Introduction
Cells and organs of the immune system emerge in the first half of gestation. While the system is largely complete at birth, immune defenses continue to expand rapidly throughout infancy and early childhood in response to the environment. These critical periods of development may be highly vulnerable to nutritional insult,Reference Palmer 1 with functional alterations that may persist throughout life. Evidence from human studies carried out over the past 25 years have documented increased mortality due to infectious diseases,Reference Moore, Cole and Collinson 2 reduced antibody response to vaccinationReference Moore, Richards and Goldblatt 3 – Reference McDade, Beck, Kuzawa and Adair 5 and altered thymic functionReference Fulford, Moore and Arifeen 6 – Reference McDade, Beck, Kuzawa and Adair 9 among individuals exposed to nutritional insults in early life. The underlying mechanisms driving this developmental programming remain an active area of research.
The most basic lymphocytes, including the innate-like B1a cells, are derived from an early wave of long-lived progenitors.Reference Dorshkind and Montecino-Rodriguez 10 These progenitors are produced exclusively by pre-hematopoietic stem cells during the embryonic, fetal and neonatal periods and cannot be replenished by adult bone marrow cells.Reference Baumgarth, Tung and Herzenberg 11 Any insults coinciding with this first wave of lymphopoiesis could therefore have a permanent impact on the population of early progenitors and, consequently, their lymphocyte progeny.Reference Palmer 1 In the B lineage, a smaller allotment of early precursors could result in a permanent deficit of B1a cells, which serve as a first line of defense in the pleural and peritoneal cavities. B1a lymphocytes constitutively secrete most of the body’s polyreactive, low affinity, germline-encoded natural IgM antibodies.Reference Avrameas 12 – Reference Casali and Schettino 14 . Natural antibodies (NAb) recognize conserved epitopes on encapsulated bacteria, such as Streptococcus pneumonia,Reference Capolunghi, Rosado, Sinibaldi, Aranburu and Carsetti 15 and are particularly important during infancy, as their protection is not dependent on prior pathogen exposure.Reference Casali and Schettino 14 Current research is also focused on the immunoregulatory functions of the B1a cells and their protein products, which may have relevance for prevention of atherosclerosisReference Tsiantoulas, Diehl, Witztum and Binder 16 and cancer.Reference Vollmers and Brandlein 17
To our knowledge, the latent influence of nutrition on early lymphopoiesis has not been previously investigated. To examine this and other outcomes, we have undertaken long-term follow-up studies of a cohort of pre-adolescent children living in the southern plains or Terai of Nepal. Children in this cohort were born to participants of a randomized, placebo-controlled trial of vitamin A or β-carotene supplementation among married women of reproductive age,Reference West, Katz and Khatry 18 during which we collected information on health and nutrition exposures throughout pregnancy and early infancy. In our first report, we have shown that circulating NAb concentrations were higher in children (aged 9–13 years) born to mothers supplemented with vitamin A than in those whose mothers received either placebo or β-carotene.Reference Palmer, Schulze, Khatry, De Luca and West 19 Our present analysis investigates associations between nutritional exposures during pregnancy and early infancy on markers of early lymphopoiesis among pre-adolescent children (aged 9–13 years). We hypothesized that exposure to undernutrition early in life – as reflected by poorer maternal nutritional status, neonatal size, and infant growth and feeding practices during the first 3 months of life – would have a permanent impact on B1a cells, as indicated by lowered circulating NAb concentrations during pre-adolescence.
Methods
This study was conducted in the southern plains district of Sarlahi, Nepal, a low-lying, population dense, rural area that forms part of the country’s border with the State of Bihar, India. Our research in this area has documented high rates of infectious morbidity and undernutrition, including a stunting prevalence (height-for-age<−2 s.d.) in excess of 50% and a 70% prevalence of underweight (weight-for-age<−2 s.d.).Reference Wu, Katz and Mullany 20 We have also reported on the high risk of protein-energy and multiple micronutrient deficiencies among pregnant women,Reference Christian, Katz and Wu 21 , Reference Jiang, Christian, Khatry, Wu and West 22 and an ~40% incidence of low birth weight.Reference Christian, Khatry and Katz 23 Although breastfeeding is nearly universal, delayed initiation and early introduction of other foods are common practices, with most infants (~73%).Reference Mullany, Katz and Li 24
Data collection protocols
Children in the present study were born between 1994 and 1997 to mothers enrolled in a cluster-randomized, controlled trial of weekly low-dose vitamin A or β-carotene supplementation delivered from preconception through lactation.Reference West, Katz and Khatry 18 Married women living in the study area were randomized by ward (n=270 wards) to receive 7000 μg retinol equivalents of vitamin A, β-carotene (42 mg) or a placebo. Local field staff delivered supplements on a weekly basis and recorded information on pregnancy and vital status, menses in the past week and supplement compliance. When participants reported a pregnancy, trained field staff collected information on household demographics, socio-economic status, dietary and morbidity histories and other exposures of interest (e.g., maternal tobacco and alcohol use, tetanus vaccination and patterns of strenuous work). Weight (Model 727 digital floor scale; Seca, Hamburg, Germany), height (standing on foot plate with adjustable centimeter tape), mid-upper arm circumference (MUAC; to the nearest 0.1 cm at the midpoint of the left arm with an insertion tape) and hemoglobin (using a portable photometric hemoglobinometer; HemoCue Inc., Mission Viejo, CA, USA) were also measured. Staff conducted two home visits over the course of the pregnancy and two additional visits at 3- and 6-months postpartum. A subset of 2263 women consented to additional health and nutritional status assessments at the study clinic (mid-pregnancy and 3 months postpartum) and at home shortly after childbirth. At clinic visits, biospecimens were collected from mothers (serum and breast milk) and infants (heel prick blood sample). We calculated gestational age as the difference in time between the pregnancy outcome and the last prospectively reported date of a menstrual period from weekly surveillance.
From 2006 to 2008, we conducted a cross-sectional follow-up of trial participants and their children. Field workers collected data on vital and health status, dietary intake, environmental exposures, activities and socio-economic status. A 7-day dietary and morbidity history was also collected. Anthropometrists measured children’s height (Harpenden 98.608 portable stadiometer; Holtain Ltd., Crymych, UK), weight (Model 881 U scale; Seca), waist circumference (Model 200 insertion tape; Seca), MUAC (standard insertion tape), and triceps and subscapular skinfold thicknesses (Model 98.610ND skinfold caliper; Holtain Ltd.) according to standard protocols.Reference Gibson 25 Venous blood was collected into a 10 ml sodium heparin-containing collection tube. Blood samples were centrifuged at 1530 g for 10 min at room temperature and separated into four 1.5-ml aliquots. Plasma samples were frozen and shipped in liquid nitrogen tanks to the Johns Hopkins Bloomberg School of Public Health in Baltimore for storage at −70°C. For the present study, we considered only first pregnancies tracked during the supplementation trial, although women may have had more than one pregnancy over the course of the trial, and singleton births. We further restricted our investigation of immune programming to a subsample of children with complete plasma archives, including maternal and infant samples collected during the vitamin A or β-carotene supplementation trial and child samples collected at the time of the cohort follow-up (9–13 years).
Laboratory methods
As field work for the cross-sectional follow-up had already been completed, we were limited in this research aim to working with cryopreserved plasma samples. We hypothesized that early life nutritional insults would impact the allotment of B1a precursors, resulting in a permanent deficit of B1a cells and their NAb protein products.Reference Baumgarth, Tung and Herzenberg 11 We therefore adopted NAb concentrations as a proxy indicator of the size of the B1a lymphocyte population. There are numerous NAb specificities. We focused on immunoglobulin M (IgM) targeting double-stranded DNA (dsDNA), as this specificity has been well characterized as part of the germline-encoded, low affinity NAb repertoire.Reference Hahn 26 Anti-dsDNA IgM is detectable at regulated concentrations in normal human subjects throughout the life course.Reference Witte, Hartung and Sachse 27 , Reference Merbl, Zucker-Toledano, Quintana and Cohen 28
Anti-dsDNA IgM was measured in serum by an enzyme-linked immunosorbent assay purchased from ALPCO Diagnostics (Catalog # 35-DSSHU-E01; Salem, NH, USA). The kit employs recombinant human dsDNA and horseradish-peroxidase-conjugated anti-human IgM for capture and detection, respectively, and human serum of known anti-dsDNA IgM concentrations (range: 0–300 U/ml; functional sensitivity: 1.0 U/ml) for calibration. All standards, controls and samples were run in duplicate. We estimated intra-assay precision using 10 replicates each of three quality control samples, representing low, medium and high concentrations. Inter-assay precision was calculated from a quality control sample run in duplicate on each plate. Our mean intra-assay coefficient of variation (CV) was 4% and our inter-plate CV was 13% across nine microplates. As there is no international reference for anti-dsDNA IgM, data are presented using arbitrary units (U/ml). As markers of current infection, we measured C-reactive protein (CRP) on an Immulite 1000 chemiluminescent analyzer (#LKCRP1; Siemens Medical Solutions Diagnostics) and α1-acid glycoprotein (AGP) by radial immunodiffusion (#123511; Kent Laboratories, Bellingham, WA, USA). All analyses were carried out in the Center for Human Nutrition Laboratory at the Johns Hopkins Bloomberg School of Public Health in Baltimore, Maryland.
Statistical analysis
NAb concentrations were positively skewed, requiring natural log transformation to approximate a normal distribution. All exploratory analyses were conducted for the full sample and then separately by child’s sex. Categorical variables were analyzed using analysis of variance, with NAb concentrations expressed as geometric means and 95% confidence intervals (CI) across each category. To explore relationships between continuous variables, we calculated Pearson correlation coefficients and generated scatterplots with overlaid linear and quadratic fits. Exposures with P<0.1 in either the overall or sex-specific analyses were flagged for possible inclusion in multivariate models.
We first explored relationships between NAb concentrations and variables measured at the time of the 2006–2008 follow-up. As the B1a cell population is established early in life and circulating NAb concentrations are tightly regulated,Reference Baumgarth, Tung and Herzenberg 11 we did not expect NAb to vary significantly with exposures measured at follow-up, that is, current infection, indicated by elevated acute phase proteins (CRP⩾5 mg/l or AGP⩾1 mg/mlReference Wärnberg, Nova and Romeo 29 ) nutritional status or socio-economic status. We then assessed relationships with early life exposures. For that purpose, we characterized maternal nutritional status by short stature (height<140 v. ⩾140 cm) and wasting (MUAC<21.5 v. ⩾21.5 cm). Season of birth was defined as post-harvest (December–February), pre-monsoon (March–May), monsoon (June–August) or pre-harvest (September–November). For infant growth, we used weight (kg), length (cm) and MUAC (cm) measurements from the ‘birth’ home visit and the 3-month postpartum clinic visit. The timing of the ‘birth’ visit ranged from 0 to 117 days of age, with a mean timing of 13.1 days and a median of 10 days (interquartile range, 8–13 days). These measurements are more reflective of, and subsequently referred to as, neonatal size. As such, we excluded 35 cases where measurements were taken at >28 days of age from all analyses of infant size and growth. Timing of the ‘3 month’ clinic visit ranged from 65 to 171 days of age, with a mean of 90.2 days and a median of 85 days (interquartile range, 83–91 days). In order to account for the variability in the length of time between neonatal and 3-month visits, we generated standardized intervals by dividing the difference in weight from the neonatal to the 3-month visit by the number of days between visits, and then multiplying by 90 days. Infant feeding was characterized by: colostrum receipt, that is, received breast milk during the first 3 days of life (no v. yes); immediate breastfeeding, that is, received milk on day zero of life (no v. yes); and breastfeeding frequency at 3 months of age (⩽10 v. >10 feeds/day).
We employed multiple linear regression for our multivariate analyses. The cluster-randomized design of the original trial may have resulted in factors other than supplement allocation clustering within wards.Reference Liang and Zeger 30 We therefore used generalized estimating equations to correct standard errors. Multivariate models were built for the following primary exposures of interest: maternal short stature and wasting, season of birth, infant size (weight, height and MUAC at the neonatal home visit and 3-month postpartum clinic visit) and growth (standardized difference in weight, height and MUAC from the neonatal to the 3-month visit), and infant feeding (colostrum receipt, immediate breastfeeding and breastfeeding frequency at 3 months). As our exploratory data analyses revealed important differences by child’s sex, we generated models for the full sample and then for girls and boys separately. First, we generated crude models for each of the continuous outcome variables and primary early life nutritional exposures. We then controlled for any factors that were associated with both the exposure of interest (i.e., maternal MUAC) and the outcome variable at P<0.2, but not on the hypothesized causal pathway. Mediating relationships were then explored using variables that were related both to the exposure and outcome, but for which the exposure preceded that relationship. Finally, we used backward stepwise regression to generate a parsimonious model, retaining variables in the model that were statistically significant at P<0.1. As we have previously demonstrated an effect of maternal vitamin A supplementation on children’s NAb concentrations,Reference Palmer, Schulze, Khatry, De Luca and West 19 a variable representing maternal supplement allocation was included in all models.
We used Stata Statistical Software: Release 10 (Stata Corporation, College Station, TX, USA2009) for all analyses and considered P<0.05 indicative of a statistically significant relationship.
Ethical considerations
The Joint Committee on Clinical Investigation of the Johns Hopkins School of Medicine in Baltimore, MD, USA, the Nepal Health Research Council of the Ministry of Health in Kathmandu, Nepal and the Teratology Society in Bethesda, MD, USA approved the original trial protocol. The follow-up study protocol was approved by the Institutional Review Board at the Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA and by the Institute of Medicine at Tribhuvan University in Kathmandu, Nepal.
Results
A total of n=2263 mothers were eligible and enrolled for the clinical protocol of the original trial, from whom we had complete biospecimens at mid-pregnancy and early infancy from n=445 (Supplementary Fig S1). Of these offspring, 98% survived and were resident in the study area at the time of the follow-up (n=438). We received consent to collect blood for 66% of the surviving, resident children (n=290). Compared with non-eligible children, the mothers of those with complete biospecimens were of similar age (24.2±0.3 years for subsample v. 24.0±0.1 years for non-eligible; P=0.4), parity (2.3±0.1 v. 2.1±0.05; P=0.09) and nutritional status (54.8 v. 55.1% with arm circumference <21.5 cm; 0.9) at baseline. However, households in the subsample tended to be of higher socio-economic status at baseline (e.g., 76.8% were low caste or non-Hindu v. 87.1% of non-eligible mothers; P<0.001).
Selected household socio-economic, maternal, infant and child characteristics assessed during the trial and at follow-up are presented for each sex, and sexes combined, in Tables 1 and 2, respectively. When assessed during early to mid-pregnancy as part of the trial, mothers were generally malnourished, with a mean height and body mass index of 150.4 cm and 19.2 kg/m2, respectively, and 66% of women having a hemoglobin concentration <11 g/dl. Mid-neonatal (median of 10 days; interquartile range, 9–12 days) anthropometric data were available for almost 90% of children, by which time mean weight was 2799 g, and 24% still had a weight <2500 g. Mean neonatal length was 47.2 cm, with girls being 0.8 cm shorter than boys (P<0.05). Household variables reflect both the realities of poverty – inadequate sanitation, education and housing – and rapidly advancing rural electrification and communications, with 25–30% of households having a cell phone and 36% owning a television; 55% of assessed pre-adolescent children were stunted (<−2 z-scores in height-for-age based on sex-specific reference data for children aged 5–19 years 31 ), with females being significantly more so than males. Roughly one quarter of our sample had elevated acute phase proteins indicative of infection.
Maternal (antenatal) and infant characteristics during original trial (1994–1997) in rural NepalFootnote a
BMI, body mass index; MUAC, mid-upper arm circumference.
a Mean timing of antenatal clinic visit was 18 weeks gestational age (range: 5–38 weeks); mean timing of neonatal home visit was 13 days of age (range: 0–117 days); infant anthropometric data missing for n=42 children.
b Mean±s.d. (all such values).
c Significantly different between males and females, P<0.05 (χ2).
Individual and household characteristics of pre-adolescent children during follow-up (2006–2008) in rural Nepal
CRP, C-reactive protein; AGP, α1-acid glycoprotein.
a Mean±s.d. (all such values).
b Based on sex-specific reference data for children 5–19 years; 31 significantly different between males and females, P<0.05 (χ2).
c CRP⩾5 mg/l or AGP⩾1 mg/ml.
Geometric mean NAb concentrations were 18.0 U/ml (95% CI: 16.7, 19.4). These differed significantly between sexes (P<0.001), with a geometric mean of 21.1 U/ml (95% CI: 19.1, 23.3) in girls v. 15.5 U/ml (95% CI: 13.9, 17.3) in boys. Bivariate analyses, whether sex specific or for the full sample, did not reveal any significant associations between NAb concentrations and either (a) season at follow-up interview (i.e., season of assessment) or (b) any of the household or individual factors shown in Table 2 (not shown). Although no relationships were apparent between NAb concentrations and current infection, based on either of the inflammatory proteins, we did note that children who reported ear discharge within the week before their blood draw had significantly (P<0.05) higher geometric mean NAb concentrations (30.0 U/ml; 95% CI: 19.7, 45.7) than asymptomatic children (17.7 U/ml; 95% CI: 16.4, 19.1). This relationship did not differ by child’s sex.
Among males, neither indicators of maternal size (height, MUAC) nor season of birth had an influence on either crude or adjusted estimates of their NAb concentration. However, subjects whose neonatal weight was <2500 kg had a significantly lower NAb concentration than those whose weight was ⩾2500 g (Table 3). In contrast to males, season of birth, possibly mediated by maternal nutritional status, influenced pre-adolescent NAb concentrations in girls (Table 4). The NAb concentration was lower among girls born outside of the post-harvest season, but especially so if born during the pre-monsoon (βGIRLS=−0.39; 95% CI: −0.69, −0.09; P=0.01) and lean pre-harvest (βGIRLS=−0.35; 95% CI: −0.67, −0.03; P=0.03) seasons. Girls born to mothers with a MUAC<21.5 cm had an unanticipated, marginally higher NAb concentration (βGIRLS=0.20; 95% CI: −0.02, 0.43; P=0.08) although this effect disappeared when adjusted by season of birth.
Pre-adolescent anti-dsDNA IgM concentration by indirect and direct early life nutritional exposure in rural Nepalese males (n=151)
dsDNA IgM, double-stranded DNA immunoglobulin M; MAUC, mid-upper arm circumference; ANOVA, analysis of variance.
a Between‐group differences in anti-dsDNA IgM concentrations assessed by ANOVA.
b β coefficient (95% confidence interval) from simple or multiple linear regression analyses.
c Measurements taken at antenatal clinic visit (mean timing was 18 weeks gestational age; range 5–38 week); adjusted model includes land ownership, maternal age, clinical anemia at antenatal visit and maternal supplement allocation.
d Defined as post-harvest (December–February), pre-monsoon (March–May), monsoon (June–August) or pre-harvest (September–November); adjusted model includes paternal literacy, land ownership, clinical anemia, fruit and vegetable intake, tetanus toxoid vaccination, breastfeeding another child during pregnancy and maternal supplement allocation.
e As timing of ‘neonatal’ home visit ranged from 5 days to 38 weeks, neonatal size and growth models were restricted to children with visit at ⩽28 days; adjusted model includes paternal literacy, land ownership, maternal age, clinical anemia and maternal supplement allocation.
f Adjusted model includes land ownership, maternal age, clinical anemia, breastfeeding another child during pregnancy and maternal supplement allocation.
**P<0.01.
Pre-adolescent anti-dsDNA IgM concentration by indirect and direct early life nutritional exposure in rural Nepalese females (n=139)
dsDNA IgM, double-stranded DNA immunoglobulin M; MAUC, mid-upper arm circumference; ANOVA, analysis of variance.
a Between‐group differences in anti-dsDNA IgM concentrations assessed by ANOVA.
b β coefficient (95% confidence interval) from simple or multiple linear regression analyses.
c Measurements taken at antenatal clinic visit (mean timing was 18 weeks gestational age; range 5–38 week); adjusted model includes land ownership, maternal age, clinical anemia at antenatal visit and maternal supplement allocation.
d Defined as post-harvest (December–February), pre-monsoon (March–May), monsoon (June–August) or pre-harvest (September–November); adjusted model includes paternal literacy, land ownership, clinical anemia, fruit and vegetable intake, tetanus toxoid vaccination, breastfeeding another child during pregnancy and maternal supplement allocation.
e As timing of ‘neonatal’ home visit ranged from 5 days to 38 weeks, neonatal size and growth models were restricted to children with visit at⩽28 days; adjusted model includes paternal literacy, land ownership, maternal age, clinical anemia and maternal supplement allocation.
f Adjusted model includes land ownership, maternal age, clinical anemia, breastfeeding another child during pregnancy and maternal supplement allocation.
*P<0.05; **P<0.01.
We further explored neonatal size, growth and infant size in both sexes (Table 5). The relationships observed with low neonatal weight in males were similarly seen between infant weight and MUAC gain from the neonatal home visit to the 3-month clinic visit, as well as infant weight, length and MUAC measured at the 3-month clinic visit. The latter associations remained significant even after adjusting for neonatal size. On a continuous scale (Fig. 1), a one kilogram increase in male neonatal weight was significantly associated (P<0.001) with 1.54 U/ml higher pre-adolescent NAb concentration. None of the associations between size or growth and NAb concentrations in pre-adolescence were statistically significant for females.
Associations of anti-dsDNA IgM concentrations with measures of neonatal size in pre-adolescent children (n=290) in rural Nepal. Line represents fit of multiple linear regression equation with 95% confidence intervals indicated by gray shading; adjusted model includes paternal literacy, land ownership, maternal age, clinical anemia and maternal supplement allocation; *P<0.05; ***P<0.001. dsDNA IgM, double-stranded DNA immunoglobulin M.
Associations between neonatal size, growth in the first 3 months of life and size at 3 months of age and pre-adolescent anti-dsDNA IgM concentration in rural Nepalese males (n=151) and females (n=139)
dsDNA IgM, double-stranded DNA immunoglobulin M; MAUC, mid-upper arm circumference; ANOVA, analysis of variance.
a β coefficient (95% confidence interval).
b As timing of ‘neonatal’ home visit ranged from 5 days to 38 weeks, neonatal size and growth models were restricted to children with visit at⩽28 days; growth was calculated as a 90-day standardized difference in measures taken at the 3 month postpartum clinic visit and the birth home visit; all infant size/growth models adjusted for paternal literacy, land ownership, maternal age, clinical anemia and maternal supplement allocation.
* P<0.05; ** P<0.01; *** P<0.001.
In neither sex did we find any evidence of association between our assessed indicators of early infant feeding practice, including immediacy of breastfeeding after birth, receipt of colostrum or breastfeeding frequency at 3 months of age, and NAb concentration in pre-adolescence.
Discussion
NAb concentrations in our study were strongly influenced by a child’s sex. This is consistent with the well-known sexual dimorphism in overall IgM concentrationsReference Butterworth, McClellan and Allansmith 32 and has been reported previously from studies specific to NAb.Reference Halliday, Salaman, Seifert, Johnson and Malcolm 33 – Reference Frostegard, Tao and Georgiades 36 Differences are most commonly attributed to sex hormones.Reference Lleo, Battezzati, Selmi, Gershwin and Podda 37 Although we did not assess whether girls had begun menstruation, data on the age of menarche from an adjacent field site suggest that a subset of girls in our sample were likely pubertal.Reference Rah, Shamim and Arju 38 Sex-specific polymorphisms may also play a role. At least one mouse model has highlighted a genetic control of B lymphopoiesis resident on the Y chromosome, although this has not yet been demonstrated in humans. This is the first study that we are aware of that considers nutritional influences on a marker of early B lymphopoiesis. The results support our original hypothesis that early life nutritional exposures – particularly the season of birth and early infant size/growth – may have a lasting impact on the emerging immune system. However, the sex-specific nature of these relationships was unanticipated. NAb concentrations measured during childhood were significantly higher among girls born during the harvest season. In boys, we found strong positive associations between neonatal size, growth during the first 3 months of life and size at 3 months of age with NAb concentrations of boys at 9–13 years of age.
In our study, girls born during the post-harvest season had significantly higher NAb concentrations than girls born in pre-monsoon, monsoon or pre-harvest seasons, possibly reflecting a seasonal influence on the efficiency or absolute number of B1a cells. There was no effect in terms of season of birth among males. The observed seasonality is in line with previous observations from Nepal,Reference Hughes, Katz and Mullany 39 where the post-harvest months are characterized by a lower burden of infections, greater food security and improved nutritional status.Reference Jiang, Christian, Khatry, Wu and West 22 , Reference Hughes, Katz and Mullany 39 , Reference Panter-Brick 40 Differences in immune defenses by season of birth have been reported elsewhere, generally with better functional outcomes observed among harvest season infants.Reference Collinson, Moore, Cole and Prentice 7 , Reference Ngom, Collinson and Pido-Lopez 41 Several hypotheses have been put forward to explain these observations, including seasonal variability in fetal undernutrition, infections, and exposure to toxins and/or pesticides.Reference Prentice, Cole, Moore and Collinson 42 It seems unlikely that our observed effect was mediated by fetal nutrient deprivation, as we would then have expected to see consistent relationships with maternal nutritional status and size at birth. After adjusting for season, we found no association between low maternal MUAC, an indicator of acute undernutrition and NAb. Furthermore, there were no relationships between any of the infant size or growth measures and NAb concentrations in girls. Seasonal differences in maternal pathogen exposure might also influence fetal lymphocyte development, although we found no evidence that maternal morbidity mediated the observed effect. Still, as morbidity data were collected only at set times points during pregnancy, they may not be fully representative of fetal exposure. Seasonal exposure to immunotoxins such as aflatoxin or pesticides may offer an alternate explanation,Reference Turner, Moore, Hall, Prentice and Wild 43 , Reference Ofordile, Prentice, Moore and Holladay 44 but were not measured in this cohort. We might alternatively consider exposures around the time of conception. Research carried out in The Gambia has demonstrated epigenetic changes among infants conceived during the hungry season.Reference Waterland, Kellermayer and Laritsky 45 An insult occurring around the time of conception could explain why the lowest NAb concentrations were observed among girls born in the pre-monsoon season, as their conception would have coincided with the peak periods for infection, undernutrition and likely environmental exposures.Reference Jiang, Christian, Khatry, Wu and West 22 , Reference Hughes, Katz and Mullany 39 , Reference Panter-Brick and Ellison 46
Our data suggested significant relationships between neonatal size, infant growth and size at 3 months of age with NAb concentrations of boys during childhood. Victora et al. Reference Victora, Bilate and Socorro-Silva 47 found a similar direct correlation between birth weight and a separate NAb specificity in a cohort of Brazilian boys aged 18 years. No girls were followed in that cohort. At least one study has reported an association between birth weight and thyroid auto-antibodies among women.Reference Phillips, Cooper and Fall 48 Interestingly, this found that women with a lower birth weight actually had higher auto-antibody concentrations. Although not significant, relationships between NAb concentrations and infant size or growth variables among girls in our sample also tended to be negative. It may not be possible to directly compare these findings, as women in the former cohort were much older, aged 60–71 years, and some had a clinical diagnosis of autoimmune disease. Thus, auto-antibodies in that setting were more likely pathogenic in nature. One final study reported no significant relationships between early life size or growth and anti-nuclear antibodies in either sex.Reference Edwards, Syddall and Goswami 49 However, that study considered only reactivity, not absolute concentrations. They also found a lower than anticipated prevalence of reactive individuals, limiting their statistical power.
Assuming that early nutritional exposures do indeed influence the body’s allotment of B1a cells and, consequently, production of NAb throughout life, the question remains as to why these influences would differ by sex. As girls already had higher NAb concentrations, it is possible that we were unable to discern the same magnitude of effect as that seen with infant growth in males (i.e., a threshold effect). However, this would not explain the apparent seasonal influence on NAb concentrations in girls. The observed sex differences may alternatively relate to the different timing of the exposures, with season of birth reflecting an earlier critical window (i.e., early gestation) for girls than for boys (i.e., late gestation and early infancy). Females are known to be more sensitive to the maternal environment, although the basis of this sensitivity is poorly understood.Reference Tarrade, Panchenko, Junien and Gabory 50 A series of animal feeding studies highlights sex-specific differences in placental gene expression when exposed to high- or low-fat maternal diets.Reference Cox, Li and Glenn 51 – Reference Mao, Zhang and Sieli 53 One of these studies specifically points to dysregulation in females of genes involved in lymphocyte signaling.Reference Gabory, Ferry and Fajardy 52 There is also recent evidence of sexual dimorphism in maternal diet-induced epigenetic changes.Reference Gabory, Ferry and Fajardy 52 , Reference Chen, Ganguly and Rubbi 54 , Reference Gabory, Roseboom, Moore, Moore and Junien 55 Although males may be more resistant to in utero programming, this may leave them with greater sensitivity to postpartum nutritional or other insults.
The functional consequences of a lower B1a cell count and, consequently, lower circulating NAb concentrations would be most acute in the first 2 years of life. Infants and young children are unable to mount an adaptive immune response to bacterial capsular polysaccharides,Reference Janeway, Travers, Walport and Shlomchik 56 relying on maternal antigen-specific antibodies and their own broad-specificity NAb for protection.Reference Carsetti, Rosado and Wardmann 57 As several NAb specificities are known to be self-reactive, the role of NAb beyond early life has historically been studied in the context of autoimmune disease.Reference Avrameas 12 , Reference Clark, Bourne, Salaman, Seifert and Isenberg 34 , Reference Victora, Bilate and Socorro-Silva 47 – Reference Edwards, Syddall and Goswami 49 , Reference Azizah, Azila, Zulkifli and Norita 58 , Reference Williams and Isenberg 59 However, more recent research suggests that the self-reactive nature of NAb serves an immunoregulatory purpose, for example, in clearing cellular debris following apoptosis.Reference Binder 60 The relevance of this to human health has been demonstrated in research on atherosclerosis, where NAb have been shown to bind oxidized low-density lipoproteins, preventing its uptake by macrophages and the subsequent formation of foam cells.Reference Tsiantoulas, Diehl, Witztum and Binder 16 Other NAb specificities have been linked with the induction of apoptosis in cancer cellsReference Vollmers and Brandlein 17 and the promotion of remyelination in multiple sclerosis.Reference Watzlawik, Wootla, Painter, Warrington and Rodriguez 61
This study provides a unique contribution to the literature, benefiting from a population-based, prospective cohort of children for whom we had a rich database of information regarding exposures throughout pregnancy and early infancy. However, our research has a number of limitations. The hypothesis that lymphopoiesis may be particularly vulnerable to early life insults would be most appropriately tested using proximal, direct measurements of precursors and lymphocytes among peripheral blood mononuclear cells using flow cytometry. Our primary limitation was the reliance on cryopreserved plasma samples, in which we measured NAb concentrations as a proxy indicator. Although NAb are constitutively produced by B1a cells,Reference Holodick, Tumang and Rothstein 62 they have not been validated as a marker for this population. It is further possible that anti-dsDNA antibody concentrations are not representative of overall NAb concentrations.Reference Fesel, Goulart and Silva Neto 63 Our measures were potentially limited as well by focusing on children 9–13 years of age, although we reasoned that any impacts would likely be permanent and thus detectable even later in life. Further limitations stem from our assessment of some exposures during the original trial, for example, birth weight or infant feeding practices, which may have caused misclassification. The present analysis was part of a broader investigation of immune programming. As such, our inclusion criteria required biospecimens from mothers and their infants/children during pregnancy, postpartum and later in life. This subsample was of somewhat higher socio-economic status, potentially limiting our ability to extrapolate findings to the broader population.
Our findings suggest that early life nutrition and consequent size and growth during infancy can have a lasting influence on the emerging immune system. Furthermore, these influences may differ by the sex of the offspring. The consequences of this influence may be highly relevant in this population, where newborns are exposed to a high infectious burden. It may also have a lasting impact on the child’s ability to maintain immune homeostasis, with potential long-term implications for immune-mediated and/or inflammatory disease.
Acknowledgments
The authors thank the participating children and their families, as well as members of the Johns Hopkins and Nepal Nutrition Intervention Project-Sarlahi (NNIPS) study teams including Parul Christian, Joanne Katz, Steve LeClerq, Margia Arguello and Veena Singh for their contributions to the study.
Financial Support
This analysis and follow-up studies were supported by grant OPP614 (Global Control of Micronutrient Deficiency) from The Bill and Melinda Gates Foundation, Seattle, WA; the original trial was supported by the US Agency for International Development (USAID), Washington DC under Cooperative Agreement No. DAN 0045A 005094-00; additional assistance was received from the Sight and Life Global Nutrition Research Institute at Johns Hopkins University; and ACP was supported by a pre-doctoral fellowship from the Procter & Gamble Company, Cincinnati, OH.
Conflicts of Interest
None.
Ethical Standards
The authors assert that all procedures contributing to this work comply with the ethical standards of the national guidelines from the United States and Nepal on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008, and has been approved by the institutional committees of the Johns Hopkins School of Medicine in Baltimore, MD, USA, the Nepal Health Research Council of the Ministry of Health in Kathmandu, Nepal and the Teratology Society in Bethesda, MD, USA (original trial protocol), as well as committees at the Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA and the Institute of Medicine at Tribhuvan University in Kathmandu, Nepal (follow-up protocol and present analysis).
Supplementary Material
For supplementary material/s referred to in this article, please visit http://dx.doi.org/10.1017/S2040174415001373
