (all P < 0.001). Only the Bolus dose increased cobalamin in milk [Bolus: 0.054 (0.047, 0.061) μg · min⁻¹ · mL⁻¹; Control: 0.041 (0.035, 0.048) μg · min⁻¹ · mL⁻¹, P = 0.039] and infant cobalamin intake [Bolus: 0.023 (0.020, 0.027) μg; Control: 0.015 (0.013, 0.018) μg, P = 0.001] compared with Control. Niacin was unaffected.

Conclusions

Maternal supplementation with LNS as a Bolus or Divided dose was similarly effective at increasing milk riboflavin, thiamin, and pyridoxal and infant intakes, whereas only the Bolus dose increased cobalamin. Niacin was unaffected in 8 h. This trial was registered at clinicaltrials.gov as NCT02464111.

Introduction

Populations in the Guatemalan Western Highlands rely predominantly on rain-fed agriculture for their livelihood and food production, and experience high rates of food insecurity and dietary micronutrient inadequacy (1–4). The Western Highlands has high rates of stunting in children and short stature in adults (5–7). Growth faltering starts in utero and continues through infancy and childhood (8). Early growth faltering and insufficient growth recovery likely result in part from poor maternal nutrition during pregnancy, continuing postpartum during exclusive then partial breastfeeding. Because the typical duration of breastfeeding is ∼17–21 mo, with ∼90%–98% of mothers reporting exclusive breastfeeding until 6 mo postpartum (1), nutrition interventions for mothers could improve both maternal and infant status while further promoting breastfeeding.

It is becoming increasingly apparent that the concentrations of B-complex vitamins, with the exception of folate, are low in the milk of women with poor status and/or inadequate intakes of micronutrients (9–13). Maternal micronutrient deficiencies and low micronutrient concentrations in milk put the infant at risk of suboptimal health and developmental outcomes such as impaired neurological function [cobalamin (14), niacin (15)], developmental regression [cobalamin (14)], beriberi [thiamin (16, 17)], anemia [cobalamin (14), riboflavin (18)], and growth stunting [cobalamin (14), riboflavin (15)]. Deegan et al. (9) have reported very low concentrations of cobalamin in milk from Guatemalan women, but there are no data on other vitamins in Guatemalan women's milk.

The Breast Milk Quality (BMQ) study demonstrated an acute impact of maternal supplementation on milk thiamin, riboflavin, and pyridoxal concentrations but no impact on milk niacin (nicotinamide) or cobalamin in Bangladeshi mothers 2–4 mo postpartum (19). However, there is a lack of evidence regarding the impact of method of supplementation on milk vitamin concentrations and on infant intake. In Cameroon, the fortification of flour with vitamin B-12 resulted in a larger increase in milk vitamin B-12 concentrations than has been observed from high-dose supplementation (20). We hypothesize this response to be due to consumption of small amounts of the vitamin multiple times per day, as compared with less efficient absorption from a larger dose given once a day due to saturation of the intestinal cobalamin receptors (21, 22).

This study sought to determine the effects of maternal consumption of equal amounts of lipid-based nutrient supplement (LNS), consumed as a single dose compared with divided into 3 interval doses per day, on milk vitamin secretion and infant vitamin intakes at 4–6 mo postpartum. We hypothesized for cobalamin, riboflavin, thiamin, and pyridoxal that smaller amounts of LNS given to the participants over the course of the 8-h study visit would be absorbed and transferred to milk more efficiently than a single bolus dose, resulting in increased cumulative infant intakes of these vitamins.

Methods

Participants and location

A total of 30 lactating women and their infants were recruited and enrolled by study staff from community health centers run by the Ministry of Health, and from nongovernmental organization–affiliated health clinics in the district of Quetzaltenango, Guatemala, Central America (14.8°N, 91.5°W, 2368 m, as measured by Google Earth). Mothers were recruited from an urban population in Quetzaltenango, as well as from rural indigenous populations from surrounding villages including Salcajá and the Palajunoj Valley.

Eligibility criteria for mothers and infants included 4–6 mo postpartum, mothers apparently healthy, between 18 and 40 y of age, BMI (in kg/m²) ≥ 18.5, midupper arm circumference (MUAC) ≥ 12.5 cm, and breastfeeding 1 child; infants needed to be apparently healthy, ≥8 breastfeeding episodes/d, singleton birth, and not premature; and no micronutrient supplementation of the mother or infant for 3 mo before enrollment. Participant recruitment, enrollment, and all study activities took place between April and August 2015.

Ethics

The study protocol was approved by the Institutional Review Board at the University of California, Davis, and by the Human Subjects Committee at the Center for Studies of Sensory Impairment, Aging and Metabolism (CeSSIAM) in Guatemala. Written informed consent was obtained from all participants before their enrollment in the study (NCT02464111). Patients were treated ethically.

Study design

The study was a randomized experimental crossover design. We chose a crossover design because the outcomes evaluated are highly variable between people and less variable within a person across time. This design gave us higher power to detect differences between treatments using a smaller sample size.

Mothers received each of 3 treatments, in a randomly assigned order: a single 30-g dose of LNS (Bolus); 30 g of LNS divided into three 10-g portions (Divided); and no LNS (Control). The unit of randomization was the individual. To develop the randomization scheme, the numbers 1–6 were assigned to the 6 possible treatment distributions across the 3 clinic visits (treatments 1–6). The treatment number was assigned a random number between 0 and 1 using Microsoft Excel's random number generator (= rand). Based on this random number, the treatments were ordered in ascending sequence and participants were assigned to a treatment order based on the date of their first milk collection visit. The principal investigator generated the random allocation sequence and assigned participants to treatment groups. The principal investigator, research staff, participants, laboratory staff, and data analysts were not blinded to the treatment allocation. Two participants received the wrong order of treatment, yielding uneven numbers of participants receiving each treatment order.

The LNS supplement for pregnant and lactating women was formulated by Nutriset for the International Lipid-Based Nutrient Supplements project (23). It contained ∼1–2 times the RDA for lactation of multiple micronutrients, including all of the vitamins examined here (Table 1). The bolus supplement was 30 g and therefore could be consumed as a quick snack. We aimed for a flexible washout period of 7 d with ≥48 h. The washout period between supplements was 2–21 d with a mean of 6 d. A minimum washout period of 48 h was used to minimize the risk of a carryover effect. Previous studies suggest that 24 h is a sufficient washout period for cobalamin (19), riboflavin (19, 24), and niacin (19), but not for thiamin (19, 25) or pyridoxal (19, 26). However, pyridoxal transfer to milk peaks 3–6 h postsupplement (26), so it seemed unlikely 48 h would be an insufficient washout period. No published data suggested that a 48-h washout period would be insufficient for the vitamins examined at the dose examined.

TABLE 1.

Maternal nutrient intake from 30 g LNS and from study diet¹

Nutrient	LNS	Control diet	RDA/AI*	LNS as % of RDA or AI*
Calories, kcal	118
Protein, g	3.9	33.9	—	—
Fat, g	14.4	30.8	—	—
Linoleic acid, g	6.7	—	13*	52*
α-Linoleic acid, g	0.7	—	1.3*	54*
Vitamin A, μg	800	90.4	1300	62
Thiamin (vitamin B-1), mg	2.8	1.0	1.4	200
Riboflavin (vitamin B-2), mg	2.8	.6	1.6	175
Niacin (vitamin B-3), mg	36	8.1	17	212
Pantothenic acid (vitamin B-5), mg	7	2.8	7*	100*
Pyridoxal (vitamin B-6), mg	3.8	1.5	2	190
Cobalamin (vitamin B-12), μg	5.2	0	2.8	186
Folic acid, μg	400	198	500	80
Vitamin C, mg	100	44.0	120	83
Vitamin D, μg	10	0	15	67
Vitamin E, mg	20	0.4	19	105
Vitamin K, μg	45	19.7	90*	50*
Calcium, mg	280	520.2	1000*	28*
Copper, mg	4	1.0	1.3	308
Iodine, μg	250	—	290	86
Iron, mg	20	8.4	9	222
Magnesium, mg	65	372.3	310–320	21
Manganese, mg	2.6	2.5	2.6	100
Phosphorus, mg	190	1135.8	700	27
Potassium, mg	200	2201.4	5100*	4*
Selenium, μg	130	6.1	70	186
Zinc, mg	30	5.7	12	250

¹The supplement contained a micronutrient mix, peanuts, milk, sugar, and vegetable oil. AIs are for lactation in women aged 19–50 y. AIs and RDAs taken from Arimond et al. (23). *AI. AI, Adequate Intake; LNS, lipid-based nutrient supplement.

Participants attended 1 blood collection visit and then three 8-h breast-milk collection visits. During the blood collection visit, maternal and infant anthropometrics were measured, a baseline maternal blood sample was collected, and demographic and food-frequency surveys were administered. Data collectors were trained in anthropometric assessment and passed a standardization exam. Infant weight was measured to within 0.01 kg using a Tanita infant scale (BD 585) and infant recumbent length to within 0.5 cm using a portable length-measuring board. Maternal weight was obtained with a Seca scale (Seca Clara 803) and height was measured by a trained data collector using a measuring tape. Maternal MUAC was measured at the midpoint of the upper arm using a measuring tape. All anthropometric measurements were collected in triplicate. The mean of each value was used in the analyses. Infant z scores were calculated according to the WHO Child Growth Standards (27).

Ten milliliters of blood was drawn from the mother's antecubital vein by a trained phlebotomist and collected into a sterile tube before being transferred into 1 of 3 tubes for hematological analysis or for serum, plasma, or RBC separation. Plasma and RBCs were collected in EDTA-coated tubes, whereas serum was collected in silicone-coated tubes. Blood and milk samples were stored at −20°C for ≤16 d and then transferred to a −80°C freezer. Samples were shipped on dry ice to the USDA, Agricultural Research Service Western Human Nutrition Research Center in Davis, CA, for analysis.

During each of the 3 breast-milk collection visits mothers arrived at the clinic fasted, provided a mid-milk sample (considered the “baseline” milk sample), then consumed the Control, Bolus, or Divided treatment. During the next 8 h additional mid-milk samples were collected from the same breast during each feeding episode while infants nursed ad libitum from the other breast. Using a Medela hand pump, foremilk was collected for 1 min, the bottle was changed, and then a 10-mL mid-milk sample was collected into a new sterilized bottle. If the mother had trouble expressing milk using the pump, she was assisted in manually self-expressing the milk sample by trained study staff. Infants were weighed on a Tanita scale (BD 585) before and after each feeding episode to quantify the amount of milk consumed, which is known as the “test-weighing” method. When a mother thought her infant wanted to feed, she or the study staff would weigh the clothed infant on the Tanita scale. After weighing, the infant would be returned to the mother to feed. Immediately after the infant stopped feeding, it was returned to the scale to be weighed again. No diaper changes or addition or removal of clothing were permitted between the pre- and postfeed weighing. Although the test-weighing procedure is prone to error (28), it outperforms the deuterium dilution technique in estimation of single feeds (29). It may underestimate milk intake compared with feeding infants weighed expressed milk by ∼3% (29), but cultural constraints prevented feeding expressed milk to infants, so we decided the test-weighing method was the best-performing method available for the context.

On 5 occasions, an error was made when measuring milk consumption by weighing the infants before or after a feed. In this case, infant milk intake was interpolated using a linear regression model that predicted feed-specific milk intake based on the infant's ID number, time since the previous feed, number of feeds per visit, and the total volume consumed per visit:

(1)

The 3 control meals per visit consisted of local foods with relatively low micronutrient content and no fortified wheat or sugar, so the vitamins of interest were derived primarily from the supplement (Table 1). Participants were provided breakfast, a snack, and lunch. Breakfast consisted of plantain, corn tortilla, black beans, and coffee. The snack was a corn “tostada” (fried tortilla) with beet, cauliflower, and coffee. Lunch contained corn tortillas, white rice, güisquil (a Guatemalan squash), potato, lettuce, cucumber, lime, onion, and hibiscus tea. Meals were provided at 08:00, 10:30, and 13:00 and LNS was provided as a Bolus at 08:00, or as Divided doses at 08:00, 10:00, and 12:00. Participants left the clinic at 15:00, or 8 h after their arrival if they had arrived earlier or later than 07:00. The food composition databases used in assessing participant nutrient intake from the control meals were USDA SRS27 (30) and the Instituto de Nutrición de Centro América y Panamá Composition Table of Central American Foods (31).

Biochemical analyses

In human milk, pyridoxal, niacin (nicotinamide), riboflavin, and FAD were measured by ultra-performance liquid chromatography (UPLC)–tandem MS using a UPLC ACQUITY system (Waters) coupled to a 4000 QTRAP MS (ABSciex) as previously described (32). Free thiamin, thiamin monophosphate, and thiamin pyrophosphate were measured by HPLC with a fluorescence detector (HPLC-FLD) after precolumn derivatization to their thiochrome esters (33). Thiamin was calculated as free thiamin + (thiamin monophosphate × 0.871) + (thiamin pyrophosphate × 0.707) and riboflavin as free riboflavin + (FAD × 0.479). Cobalamin was analyzed using the IMMULITE competitive protein binding assay (Siemens) as previously described (34).

Hematology was analyzed by Democracia Lab in Quetzaltenango, and ferritin and C-reactive protein (CRP) by CareLab in Guatemala City. Ferritin was measured using chemiluminescence and CRP was measured using Latex Serology. Cobalamin in plasma was analyzed using the Cobas e411 Bioanalyzer (Roche).

Cutoffs for biomarkers and anthropometry

The cutoff for elevated CRP as an indicator for inflammation was >6 mg/L, analyzed using Latex Serology, a latex agglutination test. Although the conventional cutoff for elevated CRP is >5 mg/L, the laboratory was only able to provide estimates of concentrations >6 mg/L. Altitude-adjusted cutoffs for microcytic anemia included hemoglobin < 131 g/L and/or hematocrit < 38%. Marginal and deficient cobalamin status were defined as plasma cobalamin concentrations of 150–221 and <150 pmol/L, respectively. Cobalamin in milk was considered low at <310 pmol/L (35). Maternal BMI was classified based on the WHO standard for adults: a BMI ≥ 25 was overweight. The maternal stunting classification used the WHO 2007 reference curves (36), using the method outlined by Lundeen et al. (37) taking the height from the WHO 2007 growth curves for 19 y, which is considered to be the age after which growth is negligible. The cutoffs for severe, moderate, and mild adult female stunting were 143.5, 150.1, and 156.6 cm, respectively.

Infant anthropometry

Length-for-age z score was classified as severely stunted (<−3), moderately stunted (−2 to −3), or normal (−2), whereas weight-for-age z score was classified as severely underweight (<−3), moderately underweight (−2 to −3), normal (−2 to 2), or overweight (>2) (27).

Data analysis

Sample size estimation

In the BMQ trial the changes in milk concentrations of thiamin, riboflavin, niacin, pyridoxal, and cobalamin in mothers supplemented with 1 RDA of B-vitamins 2–4 mo postpartum were 3.9 SD, 1.5 SD, 2.0 SD, 2.1 SD, and 0.9 SD, respectively (19). We aimed to detect a difference of 0.58 SD, which is still a fairly large effect size. With 24 mothers at a 5% level of significance and 80% power, we could detect a difference of 0.58 SD between means of micronutrient concentrations in milk. Sample size estimation was based on a 1-sided test assuming a within-subject correlation of 0.5 using SAS. Participants were recruited until 24 mother–infant dyads had provided complete biological samples and survey information.

Definitions of primary and secondary outcomes

The primary outcomes in this study were riboflavin, pyridoxal, thiamin, niacin, and cobalamin secretion in milk, and secondary outcomes were infant intakes of riboflavin, pyridoxal, thiamin, niacin, and cobalamin.

Milk vitamin secretion

The milk vitamin secretion was an 8-h estimate of milk micronutrient content using AUC, which has been previously used to estimate milk content of multiple compounds (38, 39). The AUC vitamin secretion into milk was estimated assuming that the micronutrient concentration of each feed represents the average of nutrient secretion into milk between the current feed and previous feed. Although we expect that the secretion into milk would vary during the time that milk pools in the breast, the concentration at the time of expression represents a biological average of the different transfer rates since the last time that the breast was sampled. The AUC secretion to milk was approximated by multiplying the micronutrient concentration at each feed (from the second to the last feed) by the time that had elapsed since the previous feed and then summing the products of milk concentration and time over the 8-h period. We chose to use the AUC because, in addition to a simple average of concentrations, it includes a time component, which weights the concentration of a feed that has had longer to accumulate in the breast more than those that have accumulated over a shorter time frame: this provides a time-weighted average of milk concentration. AUC is not influenced by feed frequency. One sample had milk vitamin concentrations >15 SD from the mean which were replaced with a mean of values from the previous and following feeds of that participant during the same treatment.

Milk mean concentrations

Milk was expressed as 8-h concentration means so that our data could be easily compared with previous reports which used this method. Milk mean concentrations were calculated by averaging nutrient concentrations during the 8-h clinic visit within participants.

Milk kinetics

We estimated the kinetics of the transfer of micronutrients into milk to confirm that the study design adequately captured the impact of the supplement on the milk nutrient concentrations. In the kinetic profile diagrams “interval zero” represents the geometric mean fasted milk vitamin concentration for each treatment. Subsequent “intervals” of time were calculated as time postconsumption of the first supplement, and classified in 2-h segments, yielding intervals 0, 2, 4, 6, and 8 representing time 0, 0–2 h, 2–4 h, 4–6 h, and 6–8 h, respectively. For the Control group, the 0 interval represents the time when the supplement would have been consumed. Concentrations for each interval were calculated as mean vitamin concentrations within women for all feeding episodes in an interval. When a dyad fed more than once during a 2-h time interval, the values were averaged.

Infant vitamin intake

Infant vitamin intake was calculated by multiplying the feed-specific volume by the vitamin concentration at that feed, and then summing across all feeds during the 8-h observation period. Milk volumes >99^th percentile were truncated to the 99^th percentile value (173 mL).

Statistical analysis

Data were analyzed using SAS Statistical Analysis Software version 9.4 (SAS Institute). Dyads who attended all 3 milk-collection visits were included in the analysis of milk micronutrients and infant intakes. Descriptive statistics were calculated for all variables. For infant milk consumption estimates, means were calculated within and among dyads, meaning a mean was calculated for 8-h infant milk intake across all treatments from a single participant, yielding a participant mean, and then the mean of the participant means was calculated and reported. P values < 0.05 were considered to be significant.

To normalize the distribution of outcome variables, log transformations were applied to milk AUC, milk mean, and infant intake of riboflavin, niacin, pyridoxal, and thiamin. Square root transformations were applied to milk AUC, milk mean, fasted baseline milk, and infant intake of cobalamin, and log transformations to plasma cobalamin.

Milk vitamin secretion, milk mean concentrations, and milk kinetics were analyzed using linear regression with mixed effects. Each vitamin was analyzed in an independent model that included the dyad as a random effect. In order to control for the uneven number of participants receiving each treatment order, milk vitamin secretion and milk mean concentration models were tested to see if the treatment order was significantly associated with the outcome. If the association was significant at P < 0.05, the treatment order was retained in the final model. P values were adjusted using the Tukey–Kramer method.

Milk AUC vitamin secretion controlled for the time between the first and last feeds during each clinic visit in order to remove the influence of the duration of observation on the outcome estimation. The mean percentage difference was calculated by taking the difference between an individual's response to 2 treatments, and then averaging across participants.

One outlier was identified in sensitivity analyses for each of the following outcomes: mean and AUC thiamin; mean, AUC, and infant intake of niacin; and mean milk and infant intake of cobalamin. The outlier treatment value was excluded from the final model for each outcome, but the other 2 treatment values were retained in the analyses.

We tested the association between maternal BMI and milk AUC vitamin content by calculating the Pearson partial correlation coefficient for BMI and milk AUC vitamin content controlling for treatment, time between the first and last feeds during each clinic visit, the treatment order, and the vitamin intake from the control meals. The partial R² is shown, which is calculated as the square of the partial correlation coefficient.

Linear regressions with mixed effects were used to evaluate whether the kinetics of the transfer of nutrients to milk differed between treatments across time and at specific times postsupplement. To evaluate this, the models for milk kinetics tested for a significant treatment × time interaction, splitting by time interval, and evaluated pairwise differences between treatments at the different time intervals.

Because the effects of the Divided dose persisted beyond the 8-h observation period, it is possible that the study design did not adequately capture its effects, and therefore the impact of the Divided dose may be underestimated relative to the effects of the Bolus dose. We therefore used an exploratory analysis to examine if the duration of the observation period affected the results of the impact of the treatment on milk vitamins. We created a theoretical trajectory for the Bolus dose as if it had been consumed in 3 parts, as the Divided dose was. This “theoretical divided Bolus dose” was used for additional comparison of the vitamin transfer to milk between the Bolus and the Divided dose. If the theoretical Bolus and Divided concentration kinetics did not differ, it would indicate that the observation window did not significantly affect the comparison of the Bolus and Divided dose. If the Divided dose was higher than the theoretical Bolus dose, it would indicate that the observation window limited the ability to compare the effects of the Divided and Bolus dosing methods.

Correlation between plasma and milk cobalamin

Pearson's correlation coefficient was used to assess the relation between milk cobalamin and maternal plasma cobalamin at baseline (presupplementation).

Results

Participant characteristics

A total of 30 mother–infant dyads were recruited and randomly assigned to treatment-order groups. Four participants were lost to follow-up: 1 attended the blood draw but none of the milk collection visits, 2 attended the blood draw and 1 milk collection visit, and 1 was excluded based on health concerns, so milk and blood samples were analyzed from 26 mothers (Figure 1). The primary and secondary outcomes were calculated based on data from 26 dyads. Randomization and follow-up were ongoing between April and August 2015.

FIGURE 1.

CONSORT participant flow diagram. Participants who were lost during the washout either withdrew from the study (n = 1) or were excluded based on health concerns (n = 1). CONSORT, Consolidated Standards of Reporting Trials.

The mean age of the mothers was 28 y, and they were heavy (62% overweight) for their relatively short height (Table 2). None were underweight. Nearly 80% of the mothers were stunted. The prevalence of mild, moderate, and severe stunting in mothers was 50%, 15%, and 4%, respectively. Seventeen percent of mothers reported consuming cobalamin supplements.

TABLE 2.

Baseline characteristics of mothers, households, and infants¹

Characteristics	Mean ± SD or %
Maternal
Age, y	28.3 ± 6.5
Weight,2 kg	54.9 ± 10.6
Height, cm	148.7 ± 6.1
BMI,2 kg/m2	24.8 ± 4.1
Midupper arm circumference, cm	28.0 ± 3.1
Parity	2.8 ± 2.0
Birth spacing, mo	55.8 ± 25.5
Daily expenditure on food per person, USD	1.63 ± 1.12
Literacy	88
Prevalence of overweight	62
Prevalence of microcytic anemia	8
Prevalence of CRP > 6 mg/L	8
Prevalence of deficient plasma cobalamin	4
Prevalence of marginal plasma cobalamin	23
Household
Household size	4.9 ± 2.5
Infant
Age, mo	4.5 ± 0.5
Weight, kg	7.0 ± 1.1
Length, cm	61.0 ± 3.0
LAZ	−1.3 ± 1.2
WAZ	0.0 ± 1.2
Male, %	50

¹ n = 26. Values adjusted for an altitude of 2333 m. CRP, C-reactive protein; LAZ, length-for-age z score; WAZ, weight-for-age z score.

²To calculate maternal weight and BMI, the weight of the clothing was removed: 2 kg for modern clothing and 5 kg for traditional clothing. Overweight is defined as BMI ≥ 25 kg/m².

Approximately 45% of the infants in our sample were male, and the mean age was 4.5 mo (Table 2). Overall 3%, 93%, and 3% of infants were classified as being overweight, normal weight, and underweight, respectively; two-thirds were normal length, 21% were moderately stunted, and 14% severely stunted.

Vitamin secretion into milk

When the LNS was provided as a Bolus dose, milk riboflavin AUC was higher than when the LNS was provided as a Divided dose (pairwise P = 0.008) and higher than Control (pairwise P < 0.001) (Table 3, Figure 2). When LNS was provided, milk cobalamin AUC was higher in the Bolus dose than in the Control (pairwise P = 0.039) but there was no difference between the Divided supplement and the Control (pairwise P = 0.975). Milk AUC thiamin and pyridoxal were higher in the Bolus and Divided doses than in the Control (pairwise thiamin Bolus-Control P < 0.001, Divided-Control P < 0.001; pairwise pyridoxal Divided-Control P < 0.001, Bolus-Control P < 0.001), but did not differ between the Divided and Bolus doses (Divided-Bolus pairwise thiamin P = 0.986, pyridoxal P = 0.218). The Bolus dose increased milk AUC riboflavin, thiamin, pyridoxal, niacin, and cobalamin compared with the Control by a mean of 97%, 47%, 56%, 27%, and 427%, respectively (Figure 2). Milk AUC thiamin (partial R² = 0.28) and pyridoxal (partial R² = 0.17) were positively associated with maternal baseline BMI (all P < 0.05).

TABLE 3.

AUC of vitamin secretion into milk, mean concentrations, and infant intake from milk during 8 h, by treatment¹

					P values
		AUC secretion to milk and infant vitamin intake				Pairwise	Pairwise Control-	Pairwise
Vitamin	Unit of measure, per 8 h	Control	Bolus	Divided	Treatment	Control-Bolus	Divided	Bolus-Divided
Riboflavin2	AUC secretion to milk,3 μg · min−1 · mL−1	84.5 [75.8–94.3]	154.4 [138.2–172.5]	120.9 [108.6–134.6]	<0.001	<0.001	<0.001	0.008
	Mean concentration,4 μg/L	205.9 (185.9, 228.1)	340.9 (307.3, 378.2)	283.0 (255.9, 312.9)	<0.001	<0.001	<0.001	0.035
	Infant intake,5,6 μg	34.5 (30.0, 39.6)	64.5 (56.1, 74.3)	49.4 (43.1, 56.7)	<0.001	<0.001	0.002	0.029
Thiamin7	AUC secretion to milk,3,6,8 μg · min−1 · mL−1	7.7 [7.2–8.3]	10.9 [10.1–11.7]	11.0 [10.2–11.8]	<0.001	<0.001	<0.001	0.986
	Mean concentration,4,6 μg/L	19.9 (18.5, 21.4)	25.0 (23.1, 27.0)	26.0 (24.2, 28.0)	<0.001	<0.001	<0.001	0.719
	Infant intake,5,6 μg	3.4 (2.9, 4.0)	5.1 (4.4, 6.0)	4.8 (4.1, 5.6)	0.0014	0.0029	0.008	0.858
Pyridoxal	AUC secretion to milk,3 μg · min−1 · mL−1	60.8 [55.8–66.3]	90.5 [82.8–98.9]	81.2 [74.3–88.7]	<0.001	<0.001	<0.001	0.218
	Mean concentration,4 μg/L	148.2 (136.1, 161.3)	202.9 (186.0, 221.5)	190.9 (174.9, 208.3)	<0.001	<0.001	<0.001	0.598
	Infant intake,5,6 μg	25.0 (21.4, 29.2)	39.4 (33.5, 46.4)	33.6 (28.7, 39.5)	<0.001	<0.001	0.029	0.399
Niacin8	AUC secretion to milk,3 μg · min−1 · mL−1	80.7 [70.6–92.3]	85.0 [73.9–97.8]	86.0 [74.9–98.8]	0.742
	Mean concentration,4 μg/L	194.8 (167.2, 227.0)	209.5 (179.0, 245.1)	195.6 (167.0, 229.2)	0.664
	Infant intake,5,6 μg	32.2 (27.5, 37.8)	37.3 (31.7, 43.9)	36.4 (30.9, 42.8)	0.216
Cobalamin	AUC secretion to milk,3 μg · min−1 · mL−1	0.041 [0.035–0.048]	0.054 [0.047–0.061]	0.042 [0.036–0.049]	0.026	0.039	0.975	0.064
	Mean concentration,4 μg/L	1.36 (1.33, 1.38)	1.41 (1.38, 1.44)	1.36 (1.33, 1.38)	0.010	0.020	1.000	0.021
	Infant intake,5 μg	0.015 (0.013, 0.018)	0.023 (0.020, 0.027)	0.018 (0.015, 0.021)	0.001	0.001	0.427	0.0032

¹ n = 26 dyads. Treatment differences were analyzed using mixed-effects models with a random intercept for subject and controlled for multiple comparisons. Values are least-squares means (95% CIs). The Bolus treatment was 30 g LNS, the Divided treatment was 10 g LNS provided 3 times. LNS, lipid-based nutrient supplement.

²Riboflavin = riboflavin + (FAD × 0.497 × 5.24).

³Values are mean [IQR] milk nutrient concentration from the 8-h clinic visit, for each treatment group. P values were based upon mixed-effects models accounting for pair matching. Tukey's test was used for pairwise comparisons.

⁴Milk content was calculated using AUC and represents a time-weighted average of the milk vitamin concentrations multiplied by the amount of time between the first and last feeds. Treatment differences were analyzed using mixed-effects models with a random intercept for subject. Analysis was controlled for the time between the first and last feeds. Where the order in which the treatments were provided was significantly associated with the outcome, treatment order was included as a covariate.

⁵Infant intake is the sum of the products of feed-specific milk intake and milk vitamin concentration. Treatment differences were analyzed using mixed-effects models with a random intercept for subject.

⁶Analysis was controlled for treatment order.

⁷Thiamin = thiamin + (thiamin monophosphate × 0.871) + (thiamin pyrophosphate × 0.707).

⁸Niacin = nicotinamide.

FIGURE 2.

Percentage change in AUC milk B-vitamin content. The AUC was approximated using rectangles: the milk micronutrient concentration at each feed (from the second to the last feed) was multiplied by the time that had elapsed since the previous feed. All rectangles representing products of milk concentration and time were summed over the 8-h period to calculate the AUC secretion to milk. Error bars are SEs. n = 26. Thiamin = thiamin + (thiamin monophosphate × 0.871) + (thiamin pyrophosphate × 0.707). Niacin = nicotinamide.

The mean daily (8-h) breast-milk vitamin concentrations reflected the impact of the supplement (Table 3, Figure 3): when mothers received LNS, their mean milk concentrations of riboflavin, thiamin, pyridoxal, and cobalamin (P ≤ 0.001 for all), but not niacin, differed between the Control and the supplementation methods. None of the mothers had adequate milk cobalamin, based on a cutoff of 310 pmol/L (0.42 μg/L). Fasted baseline milk cobalamin and maternal serum cobalamin were correlated, with a Pearson correlation coefficient of 0.51 (P = 0.008).

FIGURE 3.

Mean infant vitamin intakes during 8 h expressed as percentages of their AIs, assuming infants would consume 33% of their 24-h intake over this time. Error bars are SEs. n = 26. Thiamin = thiamin + (thiamin monophosphate × 0.871) + (thiamin pyrophosphate × 0.707). Niacin = nicotinamide. AI, Adequate Intake.

Kinetics of micronutrient transfer to milk

The kinetic profile of the changes in the vitamin concentrations throughout the study visit demonstrated a clear effect of the treatment on milk riboflavin, thiamin, pyridoxal, and cobalamin, but not on niacin (Table 4, Figure 4A–E). For riboflavin, pyridoxal, and cobalamin, the milk concentration peaked 4–6 h postsupplement (Figure 4A, C, E).

TABLE 4.

P values for kinetic differences between intervention treatments across intervals of time¹

Vitamin			F test	Interval 0 (0 h)	Interval 1 (0–2 h)	Interval 2 (2–4 h)	Interval 3 (4–6 h)	Interval 4 (6–8 h)
Riboflavin	Treatment		<0.001
	Time		<0.001
	Treatment × time interaction P		<0.001	0.853	0.015	<0.001	<0.001	<0.001
	Pairwise treatment × time interaction P	Bolus- Control		0.848	0.020	<0.001	<0.001	0.023
		Divided- Control		0.987	0.881	0.123	<0.001	<0.001
		Divided- Bolus		0.920	0.053	<0.001	0.197	0.016
		Theoretical Bolus vs. Divided	0.009		0.688	0.423	0.413	0.004
Thiamin	Treatment		<0.001
	Time		<0.001
	Treatment × time interaction P		<0.001	0.809	0.933	<0.001	<0.001	<0.001
	Pairwise treatment × time interaction P	Bolus- Control		0.817	0.939	<0.001	<0.001	<0.001
		Divided- Control		0.995	0.948	0.006	<0.001	<0.001
		Divided- Bolus		0.868	0.999	0.582	0.994	0.458
		Theoretical Bolus vs. Divided	<0.001		0.919	0.073	<0.001	0.001
Pyridoxal	Treatment		<0.001
	Time		<0.001
	Treatment × time interaction P		<0.001	0.800	0.703	<0.001	<0.001	<0.001
	Pairwise treatment × time interaction P	Bolus- Control		0.917	0.806	<0.001	<0.001	0.012
		Divided- Control		0.963	0.989	0.005	<0.001	<0.001
		Divided- Bolus		0.785	0.705	0.061	0.949	0.751
		Theoretical Bolus vs. Divided	0.014		0.443	0.212	0.012	0.880
Niacin	Treatment		0.185
	Time		<0.001
	Treatment × time interaction P		0.420	0.242	0.672	0.780	0.545	0.021
	Pairwise treatment × time interaction P	Bolus- Control		0.223	0.677	0.765	0.515	0.081
		Divided- Control		0.835	0.980	0.898	0.891	0.032
		Divided- Bolus		0.525	0.779	0.965	0.789	0.937
		Theoretical Bolus vs. Divided	0.090		0.889	0.170	0.029	0.061
Cobalamin	Treatment		0.002
	Time		<0.001
	Treatment × time interaction P		0.200	0.080	0.007	0.082	0.008	0.025
	Pairwise treatment × time interaction P	Bolus- Control		0.090	0.026	0.119	0.257	0.102
		Divided- Control		0.926	0.993	0.988	0.211	0.035
		Divided- Bolus		0.197	0.012	0.153	0.005	0.910
		Theoretical Bolus vs. Divided	0.551		0.561	0.765	0.834	0.177

¹ n = 26. S1, timing of supplement 1: ∼08:00, supplement (Bolus or Divided) or no supplement (Control) provided; S2, timing of supplement 2: ∼10:00, supplement (Divided) or no supplement (Bolus, Control) provided; S3: timing of supplement 3: ∼12:00, supplement (Divided) or no supplement (Bolus, Control) provided. There was some variation in the exact timing of the supplement, but no variation was present as calculated using the interval system. Interval 0: 07:00–08:00 (before supplementation), Interval 1: 08:00–10:00 (0–2 h postsupplement), Interval 2: 10:00–12:00 (2–4 h postsupplement), Interval 3: 12:00–14:00 (4–6 h postsupplement), Interval 4: 14:00–16:00 (≥6 h postsupplement). Thiamin = thiamin + (thiamin monophosphate × 0.871) + (thiamin pyrophosphate × 0.707). Niacin = nicotinamide. Treatment P values are based upon linear regression with mixed effects, accounting for pair matching. P values comparing groups are adjusted for multiple comparisons, whereas P values for theoretical comparisons are not adjusted. Tukey's test was used for pairwise comparisons. Significant at P < 0.05.

FIGURE 4.

P values for kinetic profiles of vitamin transfer to milk with the theoretical divided Bolus dose. (A) Riboflavin, (B) thiamin, (C) pyridoxal, (D) niacin, (E) cobalamin. S1, timing of supplement 1: ∼08:00, supplement (Bolus or Divided) or no supplement (Control) provided; S2, timing of supplement 2: ∼10:00, supplement (Divided) or no supplement (Bolus, Control) provided; S3: timing of supplement 3: ∼12:00, supplement (Divided) or no supplement (Bolus, Control) provided. There was some variation in the exact timing of the supplement, but no variation was present as calculated using the interval system. Interval 0: 07:00–08:00 (before supplementation), Interval 1: 08:00–10:00 (0–2 h postsupplement), Interval 2: 10:00–12:00 (2–4 h postsupplement), Interval 3: 12:00–14:00 (4–6 h postsupplement), Interval 4: 14:00–16:00 (≥6 h postsupplement). Thiamin = thiamin + (thiamin monophosphate × 0.871) + (thiamin pyrophosphate × 0.707). Niacin = nicotinamide. n = 26 for all models. Treatment P values are based upon linear regression with mixed effects, accounting for pair matching. Letters denote significant pairwise differences for the interaction between treatment and time, adjusting for multiple comparisons. Tukey's test was used for pairwise comparisons. Significance at P < 0.05.

When mothers received the Bolus or Divided dose, they had higher concentrations of thiamin and pyridoxal in their milk at 2–8 h postsupplement (intervals 2–4) than when they received the Control (Table 4, Figure 4B, C). Whereas milk cobalamin concentrations when mothers received the Bolus dose were higher than when they received the Control at 0–2 h (interval 1), by 8 h (interval 4), cobalamin concentrations when mothers received the Divided treatment were higher than concentrations in the Control group (Table 4, Figure 4E).

For milk concentrations of riboflavin and thiamin, the trajectory of the theoretical divided Bolus was lower than the trajectory of the Divided dose at 8 h postsupplement (interval 4) (Table 4, Supplemental Figure 1). For niacin and cobalamin, the trajectory of the theoretical divided Bolus dose did not differ from the trajectory of the Divided dose (Table 4, Supplemental Figure 1).

Infant milk and micronutrient intake

Linear regression with mixed effects was used to evaluate the effect of vitamin supplementation and method of dose delivery on infant vitamin intakes. The infants averaged (mean ± SD) 5.7 ± 1.0 feeding episodes during the 8-h clinic visits and consumed 38.2 ± 14.6 mL/feed. Mean ± SD total milk consumption was 206.7 ± 65.5 mL in 8 h. One outlier was identified in a sensitivity analysis of infant cobalamin intake and 1 outlier was identified in infant niacin intake, and they were excluded from the final model for each nutrient.

Infants consumed more riboflavin, thiamin, and pyridoxal (P ≤ 0.001) when their mothers received the Bolus or Divided supplement (Table 3, Figure 3). Infants consumed more cobalamin when mothers consumed the Bolus supplement than when mothers were provided the Control treatment or Divided supplement (treatment P = 0.001, Control-Bolus P = 0.001, Bolus-Divided P = 0.003). Infant niacin intake was not different when mothers consumed the Bolus or Divided supplement or when mothers received the Control treatment (P = 0.216).

The maternal Bolus LNS increased infant intakes of riboflavin, thiamin, pyridoxal, niacin, and cobalamin by a mean of 86%, 52%, 61%, 31%, and 412%, respectively, compared with the Control. Infant intakes of each vitamin in the 8-h period were also expressed as a percentage of the recommended daily Adequate Intakes (AIs) for infants this age (Figure 3, Supplemental Table 1). Assuming that during 8 h the infant would consume ≥33% of their daily intakes (one-third of the AI), the riboflavin, thiamin, niacin, and cobalamin intakes were far lower than this, even with maternal supplementation. The mean infant pyridoxal intake was 29% of the AI without supplementation, and increased to 41% after maternal LNS consumption.

Discussion

We have demonstrated that supplementation with a single day's recommended intake can increase both breast-milk content and infant intake of riboflavin, thiamin, pyridoxal, and cobalamin. The Bolus and Divided delivery methods were similarly effective at increasing milk pyridoxal over the Control, whereas the Bolus dose was more effective than the Divided dose at increasing milk content and infant intakes of riboflavin and cobalamin during the 8-h study.

To the best of our knowledge, this study is the first to compare the acute impact of supplementation method on breast-milk B-vitamin concentrations, to quantify the impact of maternal vitamin supplementation on infant vitamin intake by measuring infant milk consumption, and to report on the kinetics of B-vitamins in milk in response to supplementation beyond pyridoxal (26). Furthermore, it is the first to examine breast-milk B-vitamin concentrations in Guatemala, beyond cobalamin (9).

Although a few long-term supplementation trials using high doses of cobalamin have demonstrated that milk cobalamin can respond to supplementation (40, 41), BMQ was the first trial, to our knowledge, to examine the acute impact of supplementation on milk cobalamin and other B-vitamins. The BMQ study from Bangladesh reported an increase in milk riboflavin, thiamin, and pyridoxal, but not cobalamin or niacin, within a few hours after a single Bolus dose of 2 times RDA of these vitamins (19). Our findings are in agreement with these findings from Bangladesh for all examined vitamins except cobalamin. Unlike BMQ, we found that mean milk cobalamin and AUC cobalamin secretion to milk were higher in mothers who received LNS than in those who did not. This may be due to the difference in stage of lactation (2–4 mo in BMQ compared with 4–6 mo here), the definition of the milk cobalamin outcome (daily median in BMQ compared with daily mean and AUC here), or the fact that 17% of mothers reported taking cobalamin supplements, although supplementation was an exclusion criterion. We repeated our cobalamin analysis using medians instead of means and the treatment lost significance (P = 0.07), which suggests that daily means may be a more sensitive parameter than medians. Larger trials should be conducted in order to evaluate the effect of cobalamin supplementation on milk cobalamin concentration.

The mean total milk consumption of 207 mL in 8 h approximates one-third of the expected 24-h intake of exclusively breastfeeding infants of this age (42). Although the Bolus dose increased infant intakes of riboflavin, thiamin, pyridoxal, niacin, and cobalamin, with the exception of pyridoxal, supplementation did not come close to increasing milk vitamins to an amount that would meet the recommended intake for infants (33% of the AI) (42). The Bolus provided only 23%, 3%, 2%, and 7% of the AI for riboflavin, thiamin, niacin, and cobalamin, respectively. The lack of a response in infant intakes could have resulted from methodological limitations in estimating infant intake, the use of the AI as the target for infant intake, or the prevalence or severity of maternal micronutrient deficiencies.

The use of the test-weighing method underestimates milk intake slightly compared to feeding weighed expressed milk (29), but the test-weighing method was more appropriate for the cultural context of the present research. The AIs are based on milk values reported in the literature, and for some nutrients, their validity has been questioned (9, 43, 44). Yet, the AIs remain the best reference estimates currently available. The theory that infants will consume ∼33% of their daily milk intake during the 8-h visit is based on the assumption that infants consume equal milk volumes throughout the day and night, which has not been evaluated.

For milk vitamins to reach adequate concentrations to meet infants’ recommended intakes it is likely that mothers must first achieve adequacy of their own nutritional status (43, 45). A longer-term intervention is needed in this population to assess the efficacy of maternal supplementation for improving micronutrient concentrations in breast milk, including higher doses, a longer intervention duration, a larger sample size, and evaluation of micronutrient status in mothers. Our sample size of 26 women was small, and a larger study needs to be carried out in order to confirm the generalizability of these results.

Current programs in Guatemala that include child supplementation with LNS would benefit from including pregnant and lactating women, and women of childbearing age. Fortification may be a more effective approach than supplementation; infant status has been reported after vitamin A fortification of sugar in Guatemala (46), cobalamin fortification of flour in Cameroon (20) and thiamin fortification of fish sauce in Cambodia (47).

The duration of the observation period was not long enough to capture the entirety of the transfer of nutrients from the supplement to the milk. The lack of greater effect of the Divided dose on milk cobalamin AUC and infant cobalamin intake may have been due to the longer time needed for absorption of cobalamin; a dietary dose of cobalamin typically peaks in plasma ∼7 h postconsumption (48). To address this limitation, we evaluated the expected trajectory of milk vitamins if the Bolus dose had been provided as 3 divided doses. These theoretical kinetic results show that the milk concentrations were higher when the mothers received the Divided dose than we would expect if the Bolus dose were divided into 3 doses for riboflavin, thiamin, and pyridoxal, but not for niacin or cobalamin. This suggests a higher bioavailability or higher nutrient transfer to milk with the Divided dose than with the Bolus dose, for riboflavin, thiamin, and pyridoxal. This also demonstrates that the study duration did not affect cobalamin concentrations substantially. The lack of an effect for niacin may have been due to the fact that the control meals provided 48% of the RDA for niacin, reducing the potential to affect milk niacin.

No information is currently available to demonstrate if 48 h is a sufficient washout period after a single day's supplement. The significance of treatment order in our models indicates that 2 d may not have been a sufficient washout period.

It is important to emphasize the importance of breastfeeding for infant health and growth. Our results indicate that breast-milk B-vitamin concentrations may be low in this population. These results should be interpreted with caution and do not imply that infants should be supplemented with formula. Rather, they demonstrate the need for nutritional interventions targeting women who are at risk of malnutrition while pregnant and lactating. The benefits of breastfeeding for infants are indisputable including lower morbidity, increased survival, lower risk of developing obesity and diabetes, increased IQ scores (49), and improved microbiome composition and immune function (50, 51). Thus, there should be no debate that human milk, even from a malnourished mother, is the optimum food for the infant. Future steps in research and policy should focus on improving maternal nutrition and milk quality while continuing to encourage breastfeeding.

The amounts of the other B-vitamins consumed by the infants were similar between the Bolus and Divided doses and although maternal supplementation increased milk vitamins, the concentrations remained much too low to provide the AI for infants. Programs and policies providing vitamin supplementation to mothers for the benefit of mothers and infants could provide either Bolus supplements or Divided supplements in the form of fortified foods.

Notes

Supported by Fogarty International Center of the NIH award R25TW009343 (to JAD), the University of California Global Health Institute, a Blum Center for Developing Economies at the University of California, Davis PASS grant (to JAD), and USDA Agricultural Research Service Western Human Nutrition Research Center Project #5306-51530-019-00 (to LHA). The micronutrient supplement was provided by Nutriset.

USDA is an equal opportunity employer and provider. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH, University of California Global Health Institute, or USDA.

Supplemental Figure 1 and Supplemental Table 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/ajcn/.

Data described in the article, code book, and analytic code will be made available upon request pending application and approval.

Abbreviations used: AI, Adequate Intake; BMQ, Breast Milk Quality; CeSSIAM, Center for Studies of Sensory Impairment, Aging and Metabolism; CRP, C-reactive protein; LNS, lipid-based nutrient supplement; MUAC, midupper arm circumference; UPLC, ultra-performance liquid chromatography.