Is Frozen Human Milk That Is Refused by Mother's Own Infant Suitable for Human Milk Bank Donation?

Abstract

Background:

Infant refusal to feed previously frozen human milk is thought possibly attributable to lipase, an enzyme that cleaves fatty acids from milk triglycerides potentially changing the taste of the milk. Previous reports suggest that this milk is not harmful to infants; however, the lipase activity, macronutrient content, concentration of free fatty acids (FFAs), pH, and bacterial load of milk that meets this criterion are not fully understood.

Objective:

The objective was to determine whether refused frozen milk is different in composition from typical milk deposits received at a human milk bank.

Methods:

Frozen milk deposits previously refused by mother's own infant were collected from 16 mothers at five different time points when available (postpartum days 30, 60, 90, 120, and 150). Lipase activity, macronutrient composition, levels of FFA, pH, and bacteriology were determined. Analysis of mature donor milk and bacteriology data from the Ontario milk bank were used as controls.

Results:

The lipase activity for all samples was at or below literature values for mature human milk and lower compared with control milk (p < 0.001) for all time periods except at day 30. Macronutrient composition was not different from control values and did not change significantly over 150 days, with the exception of crude protein, which declined with milk maturity (p < 0.005). The pH for all postpartum time groups was lower (p < 0.02) in refused milk, and was inversely associated with lipase activity and FFA. FFA and bacterial counts were not different from control samples.

Conclusions:

Infant refusal of previously frozen milk may not be entirely due to endogenous lipase activity. This milk appears suitable for donation to human milk banks.

Introduction

The practice of freezing expressed human milk is common among nursing mothers and caretakers of breastfed infants, especially if mothers are caring for other children or are returning to work. The guidelines for human milk storage published by the Academy of Breastfeeding Medicine state that freshly expressed or pumped milk can be safely stored in the freezer (−18°C or colder) and used within 12 months.¹ While freezing milk can help alleviate stress and time commitments associated with breastfeeding at the breast, human milk is highly sensitive to freezing, in addition to the fact that its bioactivity persists during storage at −20°C. Freezing has been shown to decrease the levels of water-soluble micronutrients, including vitamin C and vitamin B₆, and has been shown to affect the concentration of fat and energy.^2–4 The loss of fat, and resulting generation of free fatty acids (FFAs), is thought to be a consequence of persistent activity of bile salt-stimulated lipase (BSSL), the predominant lipase in human milk.⁵

Some mothers anecdotally report that their infants will completely refuse previously frozen milk but will consume freshly expressed milk. It is suspected that higher than normal BSSL activity may contribute to an infant's refusal to feed previously frozen mother's milk; however, this has yet to be confirmed.⁴ The concentration and bioactivity of BSSL are highly influenced by both maternal genetics and stage of lactation.^6,7 Freezing does not affect the activity of BSSL; however, it damages the structural integrity of the milk fat globule membrane, increasing the accessibility of BSSL to triglycerides and therefore, the release of fatty acids.^8,9 Persistent elevated BSSL activity during freezer storage may lead to an even greater release of fatty acids, which may cause milk to smell rancid and taste acidic.¹⁰

Changes in milk pH, off-flavors, and odors may also be attributable to the presence of high bacteria. While it is known that human milk may contain abundant bacteria, higher levels may contribute to increased milk acidification and thus a change in taste.^11,12

Since mothers of infants who refuse previously frozen milk often wish to donate their milk to human milk banks, it is important to understand whether the nutrient composition and bacterial load of this milk differ from typical milk donations received.

To our knowledge, the nutritional content and bacteriological status of refused frozen milk have yet to be fully characterized. The primary objective of this study was to determine whether refused frozen milk is different from typical milk deposits received at a human milk bank. To ensure that it is adequately nutritious and not spoiled or contains excessive levels of bacteria, the macronutrient content, lipase activity, FFA composition, pH, and bacterial load of milk meeting this criterion were determined. The secondary objective was to determine any longitudinal changes in composition and explore potential relationships between outcome variables. We hypothesized that refused frozen milk would have higher lipase activity, and a corresponding higher concentration of FFA and lower pH.

Materials and Methods

Sample selection

Mothers wishing to donate their breastmilk to the Rogers Hixon Ontario Human Milk Bank (RHOHMB) at Mount Sinai Hospital, Toronto, Canada, were screened over the phone by lactation consultants between May 2016 and April 2018. Milk used for this study is that from a mother whose baby will breastfeed at the breast, drink freshly expressed milk, but refuses expressed breastmilk once it was frozen. This refused frozen milk is reported to have an aroma (e.g., sour, soapy, rancid) that is different from freshly expressed milk. A certified lactation consultant engaged in a thorough discussion with each donor, and verified that the infant refusal was independent of how the previously frozen milk was offered (e.g., bottle, sippy cup, etc.) and who attempted to offer it (e.g., father, grandmother, nanny, etc.). Written consent was obtained from each donor for the use of their milk for research purposes.

For comparison purposes, three different sets of controls were used. Mature raw donor milk (DM) samples (n = 17) were collected from individual donors at the RHOHMB, and used as controls for macronutrients and energy. Historical microbiology results of raw DM from the RHOHMB were used as controls for bacteriology (n = 487) collected over a 9-month period (May 2017–January 2018). More recent raw samples of mature DM collected in April 2018 and known to have <5 × 10⁷ CFU/L were used to compare pH (n = 15), given that freezer storage itself may affect pH.

Milk composition and bacteriological assessment

The macronutrient content was determined using a mid-infrared human milk analyzer (Miris Human Milk Analyzer, Uppsala, Sweden), calibrated against wet-chemistry techniques to measure fat, protein, and carbohydrate simultaneously. Energy content is calculated by the analyzer from macronutrient data. As an assessment of the precision of the instrument for measurement of macronutrient content, aliquots of a homogeneous pool of human DM were analyzed daily in our laboratory (n = 5) and yielded an interassay coefficient of variation (CV) for all three macronutrients <5%. The macronutrient composition was compared with control raw DM samples and referenced to current clinical practice.¹³

Lipase activity was determined using a commercially available kit (QuantiChrom Lipase Assay Kit; BioAssay Systems, Hayward, CA), validated for use in human milk.¹⁴ Precision was assessed by analyzing aliquots of a homogeneous pool of human DM each day (n = 5). Measurements yielded an interassay CV <10%. Lipase was compared with control, raw mature DM samples (n = 17), and referenced to levels previously reported by Freed et al.⁷ and Ellis and Hamosh.¹⁵

The concentrations of FFAs were quantified using a commercially available colorimetric kit (EnzyChrom Free Fatty Acids Kit; BioAssay Systems) validated for human milk. Standard addition of palmitic acid to milk samples yielded a 98% recovery. Milk samples were diluted 1:20 in 5% isopropanol and 5% Triton X-100 in water, followed by filtration through a 0.45 μm PTFE syringe filter. The lower limit of detection was 7 μM. The concentrations of FFA were compared with recent raw samples of mature DM measured in samples (n = 15) collected at random from the RHOHMB and referenced to previously reported raw human milk.¹⁶

The pH of milk samples was determined using an Orion 350 m, which was calibrated against validated pH standards before each use. The pH was referenced to typical pH of unpasteurized milk over 180 days postpartum and compared with recent raw samples of mature DM (n = 15) collected at random from the RHOHMB.¹⁷

Total bacterial counts with speciation (if applicable) were determined for each sample using both blood and MacConkey agar (incubated for 48 hours at 37°C) at the core microbiology laboratory at Mount Sinai Hospital. Bacteriology results were compared with historical data (n = 487). The local threshold for raw DM is a total bacterial count of <5 × 10⁷ CFU/L with no isolated Bacillus spp. to proceed to pasteurization.

Statistical analysis

Statistical analyses were conducted in SAS v.9.4 (SAS Institute, Cary, NC). A postpartum age (days) was assigned to each milk sample using the infant's date of birth and the pump date. If multiple samples existed for the same postpartum day, the samples were pooled. The postpartum days were chosen in 30-day intervals to ensure that they were representative of the temporal changes in milk composition. For purposes of comparison, samples were selected based on having a postpartum age, which most closely coincided with the following postpartum days: 30, 60, 90, 120, and 150. Descriptive statistics were used to characterize the nutrient composition, FFA concentration, pH, and bacterial counts. All variables were assessed to ensure that they were normally distributed using the Shapiro-Wilk statistic (PROC UNIVARIATE). If an outcome did not follow a normal distribution, nonparametric estimation methods were used. Mixed models (PROC MIXED) were used to compare the mean values for each control with refused frozen milk. If a statistically significant result was found, post hoc pairwise comparisons were conducted (LS-MEANS).

Multiple linear regression was used to model potential associations between constituents of refused frozen milk and its pH. Lipase activity, FFA concentration, and total fat were chosen as independent covariates as we hypothesized that together, they may affect milk pH. Covariates were tested for multicollinearity using a variance inflation factor tolerance of <2.5 (PROC REG). Chi-square tests for independence were used to analyze the proportion of samples with and without detectable bacterial growth between refused frozen milk and typical milk bank donations at the RHOHMB. A p-value <0.05 was considered statistically significant.

Results

A total of 16 mothers donated frozen refused milk samples (representative of a complete 24 hours expression) of varying maturity and over a minimum 3-month period. Seven to 12 samples were available for analysis at each of the planned monthly intervals with each donor mother having samples for a minimum of three of the time periods. The mean postpartum period at each group (in days) was 34, 62, 92, 120, and 152. Macronutrient levels, lipase activity, concentration of FFA, and pH for each time period, along with controls and literature references, are summarized in Table 1.

Table 1.

Longitudinal Changes in Macronutrient and Energy Composition, Lipase Activity, Free Fatty Acid Concentration, and pH Measured in Frozen Human Milk That Is Refused by an Infant

Postpartum time period
Analyte	Literature value	Control value^*	Day 30 (n = 10)	Day 60 (n = 11)	Day 90 (n = 11)	Day 120 (n = 12)	Day 150 (n = 7)
Fat (g/100 mL)	3.4–3.6^**	3.1 (0.8)	3.6 (0.8)	2.9 (1.5)	3.4 (1.5)	3.0 (1.0)	3.8 (1.7)
Carbohydrate (g/100 mL)	6.1–6.7^**	7.0 (0.3)	6.8 (0.3)	6.8 (0.4)	6.8 (0.7)	6.7 (0.2)	6.5 (0.4)
Crude protein (g/100 mL)	1.1–1.5^**	1.0 (0.2)^a	1.3 (0.3)^b	1.0 (0.2)^a	1.0 (0.2)^a	0.90 (0.1)^a	0.86 (0.1)^a
Energy (kcal/100 mL)	63–66^**	62 (7)	65 (8)	59 (14)	63 (14)	58 (9)	65 (17)
Lipase activity (U/mL)	24–43^†	53 (15)^a	45 (18)^ab	36 (15)^b	32 (18)^b	36 (18)^b	21 (17)^c
Free fatty acids (mM)	6.4–10.0^‡	16.9 (4.3)	20.5 (2.0)	17.7 (4.9)	16.7 (4.6)	18.3 (3.0)	17.9 (3.1)
pH	6.82–7.1^§	6.71 (0.15)^a	6.38 (0.05)^b	6.48 (0.10)^b	6.47 (0.11)^b	6.44 (0.14)^b	6.51 (0.08)^b

All data shown as mean (SD) unless otherwise indicated.

Mixed models (PROC MIXED) were used to compare each analyte with the appropriate control values and across each postpartum time period. If a significant result was found, post hoc pairwise comparisons were conducted (LS-MEANS).

Control values measured in raw, mature human donor milk samples (n = 17) from the Rogers Hixon Ontario Human Milk Bank, with the exception of pH and free fatty acid concentration, which was measured in (n = 15) samples.

Typical macronutrient composition of mother's milk, expressed as a mean ranging from ∼30 to 90 days postpartum.¹³

†

Typical range of lipase activity in human milk.^7,15

‡

Typical free fatty acids in expressed frozen human milk.¹⁶

Typical pH of unpasteurized milk over 180 days postpartum.¹⁷

a,b,ab,c

Statistical significance is indicated by different letters in superscript (p < 0.05).

Crude protein levels were comparable with concentrations in control milk, with the exception of day 30. At days 60, 90, 120, and 150, crude protein levels were significantly lower compared with day 30 (all p < 0.007). There were no significant changes in other levels of macronutrients compared with control milk and longitudinally. Lipase activity in refused milk was not different at day 30 compared with control raw DM but was significantly lower at days 60, 90, and 120 (p < 0.007). Day 150 refused milk had the lowest lipase activity, significantly below any other time point (all p < 0.04). All refused milk lipase activities, however, were within a range that is typically seen in human milk according to literature values. There were no overall significant differences in the concentration of total FFA compared with controls or at any time point (p = 0.25). All FFA levels in both control and refused samples longitudinally appeared greater than literature values.

Overall, pH was lower than typical literature values for raw mature human milk and was significantly lower at all time points compared with control milk (p < 0.0006). Longitudinally, there were no significant differences in pH over time. Total fat and FFA concentration were collinear, and as a result, total fat was excluded from the statistical model. Results from the multiple linear regression show that both FFA concentration (p < 0.0001) and lipase activity (p = 0.01) were significantly inversely associated with milk pH (R² = 54%).

Bacterial cultures of refused frozen milk showed some differences compared with typical milk bank donations. A higher proportion of samples with no growth (p < 0.001) was observed in refused frozen milk, while there were no differences in the proportion of samples below and above the threshold of 5 × 10⁷ CFU/L (p = 0.38). The proportion of milk that cultured positive for Bacillus spp. was similar between the two groups and ∼1%. The culture results are summarized in Table 2.

Table 2.

Culture Results from All Refused Frozen Milk Samples Compared with Typical Donations at the Milk Bank

Culture result	Refused frozen milk (n = 51)	Typical milk bank donations (n = 487)	p-Value^b
No growth	31 (60)	98 (20)	<0.001
0 CFU/L < growth <5 × 10⁷ CFU/L	9 (18)	214 (44)	NS
≥5 × 10⁷ CFU/L	11 (22)	175 (36)	NS
Bacillus spp. + ^a	1 (2)	7 (1)	NS

Data are presented as n (%) unless otherwise indicated.

Bacillus spp. CFU/L: colony forming units/liter.

χ² test of independence.

NS, nonsignificant.

Discussion

Refused frozen milk is a concern for mothers who cannot feed their infant previously expressed milk and likewise, a concern for human milk banks who may be the recipients of this milk. To our knowledge, this is the first study to systematically address this gap to ensure that this milk is not nutritionally different from typical milk donations and is of comparable bacterial load prepasteurization.

Results from this study indicate that the macronutrient composition of refused frozen milk is similar to typical values used clinically for DM and not statistically different from the DM controls used in this study. Longitudinal variability was only observed in the concentration of crude protein, which is consistent with the general consensus in the literature that protein levels decline with the maturation of milk.¹⁸ Similarly reported in the literature, variability was highest in the concentration of fat; an expected finding given its association with maternal diet and diurnal secretion.^19–21 Moreover, this variation may be attributable to our small sample size, which may be unrepresentative of typical DM batches. In addition, some loss of fat may have resulted from container changes in the laboratory.

Lipase activity of all refused milk samples was either not different from or lower than control samples, and all were similar to literature estimates. In refused frozen milk, the activity of lipase over the period of lactation studied showed a longitudinal declining trend; a finding that is also consistent with previous reports in the literature.⁷ This does not support the hypothesis that refused milk has higher lipase activity.

Independent of lipase activity, the release of triglycerides and FFAs from the milk fat globule may in fact be a consequence of the variations in freezer conditions (e.g., duration, temperature), thus impacting the physical structure of the milk fat globule. Interestingly, we found that the concentration of FFA did not differ between the refused frozen milk and control DM, but was higher than that reported in the literature for frozen milk.¹⁶ We also found that the concentration of FFA and lipase activity were independently inversely correlated to milk pH—there was no significant correlation between the concentration of FFA and lipase activity. Consistent with the previous literature, there is evidence to suggest that regardless of lipase activity, FFAs may be generated when milk is stored in the freezer, which can ultimately lower milk pH and render the milk unpalatable to the infant.²²

pH was used as a surrogate measure of milk palatability. Freshly expressed human milk is thought to have a slightly alkaline pH of ∼7.1–7.2.²³ Freezing milk has been shown to decrease pH in a time-dependent manner, and has been proposed as an indicator of milk quality given its relationship with commensal lactic acid producing bacteria present in the milk.^10,24 In this study, the pH of refused frozen milk was on average lower than literature values for fresh milk, in addition to being significantly lower at day 30 (p < 0.001) compared with control DM received at the RHOHMB (Table 1). Change in milk pH, assuming pH is related to palatability, is multifactorial and requires further investigation.

While our findings may be more closely related to storage issues and not necessarily to higher than normal lipase activity as previously thought, we acknowledge that an infant's individual taste preference may also affect his or her refusal.²⁵ It is also possible that an infant's aversive reaction to previously frozen milk might be associated with the known sensory deterioration of human milk during frozen storage, given that infants have excellent olfactory development.^26,27 Polyunsaturated fatty acids (PUFAs) in human milk are prone to auto- or enzymatically mediated oxidation, which has been shown to produce odorant products with fishy or metallic aromas.²⁸ Milk with higher concentrations of PUFAs might have an increased susceptibility to oxidation when frozen, producing more unpleasant odorant molecules; however, this is yet to be proven. Future research into refused frozen milk should investigate the levels of PUFA-derived oxidative products and the potential off-tastes and poor aromas associated with them.

Contrary to the hypothesis, only 40% of refused frozen milk had detectable bacterial growth, significantly lower than literature values and typical donations received at the RHOHMB.²⁹ A smaller sample of refused frozen milk compared with larger, more robust historical data may account for this difference. Refused frozen milk had comparable proportions of samples above the threshold for pasteurization, which cultured positive for Bacillus spp. There is thus insufficient evidence to suggest that excessive bacterial growth is the reason for infant refusal of milk.

Until recently, the RHOHMB did not accept milk donations if the mother reported infant refusal to feed previously frozen milk. This study precipitated a change in clinical practice and has guided the RHOHMB to update its screening policy; refused frozen milk can now be donated and may enter the stream of milk that produces DM.

Conclusion

We conclude that infant refusal to feed previously frozen milk may be due to several factors, which may include lipase activity. While lipase activity was not necessarily elevated in this study as we had hypothesized, conditions of freezer storage (temperature, duration) may affect the release of triglycerides and result in the release of poor tasting FFAs. Importantly, we show that refused frozen milk is adequately nutritious and does not contain bacterial load beyond what is expected for human DM. Since human milk that meets this criterion is not systematically different from typical donations, in terms of nutrition and bacteriology, it is suitable for milk bank donation. In doing so, this may increase the number of eligible donors for human milk banks.

Footnotes

Acknowledgments

The authors gratefully acknowledge all the milk donors who participated and all staff at the Rogers Hixon Ontario Human Milk Bank who facilitated donor screening and sample collection. They also disclose receipt of the following financial support for the research, authorship, and/or publication of this article: MaxiMOM, FDN-143233 (funding reference number as assigned by the funding agency, Canadian Institutes of Health Research) and the Mount Sinai Hospital Foundation (O'Born Family Donation).

Disclosure Statement

No competing financial interests exist.

References

Eglash

, Simon

. ABM Clinical Protocol #8: Human Milk Storage Information for Home Use for Full-Term Infants, Revised 2017. Breastfeed Med, 2017; 12:390–395.

Buss

, McGill

, Darlow

, et al. Vitamin C is reduced in human milk after storage. Acta Paediatr, 2001; 90:813–815.

Friend

, Shahani

, Long

, et al. The effect of processing and storage on key enzymes, B vitamins, and lipids of mature human milk. I. Evaluation of fresh samples and effects of freezing and frozen storage. Pediatr Res, 1983; 17:61–64.

Lawrence

, Lawrence

The Collection and Storage of Human Milk and Human Milk Banking Breastfeeding, 7th ed. Philadelphia, PA, Saunders. 2010, pp. 689–717.

Blackberg

, Hernell

. Bile salt-stimulated lipase in human milk. Evidence that bile salt induces lipid binding and activation via binding to different sites. FEBS Lett, 1993; 323:207–210.

Landberg

, Huang

, Stromqvist

, et al. Changes in glycosylation of human bile-salt-stimulated lipase during lactation. Arch Biochem Biophys, 2000; 377:246–254.

Freed

, Berkow

, Hamosh

, et al. Lipases in human milk: Effect of gestational age and length of lactation on enzyme activity. J Am Coll Nutr, 1989; 8:143–150.

García-Lara

, Escuder-Vieco

, García-Algar

, et al. Effect of Freezing Time on Macronutrients and Energy Content of Breastmilk. Breastfeed Med, 2012; 7:295–301.

Berkow

, Freed

, Hamosh

, et al. Lipases and lipids in human milk: Effect of freeze-thawing and storage. Pediatr Res, 1984; 18:1257–1262.

10.

Vazquez-Roman

, Escuder-Vieco

, Garcia-Lara

, et al. Impact of Freezing Time on Dornic Acidity in Three Types of Milk: Raw Donor Milk, Mother's Own Milk, and Pasteurized Donor Milk. Breastfeed Med, 2016; 11:91–93.

11.

Engur

, Cakmak

, Turkmen

, et al. A milk pump as a source for spreading Acinetobacter baumannii in a neonatal intensive care unit. Breastfeed Med, 2014; 9:551–554.

12.

Gomez-Gallego

, Garcia-Mantrana

, Salminen

, et al. The human milk microbiome and factors influencing its composition and activity. Semin Fetal Neonatal Med, 2016; 21:400–405.

13.

Gidrewicz

, Fenton

. A systematic review and meta-analysis of the nutrient content of preterm and term breast milk. BMC Pediatr, 2014; 14:216.

14.

Christen

, Lai

, Hartmann

, et al. The Effect of UV-C Pasteurization on Bacteriostatic Properties and Immunological Proteins of Donor Human Milk. PLoS One, 2013; 8:e85867.

15.

Ellis

, Hamosh

. Bile salt stimulated lipase: Comparative studies in ferret milk and lactating mammary gland. Lipids, 1992; 27:917–922.

16.

Chuang

, Yeung

, Jim

, et al. Comparison of free fatty acid content of human milk from Taiwanese mothers and infant formula. Taiwan J Obstet Gynecol, 2013; 52:527–533.

17.

Morriss

Jr ., Brewer

, Spedale

, et al. Relationship of human milk pH during course of lactation to concentrations of citrate and fatty acids. Pediatrics, 1986; 78:458–464.

18.

Bauer

, Gerss

. Longitudinal analysis of macronutrients and minerals in human milk produced by mothers of preterm infants. Clin Nutr, 2011; 30:215–220.

19.

Mitoulas

, Kent

, Cox

, et al. Variation in fat, lactose and protein in human milk over 24 h and throughout the first year of lactation. Br J Nutr, 2002; 88:29–37.

20.

Keikha

, Bahreynian

, Saleki

, et al. Macro- and Micronutrients of Human Milk Composition: Are They Related to Maternal Diet? A Comprehensive Systematic Review. Breastfeed Med, 2017; 12:517–527.

21.

Gunther

, Stanier

. Diurnal variation in the fat content of breast-milk. Lancet, 1949; 2:235–237.

22.

Dill

, Chen

, Alford

, et al. Lipolytic Activity During Storage of Human Milk: Accumulation of Free Fatty Acids. J Food Prot, 1984; 47:690–693.

23.

Lawrence

, Lawrence

Biochemistry of Human mIlk. Breastfeeding 7th Edition Saunders, 2010:105–170.

24.

Vazquez-Roman

, Garcia-Lara

, Escuder-Vieco

, et al. Determination of Dornic Acidity as a Method to Select Donor Milk in a Milk Bank. Breastfeed Med, 2013; 8:99–104.

25.

Hung

, Hsu

, Su

, et al. Variations in the rancid-flavor compounds of human breastmilk under general frozen-storage conditions. BMC Pediatr, 2018; 18:94.

26.

Spitzer

, Doucet

, Buettner

. The influence of storage conditions on flavour changes in human milk. Food Qual Preference, 2010; 21:998–1007.

27.

Soussignan

, Schaal

, Marlier

, et al. Facial and autonomic responses to biological and artificial olfactory stimuli in human neonates: Re-examining early hedonic discrimination of odors. Physiol Behav, 1997; 62:745–758.

28.

Spitzer

, Buettner

. Characterization of aroma changes in human milk during storage at −19°C. Food Chem, 2010; 120:240–246.

29.

Landers

, Updegrove

. Bacteriological screening of donor human milk before and after Holder pasteurization. Breastfeed Med, 2010; 5:117–121.