Development of home‐based methods to defat human milk for infants with chylothorax: An experimental study

Abstract

Background

Chylothorax is a postoperative complication for infants with congenital heart defects; with high nutrition risk. Defatted human milk is recommended; however, refrigerated centrifugation to process milk poses accessibility barriers for many hospitals and families at home. Creation of a simplified home‐based defatted milk protocol allows infants with chylothorax to be provided the immunological benefits of human milk postoperatively.

Methods

Milk from 20 mothers was tested to compare refrigerated centrifugation as the standard defatting technique against gravity‐based methods: syringe tip‐down and gravy separator. Two timeframes, 24 h and 48 h, were tested to determine if additional time had a significant impact on fat reduction. The MIRIS human milk analyzer provided results for fat, true protein, carbohydrate, and energy content. One‐way analysis of variance was used to determine a significant difference on fat content among methods.

Results

All methods had a significant reduction in fat content, with centrifugation having the largest mean decline from 3.4 to 0.5 g/100 ml (P < 0.0001). The second most effective method to defat milk was 48‐h gravy separator with a mean decline to 0.7 g/100 ml (P < 0.0001). Postpartum age of milk impacted the degree of fat removal in all methods. True protein content remained the same as baseline in all methods.

Conclusion

A simplified home‐based gravity separation method over 48 h reduced human milk fat by 80%. This is the first protocol to defat human milk without use of the more resource‐intensive centrifugation method, that shows significant fat reduction with easy‐to‐use and accessible equipment for management of infants with chylothorax.

CLINICAL RELEVANCY STATEMENT

A simplified home‐based human milk defatting protocol ‐ 48h refrigerated gravity based gravy sepatrator, has been developed, which will be of potential use for families to use at home to provide infants with human milk, while managing chylothorax.

INTRODUCTION

Infants with congenital heart disease (CHD) are a vulnerable population at high risk for feeding–related morbidity and mortality. The benefits of receiving human milk are well established among this population. ¹ Chylothorax is the accumulation of lymphatic fluid rich in triglycerides within pleural cavities in neonates and infants. ² Chylothorax is usually secondary to cardiothoracic surgery due to congenital anomalies, and can lead to prolonged admission time, fluid and electrolyte imbalance, malnutrition, and poor growth outcomes. ² Standard of practice for chylothorax is to remove all human milk feeding because of its high fat content, and start a low‐fat cow's milk formula for a treatment period of 2–6 weeks. ³

Recently, exploration of defatted human milk for feeding infants has increased in interest because of several human milk health benefits, including immune responses. ⁴ , ⁵ , ⁶ , ⁷ Various methods to remove fat from human milk exist and include refrigerated centrifugation which is the current standard, electric cream separator, or refrigeration with separation using gravity. ⁸ Refrigerated centrifugation, requires specific models of centrifuges and are more suited for well‐established milk banks and hospital milk handling rooms. ⁹ Furthermore, centrifugation methods are not easily transferred to home‐based therapies after discharge. ¹⁰ , ¹¹

There is a clear need to develop simple, home‐based methods for defatting human milk with a robust comparison against the current standard method using reliable human milk analyzers. Current gold‐standard for human milk analysis is the MIRIS Human Milk Analyzer (HMA), that uses an infrared spectroscopy technique and provides more accurate human milk fat and true protein values. It is important that the comparable method meets the clinical requirements for treatment of chylothorax, while providing an accessible and efficient method for families of all demographics. The objective of this study was to compare the refrigerated centrifugation method against simple gravity‐based methods—syringe tip down and gravy separator using the MIRIS HMA.

METHODS

We performed an experimental study on human milk to develop home‐based methods to defat human milk for infants with chylothorax. Human milk used in this study was from the local human milk banks' pool of donated research milk. Human milk may be consented for research if it is deemed ineligible for use in infants, as determined by the local bank's guidelines. No review board or consent forms were completed because of the deidentified nature of all milk samples. Only basic demographic information was attached to the milk samples.

Sample size and protocol set‐up

Prior to starting the study, pilot work inspected various methods of gravity defatting in comparison with refrigerated centrifugation using donor human milk. This included syringe‐tip up vs syringe tip‐down, open vessel (bottle), and three models of gravy separators (commercial separators for gravies and sauces). Surface fat removal was attempted using spoon and syringes on relevant containers. ⁹ Gravity methods syringe tip‐up, open vessel, and two models of gravy separators were excluded because of messiness in separation and the potential for human error. Furthermore, these methods also had remixing of fat during collection. Time frames examined for the refrigeration included 24 h, 48 h, and 72 h after thawing of previously frozen milk. Final methods selected for the full comparison were those in which the fat would remain in the container during collection, and no manual fat removal would be needed. At this development stage, 72 h was removed as a time point because of the logistics of having milk in the fridge for prolonged duration with an increased risk of expiration/contamination. ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶

Twenty unique individuals' donated milk was used for this study with a developed standardized protocol to test the refrigeration gravity method over two time points (24 and 48 h) vs refrigerated centrifugation (Figure 1). Human milk was provided frozen and unpasteurized by the local human milk bank. Analysis was performed on a baseline (unprocessed) sample and a control (refrigerated centrifugation) sample from each individual donor to compare effectiveness of method and timepoints. All samples were analyzed in triplicate for fat, protein, carbohydrate, and energy content using the gold standard MIRIS HMA. Since the goal of this study was to focus on simulating an at‐home environment, the number of transfers between containers and temperatures used were envisioned to mirror an at‐home environment (Figure 1). To more accurately reflect the fat content lost in these methods that would be observed at home, samples underwent minimal transfers, one freeze‐thaw cycle, and were kept at standardized temperatures.

Figure 1.

Flowchart demonstrating methods of HM processing for a single milk donor, mirrored to patient experience in an at home environment. In a laboratory, measurements (analysis) are performed in triplicate for macronutrients with the MIRIS Human Milk Analyzer. HM, human milk.

Techniques for defatting human milk

Three methods for defatted human milk were used (Figure 2): refrigerated centrifugation (control), gravity separation by 50 ml syringe with the tip down, and gravity separation by 1 L gravy separator. All samples were refrigerated at 4°C during the separation process and were compared with unmodified human milk for each donor. Donor milk was provided frozen in individual pumping bags across multiple collection days. This milk was thawed from −20°C at room temperature for 2 h as detailed by the MIRIS HMA manual for small sample volumes, ¹⁴ pooled, and homogenized before aliquoting to the three methods. In a home or clinical setting, all milk for feeding should be thawed in the fridge to prevent bacterial colonization when fresh milk is not available. ¹² , ¹³ , ¹⁴ Before collection of defatted milk, the volume of visible fat layer was measured off the container's volumetric markings. After defatting, 10 ml of milk from each sample was used for MIRIS HMA analysis.

Figure 2.

Demonstration of human milk defatting methods. (A) refrigerated centrifugation; (B) syringe tip‐down; (C) gravy separator; (D) removal of defatted milk using a pasteur pipet; (E) removal of defatted milk through dispensing nozzle; (F) dispensing nozzle secured with a paperclip.

Refrigerated centrifugation: 50 ml of milk per donor was collected into a 50‐ml Falcon tube and spun in refrigerated centrifuge at 3000 rpm for 15 min at 2°C. ⁹ After centrifugation, defatted milk was collected from beneath the stiff fat layer using a 3‐ml BD Falcon pasteur pipet. The pipet was used to make a hole through the fat against the tube, and a clean pipet was then inserted and used to collect the defatted milk underneath.

Syringe tip‐down, gravity: two portions of 50 ml of milk per donor was drawn up into two 50‐ml BD MED‐RX oral syringes and capped. Syringes were stored tip‐down in the refrigerator, one portion each for 24‐ and 48‐h time frames, in a tube rack. At each time point the single portion of defatted milk was slowly expelled from the syringe tip into a beaker and homogenized by gentle swirling.

Gravy separator, gravity: two portions of 500 ml of milk per donor were poured into two OXO Good Grip gravy separators; the rubber dispensing nozzle on the bottom was sealed with a paperclip and covered with tinfoil. Separators were stored in the refrigerator with one portion for each 24‐ and 48‐h time frame. At the allocated time point, the paperclip was removed from one portion, and defatted milk was collected into a beaker and homogenized by gentle swirling.

Human milk analysis, MIRIS

Measurements for macronutrients (fat, true protein, carbohydrate) and energy for each sample were analyzed using the MIRIS HMA. The MIRIS system uses midinfrared (MIR) transmission spectroscopy, in which milk is exposed to MIR radiation; the reduction in specific waveband signals are unique and proportional to each macronutrient. MIRIS technology for donor human milk has been previously described ¹⁷ , ¹⁸ , ¹⁹ and validated ²⁰ , ²¹ , ²² and used by our laboratory in earlier studies. ²³ , ²⁴ For each measurement, 3 ml of milk warmed to 40°C and sonicated (1.5 s/ml) was injected. ¹⁴

Statistical analysis

MIRIS HMA measurements for fat, true protein, and carbohydrate are provided in grams per 100 ml; energy as kilocalories per 100 ml. All measurements were done in triplicate for each sample, results are provided as mean (SD) with minimum to maximum range values. Statistical significance for fat reduction was determined with a one‐way analysis of variance (ANOVA), with Dunnett multiple comparisons test at P < 0.05 (95% CI) using baseline (unprocessed) human milk as the control group. Subanalysis of all methods against each other for all donors was done using a two‐way ANOVA with Tukey multiple comparisons test P < 0.05 (95% CI). All analysis were performed using GraphPad PRISM 6.07 (GraphPad PRISM Software Inc).

RESULTS

Human milk fat content

Human milk from 20 unique mothers were included with 180 milk samples tested. Demographics of the mothers who donated milk can be found in Table 1. All methods had statistically significant fat reduction by one‐way ANOVA Dunnett multiple comparisons (Table 2). To be clinically significant, fat content would need to be ≤1.7 g/100 ml based on the long‐chain triglyceride (LCT) content of an earlier clinically tested recipe (defatted milk was 0.65–1.25 g/100 ml via the Folch technique with the addition of several modulars that contributed LCT). ⁷ This amount of fat reduction was achieved for all milk defatted for 48 h in the gravy separator and via refrigerated centrifugation. Milk fat content of 1.7 g/100 ml, which does not account for added medium‐chain triglycerides (MCTs) (as they do not impact chyle production ² , ²⁵ ), was shown to be sufficiently low to resolve pleural effusions in chylothorax infants using defatted human milk. ⁷ Refrigerated centrifugation remains the most effective method with a fat reduction of 2.9 g/100 ml (P < 0.0001); fat content went from an average of 3.4 g/100 ml at baseline to 0.5 g/100 ml. The second most effective method to defat was 48 h gravy separator with an average decline of 2.7 g/100 ml (P < 0.0001); fat content was reduced to 0.7 g/100 ml compared with baseline.

Table 1.

Characteristics of 20 (n = 20) human milk donors.

	Mean (SD)	Min‐max
Age, years	32.0 (4.0)	23.0–39.0
Postpartum age,a weeks	8.8 (10.0)	1.0–36.0
Time milk frozen,b weeks	25.8 (13.9)	2.0–55.5

^a Calculated as the average between date of birth and first/last pumping date.

^b Calculated as the average between date of thaw in laboratory and first/last pumping date.

Table 2.

Macronutrient content from unpasteurized donor human milk measured with the MIRIS Human Milk Analyzer across five defatting methods compared with baseline (unprocessed) milk.

	Fat, g/100 ml		True protein, g/100 mla		Carbohydrate, g/100 ml		Energy, kcal/100 ml
Method	Mean (SD)	Min‐max	Mean (SD)	Min‐max	Mean (SD)	Min‐max	Mean (SD)	Min‐max
Baseline	3.4 (0.8)	2.1–5.5	1.1 (0.3)	0.5–1.6	7.3 (0.5)	6.0–8.4	66.6 (8.6)	55.0–87.7
Refrigerated centrifuge	0.5 (0.1)	0.3–0.8	1.1 (0.3)	0.7–1.5	7.3 (0.4)	6.4–8.2	39.4 (3.1)	33.0–44.0
24‐h syringeb	1.7 (0.6)	0.8–2.5	1.1 (0.3)	0.7–1.6	7.6 (0.3)	7.1–8.1	52.4 (6.4)	43.3–62.7
48‐h syringeb	1.2 (0.5)	0.5–2.0	1.1 (0.3)	0.7–1.5	7.7 (0.4)	7.0–8.3	47.6 (5.7)	40.0–56.0
24‐h gravy separatorb	1.1 (0.5)	0.4–2.1	1.1 (0.3)	0.7–1.6	7.6 (0.3)	7.1–8.1	47.0 (6.1)	39.0–58.3
48‐h gravy separatorb	0.7 (0.3)	0.3–1.3	1.1 (0.2)	0.7–1.5	7.6 (0.3)	7.1–8.4	43.5 (4.1)	38.0–49.0

^a The MIRIS HMA system presents crude and true protein, in which crude protein is measured as the total amount of nitrogen in a sample. True protein is a measure removing nonprotein nitrogen compounds and is a more accurate value of actual protein content in a sample. ¹⁴

^b Gravity methods, with 24‐ and 48‐h storage at 4°C to allow fat separation from human milk.

Variables influencing methods

Our findings confirmed 48‐h methods being more effective than the 24‐h time frame. Although having a statistically significant fat decline from baseline, the other methods of 24‐ and 48‐h syringe and 24‐h gravy separator had only 9, 16, and 16 donors of 20 under 1.7 g/100 ml, respectively, and would not be optimal for clinical use (Figure 3). Of further clinical significance is the variability of fat content from each method. Centrifugation had the smallest range of 0.3–0.8 g/100 ml, and the second method with minimal variability was 48‐h gravy separator with 0.3–1.3 g/100 ml. The other methods had a potential maximum fat content of 2.0–2.5 g/100 ml, which is likely clinically significant fat‐content level and would not be recommended in management with a low‐fat diet. Energy and protein results followed expected trends with energy decreasing with fat removal (P < 0.0001 for all methods) and protein content remaining unchanged. Carbohydrate levels were noted to numerically increase in content with greater fat removal in the gravity methods (Table 2).

Figure 3.

Impact of defatting on macronutrients measured with MIRIS Human Milk Analyzer analysis. Values presented as distributions with means (SD). (A) Fat, dotted line represents the 1.7 g/100 ml clinical maximum for feeding infants with chylothorax. ⁷ (B) True protein. (C) Carbohydrate. (D) Energy.

Postpartum age of milk

Of note was the difference in fat removal between early and late milk defined as milk pumped <28 days postpartum age and milk pumped >28 days postpartum age, respectively. Donor characteristics by postpartum age group can be found in Table S1.1. With similar baseline fat values (mean early 3.4 g vs late 3.4 g/100 ml), refrigerated centrifugation had 0.6 g fat/100 ml remaining when it was early milk compared with 0.4 g fat/100 ml with late milk (Table S1.2; Figure S2.1). Gravy separator at 48 h had a decrease with 1.0 g fat/100 ml (range 0.8–1.3 g/100 ml) remaining in the early milk group compared with 0.5 g fat/100 ml (range 0.3–0.8 g/100 ml) with late milk. The differences in fat reduction were also physically visible by the size of fat layer formed for each method. All methods experienced a percentage in fat difference between the early and late age milk, with a mean percentage of fat of 9.3% early and 9.9% late for refrigerated centrifugation and 20.3% early and 17.0% late for 48‐h gravy separator (Table S1.3). Two‐way ANOVA Tukey multiple comparisons test was performed to evaluate the impact of each donor independently between all methods because of the large spread of postpartum age in the donors. As indicated with the one‐way ANOVA, refrigerated centrifugation remained the gold standard with 19 of the 20 donors having significant fat reductions, followed by the 48‐h gravy separator with 18 of the 20 donors seeing a significant fat reduction. The two donors who did not see a significant result, had a low baseline fat concentration (2.1 and 2.4 g/100 ml) compared with the other donors.

DISCUSSION

Methods to defat human milk, especially outside of the hospital, and understanding of the optimal fortification required are not sufficient enough to permit widespread use in the clinical care of infants with chylothorax. The current study aimed to address the primary gap of establishing a simple home‐based method. The method was analyzed and verified by the gold standard human milk analyzer (MIRIS HMA). Home methods need to be cost‐effective and labor‐effective, accessible to all communities, and as capable as refrigerated centrifugation. Our results show that the refrigerated gravy separation using gravity over 48 h provides adequate fat separation (on average 0.7 g fat/100 ml) comparable to the more complicated refrigerated centrifugation (on average 0.5 g fat/100 ml) method. The method outlined using the gravy separator is inexpensive and, with step‐by‐step instructions, can be useful for families to feed infants with chylothorax at home.

Our study was based on earlier work by Barbas et al, ⁸ in which they compared similar methods; however, our results have the benefit of MIRIS HMA analysis for the final milk composition. In contrast to their alternative fat separation, cream separator, we decided to test a more accessible and less cumbersome tool and trialed three off‐the‐shelf gravy separators, commonly used for holiday meals. During protocol development the OXO brand with a dispensing nozzle on the bottom of the container performed the best, which was less prone to spillage and fat remixing and therefore produced the lowest fat milk. Using this model gravy separator and inclusion of two time points (24 and 48 h) allowed a full exploration of influence of gravity over time within the limits of optimal human milk storage guidelines.

The gravy separator also has some additional practical advantages compared with the syringe tip‐down method; it took up the least amount of space in the fridge and is therefore likely to be sustainable at home. The container is high‐temperature resistant for sterilization and reduces waste as one separator can be reused within a single family, whereas each syringe is labeled single use. Furthermore, the gravy separator had better visibility of fat layering, presumably because of the larger surface area. Using the syringe‐tip down method for higher baseline fat donors, the fat deposit on top was difficult to view separate from the defatted milk. For the OXO gravy separator, however, the dispensing nozzle on the bottom is cone shaped, thus resulting in a funnel being produced during drainage. Thus, a safety margin of 50–100 ml was required to be left under the visible fat layer to prevent any remixing and increasing the fat content of the separated milk. The minimal amount of human milk possible to use in the gravy separator would be 250 ml (which equals a mother pumping 60 ml, or 2 ounces, four times a day). This minimum amount allows the safety margin below the fat layer to ensure none of the separated fat is included in the siphoned milk.

In an earlier 2016 study evaluating clinical outcomes, in postoperative neonates receiving a recipe containing additional fat sources mixed into defatted milk based on reaccumulating chyle, 1.7 g fat/100 ml was suggested to be adequate. ⁷ In a more recent 2020 study at the same center, a long‐chain fat provision of 1.2 g/100 m (defatted milk by refrigerated centrifuge mixed into a recipe that included soybean oil) had no negative impact. ²⁶ Considering the fat provision in earlier completed patient‐based trials, we suggest that the 48‐h gravy separator method is the ideal home‐based method. The 48‐h gravy separator method had a 75th percentile IQR at 1 g/100 ml with no milk samples going over 1.7 g/100 ml and gives a bit more confidence in using the method at home.

Compared with two low‐fat formulas available—Monogen (Nutricia) and Lipistart (Vitaflo; Nestlé Health Science)—there may be good reason to consider inclusion of defatted human milk in the neonatal (<28 days) population. The LCT content of Lipistart is roughly double of Monogen (0.66 g vs 0.35 g LCT/100 ml, respectively). The rationale for the use of a higher fat formula is that Lipistart is more appropriate to treat LCT malabsorption conditions in neonates who have higher fat requirements to support development. Our early postpartum age milk sample was reduced to 1.0 g/100 ml fat, which remains below the 1.2 g/kg intake of later studies ²⁶ and may be protective for neonatal neurodevelopment compared with the later milk reduction to 0.5 g/100 ml which falls below Lipistart fat content. With all defatted milk, additional MCT oil is necessary to meet the demands from fat energy for these highly vulnerable infants.

Our study has a few limitations to consider. This study was conducted by a trained research team member who is comfortable with performing such procedures, and therefore might not match the broad range of education and training of families required to perform the defatting task. In addition, the finding of differences between early and late gestational age milk potentially needs to be considered in future in a clinical trial. Thus, our study findings may not be generalizable to all infants. Furthermore, milk may have been pumped across different lactation stages, and full details were not available from all mothers. However, the BC Women's Provincial Milk Bank provides potential donors with strict guidelines for the safe collection of milk for donation; therefore, potential for negative/positive fat or protein bias from medications and blood contamination ¹⁴ was reduced. Finally, our study used previously frozen milk. Effects of freezing human milk are inconclusive for potential changes in macronutrient content ²⁷ , ²⁸ , ²⁹ and for the impact of storage time on fat globules ³⁰ and would need to be evaluated in the context of defatted milk for clinical use. Since fat content may decrease over storage time, defatting procedures should be tested using fresh milk to fully understand the impact of freezing.

CONCLUSION

Human milk is the optimal food for infants and ideal methods for providing defatted human milk to infants with chylothorax are necessary to be developed. The methods to be used at home must be practical and easy to use to ensure parents can continue to provide the best possible nutrition to these vulnerable infants. Among the tested methods for a home‐based system, the results recommend using 48‐h gravy separator, which relies on gravity with refrigeration, for families who express interest in continuing to offer human milk during low‐fat diet therapy for chylothorax. The proposed method, as outlined in our methods section, is fairly easy to use and can be adapted at home. Use of the MIRIS HMA for analysis of fat reduction provides confidence that the 48‐h gravy separator method is practical for use with families. Further research is required to ensure clinically important outcomes are evaluated (ie, infant growth, family coping). These findings have the potential to improve family‐centered care in a complex postoperative condition and bring equitable nutrition to families globally.

AUTHOR CONTRIBUTIONS

Kaitlin Berris, Kendall Plant, Frances Jones, and Rajavel Elango contributed to the design of the research; Kaitlin Berris was the project administrator and Rajavel Elango supervised; Kaitlin Berris, Frances Jones, Diego Marquez, and Vicki Hsieh contributed to the clinical relevance of the methods; Kendall Plant performed the benchwork and statistics; Kendall Plant, Kaitlin Berris, and Rajavel Elango verified the data presented; Kaitlin Berris, Kendall Plant, and Rajavel Elango analyzed the data; and Kaitlin Berris and Kendall Plant shared equal contribution to the preparation of the manuscript. All authors read and approved the final manuscript.

CONFLICT OF INTEREST STATEMENT

None declared.

Supporting information

Supplemental Material For Review ‐ JPEN ‐ Nov 2024.

Berris K, Plant K, Jones F, Marquez D, Hsieh V, Elango R. Development of home‐based methods to defat human milk for infants with chylothorax: an experimental study. J Parenter Enteral Nutr. 2025;49:724‐731. 10.1002/jpen.2782

DATA AVAILABILITY STATEMENT

Individual participant information is not available, as it was not available from the Milk Bank. Method protocols and individual data will be made available after study publication, based on a reasonable request to the corresponding author.