Long Chain Polyunsaturated Fatty Acid Levels in U.S. Donor Human Milk: Meeting the Needs of Premature Infants?

Abstract

Objective

To determine fatty acid levels in the US donor milk supply.

Study Design

Donor human milk samples from Iowa (n=62), Texas (n=5), North Carolina (n=5), and California (n=5) were analyzed by gas chromatography. Levels in Iowa donor milk were compared before and after pasteurization using Student’s t-test. Docosahexaenoic acid (DHA) and arachidonic acid (ARA) levels were compared among all milk banks using ANOVA.

Results

ARA (0.4 pre, 0.4 post, p=0.18) and DHA (0.073 pre, 0.073 post, p=0.84) were not affected by pasteurization. DHA varied between banks (p <0.0001), whereas ARA did not (p = 0.3). DHA levels from all banks were lower than published values for maternal milk and infant formula (p<0.0001).

Conclusion

Pasteurization of breastmilk does not affect DHA or ARA levels. However, DHA content in US donor milk varies with bank location and may not meet the recommended provision for preterm infants.

Introduction

Long chain polyunsaturated fatty acids (LCPUFA) including docosahexaenoic acid (DHA) and arachidonic acid (ARA), are essential for normal growth, neurodevelopment and health. (1–3) Preterm infants are relatively deficient in these essential fatty acids and have a higher incidence of neurodevelopmental problems later in life. (4–6) Furthermore, animal and epidemiologic data suggest that DHA plays a protective role in the prevention of bronchopulmonary dysplasia (BPD), retinopathy of prematurity (ROP), and necrotizing enterocolitis (NEC), all associated with significant morbidity and mortality in these infants. (7–11) During the last trimester when a fetus undergoes rapid growth and brain development, maternal and fetal DHA levels are high, allowing increased incorporation into tissues. (1, 3, 12, 13) Infants born before this process is complete have interruption in the normal LCPUFA accretion. In very preterm infants this deficit persists due to decreased fat stores, poor nutritional provision and ongoing inability to convert precursor fatty acids to ARA and DHA in vivo. (14) For these reasons, a postnatal dietary source is even more critical to support normal growth, neurodevelopment and health of this at risk population. (15–20)

Mother’s own milk is the optimal diet because of its many benefits including provision of essential fatty acids. Further advantages of human milk in extremely low birth weight (ELBW) infants include decreased risks of infection, NEC and feeding intolerance. (21–23) Mother’s own breastmilk is not available to all preterm infants in the Neonatal Intensive Care Unit (NICU). Alternatives include pasteurized donor human milk or ARA/DHA supplemented formula. Because of its recognized benefits and increasing availability, pasteurized donor human milk is increasingly used for preterm infants in the NICU. (23)

LCPUFA concentrations in the U.S. donor milk supply and the effect of pasteurization on these nutrients are not well studied. The specific aims of this study were to analyze the effects of Holder pasteurization on LCPUFA content and to determine the variance of ARA and DHA among multiple U.S. donor milk banks to further assess LCPUFA provision from this neonatal nutrition option.

Materials and Methods

Donor Human Milk Collection before and after Pasteurization

Under the approval of the University of Iowa’s Institutional Review Board (IRB), samples of pooled donor breastmilk from the Mother’s Milk Bank of Iowa (MMBI) were collected for analysis before and after Holder pasteurization. Donors remained anonymous and samples were analyzed only if each donor consented to use of their milk for research purposes. Milk was donated frozen within 6 months of expression. By usual protocol, donor human milk was pooled in batches from 3 to 10 donors. Thereafter, pooled milk was processed via Holder pasteurization at 62°C for 30 min, according to standards published by the Human Milk Banking Assoc of North America (HMBANA). (24) For our study, 1 ml of thawed, pooled donor human milk was collected before pasteurization in 2 ml amber glass vials with Teflon coated caps. A corresponding sealed bottle of donor human milk was obtained from the same pooled batch after pasteurization. All samples were frozen at −20°C until processing.

Fatty Acid Analysis

Direct transesterification of milk samples was accomplished within 30 days of collection as described by LePage. (25) Thawed, 0.1ml milk samples were pipetted into glass vials with Teflon caps. One ml of methanol:benzene 3:2 v:v containing 0.5mg/ml of 17:0-heptadecanoic acid as internal standard and 1ml of freshly prepared acetylchloride:methanol 5:100 v:v were added to each tube. Methanolysis at 95–100°C for one hour was performed to create fatty acid methyl esters (FAMEs). One ml of distilled water and 1 ml of hexane with 0.5mg/ml 15:0-methylpentadecanoate as external standard were added. Samples were centrifuged and stored at 4°C until analysis by gas chromatography (GC). One microliter of the FAME containing upper phase was injected onto an HP5890 GC equipped with a Sigma Aldrich Omegawax 250 Capillary Column (30 m x 0.25 mm x 0.25 μm film thickness.) Nitrogen was the carrier gas. Fatty acids of carbon length 12 to 24 were detected by flame ionization. A Gilson 506C interface system recorded data. FAME peaks were identified using 37 FAME mix authentic standard (Sigma Aldrich, Bellefonte, PA) and integrated using Unipoint 3.0 Software, to report peaks on a wt:wt% basis.

HMBANA Donor Milk Sample Collection and Analysis

Pasteurized milk from other HMBANA banks was analyzed to determine variability of these nutrients in the U.S. donor milk supply. With IRB approval, milk samples were obtained from The Mother’s Milk Bank at Austin (Texas, n=5), WakeMed Mother’s Milk Bank (N. Carolina, n=5), and The Mother’s Milk Bank (San Jose, California, n=5). Samples of pooled, pasteurized and frozen donor human milk were shipped on dry ice and stored at −20°C until analysis. The Mother’s Milk Bank of Austin provided donor human milk samples in three caloric concentrations: 20 kcal/oz (n=2), 22 kcal/oz (n=2), and 24 kcal/oz (n=1). Milk from other HMBANA banks was not routinely tested for nutrient content but assumed to be 20 kcal/oz. Samples were analyzed by the same methodology described above.

ARA and DHA analysis of infant formula and maternal human milk samples

To allow direct comparison between formula, mother’s own milk and donor breastmilk, we subjected samples (n=2) of Enfamil Lipil (Mead Johnson, Evansville, IN) and maternal human milk samples (n=6) to the same methodology as our donor milk samples. To simulate the feeding experience of most ELBW infants in the NICU, the human milk samples were collected from individual mothers in Iowa at the time of pumping and frozen at −20C for greater than two weeks, thawed and then processed by the aforementioned analysis in a blinded fashion. Because these were anonymous samples from individual mothers living in Iowa, maternal factors other than region of residence are not reported.

Statistical Analysis

Fatty acid content was reported as mean wt:wt% of each sample ± the standard deviation (SD). Data were analyzed using a paired t-test for pre and post pasteurized human milk comparisons, defining significance at p<0.05. Nonparametric testing was also completed for LA due to a few outlying data points but revealed unchanged results. Variability of fatty acid levels in donor human milk between different batches and poolings over time was evaluated by calculating mean, median, range and coefficient of variation for individual fatty acids using 62 samples from 31 batches and 32 donors collected over a 15 month period from the MMBI. LCPUFA levels among various HMBANA banks were compared via ANOVA using the same level of significance (p<0.05).

Results

Effects of pasteurization on fatty acid content

To study the effects of pasteurization on LCPUFA content in breastmilk, 31 samples of pooled, donor human milk from the MMBI were analyzed before and after Holder pasteurization. Mean wt:wt % (SD) of ARA, DHA, and their respective precursors LA and ALA, are reported in Table 1. Other than a 3% decrease in ALA (p=0.04), there was no significant difference in fatty acid composition with pasteurization of donor human milk. Relative concentrations of LA, ARA and DHA did not change with Holder pasteurization.

Table 1.

Effects of Holder pasteurization on polyunsaturated fatty acid levels in Iowa donor human milk.

	Pre pasteurized (n=31)	Post pasteurized (n=31)	p value
LA	18.04 (1.25)	18.00 (1.10)	0.8
ALA	1.37 (0.24)	1.33 (0.22)	0.04*
ARA	0.402 (0.07)	0.396 (0.68)	0.18
DHA	0.0732 (0.01)	0.0729 (0.01)	0.8

Fatty acid levels expressed as wt:wt% (SD) of sample. All samples were taken from the MMBI,

^*p<0.05

LA, linoleic acid (18:2n6); ALA, α linolenic acid (18:3n3); ARA, arachidonic acid (20:4n6); DHA, docosahexaenoic acid (22:6n3).

Variability of fatty acid content across poolings over time

Utilizing the methodology outlined above, 62 samples of pooled donor human milk from the MMBI were collected and analyzed for fatty acid content and variability. Mean composition of select fatty acids, expressed as a wt:wt%, and summary statistics are listed in Table 2. Samples analyzed represent milk collected from 32 Iowa donors in 7 pools and 31 batches over a 15 month period of time. DHA and ARA content of MMBI donor milk did not vary significantly across poolings from batch to batch over time.

Table 2.

Summary statistics for fatty acid analysis in Iowa donor human milk.

	LA	ALA	ARA	DHA
Mean	18.02	1.35	0.40	0.073
Range	16.14–20.44	1.01–1.98	0.28–0.53	0.05–0.10
Median	17.99	1.32	0.40	0.07
Mode	17.55	1.34	0.35	0.05
SD	1.16	0.23	0.06	0.01
Coefficient of Variation	0.06	0.17	0.17	0.17

Fatty acid levels expressed as wt:wt% of sample, n=62 from the MMBI.

LA, linoleic acid (18:2n6); ALA, α linolenic acid (18:3n3); ARA, arachidonic acid (20:4n6); DHA, docosahexaenoic acid (22:6n3).

Comparisons among HMBANA milk banks

The same methodology was utilized to compare fatty acid content in pasteurized donor human milk from different U.S. HMBANA banks. Samples analyzed were from various batches of pooled, pasteurized donor human milk from Texas, North Carolina, California and Iowa. ARA and DHA content from various milk banks are reported in Table 3. as mean wt:wt% (SD). DHA content from the Mother’s Milk Bank of Austin increased monotonically with caloric content, with DHA means of 0.14, 0.17 and 0.35wt:wt% for 20, 22 and 24 kcal/oz samples respectively. Other centers did not analyze caloric content of submitted samples and therefore are assumed to be 20kcal/oz. Mean DHA levels varied significantly in a range between 0.07wt:wt% and 0.20wt:wt% among studied U.S. banks (p <0.0001). Even if higher caloric content samples from Texas are excluded, variation in DHA content among the banks exists (p<0.001). By contrast ARA levels, ranging between 0.40 and 0.46wt:wt%, did not vary significantly among studied milk banks (p = 0.3).

Table 3.

Comparison of LCPUFA levels in pasteurized, donor human milk among studied U.S. milk banks.

	Iowa (n=31)	Texas (n=5)	N. Carolina (n=5)	California (n=5)	p value
ARA	0.40 (0.06)	0.41 (0.10)	0.46 (0.03)	0.41 (0.11)	0.3
DHA	0.073 (0.01)	0.20 (0.09)	0.15 (0.05)	0.14 (0.08)	<0.0001*

Fatty acid levels expressed as wt:wt% (SD) of sample,

^*p<0.05.

Samples collected from the Mother’s Milk Bank of Iowa (Iowa), The Mother’s Milk Bank at Austin (Texas), WakeMed Mother’s Milk Bank (N. Carolina), and The Mother’s Milk Bank (California).

DHA analysis of infant formula and maternal human milk samples

Post hoc analysis of infant formula and mother’s own milk was done for comparison of DHA content to donor human milk. Enfamil Lipil revealed DHA levels (0.30wt:wt%) similar to the published product information (0.32wt:wt%). Maternal milk from individual mother’s in Iowa was analyzed after pumping, freezing and thawing to simulate the process used for most feedings in the NICU. The average DHA content (0.08 +/− 0.03 wt:wt%) was similar to that found in donor human milk from the same region (MMBI). However, individual variation was much greater in these non- pooled samples with a range of 0.03–0.21wt:wt%.

Discussion

Using a large number of samples from the MMBI, we found that Holder pasteurization does not significantly affect ARA or DHA levels in donor human milk. We also determined that LCPUFA levels are consistent at a single bank between batches of pooled donor milk over time. However, DHA levels in Iowa breastmilk were much lower than expected, prompting a smaller scale comparison which revealed significant differences in DHA levels among milk banks from different regions of the U.S. Overall, DHA content in the U.S. donor human milk supply does not meet the recommended dietary provision for infants (26) nor does it approximate what is needed to replenish the deficit found in premature infants. (14, 27)

Congruent with published findings (28) (29) (30), pasteurization did not alter LCPUFA levels in donor human milk. Our study builds on smaller studies that evaluated the effects of Holder pasteurization on human breastmilk in 6 to 12 samples, mostly from individual donors. Because HMBANA banks routinely pool donor human milk prior to pasteurization to decrease individual variation of nutrients, our study compared only pooled donor human milk samples. Additionally, we used a larger number of samples and determined the variability of LCPUFA levels in pooled donor milk from a single milk bank (MMBI) over 15 months time. This information has not been previously reported in the literature, and is important because DHA content in breastmilk varies significantly among individual mothers with time. (31) Our results show that pooling donor milk limits variation from batch to batch over time. Therefore, we conclude that the pooling and pasteurization process used by HMBANA results in a product containing consistent quantities of LCPUFA in milk dispensed from individual banks.

Although HMBANA processing created consistent LCPUFA concentrations in dispensed donor human milk, we demonstrated that DHA, but not ARA levels, varied significantly between individual banks. This was not entirely surprising because DHA varies in expressed breastmilk based on multiple maternal characteristics including timing of expression, dietary and regional factors. (32) Among the banks studied, Iowa had the lowest (0.07wt:wt%) and Texas had the highest (0.20wt:wt%) DHA content. Because Texas submitted donor milk samples with increasing caloric content, the mean level may be falsely elevated. If only 20kcal/oz milk from Texas was included (0.14wt:wt%) there would be less, but still significant, variation in DHA among the studied banks. Nonetheless, DHA levels in pasteurized donor milk from Texas, North Carolina, and California, which are in nearer proximity to the coast, were higher than donor milk from Iowa. The Mother’s Milk Bank of Ohio, another inland location, reported DHA levels in donor milk that were closer to those found in Iowa. (30). We speculate that the donor population of Iowa consumes smaller amounts of omega-3 laden fish than do donors in more coastal areas, due to limited availability and higher cost of these fish in the Midwest. Also, prenatal vitamins prescribed in Iowa during the time of this study were not typically enriched with omega-3 fatty acids.

Despite regional variation, mean DHA content measured in donor milk samples from all HMBANA banks studied (0.07 to 0.20wt:wt%) was lower than mean levels reported in other neonatal nutrition options. An illustrative comparison of these differences is represented by Figure 1. Current recommendations for DHA intake in all infants is based on a worldwide mean DHA content in expressed breastmilk, as published by Brenna et al. (26, 32) This meta-analysis of 65 studies, including 2,474 women from 34 countries, showed a worldwide mean DHA level in breastmilk of 0.32+/−0.22wt:wt%, with notable regional variation ranging between 0.06 to 2wt:wt%. Several banks (Texas, North Carolina and California) had DHA content congruent with published North American means (0.14wt:wt%). (32) Iowa levels were 50% lower, similar to those found in expressed breastmilk in rural Pakistan. (33) Currently, DHA is added to most infant formula in the range of 0.14–0.32wt:wt%.

Figure 1.

Graphic representation of mean DHA levels in neonatal nutrition options. Levels in donor human milk are significantly lower than in published levels found in maternal milk or infant formula.

Fatty acid levels are shown as mean wt:wt% (SEM). Donor milk includes measurements taken from the Mother’s Milk Bank of Iowa (Iowa, n=31), The Mother’s Milk Bank at Austin (Texas, n=5), WakeMed Mother’s Milk Bank (N. Carolina, n=5), and The Mother’s Milk Bank (California, n=5). Maternal milk includes means reported in worldwide metanalysis by Brenna et al. Infant formula includes level as measured in Enfamil Lipil.

Overall, pasteurized donor human milk from all banks studied had DHA content lower than the worldwide mean found in maternal breastmilk. We speculate that the DHA content in donor human milk is low, in part, because of inherent systematic factors. Women who donate breastmilk have typically been lactating for several months at the time of donation, thus further removed from when maternal levels peak in the third trimester of pregnancy. (34) Donors in Iowa have been lactating for an average of six to twelve months at the time of donation, which may contribute to lower DHA content. (21, 35, 36) In all HMBANA banks, breastmilk can be donated frozen up to 6 months after expression. Another source of low DHA may be freezing and storing expressed breastmilk which may allow degradation of DHA. (21, 32) DHA levels in human milk from individual Iowa mothers, evaluated after expression, storing and freezing were similar to samples from the MMBI, possibly consistent with freezing and storage as a source of DHA loss, though shared maternal characteristics with donors is also possible. It is estimated that ELBW infants need 17mg/kg/d or 1.5% of fatty acid intake as DHA to catch up from the relative deficiency at birth which worsens in the first month of life with current neonatal nutrition regimens. (14) No current neonatal nutrition option, including donor human milk, alone can meet the needs of this at risk population.

Limitations of our study include the inability to assess donor characteristics for each pooled batch of milk. Confounding factors, such as dietary characteristics, days of lactation, length of time since pregnancy, time of expression, or the specific storage conditions for donor milk or maternal milk samples are not known. Although there seemed to be a regional difference between Iowa and other studied HMBANA milk banks, the sample size from non-Iowa banks was limited increasing the risk of Type II error. Although sample sizes from milk banks other than the MMBI were small, these samples were pooled and pasteurized, not from individual mothers. Because this study shows that pooling milk provides consistent LCPUFA levels between batches over time, these samples can be considered representative of regional differences rather than individual donor differences, substantiating previously published regional variation reported in mother’s own milk from various places around the world. (32)

In conclusion, with the exception of a minor decrease in ALA, Holder pasteurization does not affect LCPUFA content of donor human milk. As is true with mother’s own milk, DHA levels vary significantly between milk banks, but ARA levels do not. Donor human milk has lower DHA content than worldwide levels of mother’s own milk or formula and does not meet the recommended nutritional guidelines for optimal provision of LCPUFA to infants. Furthermore, without supplementation the average DHA provision from donor human milk in this study (0.14wt:wt%) provides one tenth of the estimated needs of ELBW infants in the NICU. (14) Despite low DHA levels, donor human milk provides other health benefits to ELBW infants in the NICU over formula (21) (37), making it an increasingly popular alternative for very preterm infants. (23) To provide adequate growth for this population of patients, donor human milk is typically fortified to provide more calories, protein, calcium and phosphorous. Given the growing body of evidence about the importance of LCPUFAs in neonates, the addition of DHA to fortification of donor human milk should be further studied to support the overall health and neurodevelopment of this at risk population.