When is Helicobacter pylori acquired in populations in developing countries? A birth-cohort study in Bangladeshi children

ABSTRACT

Helicobacter pylori colonization is prevalent throughout the world, and is predominantly acquired during childhood. In developing countries, >70% of adult populations are colonized with H. pylori and >50% of children become colonized before the age of 10 years. However, the exact timing of acquisition is unknown. We assessed detection of H. pylori acquisition among a birth cohort of 105 children in Mirzapur, Bangladesh. Blood samples collected at time 0 (cord blood), and at 6, 12, 18, and 24 months of life were examined for the presence of IgG and IgA antibodies to whole cell H. pylori antigen and for IgG antibodies to the CagA antigen using specific ELISAs and immunoblotting. Breast milk samples were analyzed for H. pylori-specific IgA antibodies. Cord blood was used to establish maternal colonization status. H. pylori seroprevalence in the mothers was 92.8%. At the end of the two-year follow-up period, 50 (47.6%) of the 105 children were positive for H. pylori in more than one assay. Among the colonized children, CagA prevalence was 78.0%. A total of 58 children seroconverted: 50 children showed persistent colonization and 8 (7.6%) children showed transient seroconversion, but immunoblot analysis suggested that the transient seroconversion observed by ELISA may represent falsely positive results. Acquisition of H. pylori was not influenced by the mother H. pylori status in serum or breastmilk. In this population with high H. pylori prevalence, we confirmed that H. pylori in developing countries is detectable mainly after the first year of life.

Keywords:

• acquisition
• transmission
• epidemiology
• developing countries
• CagA
• seroprevalence
• seroconversion
• breast feeding
• serological diagnostic testing

Introduction

Helicobacter pylori is most prevalent in developing countries^1-3 since the incidence of its acquisition has fallen dramatically in developed countries.^4-9 Colonization of adults with H. pylori is persistent and usually asymptomatic.^5,10 In contrast, both transient and persistent colonization has been observed in children.^11-13 Acquisition of H. pylori occurs predominantly in childhood^11,14–16 and intra-familial clustering has also been described.^17-19 However, the exact timing of acquisition is still not well-understood.

Our prior prospective study in Apache children showed that detection of H. pylori acquisition occurs mainly in the second and third year of life.²⁰ A parallel trend was observed in Bangladeshi children in whom there was 33% H. pylori prevalence at age 10–15 month and 84% at age 5 to 8 years.²¹ Similarly, in Gambian children, H. pylori prevalence was 19% at 3 months of age and 84% in those older than 30 months.²² In Madagascar, prevalence was 18.1% in children <1 year, but 61.5% for children >10 years.²³ In developed countries, although overall prevalence is lower, in German children, prevalence was 8.9%, 36.4% and 31.9% in 1-, 2-, and 4-year olds.²⁴ In Japanese children, prevalence was 1.9%, 7.1%, 6.7%, and 7.6% at <1, 1–3, 4–6, and 7–9 years.²⁵ While all of these studies indicate that acquisition of H. pylori is detected mostly after the first year of life, they did not determine maternal H. pylori status at time of birth, nor were the specific bacterial antigens that are the target of the immune response identified. The humoral immune responses against H. pylori have been used as markers of acquisition as well as a diagnostic tool in humans,^26,27 mice,²⁸ and rhesus macaques.²⁹ However, there is no indication that the specific humoral responses to H. pylori are protective³⁰ and reinfection following anti-microbial treatment is not unusual.³¹ Pepsinogen I (PGI) is a proenzyme of pepsin secreted by chief cells in the gastric fundus.³² With H. pylori acquisition, serum PG1 levels in adults rise, remain high and can predict later changes in the gastric mucosa.³³ In children, PG1 levels display a sharp but transient increase.²⁰ Breastfeeding is important for the development of the infant microbiota and immune system³⁴ and is known to be protective against many organisms.³⁵ There are indications that nursing provides protection for H. pylori colonization in both developed^36,37 and developing countries.³⁸

The purpose of the present study was to assess when H. pylori acquisition can be detected among children in developing countries, to characterize antigenic targets that are potentially associated with early seroconversion, to confirm whether PGI levels reflect gastric physiologic changes associated with H. pylori acquisition, and to determine the effect of breastfeeding on H. pylori acquisition.

Results

Maternal antibodies in the birth cohort

We first estimated the prevalence of H. pylori in the maternal population by assessing the immune response to H. pylori antigens in the cord blood samples, as an indicator of the mother's status (Table 1). As in previous studies^39,40 we used the presence of either H. pylori specific IgG or CagA specific IgG as our gold standard for H. pylori positivity. Of 98 cord blood samples examined, the prevalence of IgG to water-extracted whole cell antigen (WEWCA) (70.4%) and CagA (78. 6%) were similar. In total, 85 (86.7%) of the 98 mothers were H. pylori-positive when defining positivity as the presence of serum antibodies to either antigen. Assuming that H. pylori-positive mothers would display H. pylori or CagA-pecific IgA antibodies in their milk, we also used the presence of IgA in the mother's breast milk as an indication of maternal status (Table 1). Of 43 mothers who had detectable breast milk antibodies to H. pylori antigens, 33 (76.7%) also had serum antibodies to either the H. pylori or CagA antigen. Subsequently, we considered the presence of serum or breast milk antibodies as evidence of H. pylori positivity in mothers⁴¹ to exclude the possibility of false negative results based on the serum IgG, which has been reported.^40,42–44 Thus, 91 (92.8%) of the mothers had evidence for carriage of H. pylori; of these mothers, 93.4% were CagA positive. Thus, the mothers in this study represented a population with very high H. pylori prevalence, as reported from other developing countries.⁴⁵

Table 1. ELISA results to assess H. pylori status in mothers using cord blood from their children or breastmilk.

	Cord blood (n = 98)		Breast milk (n = 104)
Antigen detected	n	%	n	%
Whole cell antigen^*	69	70.4	40	38.5
CagA	77	78.6	38	36.5
Either	85	86.7	43	41.3

* WEWCA (water-extracted whole cell antigen), as reported.²⁰

^**recombinant CagA antigen, as reported.^5,83

Final H. pylori status of children

Next, we determined the H. pylori status of the children at the end of the study period (24 month of age). Again, we considered presence of either H. pylori- or CagA-specific IgG as the standard for H. pylori positivity and then combined the results of the serum IgA determinations in order to exclude falsely negative results. Of the 105 children, 55 (52.4%) were negative in all ELISA tests determinations, whereas 50 children (47.6%) were positive for H. pylori in one or more of the ELISAs during months 12 to 24 (Table 2). The sero-prevalence of H. pylori (47.6%) was substantially lower in the 2–year old children than in their mothers (86.7%). However, as with the mothers, CagA sero-positivity [39 (37.1%) of 105] exceeded H. pylori IgG sero-positivity [33 (31.4%) of 105]. Of the 50 H. pylori positive children, 39 (78%) were CagA-positive, similar to the proportion of their mothers who were CagA seropositive (78.6%).

Table 2. Serologic markers of H. pylori status in 105 birth-cohort children at 24 months of age.

Whole cell antigen^*		CagA antigen^**
IgA	IgG	IgG	Number of children in category	% of total
−	−	−	55	52.4
−	−	+	20	19.0
−	+	−	9	8.6
+	−	−	2	1.9
+	−	+	2	1.9
−	+	+	4	3.8
+	+	+	13	12.4

* WEWCA (water-extracted whole cell antigen), as reported.²⁰

** recombinant CagA antigen, as reported.^5,83

The relationship of maternal H. pylori status to the child's status

Next, we evaluated the role of maternal status in the child's early life acquisition of H. pylori (Table 3). We expected that children from H. pylori-positive mothers would have a higher likelihood of being positive, therefore; we compared the final maternal status with the final status of their children. Of the 98 mothers, only 7 were H. pylori-negative and four (57.1%) of their children also were negative but the other three (42.9%) showed positive ELISA results. However, despite the high H. pylori prevalence in mothers, only 47 (51.6%) children of the 91 H. pylori⁺ mothers were sero-positive. Thus, despite low power for calculations (only 7 negative mothers) the H. pylori sero-status of the mother seems not to be a predictor for the acquisition and maintenance of H. pylori in 24-month old children (OR 1.05, 95% CI 0.22-4.94, p = 0.96).

Table 3. Relationship between maternal H. pylori status and child's H. pylori status at 24 months of age.

Maternal H. pylori status^a		Number of mothers	% of children positive^b
Serum:	negative	13	54.8
	positive	85	47.1
Breast milk:	negative	61	47.5
	positive	43	46.5
Combined:	negative	7	42.9
	positive	91	48.4

^a Based on H. pylori-specific antibodies (serum IgG, or breastmilk IgA).

^b Based on H. pylori or CagA-specific antibodies (IgA and/or IgG).

To extend this observation, we repeated the analysis examining serum IgA to H. pylori in the mothers. However, only 8 (8.2%) of the 98 mothers were positive for H. pylori-specific serum IgA. Among these children, 3 (37.5%) were sero-positive and 5 (62.5%) sero-negative; this was similar to the sero-positivity in the 90 IgA-negative mothers (48.9%; p = NS).

Next, we evaluated the utility of H. pylori-specific IgA in maternal breast milk as a predictor of childhood outcome. We hypothesized that H. pylori-specific IgA in breast milk would protect children from H. pylori acquisition. Of 104 mothers (Table 3), 43 (41.3%) were positive for H. pylori- or CagA-specific IgA in their breast milk. Among these mothers, 20 (46.5%) of their children were H. pylori-sero-positive at age 2. Of the 61 mothers who did not have H. pylori-specific IgA-detectable in their breast milk, 29 (47.5%) had sero-positive children. The presence of IgA in the breast milk was not a predictor for the H. pylori status of children (OR 0.96, 95% CI 0.44-2.10, p = 0.92); thus, H. pylori-specific IgA in breast milk does not appear to protect breast-fed children from H. pylori colonization.

H. pylori-seroconversion in children

We assessed the number of children who seroconverted to H. pylori positivity during the first two-years of life (Table 4). The majority [50 (86.2%)] of the 58 children who seroconverted to H. pylori antigens remained sero-positive at age 2 years which is in line with other studies.¹⁶ For the 8 children who had transient seroconversion, the mean number of positive tests was significantly lower than in the children who were sero-positive at age 24 months (Student's t-test; p = 0.05) (Table 4).

Table 4. Response to H. pylori antigens in 105 Bangladeshi birth-cohort children, at 24 months of age.

		IgA	IgG
Seroconversion status	Number of children	WEWCA^*	WEWCA^*	CagA	Mean number of positive assays
Sero-positive^a	50	17	26	39	1.6 ± 0.4
Transient^b	8	1	3	5	1.1 ± 0.1
None	47	0	0	0	NA^c

* WEWCA: water-extracted whole cell antigen, as reported²⁰.

^a Among the 50 sero-positive children at 24 months of age, 32 (64%) seroconverted at 12 or 18 months, and 18 (36%) seroconverted at 24 months.

^b Among the 8 transiently-positive children, 4 seroconverted at 12 months and 4 serconverted at 18 months, but subsequent samples were negative.

^c NA, not applicable.

Next, using these assays we determined when H. pylori seroconversion and presumably acquisition occurred (Table 5). Since we were unable to ascertain H. pylori status based on IgG antibodies in the first 6 months of life, we asked whether there was IgA positivity. During the first 6 months of life, we found that only 1 (0.95%) of the children became positive, and we found evidence for continued seropositivity at later times in that child.

Table 5. Calculated seroconversion rates for H. pylori persistence in 105 Bangladeshi birth-cohort children during the first two years of life.

Age of child (months)	Number sero-negative children^*	Number becoming positive	Annualized seroconversion rate (%)^**
0	98	NA^***	NA^***
1-5	98	0	0.0
6-11	98	14	28.6
12-17	81	16	39.6
18-23	67	20	59.8

* Number of children enrolled at the start of the time period.

** Based on the number of IgG seroconversions to either WEWCA or CagA, divided by the total number of sero-negative children evaluable at the start of the particular six-month period; rate is annualized.

*** NA; Not applicable, because of maternal IgG interference with the serological determination.

Between 6–11 months of life, 14.3% of the 98 children analyzed became sero-positive; for an annualized rate of 28.6%. During the next 6 months of life, 19.8% of those who had remained sero-negative showed evidence of positivity (annualized rate 39.6%). For the final 6 months of follow-up, 29.9% of those remaining sero-negative showed evidence of H. pylori acquisition (annualized rate of 59.8%). Thus, despite the high H. pylori prevalence in mothers, we detected most [36/50 (72%)] H. pylori acquisition after the first year of life, which is in line with our prior study in Native American children.²⁰

Identification of H. pylori antigens by immunoblotting

Next, we assessed the type and number of H. pylori antigens that children recognize as result of seroconversion. Using immune human serum or rabbit serum to detect the specific antigens present, we first characterized three antigen preparations, including an H. pylori whole cell (WC) preparation, an H. pylori water-extracted whole cell antigen (WEWCA), and a recombinant CagA protein (Figure 1). When the H. pylori WC preparation was used as antigen, high background signal was obtained even with the serum from an H. pylori-negative human (Figure 1, panel A, lane 3). In particular, two intense bands – at ∼150 kDa and ∼66 kDa – were observed. When strips with the same antigen preparation were incubated with an anti-urease rabbit serum⁴⁶ (not shown), these two bands also were present, suggesting that the cross-reactivity is for the HSP60 chaperones that are part of the urease complex.⁴⁷ The water-extracted preparation (WEWCA) does not contain the CagA antigen since the rabbit α-CagA serum did not detect a band. This observation helps explain why sera from some persons carrying a CagA⁺ strain, may not react with sufficient intensity to be considered as positive (Figure 1, panel B, lanes 1 and 4). However, the CagA antigen was recognized in the WC preparation of H. pylori 26695 (Figure 1, panel A, lanes 1 and 4) and when purified truncated CagA protein (Figure 1, panels C, lanes 1 and 4) was used in immunoblotting (at ∼120 kDa, as well as possible degradation products). The nonspecific and low molecular weight bands in the CagA preparation are most likely due to cross reactivity with Escherichia coli proteins⁴⁷ since most of those bands disappear if serum is absorbed with E. coli whole-cells as described⁴⁸ or represent degradation products of the recombinant CagA.⁴⁹ These results indicate the utility of using the WEWCA H. pylori and CagA antigens for serological testing, but not the whole cell antigen, which lacks specificity, which confirms a wide body of prior studies.^40,50

Figure 1. Western blot of H. pylori antigen preparations with human and rabbit serum samples. Panel A. Whole cell preparation of H. pylori 26695; Panel B. Water extracted whole cell antigen (WEWCA), as described.²⁰ Panel C. Recombinant CagA protein, as described.⁸³ Sera are: Lane 1, H. pylori⁺/CagA⁺ patient (H, human serum), lane 2, H. pylori⁺/CagA⁻ patient. Lane 3, H. pylori⁻/CagA⁻ patient, and Lane 4, serum from rabbit immunized with purified H. pylori CagA antigen (R, rabbit).

Analysis of H. pylori acquisition, using immunoblotting

We next assessed the type and number of antigens that children recognized as a result of seroconversion, using the three antigen preparations mentioned above. We originally included the WC antigen preparation; however, since this showed very high background, and the recognition of CagA was not consistent (data not shown), we did not use it for further studies, focusing on the WEWCA and CagA preparations, which showed only minor background. To understand the range of responses, we examined serum from 10 of the persistently colonized children. For each assay, all serum specimens including cord blood samples were tested. Supplementary Figure 1, panel A shows three positive children with the most commonly observed band pattern at seroconversion. The immunoblot results obtained in all children tested who had persistent seroconversion correlated with their ELISA results, since most of the H. pylori-positive sera showed recognition of a cluster of protein bands between 95 and 50 kDa, and we also confirmed seroconversion to the CagA antigen (panel A).

We then assessed the serum samples of 3 children who showed transient seroconversion (panel B). No particular band was correlated with the transient seroconversions, although there was an increase in non-specific band recognition, particularly around 60 kDa, suggesting that this finding may be an indicator of a falsely positive result. There was no recognition of the band associated with the CagA antigen (panel B). For 6 persistently H. pylori-negative children, only bands around 60 kDa were observed, without reactivity to the CagA antigen (panel C). Performing further immunoblots using a purified VacA antigen, we did not observe recognition in any of the serum samples tested (data not shown). These results differentiate the nature of the antigenic recognition in those subjects with persistent and transient serologic responses.

Physiological changes as a result of H. pylori acquisition

We determined serum PG1 levels as an indication of gastric physiology in relation to H. pylori acquisition (Figure 2, panel A). The baseline levels of the uninfected children (months 6–18; mean ± SEM 9.12 ± 1.20 ng/ml) were significantly lower compared to the levels of children after H. pylori acquisition (months 12–18; mean±SEM 18.24 ± 2.38 ng/ml) (Mann-Whitney test, p = 0.001). Values of these children before seroconversion (6-18 months, mean ± SEM 12.83 ± 1.80) were not significantly different from the values of the non-seroconverters. For children who did not show H. pylori seroconversion, the serum PGI levels progressively increased over the first year of life, but remained relatively stable after 12 months (Figure 2, panel B), similar to a prior report.²⁰ In contrast, in serum samples from children who had persistent seroconversion, there was a transient increase in the PGI level that correlated with the time of seroconversion (Figure 2, panel C)

Figure 2. Serum pepsinogen I (PGI) levels of 18 birth-cohort children. Panel A. PGI levels of samples (n = 20) from children who never seroconverted compared to children with seroconversion (split into samples from children before (n = 21) and after (n = 23) seroconversion). Horizontal bars indicate the mean.**p = 0.001; Panel B. Serum PGI levels over time (mean±SEM ng/ml) in five children who were persistently H. pylori sero-negative and were followed from birth; n = 5 per time point. Panel C. Changes in PGI levels (mean±SEM ng/ml) associated with H. pylori seroconversion in children who became persistently sero-positive. n = 9, 12, 13, 9, 3 for the indicated time points respectively;

.

Discussion

To identify when H. pylori acquisition occurs in human children, the optimal approach is to perform longitudinal birth cohort studies that can document the time point of acquisition. Such studies are not common, and studies of recent years mostly utilize stool antigen detection as the sole diagnostic tool, which lacks accuracy, especially in young populations.⁵¹ Moreover, such studies did not study children starting from birth and mostly do not assess the maternal H. pylori status, as recently reviewed.¹⁶ The current study provides unique insights into when H. pylori is acquired in children from developing countries by following a cohort of children from birth from mothers of known H. pylori status. Moreover, there is controversy about which method to use to confirm H. pylori acquisition in infants and children.^52-55 This question is relevant because there is need for a reliable, reproducible, and technically feasible non-invasive method for detecting H. pylori acquisition.

Serum antibody tests are reliable for adults based on the fact that seroconversion occurs in less than one month,^27,56 but may not be reliable in children because immune responses to H. pylori infection may be delayed.^22,57 Other studies have used as gold standards the urea breath test or stool H. pylori antigen detection, but these too also have been reported to have limited reliability in early life,^58,59 and breath tests are difficult to do with very young children.⁶⁰

However, measurement of IgA and IgG serum antibodies by ELISA is a reliable method with more than 95% sensitivity and specificity both in adults and in children.^44,61,62 We combine these results to minimize the rate of false-negative events. Moreover, we have previously used this method to show that both transient and persistent H. pylori colonization can be determined by serological methods in both human children²⁰ and rhesus monkeys.^63,64

In the present report, in a birth-cohort of 105 infants to 24 months of age, H. pylori prevalence gradually increased with subject age. On the basis of persistently positive H. pylori results, this study confirmed that acquisition of H. pylori can be detected mainly after the first year of life.^20,22,24,25 We also avoided measuring the transplacental transfer of IgG antibodies that others have reported,^65-67 since we used the serum sample from each infant at six-months of age, and not before, as our baseline specimen.

An unresolved issue in the assessment of H. pylori acquisition is the significance of transient responses, and the difficulties determining whether the transiency represents true colonization or false positive results.^12,20,67 In this study, we observed a low incidence (7.6%) of transient responses, but in testing the serum samples by immunoblot to identify which antigens are the main targets in those transient responses, no specific antigens were detected. Although the bands observed may represent heat shock protein (Hsp60) or urease,^46,47 their identity was not confirmed. In total, these transient antibody responses may represent false positive events due to cross-reactivity from other infections. Alternatively, the transience may reflect H. pylori suppression or elimination after antibiotic treatment for other infections, or the acquisition of other competing organisms, before a full serological response is formed.

Intra-familial spread of H. pylori appears important, with mothers playing the pivotal role in H. pylori transmission.^9,17,68 Our finding that maternal H. pylori status did not predict H. pylori status of the infant reflects the low power of this study since nearly all mothers were positive. In another study performed in Bangladesh, isolates from mothers and their youngest children were strongly related, with weaker associations with other siblings,⁶⁹ which could reflect on-going genomic changes in H. pylori during colonization.⁷⁰ Another hypothesis for the lack of association between maternal and infant H. pylori status, is that in the developing world, horizontal rather that familial H. pylori transmission may be favored.⁷¹ In the population we studied, frequent care-giving by grandmothers or aunts may introduce other sources for the organism. Although we assessed acquisition until the age of 2 years, this may not be a sufficient period for the totality of childhood transmission.^17,69 Detectable H. pylori gastric colonization may perturb gastric physiology,^72,73 including effects on PGI levels²⁰ and we confirm the sharp but transient increase in PGI associated with detecting H. pylori acquisition.²⁰ Confirmation of this finding is important and might be of value for detection of H. pylori acquisition, and also might provide an indicator of the future health or disease in colonized children.

Infants may be protected from early H. pylori colonization by breastfeeding,^36,37 and by the presence of specific human milk antibodies to H. pylori.⁶⁵ However, our results showed no protection against H. pylori colonization in extensively (∼1 year) breast-fed infants by mothers with high breast milk antibody levels against H. pylori antigens. A major limitation in this study for assessing potential protective roles of breastfeeding against H. pylori acquisition is that extended breastfeeding (two or more years) is a universal practice in this community. However, that in few of the infants was H. pylori detected earlier than one year after birth suggests that either specific or total IgA antibodies or dietary components in milk could suppress H. pylori colonization^74-78 which reduces the host immunological responses that we use to detect the organism's presence. Another scenario could be that breastfeeding delays the introduction of solid food and weaning, or pre-masticated foods from the mother subsequently reducing frequent horizontal H. pylori exposure.⁷⁹ A third scenario is that breastfeeding selects for microbiota, e.g. lactobacilli^74-77 hat may inhibit H. pylori.

If H. pylori is truly acquired mostly after age 1, is it that the stomach is an inhospitable niche in the first year of life, and that transmission from the mother and other sources is ineffective until the stomach can appropriately mature? An alternative possibility is that H. pylori is transmitted from the mother at birth, but remains in very small populations until the stomach becomes sufficiently hospitable. At that point, the H. pylori population blooms, and then is detected by the host immune response, detection of its urease activity, and other markers of the presence of large numbers of organisms. This study could not distinguish between these possibilities, but both remain feasible.

Conclusions

In conclusion, this study confirms that in developing countries in which H. pylori prevalence is very high, the organisms are not commonly detected in the first year of life, followed by very high seroconversion rates. Detection of transient responses may reflect true transience due to lack of adaptation of the organism or loss due to collateral antibiotic use, which is widespread in this population.⁸⁰ Future studies should focus on when the organisms are actually acquired. Longitudinal studies as well as studies that determine the H. pylori status of the mother are not common.¹⁶ Since this study utilized samples obtained 20 years ago, the results provide an excellent basis for comparisons to present-day cohorts.

Materials and methods

Study group

A prospective birth-cohort study was conducted from October 1993 through September 1994 in Mirzapur Upazila, a rural community located 63 kilometers from Dhaka. Initially, 288 children were enrolled but 36 were excluded because of death or out-migration, and 252 children were enrolled in a study of the cause of diarrhea.⁸¹ Of those children, 105 who completed the follow-up and from whom more than one serum specimen was collected were included in the present study. For 79 (75.2%) of those children, five samples were collected: birth (cord blood), 6, 12, 18, and 24 months of life. We tested a total of 491 serum samples (mean±SD number of samples per child = 4.7±0.7). Written informed consent was obtained from the parents of the children before enrollment. The study was approved by the Research Review Committee (RRC) and the Ethical Review Committee (ERC) of the International Centre for Diarrhea Disease Research (ICDDR, B).^81,82

Serum antibodies to H. pylori antigens

Antigen-specific enzyme-linked immunosorbent assays (ELISAs) were used to determine antibody responses in the study group. Each serum sample was tested for the presence of IgG and IgA antibodies to a water-extracted whole cell antigen (WEWCA) preparation of H. pylori, as reported.²⁰ In brief, ELISA plates were coated with WEWAC, washed and incubated with serum samples diluted 1:200 and 1:800. After removal of the secondary antibody (goat anti-human IgG/IgA alkaline phosphatase conjugated antibodies) and ELISA development, optical density (OD) values were calculated from the mean absorbance of two assays per samples. On each plate, the positive control was used to express the results as OD-ratios (ODRs) for normalization between the plates. Sensitivity and specificity of this ELISA is >95%. The presence of IgG antibodies to a recombinant CagA antigen was determined as previously reported.^5,39,83 In brief, ELISA plates were coated with recombinant CagA, washed, incubated with human serum (1:100 diluted), and developed and analyzed as described above. The ELISA has a sensitivity of 94.4% and a specificity of 92.5%. The cut-off values used to determine a positive or negative serology (for WEWAC and CagA) were based on our previous work with a birth-cohort of Native American children.²⁰ The pre-defined cut-off values were 0.7 for IgG WEWAC, 0.4 for IgA WEWAC and 0.3 for IgG CagA.²⁰

Immunoblot assays

To confirm the H. pylori status determined by ELISA serology in a subset of children, immunoblots were performed using 3 different antigen preparations including: H. pylori strain 26695 whole cell (WC) preparation, the WEWCA, or the recombinant CagA preparation,⁸³ to determine which proteins are recognized by the children's sera (1:100 dilutions). SDS-PAGE gels and nitrocellulose strips were prepared as described.⁸⁴ Using the WEWCA, we examined serum samples of 23 children that included 10 from the birth cohort study who had persistent H. pylori seroconversion, 7 who were transient seroconverters and 6 children who did not seroconvert to H. pylori antigens during the study period. Since multiple sera from each child were available, a total of 104 serum samples were tested. Sera of 3 children from each group also were tested against the CagA and the 26695 WC antigen preparations.

Determination of serum pepsinogen I

We tested PGI levels in 84 serum samples from 18 children. Five children who did not seroconvert to H. pylori were used to determine the baseline levels of PGI and we studied 13 children (mean±SD 9.2±3.9 number of samples per time point) who persistently seroconverted to assess the PGI level as an indicator of perturbation on the gastric mucosa. All determinations were assessed at least in duplicate using the BioHit ELISA kit (BioHit, Helsinki Finland) following the manufacturer's recommendations. One outlier specimen was excluded from the statistical analysis (seroconverter group at 12 months, value: 76.2 ng/ml, p = 0.018, Dixon´s test).

Breast milk ELISA

Breast milk samples were obtained from nursing mothers (n = 104) at 0, 3, 6, and 9 months (Mean = 3.0±0.9 times per mother). We tested for specific IgA antibodies to H. pylori WEWAC and CagA antigen preparations as described above by using 1:100 dilutions of breast milk. The thresholds for positivity were estimated from breast milk samples from 12 mothers that were negative in both H. pylori serological assays (WEWAC and CagA) and whose children were persistently negative for H. pylori. The thresholds were 0.47 for IgA WEWAC and 0.13 for IgA CagA, which represent the mean plus two SD of the values from the negative mothers. Because of natural variation in breast milk concentrations, specific IgA titers were adjusted to total IgA, as reported.⁶⁷ Total secretory IgA was calculated in each sample by ELISA as reported⁸⁵ by using an rabbit polyclonal anti-IgA antibody as coating antibody.

Study definitions

We have previously reported that maternal antibodies to H. pylori that pass via the placenta remain in the child's circulation until 6 months after birth.²⁰ Therefore, the following criteria were used in the characterization of H. pylori status of the subject children: Never positive: all serum samples at 12, 18, and 24 months were negative; Transient positive: any serum sample positive at 12 and/or 18 months of life but negative at 24 months. Persistent-positive was defined as a child with positivity remaining at 24 months of life.

Contributors

GIPP and MJB designed the study. SK, GIPP, AZP, PB, SAS, KZH, and RBS collected samples and performed experiments. SK and GIPP analyzed the data. SK, GIPP, and MJB wrote the paper. All authors have read and were involved in critical revisions of the final paper.

Competing interest

None declared.

Ethics approval.

Each reporting center gave local approval as mentioned in text.

Acknowledgments

We dedicate this paper to the memory of our colleague, R. Bradley Sack, M.D., Ph.D., who sadly did not survive until its publication. We remember Brad as a dedicated scientist and a great friend.

Additional information

Funding

HHS | NIH | National Center for Research rdf:resources (NCRR), P60 MD000538 HHS | NIH | National Center for Research rdf:resources (NCRR), RO1 GM63270 C&D Fund NYU Center for the study of Asian American Health Thrasher Reserach Fund Award, 02816-9.