Breast Milk and Saliva Lactoferrin Levels and Postnatal Cytomegalovirus Infection

Abstract

Objective

Very low birth weight preterm infants are at risk for life-threatening infections in the NICU. Breast milk protects against infections but carries the risk of infection by cytomegalovirus (CMV) shed in mother’s milk. Lactoferrin is a breast milk and saliva protein with potent neutralizing activity against CMV.

Study Design

VLBW, maternal breast milk fed infants in the NICU and their lactating mothers were enrolled and followed for 3 months/discharge. Breast milk and infant saliva samples were collected biweekly. Maternal CMV status was determined on breast milk. CMV was measured using quantitative polymerase chain reaction and lactoferrin by enzyme-linked immunosorbent assay.

Results

In an in vitro neutralization assay, the IC₉₀ of purified human lactoferrin against CMV was 2.08 ng/mL. Bovine lactoferrins were more potent, IC₉₀s > 10-fold higher. Lactoferrin was detected in all breast milk (median: 3.3 × 10⁶ ng/mL) and saliva (median: 84.4 ng/swab) samples. Median CMV load in breast milk was 893 copies/mL. There was no correlation between breast milk lactoferrin concentration and CMV load. Five infants acquired postnatal CMV. There was no difference in saliva or breast milk lactoferrin concentration for mother–infant pairs and postnatal CMV acquisition.

Conclusion

Lactoferrin neutralizes CMV in vitro, but concentrations in breast milk and saliva are likely too low for effective neutralization in vivo.

Preterm very low birth weight (VLBW< 1,500g) infants are at high risk for life-threatening infections in the neonatal intensive care unit (NICU).¹ Breast milk provides protection against devastating infections, including bacterial sepsis and necrotizing enterocolitis (NEC), and is recommended for all preterm infants.^2–6 However, fresh breast milk carries the risk of infection by cytomegalovirus (CMV) shed in maternal milk.^7,8

Congenital CMV is the leading infectious cause of neurodevelopmental impairment and the leading nongenetic cause of hearing loss in developed countries.^9–12 Infection in infants occurs in utero (congenital infection) or via mucosal exposure (postnatal infection).^13–15 Although generally harmless in full-term, immunocompetent infants, postnatal CMV (pCMV) infection in VLBW infants can result in a severe sepsis-like illness characterized by pneumonitis, enteritis, hepatitis, and thrombocytopenia.^16,17 In addition, studies suggest that pCMV is associated with an increased postnatal age at discharge and lasting sequelae, including neurodevelopmental impairment, bronchopulmonary dysplasia (BPD), NEC, and hearing loss.^13,18–25 Historically, CMV was transmitted to hospitalized infants through blood transfusions; however, the use of CMV-seronegative or leukoreduced blood has significantly reduced this mode of transmission. Exposure to breast milk from CMV-infected mothers is now the most common means of transmission to preterm infants.^13,26

Although CMV is shed in breast milk of the majority of CMV-seropositive mothers, transmission occurs in only 30 to 50% of exposed preterm infants.²⁷ This natural protection of the majority of infants suggests that immunologic factors in the infant or breast milk can impede infant CMV acquisition in the gastrointestinal tract. A previous study found no substantial differences in the CMV-specific immune responses in the milk of pCMV-transmitting and nontransmitting mothers,²⁸ suggesting innate antiviral factors may be more important than adaptive immune responses. Several innate antiviral immune factors in breast milk could have an impact on vertical virus transmission.^29–35 Lactoferrin is a breast milk and saliva protein with potent activity against CMV.^36–41 Oral lactoferrin prevents severe disease in mice following murine CMV challenge.⁴² In addition, a recent clinical trial found that enteral bovine lactoferrin was safe and protective against bacterial sepsis in preterm infants.^43,44 Identifying an association between postnatal CMV acquisition and lower levels of lactoferrin could provide evidence of a potential role for lactoferrin as a prophylactic agent against postnatal CMV.

In this study, we prospectively enrolled preterm VLBW infants receiving their mother’s own milk in the Duke NICU. We collected and analyzed breast milk and infant saliva to determine lactoferrin concentration, CMV load, and whether lactoferrin concentration was associated with maternal CMV shedding or pCMV transmission.

Materials and Methods

Study Design

Infants born at Duke University Medical Center, Durham, NC and admitted to the NICU from October 2015 to September 2017, and their lactating mothers were enrolled prospectively. Infants ≤32 weeks of gestational age or ≤1,500 g, <21 days of life (DOL), and whose mothers planned to provide maternal breast milk were eligible for inclusion. Infants with known congenital CMV (cCMV), major congenital anomalies, or gastroschisis were excluded. The mother–infant pair was excluded if the infant tested positive for CMV in the initial saliva sample collected <21 DOL. The study was approved by the Duke Institutional Review Board, and written informed consent was obtained from a parent of all participating infants. Infant saliva and maternal breast milk samples were collected until 36 weeks of postmenstrual age, death, transfer, study withdrawal, or if the infant was no longer receiving maternal breast milk. In addition, demographic, laboratory, and clinical outcomes were recorded. Mother–infant pairs with at least one breast milk and one saliva sample were included in the study. In cases of multiples, a separate maternal sample was collected for each infant, timed to coordinate with the infant’s sample, and each mother–infant pair was analyzed separately.

Specimen Collection

Saliva:

Saliva samples were collected at enrollment then biweekly until study completion. A sample was collected at <21 DOL to exclude cCMV. Saliva samples were collected by swabbing the inside of the infant’s cheek prior to feeding (as long as feasible from a previous feed) to reduce the risk of breast milk contamination of samples. Two saliva swabs were collected: (1) sterile polyester tipped swab (VWR) for quantitative polymerase chain reaction (qPCR) and (2) sterile Weck-Cel swab (Beaver-Visitec International, Waltham, MA) for lactoferrin enzyme-linked immunosorbent assay (ELISA). The polyester-tipped swabs were stored at room temperature and the Weck-Cel at −80°C and transported to the laboratory in batches for processing. Saliva was used instead of urine for CMV testing because of ease of collection in VLBW infants and because saliva testing was thought to be as sensitive as urine for CMV testing at the time of study design.^45–47

Breast milk:

Fresh or thawed maternal breast milk (1–2 mL) the infant would receive that day was collected at enrollment then biweekly until study completion, on the same day or as close as possible to infant saliva collection. Breast milk was stored at −80°C and transported in batches to the laboratory for processing.

Specimen Processing and Testing

Saliva processing:

After thawing at room temperature, 200 μL of PCR grade water was added to each swab and vortexed thoroughly. Each swab was then either incubated at room temperature for 30 minutes or at 95°C for 5 minutes.

Breast milk processing:

Breast milk was delipidized by centrifugation at 23,000 × g at 4°C for 20 minutes. The viral pellet was separated by centrifugation at 55,000 × g at 4°C for 2 hours. The aqueous layer was used for ELISA. The viral pellet was resuspended in 200 μL of phosphate-buffered saline (PBS), and viral deoxyribonucleic acid (DNA) was extracted using High Pure Viral Nucleic Acid kit (Roche Life Science, Mannheim, Germany) for quantitative PCR (qPCR).

Quantitative PCR:

We added 5 μL of sample to duplicate wells containing a 25-μL mixture of 15 μL of SybrSelect (Invitrogen, Carlsbad, CA), 300nM of primers (Integrated DNA Technologies, Coralville, IA), and water. Primers amplified a 126-bp region of the immediate early-1 gene loci of CMV (IE1 Forward – CAAGCGGCCTCTGATAACCA, IE1 reverse – ACTAGGAGAGCAGACTCTCAGAGG). The qPCR conditions consisted of a 2-minute cycle at 50°C followed by 10minutes at 95°C, 40 cycles of denaturation at 95°C for 15 seconds, and combined annealing/extension at 63°C for 1minute. Serial dilutions of a plasmid with the amplification region were used to generate a standard curve for quantification of CMV DNA. The sample was considered positive for CMV by qPCR if at least two out of six replicates had virus detected. Virus could be accurately quantified if more than 250 copies/mL were detected. Samples were repeated until two replicas were positive or six replicates were completed, whichever came first.

Maternal CMV status:

Maternal CMV status was determined using a CMV glycoprotein B (gB) ELISA on breast milk. This correlates 100% with CMV gB-specific immunoglobulin G (IgG) detection in plasma in a cohort of known CMV seronegative and seropositive women (►Supplementary Table 1 [available online only]). Plates were coated with 30ng/well of CMV gB protein (Sanofi, Paris, France) overnight at 4°C. Following incubation, plates were blocked for 2 hours with blocking solution (80.5% PBS+, 4% whey protein, 15% goat serum, and 0.5% Tween-20) then serial dilutions of breast milk (1:3–1:3,000) were added to wells in duplicate for 1.5 hours. Plates were washed two times and incubated for 1 hour with Goat Anti-Human IgG horse radish peroxidase (HRP) (1:10,000). After four washes, SureBlue Reserve TMB Microwell Peroxidase Substrate (KPL) was added for 10minutes, and the reaction was stopped with 1% HCl solution. Plates were read at 450nm. The lower threshold for positivity was greater than three times the average optical density of the blank wells on the same plate in the 1:3 dilution. Mothers were considered CMV positive if they tested positive in breast milk by CMV gB ELISA, CMV qPCR, or both.

Lactoferrin ELISA:

A commercial ELISA kit for human lactoferrin (AssayPro, St. Charles, MO) was used to determine lactoferrin concentration in saliva and breast milk. The concentration was determined using the provided standard control. For saliva, lactoferrin was reported as ng/swab because the volume of saliva collected on each swab was too small to reliably quantitate.

Neutralization assay:

The neutralization of lactoferrin was measured by a green fluorescent protein (GFP)-based assay. Briefly, ARPE-19 cells were seeded into 384-well flat-bottom plates and incubated for 1 day at 37°C and 5% CO₂ to achieve 100% confluency. Once confluent, dilutions of 1:4, 1:10, and a threefold, 10 series dilution (1:30–1:590,490) of lactoferrin in infection media were incubated with a multiplicity of infection of 2.0 of AD169r-GFP (Bad-UL131) for 60minutes at 37°C and 5% CO₂.⁴⁸ Four commercially available lactoferrin preparations: Human milk purified (Sigma, St. Louis, MO), Human milk recombinant (Sigma), Bovine (Bioferrin Glanbia) and Bovine (Jarrow) were used. Cells were incubated for 42 hours at 37°C and then fixed for 10 minutes with 3.7% paraformaldehyde. Cell nuclei were stained with DRAQ5 (1:5,000) for 10 minutes. GFP expression was read on a Cellomics Arrayscan (ThermoFisher Scientific, Waltham, MA). Neutralization titers (IC₅₀ and IC₉₀) were calculated according to the method of Reed and Muench using the relative light unit, indicating the lactoferrin concentration that resulted in a 50% (IC₅₀) and 90% (IC₉₀) reduction in fluorescence signal compared with control wells infected with virus only.

Statistical analysis.

Mann–Whitney’s t-test or ANOVA were used to test for significant differences between groups. Spearman’s rank-order correlation was used to test for correlations. Analyses were performed with GraphPad Prism 7.04 and R Studio.

Results

Potency of Human and Bovine Lactoferrin against Human CMV

We first assessed the neutralization activity of lactoferrin against CMV in vitro. Four commercial preparations were tested; two purified bovine lactoferrin products (Glanbia, Jarrow), human-purified lactoferrin (Sigma), and human recombinant lactoferrin (Sigma). All lactoferrin products neutralized HCMV strain AD169 repaired with IC₉₀s from 0.11 to 2.08 ng/mL (median: 0.14 ng/mL; ►Fig. 1). Interestingly, human-purified lactoferrin was significantly less effective, with an IC₉₀ >10-fold higher than the bovine and recombinant human lactoferrin preparations tested. Human recombinant lactoferrin was the most potent (IC₉₀ of 0.11 ng/mL) but was similar to the two bovine lactoferrin preparations (0.13 and 0.14 ng/mL).

Fig. 1.

Neutralization potency of human and bovine lactoferrin against human cytomegalovirus. Neutralization of HCMV AD169 by human purified, human recombinant, and two bovine-purified lactoferrin preparations. The dotted lines represent the IC₅₀ and IC₉₀ (ng/mL). HCMV, human cytomegalovirus.

Lactoferrin Concentration and CMV Load in Breast Milk and Saliva

Next, we characterized the levels of lactoferrin and CMV in breast milk and saliva over time and assessed for correlation between lactoferrin concentration and the magnitude of viral shedding in breast milk. Lactoferrin was detected in all breast milk and saliva samples. The breast milk concentration ranged from 1.8 × 10⁴ to 3.0 × 10⁷ ng/mL (►Fig. 2A). While the median lactoferrin concentration (3.3 × 10⁶ ng/ mL) was slightly above the IC₉₀ for human-purified lactoferrin, many samples fell below the human-purified lactoferrin IC₉₀. Lactoferrin concentration in infant saliva was lower and more variable from <1 to 2.1 × 10⁵ ng/swab (►Fig.2B). Saliva lactoferrin concentration was considerably below the IC₉₀ for human-purified lactoferrin for all samples. This variable concentration could be related to inconsistent amounts of saliva collected or natural infant variability. Of CMV positive mothers, 41 of 43 (95%) had detectable virus in breast milk for at least one time point. The viral load varied from the limit of detection to over 5.8 × 10⁷ copies/mL (median: 893.3 copies/mL; ►Fig. 2C). There was no correlation between the CMV load and lactoferrin concentration in breast milk (Spearman’s correlation, r = 0.003; p = 0.98; ►Fig. 2D).

Fig. 2.

Lactoferrin concentration and cytomegalovirus load in maternal breast milk and preterm infant saliva. Each open circle represents one sample and the black-dashed line connects samples from the same subject over time. (A) Breast milk lactoferrin concentration over time, n = 47. The red line represents the median. The black line is the IC₉₀ of human-purified lactoferrin against HCMV AD169. (B) Infant saliva lactoferrin concentration over time, n = 47. The red line represents the median. The black line is the IC₉₀ of human-purified lactoferrin against HCMVAD169. (C) Breast milk CMV load over time, n = 43 The red line represents the median. (D) Lactoferrin concentration versus CMV load in breast milk from samples with detectable virus, n = 78 samples from mothers of 39 subjects. Using Spearman’s correlation, r = 0.025, p = 0.83. HCMV, human cytomegalovirus; LOD, limit of detection; LOQ, limit of quantification; UD, undetected.

Lactoferrin Concentration and Postnatal CMV Acquisition

Next, we compared the level of milk and saliva lactoferrin in mother–infant pairs where the infant acquired pCMV and those that did not to determine if there is a role for lactoferrin in preventing pCMV transmission to preterm infants. Demographic characteristics of the infants enrolled (n = 62), those with exposure to maternal CMV (n = 47), and those that acquired pCMV (n = 5) are shown in ►Table 1. Eight infants tested positive for CMV in the first saliva sample (collected to exclude cCMV), so these mother–infant pairs were excluded from further testing/analyses. Three of these infants (three-eighths, 38%) tested positive for CMV in only one sample, and thus could represent transient oral infection⁴⁹ or breast milk contamination. In total, 47 of 62 (76%) mothers had detectable CMV IgG or virus in their milk, indicating seropositivity (►Fig. 3A). Of the exposed infants, 5 of 47 (11%) acquired pCMV (►Fig. 3A). Of note, the CMV load in saliva of postnatally infected infants was much lower than that detected in cCMV infections, and two postnatally infected infants had virus detected in saliva inconsistently over time (►Fig. 3B). The milk lactoferrin concentration was lower at several time points in mothers that transmitted CMV to their infant (weeks 4–6 and 8–10), but there was no significant difference between the two groups (p = 0.2–0.4; ►Fig. 3C). The saliva lactoferrin concentration was also lower at several time points in infants with pCMV (weeks = 1–5 and 7), but not significantly different between infants with and without pCMV (p = 0.4–0.7; ►Fig. 3D).

Table 1.

Demographics of infant cohort. Demographic characteristics of (1) infants from mother-infant pairs that had at least one maternal and one infant sample collected, (2) infants of mothers who tested positive for cytomegalovirus (immunoglobulin G or quantitative polymerase chain reaction) in breast milk, and (3) infants positive for postnatal cytomegalovirus

	Total n = 62 (%)	Mother’s milk CMV + n = 47 (%)	Postnatal CMV + n = 5 (%)
Gestational age (wk)
≤24	6 (10)	6 (13)	1 (20)
25–28	27 (44)	23 (49)	1 (20)
29–32	29 (62)	18 (38)	3 (60)
Birth weight (g)
<500	2 (3)	2 (4)	1 (20)
500–749	11 (18)	10 (21)	2 (40)
750–999	12 (19)	11 (23)	1 (20)
>1,000	37 (60)	24 (51)	1 (20)
Male	27 (44)	18 (38)	0
Race/ethnicity
White	33 (53)	25 (53)	3 (60)
African American	21 (34)	14 (30)	1 (20)
Other	8 (13)	8 (17)	1 (20)

Abbreviations: CMV, cytomegalovirus; qPCR, quantitative polymerase chain reaction.

Fig. 3.

Lactoferrin concentration and saliva cytomegalovirus load in mothers and infants with and without postnatal cytomegalovirus. (A) Schematic of infant CMV screening in this cohort. (B) Saliva CMV load in infants that acquired postnatal CMV infection (black, n = 5) and a representative sample of those with congenital CMV infection (blue, n = 4). Each line represents one infant. (C) The median lactoferrin concentration in maternal breast milk in mothers that transmitted postnatal CMV (red, n = 5) to their infant and those that did not (black, n = 42). Filled circles represent the median and error at each time point. The black line is the IC₉₀ of human-purified lactoferrin against HCMV AD169. (D) The median saliva lactoferrin concentration in infants that acquired postnatal CMV (red, n = 5) and those that did not (black, n = 42). Filled circles represent the median and error at each time point. The black line is the IC₉₀ of human-purified lactoferrin against HCMV AD169. CMV, cytomegalovirus; HCMV, human cytomegalovirus; LOQ, limit of quantification.

Clinical Outcome of Postnatally CMV Infection Preterm Infants

Finally, we reviewed clinical data to determine if infants identified with pCMV by prospective screening had discernable sequelae of infection. Of the infants positive for pCMV infection in saliva, three-fifths (60%) had a rule out sepsis event (blood culture, ± urine culture, antibiotics) performed within 2 days of the first positive sample (before or after). One infant has documentation that the work up was for abdominal distension, one infant’s work up was for bradycardia and desaturations, and the other had no reason documented. All bacterial cultures were negative. Another infant with pCMV developed medical NEC in thesecond week of life, prior to testing positive for pCMV on DOL 54 in this study. None of the identified infants with pCMV infection were diagnosed with BPD at discharge. Finally, all five postnatally CMV-infected infants passed a hearing screen prior to discharge. No infants were tested clinically for pCMV.

Discussion

In this study, we found that four commercially available preparations of lactoferrin neutralize CMV in vitro (IC₉₀: 0.11–2.08 ng/mL). We followed VLBW infants and their lactating mothers in the Duke NICU for their levels of this innate antiviral protein in mother’s milk and infant saliva and identified five infants with pCMV infection. We then compared the neutralizing lactoferrin concentration to that in breast milk and infant saliva, finding that the lactoferrin concentration in many breast milk and all saliva samples was much lower than the measured anti-CMV IC₉₀ of human-purified lactoferrin. Although milk lactoferrin levels were lower at many time points in women that transmitted pCMV, it did not differ significantly between pCMV-transmitters and nontransmitters. Saliva lactoferrin concentration was more variable and also lower at initial time points in infants that acquired pCMV, but did not reach significance in this small study.

Previous studies have demonstrated the anti-CMV activity of lactoferrin, with the potency varying with lactoferrin preparation (amount of iron bound, charge), CMV strain and cell type used (no activity to IC₅₀s of 0.15–1.75 ng/mL).^{36–38,41,50} While all preparations of lactoferrin tested had neutralizing activity against HCMV, human-purified lactoferrin was the least effective, with an IC₉₀ >10-fold higher than the other lactoferrin preparations. Previous studies have also found bovine lactoferrin to have higher antimicrobial activity than human lactoferrin, but to our knowledge, this has not been shown for CMVand the reasons behind this are unknown.^51,52 This finding suggests that bovine lactoferrin might be more protective against CMV transmission than the natural lactoferrin present in human breast milk.

The lactoferrin concentration in breast milk of these mothers of preterm infants was consistent with previous reports.^53,54 We found no correlation between milk CMV load and lactoferrin concentration, consistent with previous findings in small study of 23 preterm infants and their lactating mothers.⁵⁵ Finally, we found no major difference in lactoferrin concentration in milk of women that transmitted CMV and those that did not. This comparison was limited by the small number of infants with pCMV (5/47, 11%) and therefore is not powered to detect small differences. As mentioned previously, lactoferrin has neutralizing activity against CMV, as well as numerous other microbial pathogens. It is possible that this antimicrobial activity does not extend to CMV transmission via breast milk feeding. Another explanation is that no difference was found because the lactoferrin concentration in most breast milk is not adequate for effective CMV neutralization. Although the median lactoferrin concentration in breast milk (3.3 × 10⁶ ng/mL) was slightly above the IC₉₀ for human-purified lactoferrin (2.08 × 10⁶ ng/mL), the concentration in many samples fell below the IC₉₀. In addition, the concentration of lactoferrin that reaches the proximal and distal gastrointestinal tract, where CMV infection could occur, is likely much lower than that measured in expressed milk. Given this, there are likelyother proteins and innate immune molecules responsible for preventing transmission of pCMV to preterm infants. However, given the potent neutralizing activity of lactoferrin against CMV, a higher concentration of lactoferrin could improve protection against pCMV transmission.

A major limitation of this study was the low number of infants that acquired pCMV (11%), which is on the lower end of what was reported (11–32%) in a recent meta-analysis.¹⁶ This low rate could be due to the low enrollment (~50%) of extremely low birthweight (<1,000 g) infants in this study, who are at the highest risk for pCMV infection.^24,56,57 This could also be because of limitations of the accuracy of saliva CMV qPCR testing in VLBW infants. We noted significantly lower viral loads in VLBW infants with pCMV compared with those with confirmed cCMV. Many preterm infants have a small oral cavity, are intubated, have a feeding tube inserted and have small amounts of saliva. Infants could test falsely negative if not enough saliva, and therefore not enough CMV DNA is collected for detection. This could limit the wide-spread use of saliva CMV PCR for diagnosis of pCMV in preterm infants. Most clinical laboratories do not use an internal standard to ensure appropriate sample volume collection for saliva samples. In fact, a recent study found urine superior to saliva for detection of pCMV in preterm infants, but they did not explore the reasons behind this finding.⁵⁸ Future and confirmatory studies should use urine for CMV testing in VLBW infants.

It is important to note that the majority of infants (three-fifths, 60%) with pCMV infection had symptoms that prompted the medical team to test for bacterial sepsis and start antibiotics around the time of the first positive sample. None of the infants were tested clinically for CMV at that time, and bacterial cultures were all negative. The clinical symptoms could have been secondary to CMV infection or other sequelae of prematurity. VLBW infants in the NICU undergo numerous sepsis rule outs, and the contribution of pCMV tothis is unknown but could be significant. None of the infants in our study with pCMV developed BPD or had a failed hearing screen, and one had NEC, which has been associated with pCMV infection, but is unlikely to be associated in this patient given the timing.^59,60 Given the relatively mild symptoms reported in these infants, it is not surprising that they did not have long-term complications attributable to pCMV, which have mostly been associated with clinically identified pCMV, not prospective sampling.^21,23

Postnatal CMV is a potentially serious and life-threatening infection in preterm infants. Studies are needed to fully define the morbidity associated with pCMV and determine if treatment reduces morbidity. However, concurrent studies are needed to find ways to prevent pCMV acquisition while safely allowing for fresh breast milk feeding in this vulnerable population. While we did not find significant differences in the lactoferrin concentration in the breast milk of pCMV transmitting and nontransmitting mothers of preterm infants, our data suggest that the lactoferrin concentration in breast milk is not high enough to effectively neutralize CMV. Studies are in need to determine if a higher concentration of lactoferrin improves protection against pCMV transmission. As the edge of viability expands and NICU populations increase, strategies to reduce preventable morbidity such as pCMV will become increasingly important.