Maternal serum but not breast milk IL-5, IL-6, and IL-13 immune markers are associated with scratching among infants

This study was derived from two longitudinal studies, which were approved by the University of South Carolina, the Medical University of South Carolina, and the Palmetto Health Institutional Review Boards for Human Subjects. All participants signed a written consent form either in English or Spanish.

Expecting mothers were enrolled between April 2008 and January 2010 in Columbia and Charleston, South Carolina in obstetric clinics and prenatal classes. Eligibility criteria included (1) aged 18 years or older, (2) no chronic illness (diabetes, thyroid or adrenal disorders, or chronic infections), (3) planned to stay in the area for at least 9 months, and (4) willingness to provide a breast milk sample 2 weeks after delivery.

Data collection was described elsewhere [25]. In brief, women took part in four telephone or in-person interviews: a core demographic and baseline interview conducted before delivery, and three interviews at 2 weeks, and 6 and 12 months after delivery, respectively. In the baseline interview, information was obtained about women’s race (African American, European American or other), maternal age, marital status, cigarette use prior to and since the beginning of pregnancy, whether cigarettes were smoked inside the home (categorized into yes vs no), education level (less than high school, some college, college graduate, or graduate school). In addition, maternal and paternal history of asthma, wheezing and whistling in the chest was obtained by asking: “Have you ever had asthma?” and “Have you ever had wheezing or whistling in the chest at any time in the past?” A history of allergic rhinitis was assessed by two questions: “Have you ever had hay fever?” and “Have you ever suffered—in the absence of a cold—from an itchy stuffy or runny nose and/or swollen, itchy eyes?” Eczema history was assessed by the question: “Have you ever had an itchy rash, which was coming and going for at least 6 months?”

The second interview was conducted 2–4 weeks after delivery to collect data on the specifics of delivery (spontaneous vaginal delivery, after induction vaginal delivery, Cesarean section), gestational age (weeks), maternal and household smoking during pregnancy period (yes, no), gender of the offspring, maternal use of antibiotics and acetaminophen during pregnancy, and history of vaginal infections/pelvic conditions during pregnancy. The date of delivery was used to define exposure to pollen in South Carolina [25].

The third and fourth interviews were based on the International Study of Asthma and Allergies in Childhood (ISAAC) questionnaire [28] and additionally ascertained scratching at ages 6 and 12 months. The recall period was restricted to the respective previous 6 months. At ages 6 and/or 12 months scratching (dichotomized) was based on a positive answer to the following question: “How often has your child scratched his/herself in the last 6 months?” [options: almost always; very often; sometimes; never]. In addition we collected information on a reported doctor’s diagnosis of eczema and eczema-like manifestations. A diagnosis of eczema was ascertained by using the question “In the last 6 months, was a doctor’s diagnosis made in your child of atopic dermatitis or eczema?” Regarding other skin manifestations of the child in the last 6 months, the following question was asked: “Have you noticed one or more of the following skin disorders in your child (1) cradle cap, (2) cheek-eczema, (3) ear eczema, and (4) eczema at other sites of the body. In addition, the interview collected information on duration of breastfeeding (weeks) as well as information on smoking in the house and pet exposure in the last 6 months.

Women were asked to provide one blood sample before delivery and one breast milk sample about 2 weeks after delivery. Ten milliliters of blood were taken by venipuncture from each woman in the last trimester of pregnancy (range 0–13 weeks before delivery, median 3 weeks before delivery). All serum samples were collected in sterile tubes (BD Vacutainer®, 10 mL), and centrifuged within 1 h of collection at 3500 revolutions per minute (rpm) for 10 min (minutes) at 4 °C. The separated serum samples were stored at −20 °C then after 24 h were transferred to −80 °C where the samples were stored until needed for analysis.

All women followed a detailed breast milk collection protocol. Each participant expressed approximately 15 mL of breast milk, on average 3 weeks after delivery (range 1–8 weeks), in the morning after putting the baby to breast using an electric breast pump provided by the study. All breast milk samples were collected in sterile plastic bottles (Medela 80 ml [2.7 oz]). Research staffs picked the breast milk sample up at the participant residence or a place of convenience. Within 1 h of collection, all samples were transferred to sterile centrifuges tubes and spun at 2900 rpm for 10 min at 4 °C. Fat was removed; centrifugation and fat removal steps were repeated until all fat was taken out. Finally, the cell pellet was removed. The isolated whey and serum were aliquoted and stored in a −80 °C freezer until preparation for immunoassays.

The concentrations of IL-1β, IL-4, IL-5, IL-6, IL-8 (CXCL8), IL-10, IL-12 (p70), IL-13, CXCL10 (IP-10), CCL11 (eotaxin-1), and IFN-γ in both serum and whey were assayed using the Bioplex Protein Array system (Bio-Rad Laboratories, Inc., Hercules, CA, USA). This multiplexes system allows for the assessment of several IMs in the same Bio-Rad custom-made bioplex pro human cytokine, chemokine, and growth factor multiplex plate. ELISAs (enzyme-linked immunosorbent assay) were used to determine concentrations of IgA (Immunology Consultants Laboratory, Portland, OR, USA) and TGF-β1 (R&D Systems, Minneapolis, MN, USA). To activate latent TGF-β1 to immuno-reactive TGF-β1 detectable by the Quantikine TGF-β1 immunoassay, we followed the manufacture procedure. All assays were conducted according to the manufacturer’s kit instructions. Each sample, including standards and the blank, was assayed in duplicate.

A total of 15 multiplexes and ELISAs were performed to determine the concentration of the IMs in serum and whey. For each plate we determined the limit of detection (LOD) by multiplying the standard deviation of the blank by three. Those samples that had concentrations below the detection limit were assigned a value corresponding to half the LOD.

The exposures were IMs levels in maternal serum before delivery and in whey. The following eight IMs were assessed: T-helper type 1/pro-inflammatory cytokines/chemokines (IFN-γ, CXCL10, IL-6, and CXCL8), T-helper type 2/pro-allergic cytokines (IL-5 and IL-13), T-regulatory/anti-inflammatory cytokine (TGF-β1) and IgA levels [25]. The concentrations were recorded in pg/mL, except for IgA, which was measured in mg/mL. Due to their distributions, we categorized IL-6, CXCL8, IL-13, IgA, and TGF-β1 into tertiles (high, median vs low) and IL-5, IFN-γ, and CXCL10 in both serum and whey, which were dichotomized (high vs median and low). Either the rank procedure was used to group IMs into tertiles, or the IMs were dichotomized (low vs high). Hence the lower group contains a combination of low values, the imputed ½ LOD values and other lower values.

We ran separate analyses for maternal serum and breast milk IMs. Confounders controlled for in the statistical analyses included maternal characteristics such as race, age at pregnancy, marital status, smoking during pregnancy, household cigarette use at ages 6 and 12 months, maternal and paternal history of eczema, consumption of acetaminophen during pregnancy, and vaginal or urinary infections during pregnancy. Offspring confounders comprised of gender and season of birth. Of note, we did not control for breastfeeding duration since the IMs are intervening variables between breastfeeding and scratching (i.e., only those infants who are breastfed will receive IMs in breast milk).

Regarding repeated measurement of scratching at ages 6 and 12 months, generalized estimating equations (GEE) were employed to determine the role of the maternal IMs. Adjusting for within-participant effects using the regular maximum likelihood method, we started with an unstructured covariance matrix, which requires the least amount of constraints. Other covariance matrices, including compound symmetry and autoregressive, were considered and most of the models presented the same QIC goodness of fit statistic [29].

A total of 16 adjusted models (eight for serum and eight whey markers) were run to analyze the effect of IMs on scratching. Initially, we considered analyzing serum and whey IMs in one model; however, this led to collinearity problems [30], since most of the IMs presented correlations above 0.5 (e.g., CXCL10 and CXCL8 in whey r² = 0.67; p < 0.0001).

We may consider that immune markers (IMs) in whey (after delivery) may result from maternal immune responses indicated by immune markers in maternal serum measured earlier during pregnancy. In this case, IMs in whey are the consequence of IMs in serum. Hence, they are in the same pathway with IM_whey as intervening variable: IM_serum → IM_whey → scratching. However, in regression models, it is not appropriate to include both serum IM and whey IM in the same model, since controlling for an intervening variable as a confounder would split the initial association between the risk factor and the outcome into two associations, destroying the potential pathway. For these reasons, we statistically analyzed IMs in serum and whey separately. Although whey IM did not gain statistical significance for scratching, they may gain importance in path-analytical models.

We detected that the IMs had non-linear relations with scratching. We tested for linearity by cross-tabulating groups with increasing immune marker levels with the outcome. Allowing non-linear associations, we therefore analyzed five IMs by comparing risk ratio (RR) estimates for scratching, contrasting the upper tertiles with the lowest tertile (low concentration) as reference group (IL-6, CXCL8, IL-13, IgA, and TGF-β1). IL-5, IFN-γ, and CXCL10 in both serum and whey were dichotomized (high vs median and low). Also we assessed the use of quartiles; however, there were not enough observations per category due to the sample size. Since we ran 16 individual adjusted models, we applied false discovery rate (FDR) (p = 0.05) for serum and whey IMs to adjust for multiple testing and avoid false positive associations [31]. We estimated the 95 % confidence interval (CI) and also presented p-values to identify associations that survive penalizing for multiple testing.

All confounders were simultaneously entered as indicator variables into the models. A backward elimination process was used to retain confounders in the final model: confounders were those variables which changed the effect of the main association by 10 % or more when adding this factor into the model. All statistical analyses were performed using SAS version 9.3.

Our results suggest that high levels of two Th-2 or type 2 immune response markers (IL-5 and IL-13) in maternal systemic circulation (serum) were a risk for the occurrence of scratching in infants 6–12 months later. Th2 cells play a critical role in the pathogenesis of allergic diseases including atopic dermatitis [32]. The Th2 cells are critical players in allergy because: (1) IgE antibody responses are Th2 (i.e., IL-4) dependent; (2) IL-5 is critical player in eosinophilia; (3) IL-13 is critical in mucus production and also enhances IgE responses; (4) these cytokines can also be produced by mast cells, basophil and eosinophil that can result in positive feed-back enhancement of chronic allergic reactions such as atopic dermatitis [33]. Our novel findings that higher maternal circulating levels of IL-5 and IL-13 cytokines enhances risk of scratching in infants reflects the possibility that such maternal environment conditions the fetus for a future disease phenotype upon delivery [23, 33, 34]. Prior investigations suggest that maternal cytokines play a role in the fetal programming, with a significant impact on fetal immune outcomes [34].

IMs related to T helper 1 (Th1) or type 1 immune responses are linked to cell-mediated and phagocyte-dependent protective responses and are generally considered as anti-allergic immune responses; it was surprising that we did not find a significant association of prototypic Th1 cytokine IFN-λ in the maternal circulation with protection from scratching among infants; however a type 1 promoting chemokine, CXCL10 (or IP-10) showed weak association before FDR analysis (Table 4) [35, 36].

Recent studies have identified IL-31 and IL-33 as a critical pruritogenic cytokines in humans and in mice because over expression of IL-31 and IL-33 in the skin enhances pruritus and atopic dermatitis phenotypes in mice [5]; and atopic skin lesions in both humans and in mice express very high levels of these two cytokines [4]. Most importantly, studies show that IL-33 positively interacts with IL-5 and IL-13 in exhibiting its pruritogenic function [5]. Here we report a significant relationship of maternal Th2 cytokines IL-5/IL-13 with infant scratching behavior. This suggest that future work is needed to evaluate whether maternal levels of pruritogenic cytokines IL-33 (and IL-31) could also be critical risk factors for infant scratching [4‐6, 37].

We unexpectedly found that the prototypic pro-inflammatory cytokine IL-6 in the maternal circulation was highly associated with scratching behavior among infants. Results of the repeated measurements suggest a risk for scratching 6–12 month later when the levels of IL-6 in maternal serum were higher. We showed that IL-6 cytokine in serum correlate positively with type 2 markers (IL-13, IL-5) and negatively with TGF-β1 [25]. To the best of our knowledge, no studies have considered the effect of this combination of IMs in maternal serum on scratching episodes. Furthermore, IL-6 has not been reported to be a critical player in allergic disease in general although a recent report has linked elevated cardiac IL-6 levels to tree nut induced systemic anaphylaxis for the first time in a mouse model [38]. Neverthess, IL-6 is considered to modulate pruritus because of its pro-inflammatory activity. For example, IL-6 levels were significantly higher in hemodialysis patients with uremic pruritus, compared to hemodialysis patients without uremic pruritus [2]. Further assessment is necessary to clarify the interrelation of IL-6 with proto-typic Th2 cytokines such as IL-5 and IL-13 on scratching and the possibility that scratching is more than simply a Th2 driven response.

In contrast to maternal serum Th2 cytokines and IL-6 that are linked to infant scratching behaviour in this study, no link with breastmilk IMs was found. This finding was surprising, and suggests that the half life of breast milk IM’s makes their quantification in a study such as ours difficult, or that systemic immune markers do not always correlate with mucosal (breast) immune markers.

There is a lack of agreement between reported scratching and reported doctor’s diagnosis of eczema but a strong association with cradle cap and weak associations with maternal observance of specific eczema location between 7–12 months. It is possible that the observed scratching is a response to different etiologies. Other triggering factors modulating pruritus may include contact and aero allergens, microbial agents, food, exogenous irritants, and endogenous factors [26]. However, maternal reported scratching can be considered as a more sensitive (but less specific) marker for childhood atopic dermatitis than physician diagnosed eczema. Further investigations are necessary to determine the agreement doctor’s diagnosed eczema and scratching episodes.