Perinatal Whole Blood Zinc Status and Cytokines, Adipokines, and Other Immune Response Proteins

(1) Background: Zinc is an essential micronutrient and zinc deficiency is associated with immune dysfunction. The neonatal immune system is immature, and therefore an optimal neonatal zinc status may be important. The aim of this study was to investigate the possible association between neonatal whole blood (WB)-Zinc content and several immune markers. (2) Methods: In total, 398 healthy newborns (199 who later developed type 1 diabetes and 199 controls) from the Danish Newborn Screening Biobank had neonatal dried blood spots (NDBS) analyzed for WB-Zinc content and (i) cytokines: Interleukin (IL)-1β, IL-4, IL-6, IL-8, IL-10, IL-12 (p70), interferon gamma, tumor necrosis factor alpha, and transforming growth factor beta; (ii) adipokines: leptin and adiponectin; (iii) other immune response proteins: C-reactive protein (CRP), and mannose-binding lectin (MBL), and soluble triggering receptors expressed on myeloid cells1 (sTREM-1). WB-Zinc content was determined using laser ablation inductively coupled plasma mass spectrometry. For each analyte, the relative change in mean level was modelled by a robust log-normal model regression. (3) Results: No association was found between WB-Zinc content and all the immune response markers in either the unadjusted or adjusted models overall or when stratifying by case status. (4) Conclusions: In healthy Danish neonates, WB-Zinc content was not associated with cytokines, adipokines, CRP, MBL or sTREM, which does not indicate a strong immunological function of neonatal zinc status.


Introduction
Zinc (Zn) is an essential trace metal for a plethora of metalloenzymes involved in growth, cognition, reproduction, and immune function [1]. Zinc serves as an antioxidant through several mechanisms [2]. For examples, the zinc ion has a structural role in stabilizing copper/zinc superoxide dismutase, thereby increasing the redox potential of copper ion (Cu (II)) resulting in the breakdown of free radicals, but zinc is not redox active and cannot catalyze Fenton chemistry [3]. Further, zinc induces the cysteine-rich zinc-binding protein metallothionein, which serves as an antioxidant [4]. Individuals were randomly selected from an original case-control study, conducted by our research group, aiming at examining predictors of childhood and adolescent T1D [25]. The focus of the current study is the association between zinc and a large number of cytokines, adipokines and other immune response proteins in healthy/non-infected newborns. Cases were those who later developed T1D, and controls were those who did not. Briefly, T1D diagnoses were retrieved from the Danish National Patient Registry based on International Classification of Diseases (ICD )codes (ICD-8 249 and 250. ICD- [10][11][12][13][14]x) and were cross-validated using the Danish Childhood Diabetes Registry (DanDiabKids) to secure a T1D diagnosis [26]. DanDiabKids, established in 1996, is a nationwide register, collecting data from all Danish pediatric diabetes clinics treating patients with T1D aged 0-18 years.
Cases were matched with controls based on date of birth. In total, 398 participants (199 cases and 199 controls) were included in the study, though 17 controls and 1 case were excluded from the primary pre-selected analysis due to missing covariate data ( Figure 1).

Assessment of Whole Blood Zinc Content
The NDBSs, measuring 3.2 mm in diameter and containing~3.4 µL of dried capillary blood, from each participant, were analyzed for zinc ( 66 Zn) using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) [19]. Additionally, the NDBSs were analyzed for potassium ( 39 K) in the same run. 39 K was used to adjust for possible differences in blood volume in the NDBSs and the hematocrit value in the newborn [27,28]. The NDBSs were analyzed by continuous line scans through the center of each sample, 12 consecutive points on each NDBS, which aimed to increase the precision of the WB-Zinc content. All of the samples were analyzed in the same analytical run, thereby minimizing operational variation.
During the study period (1991-1998) the American filter paper 903 (Whatman BFR, St. Louis, MO, USA) were used. By only including cases and controls from this period, we removed the risk of including filter paper, by different manufacturers, with possible zinc content differences. For further details regarding analytic setup, see Kyvsgaard et al. [19].
In all assays, matched pairs were run together to avoid batch effects/interassay variation [30]. Biomarker analyses are described in detail elsewhere [29]. Quality control of the analysis was made using mouse IL-6 as an internal analyte added to the extraction buffer to detect pipetting errors, and biotinylated beads to detect signal errors (a more detailed description are found in Skogstrand [31]). Calibration curves were used on each plate together with one high and two low controls. Samples, calibrators, and controls were analyzed in duplicates.
The absolute levels of the cytokines, adipokines and other proteins involved in the immune response stratified by case status are presented elsewhere [32].

Other Variables
A number of descriptive variables and possible confounders were available for inclusion in our analyses. These variables are listed and the coding of them is presented in Table 1.

Statistical Analysis
The random sampling of 400 individuals (200 cases and 200 controls) from the larger cohort of 2086 cases and 4172 controls [25] was performed by a random number generator in Statistical Analysis Software (SAS) version 9.3 using the floor-statement and was performed on the cases born from 1991-1998 (as the same filter paper was used in this period). Each case was originally matched with 2 controls based on date of birth but, for this nested study, we only included the first control found for each case.
Neonatal WB-Zinc content was log2-transformed, due to a log-normal distribution. Results are therefore interpreted as relative change (RC) in mean levels of immune response markers for each doubling in neonatal WB-Zinc content. We analyzed the data overall and stratified by case status.
Neonatal WB-Zinc content was modeled by a robust log-normal model regression, taking into account: (i) that measurements are potentially both left and right censored; and (ii) correlation within immunoassay. To account for correlation within the immunoassay, an inference was based on a working independence generalized estimation equation (GEE) approach.
The simultaneous evaluation of neonatal WB-Zinc content on all analytes was performed using the model stacking approach detailed in Pipper et al. [33].
Subsequent adjustment for multiple testing and familywise 95% confidence bands (95% CI) were calculated using the single-step procedure by Hothorn et al. [34]. GEE estimates of mean ratios and accompanying confidence limits were calculated on a log-scale and transformed back to the original scale.
Model selection was performed prior to statistical analyses based on our previous work using these data: (i) univariate models; and (ii) primary adjusted models using possible confounders (covariates associated with neonatal WB-Zinc content (birth weight and birth year), see Kyvsgaard et al. [19]). Due to the construction of our study sample, we also included case status in the multivariate models.
Overall functional misspecification by including neonatal WB-Zinc content as a trend (linear variable) was assessed by a lack-of-fit test. Specifically, we included a quadratic term of neonatal WB-Zinc content and tested its significance by a robust Wald test.
p-values were evaluated at the two-sided 5% significance level and we decided prior to our statistical analyses both to include results with and without adjustment for multiple testing.
All analyses were made using the statistical software package R version 3.5.1 (the R foundation for statistical programming, Vienna, Austria) and the add-on packages, survival and multcomp.

Ethics
The study was performed in accordance with the Helsinki II Declaration. Furthermore, the study was approved by the Danish Ethical Committee (H-2-2014-007) and by the DNSB Steering Committee. According to Danish law, anonymous studies do not require further informed consent.

Unadjusted Models and Adjusted Models
No association between neonatal WB-Zinc content and cytokines, adipokines and other proteins involved in immune response were found in the unadjusted and adjusted analyzes without correction for multiple testing ( Table 2). Results remained the same after correction for multiple testing (Supplementary Table S1). Non-linear models were considered for the above-mentioned primary exposures. A quadratic term did not improve model fit, and so these exposures were retained as linear predictors.

Unadjusted Models with Statification on Case Status
We found no signs of effect modification by case status (neonates who later developed T1D versus neonates who did not) on our main effects. Results are presented in Table 3. Table 3. Relative change in mean neonatal levels of cytokines, adipokines and other proteins involved in the immune response with a 95% confidence bands linked with every doubling in neonatal whole blood zinc content-results from univariate models with stratification by case status and no correction for multiple testing.

Comparison with Other Studies
A randomized controlled trial included newborns with sepsis, and one group was zinc-supplemented (i.e., 3 mg/kg of zinc sulfate monohydrate twice a day orally for 10 days) on top of standard treatment with antibiotics (n = 134, age < 28 days). The zinc-supplemented group of neonates had significantly higher serum zinc levels and a significant decline in the concentration of the inflammatory cytokines IL-6 and TNF-α. The control group only had a significant reduction in IL-6 after treatment with antibiotics. Mortality was lower in the zinc group, though it was not statistically significant (p = 0.12) [35]. This study suggests that zinc has anti-inflammatory properties. The study differs from ours by including infected newborns and by using zinc supplementation as intervention. Zinc supplementation lead to a higher variation in serum zinc levels between the two groups of newborns, thereby increasing the power and limiting the risk of type 2 errors (false negative results). A murine study found that the IL-6 functions as an enhancer of the zinc importer zinc transporter protein (Zrt)-, iron-regulated transporter (Irt)-related protein (ZIP)14 expression on hepatocytes, resulting in lower plasma concentrations of zinc when inflammation was induced [36].
A study, which included a cohort of healthy breast-fed Indonesian infants, had their serum zinc analyzed at age 3, 6 and 9 months. Further, biomarkers for inflammation (i.e., CRP and α-1-glycoprotein) were measured. This study showed that infants with lower zinc status had higher CRP levels. This study also highlights that zinc deficiency is overestimated when the infants are in a condition with inflammation [9], which, as mentioned above, may be mediated by IL-6, etc. [36]. We could not confirm an association between levels of neonatal WB-Zinc content and CRP, perhaps due to the differences in age between our studies. At 3-9 months of age, zinc status is primarily dependent on the infant's diet and not dependent on the mother's nutritional status, as is the case when a fetus [37]. In addition, the cohort of infants were from Indonesia, where approximately 30% of the population have an inadequate zinc intake, compared to only 6% in Scandinavia [5]. Maternal zinc status is dependent on the nutritional status and/or chronic infection [38]. Maternal zinc deficiency can lead to a reduction in placental zinc transport to the fetus, thereby inducing zinc deficiency in the fetus. However, this is only observed when the maternal zinc deficiency is severe due to the fact that the active transport of placental zinc to the fetus is maintained even though maternal zinc levels are lowered [1].
A study, including 88 healthy mothers who gave birth to healthy newborns with a gestation age of 38-42, showed that maternal serum leptin levels were positively correlated with maternal zinc levels. However, serum leptin levels in cord blood were not found to be associated with serum zinc in cord blood [39]. The latter result is in line with findings from our study.
In elderly individuals, zinc supplementation decreases circulating inflammatory cytokines, CRP and other plasma oxidative stress markers [12,13,15]. A randomized, double-blind, placebo trial included healthy elderly individuals (n = 40, aged 56-83). The intervention group received 45 mg zinc/day for 6 months, which lead to an increase in plasma zinc, and a decrease in CRP and IL-6 levels [15]. A similar randomized, double-blind, placebo-controlled study included healthy elderly individuals (n = 50, aged 55-87). The intervention group received 45 mg zinc/day in total in 12 months. Zinc supplementation lead to a significantly lower incidence of infections, an increase in plasma zinc, and a decrease in TNF-α levels and oxidative stress markers [13]. Further, this study showed that older individuals had significantly lower plasma zinc and increased oxidative stress compared with a group of younger adults (n = 31, aged 18-54) [13]. Furthermore, a cross-sectional study included healthy Rural Korean Adults (n = 1,055 i.e., 404 men and 651 women, aged > 40). These subjects had their dietary zinc intake analyzed. No significant inverse relationship between dietary zinc intake and inflammatory cytokines/markers IL-6, TNF-α, and CRP were shown [12]. These studies [12,13,15] suggest that the elderly population differ from newborns due to the increased level of oxidative stress and higher risk of zinc deficiency, which can explain why studies that include elderly individuals are able to show a significant decrease in levels of inflammatory cytokines and other inflammatory markers when zinc supplemented. Further, randomized controlled trials are more powerful compared to observational study designs.

Strengths and Limitations
This study is larger than many of the studies previously published on the association of early life zinc exposure and immune response markers [9,14,16,17,35,39]. Further, we report findings from a population of neonates with no sign of serious infections (CRP range of 0.01-20.1 mg/L). A limitation is pre-analytical factors, which could falsely elevate levels of zinc, such as hemolysis, starvation, and diurnal variation [1]. In particular, dried blood spots are prone to hemolysis and, unfortunately, we did not have serum values. The zinc level for each participant was divided with the potassium level, thereby adjusting for hematocrit, thereby limiting hemolysis bias [28]. The study is cross-sectional without serial data on zinc and cytokines, adipokines and other proteins involved in the immune response, which could have given further information on long-term associations. Also, we did not have information on the mothers' zinc or nutritional status, thereby we were not able to investigate the link between the mothers' and newborns' zinc status and immune response. A reason for why no associations were found might be due to the fact that zinc deficiency is rare in Denmark~6% [5], but our intent was also to examine associations within the normal physiological span. Residual confounding cannot be excluded either.

Future Perspective
No link between neonatal WB-Zinc content and cytokines, adipokines and proteins involved in the immune response was found in this study of healthy neonates. In future research, it would be interesting to conduct serial analyses of zinc status including other micronutrients during pregnancy and in the neonatal period in relation to immune response markers, due to possible complex interactions between micronutrients and the immune system. It may be relevant to include subgroups of newborns with an increased risk of zinc deficiency (i.e., premature [1,40]) or newborns with low birth weight [41] in future research.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2072-6643/11/9/1980/s1, Table S1: Relative change in mean neonatal levels of cytokines, adipokines and other proteins involved in the immune response with 95% confidence bands linked with every doubling in neonatal whole blood zinc content-results from models with correction for multiple testing.
Author Contributions: J.N.K., J.S., and S.U.T. designed the research (project conception, development of overall research plan, and study oversight). S.U.T, J.N.K., F.P., and J.S. were all majorly involved in data acquisition (hands-on conducting of the experiments and/or data collection). C.B.P., S.U.T. and C.E. analyzed data and performed statistical analysis. J.N.K., S.U.T., C.E., E.B.L., C.B.P., F.P., and J.S. wrote the paper (major contribution). J.N.K., S.U.T., C.E. and J.S. had primary responsibility for the final content. All authors have read and approved the final manuscript. Finally, all authors have agreed to be accountable for all aspects of the work. Søren Husted for helping with the trace elements analyses. We thank all members of DanDiabKids for collecting data on cases with type 1 diabetes. This research was supported by the Danish Diabetes Association.