Point-of-care human milk concentration by passive osmosis: comprehensive analysis of fresh human milk samples

Mother’s own milk (MOM) provides critical nutritional and health benefits for preterm infants, reducing morbidity and mortality and improving development. Despite its benefits, MOM alone often does not meet the elevated nutrient needs of preterm infants, leading to common use of bovine- or donor human milk-derived fortifiers, which may alter human milk components, carry higher costs, replace substantial MOM volume, and potentially increase certain risks. A point-of-care approach using passive osmotic concentration could increase MOM nutrient and bioactive densities without heat or pressure and without displacing MOM. This study hypothesized that fresh, never frozen/thawed HM can be concentrated via passive osmosis to increase sensitive components without damage, maintaining acceptable pH and osmolality and preserving cell viability.

Design and oversight: The New England IRB approved the study. Recruitment used flyers, collaboration with Mother’s Milk Bank Northeast, and Baby Café participants with surplus milk. Informed consent was obtained. Sample collection: Fresh HM samples (n=19) were collected 30 min to 20 h (average 11.5 h) after expression, stored at 4 °C by donors. Samples were gently swirled before aliquoting. Baseline (unconcentrated) aliquots were stored at 4 °C. Passive osmotic concentration: HMC device was rinsed with filtered water at 38 °C and placed into 75 mL of fresh HM in an 80-mL polypropylene bottle. Bottles were capped and stored at 4 °C for 3 h. The device was removed and masses/volumes of HM (final) and device were recorded. Immediate analyses included cell viability and MIRIS macronutrients; remaining matched samples (unconcentrated and concentrated) were gently swirled, aliquoted, and shipped overnight on dry ice to specialized laboratories. Concentration calculation: Percentage volume reduction (HMcon) was computed as (Δvol/Vinitial) × 100, where Δvol = Vfinal − Vinitial. Analytical methods: - Mid-infrared macronutrients: Energy, fat, carbohydrates, crude protein, and true protein were measured with Miris Human Milk Analyzer per protocol. Samples were heated to 40 °C, homogenized, analyzed in duplicate; true protein was calculated assuming 20% of crude protein is non-protein nitrogen. - Proteins and small solutes: Total protein by BCA assay (triplicate); lactose by enzymatic assay after protein/fat removal with Carrez solutions; lactoferrin by ELISA; active IgA against E. coli antigens by ELISA; sodium by ion-selective electrode. Reported intra/inter-assay CVs were provided for each assay. - Enzymatic activities: Lysozyme, PAF acetylhydrolase, catalase, and glutathione peroxidase were quantified using commercial kits with specified dilutions. Bile salt-stimulated lipase (BSSL) was measured via p-nitrophenyl myristate hydrolysis with spectrophotometry at 405 nm at 37 °C, using a p-nitrophenol standard curve. - Human milk oligosaccharides (HMOs): Nineteen HMOs with available standards were quantified using HPLC methods per a published protocol, including 2′-FL, 3FL, LNnT, LNT, LNH, LSTb, LSTc, 3′-SL, 6′-SL, DSLNT, DSLNH, DFLac, FLNH, DFLNH, LNFP I/II/III, FDSLNH, DFLNT. - Choline-related compounds and lipids: Free choline, phosphocholine, and betaine were measured by HPLC-MS/MS with stable isotope internal standards; phosphatidylcholine and sphingomyelin were measured separately by modified HPLC-MS/MS methods. - Fatty acids: Total and individual fatty acids were measured by GC-FID with internal standards (C9:0, C17:0), with linearity verified over typical HM ranges. - pH and osmolality: Measured post-thaw using a calibrated pH/ion meter and freezing point osmometer; osmolality measured in duplicate (third reading if >3 mOsm difference). - Cell viability: Cells isolated by centrifugation and PBS washes; viability assessed by trypan blue exclusion using an automated cell counter. Statistics: JMP 17.2 was used. Outliers were identified by robust fit (Huber M-estimation). Paired t tests determined differences between unconcentrated and concentrated samples (p < 0.05).

• The HMC device achieved an average HM volume reduction of 16.3% ± 3.8% after 3 h at 4 °C.
• pH: Decreased from 7.37 ± 0.28 to 7.08 ± 0.27; not significant (p > 0.05).
• Osmolality: Increased by ~33%, from 295 ± 3.44 mOsm to 392 ± 28.7 mOsm (p < 0.05), remaining within neonatal feeding parameters.
• Cell viability: Mean live cell percentage did not differ significantly (7.63% vs 5.68%, p > 0.05).
• Across 41 components, 31 increased significantly; none decreased significantly. Ten did not differ significantly.
• MIRIS macronutrients (carbohydrates, crude and true protein, total fat, energy, total solids) all increased significantly within the expected range based on volume reduction.
• Specific analytes: Sodium, protein (BCA), lactose, lactoferrin, and active IgA increased significantly; active IgA rose less than expected (5.9% ± 10.2%).
• Enzymes: BSSL activity increased significantly (21.5% ± 39%, p < 0.05), within expected range. Lysozyme, PAF acetylhydrolase, catalase, and glutathione peroxidase showed no significant change.
• Small molecules: Total fatty acids and sphingomyelin increased within expected ranges. Betaine (+32.4% ± 22.3%) and free choline (+28.3% ± 9.6%) increased more than expected. Phosphocholine did not increase significantly.
• HMOs: 14/19 HMOs increased significantly post-concentration. LSTc, DFLNT, 6′SL, LNFP II, and 3FL did not significantly change; 3FL showed a nonsignificant decrease. Increases beyond expected were observed for LNFP III, DFLNH, DSLNH, and DSLNT; DFLac increased less than expected.
• Caloric density and protein example: From ~22 kcal/oz and 1.16 g/dL protein at baseline to ~26 kcal/oz and 1.39 g/dL protein post-concentration (~20% increases).

Results demonstrate that passive osmotic concentration at the point of care can selectively remove water from fresh human milk, increasing concentrations of macronutrients, many bioactive molecules, and certain enzymes without significantly impacting pH or live cell viability. The significant rise in osmolality and macronutrients fell within clinically acceptable parameters, suggesting feasibility for neonatal feeding. Compared to fortification with bovine- or human milk-derived products, which can degrade some bioactive activities and displace MOM volume, this approach preserves native HM components and avoids heat or pressure. Enhanced nutrient and bioactive delivery from concentrated MOM may improve absorption and clinical outcomes for preterm infants, while potentially streamlining NICU feeding workflows and supporting parental empowerment in providing MOM.

Point-of-care passive osmotic concentration effectively enriches fresh human milk by selectively removing water while maintaining pH and cell viability within neonatal parameters. This novel approach can increase caloric and protein density and enhance bioactive and enzymatic components without adding fortifiers, offering a potential alternative or adjunct to current fortification practices. Future research should establish clinical efficacy and safety, including standardized FDA-indicated analyses (leachables, toxicology, microbiology), evaluate short- and long-term outcomes (growth, tolerance), characterize effects on additional HM components (e.g., hormones such as melatonin), and further elucidate component–membrane interactions and assay performance in concentrated HM.

• Small sample size due to focus on fresh HM and donor availability.
• Variable donor handling, storage times (some >20 h), and self-reported expression times likely contributed to low baseline cell viability and between-sample variability.
• Accuracy of MIRIS HMA in concentrated HM has not been specifically validated, although concordance with BCA protein suggests reasonable performance.
• Potential membrane interactions may differentially affect HMOs with specific charge/linkages (e.g., 6′SL, LSTc), contributing to variable enrichment.
• Fatty acid quantification variability possibly related to concentrations outside the validated linear range in GC-FID; extrapolation of linearity may be acceptable but unconfirmed across the entire range.
• Phosphocholine analysis showed poorer reproducibility and potential chromatographic interactions, contributing to nonsignificant change.
• For some samples, certain components did not change significantly after incubation, reflecting biological and methodological variability.