Non-invasive Genotyping of Helicobacter pylori cagA, vacA, and hopQ from Asymptomatic Children

Abstract

Background

H. pylori infection is usually acquired in childhood, but little is known about its natural history in asymptomatic children, primarily due to the paucity of non-invasive diagnostic methods. H. pylori strains harboring cagA and specific alleles of hopQ and vacA are associated with increased risk for gastric cancer. Many studies of H. pylori virulence markers in children have the bias that symptomatic subjects are selected for endoscopy, and these children may harbor the most virulent strains. Our aim: to genotype cagA, hopQ and vacA alleles in stool DNA samples of healthy Colombian children residing in an area with high incidence of gastric cancer, in order to avoid selection bias resulting from endoscopy.

Methods

H. pylori status of 86 asymptomatic children was assessed by ¹³C-Urea Breath Test (UBT) and PCR. H. pylori 16S rRNA, cagA, hopQ and vacA genes were amplified from stool DNA samples and sequenced.

Results

UBT was positive in 69 (80.2%) of 86 children; in stool DNA analysis, 78.3% were positive by 16S rRNA PCR. cagA, vacA and hopQ were detected in 66.1%, 84.6%, and 72.3% of stool DNA samples from 16S rRNA positive children. Of the children's DNA samples which revealed vacA and hopQ alleles, 91.7% showed vacA s1 and 73.7% showed type I hopQ. Type I hopQ alleles were associated with cagA-positivity and vacA s1 genotypes (P<0.0001).

Conclusions

Using stool DNA samples, virulence markers of H. pylori were successfully genotyped in a high percentage of the asymptomatic infected children, revealing a high prevalence of genotypes associated with virulence. Type I hopQ alleles were associated with the presence of cagA and the vacA s1 genotype.

INTRODUCTION

Helicobacter pylori infection is primarily acquired in childhood and persists as an asymptomatic infection for decades in the majority of colonized individuals. Only a small fraction of infected persons develop clinically significant outcomes, such as peptic ulcer or gastric adenocarcinoma [1]. However, infection early in life has been linked to a greater risk for gastric cancer [2–4]. The epidemiology of infection with H. pylori is characterized by a linear increase in prevalence with age in Western industrial countries and by a steep rise in childhood, followed by a stable high rate in adults in developing countries. This pattern results in a large number of asymptomatic children being infected [5]. Colonization with H. pylori at an early age may be transient, and spontaneous elimination or loss of infection due to use of antibiotics may occur [6–8]. Studies in Peru and Nicaragua suggest that infection status in children within the first five years of life may change frequently with subjects losing and regaining the H. pylori bacterium [9, 10]. Other investigators have questioned the transient nature of infection in children, and raised the issue of reliability of diagnostic methods [11, 12].

The most widely used methods for diagnosis of H. pylori infection, such as culture and histology, are sensitive and highly specific tests, but they require invasive procedures, which are not indicated in asymptomatic children. Among non-invasive methods, serology, in spite of its high specificity, shows low sensitivity in children and may not always indicate current infection with H. pylori. The ¹³C - Urea Breath Test (UBT), considered very sensitive and specific, requires expensive equipment and is not always routinely available in clinical settings. The Stool Antigen Test has been found to be useful in children; however, like the UBT, it indicates only the presence of the bacterium, but does not allow characterization of the H. pylori strains with which individuals are infected. Another option is the string test, which employs a minimally invasive non-endoscopic procedure for harvesting gastric juice samples [13–15]. This procedure requires the swallowing of a capsule attached to a string, which is left in the stomach for an hour; the capsule dissolves, and the string is drawn out and used for culture and/or PCR detection of H. pylori from the stomach. The diagnostic efficiency of the string test when used with PCR may be comparable with ¹³UBT in detecting H. pylori infection, though results may vary depending on the performance and processing technique [16, 17]. The capsule is easily swallowed, but parents may hesitate to subject their children to this test, creating a selection bias. Molecular assays based on detection of H. pylori DNA and of H. pylori virulence markers in stool samples from children may be considered an alternative that avoids selection bias that may result when children must be subjected to endoscopy or other invasive techniques. While studies detecting the presence of H. pylori DNA in stool are common (reviewed in [18]; see also [19–29]), those characterizing virulence determinants, or markers associated with virulence, are fewer [30] and mostly concern cagA [31–34].

Several H. pylori genes, such as cagA, vacA, and hopQ, are associated with specific disease outcomes. The cagA gene and its encoded CagA protein are associated with peptic ulcer disease and with an increased risk of gastric adenocarcinoma [2, 35]. The vacA gene is present in all H. pylori strains, but only vacA s1 type strains secrete an active vacuolating cytotoxin; the s1 allele is associated with a higher risk for peptic ulcer disease and gastric adenocarcinoma [35, 36]. The less well-studied hopQ gene encodes HopQ [37], an outer membrane protein (omp-27) that can modulate the adherence of some H. pylori strains to gastric epithelial cells [38] and thus may play an important role in the initial colonization and long-term persistence of the bacterium in the stomach. The hopQ gene is present in 2 forms: types I and II. hopQ type I alleles were found significantly more commonly in cagA positive vacA s1 strains from patients with peptic ulcer disease than in cagA negative vacA s2 strains from patients without ulcer disease [39–41]. HopQ allele prevalence and relationships with other disease-associated genes have not been studied in H. pylori-infected asymptomatic populations residing in areas of high gastric cancer incidence. The 16S rRNA gene has been used previously for H. pylori identification in stool DNA samples from children and adults [8, 42].

Our previous studies of H. pylori infection in rural Colombian residents have relied on gastric biopsies from symptomatic adult volunteers [43, 44]. H. pylori strains from such biopsies have shown a high proportion of the disease-associated markers cagA and vacA s1 (from 87.0 to 90.4% and from 93.2 to 96.1% respectively). To reduce the possible selection bias in use of symptomatic subjects, in the present study, we aimed to investigate the prevalence of H. pylori infection and of the cagA gene, vacA genotypes, and hopQ genotypes using a non-invasive analysis of stool DNA samples from asymptomatic Colombian children residing in a high risk area for gastric cancer. In addition, we intended to characterize the diversity of hopQ alleles in healthy children and investigate a potential relationship of hopQ to cagA and vacA genotypes.

MATERIALS AND METHODS

Study population

The population consisted of 86 asymptomatic children (41 males; 45 females) ages 4.1–8.7 (mean 6.4±1.3) years, from 79 families. All children were from the rural villages of Nariño or Genoy, in the State (“departamento”) of Nariño, located at a high altitude in the Andes Mountains of southwestern Colombia. A high gastric cancer incidence has been documented in this area [45]. The children were predominantly mestizos of Spanish-Amerindian ancestry from rural agricultural families. Participants were recruited by door-to door outreach, by a social worker who obtained informed consent from the parents. The inclusion criterion for initial recruitment into the study was that the child be from 4 to 8 years old, and no children in this age range were excluded. None of the children had received antibiotics within the six weeks before stool collections. As part of an earlier study, all of the children had received a course of anti - H. pylori therapy either in 2001 or in 2004, at least six months before our first molecular investigation. All parents gave written consent for the study. The study was approved by the Ethics Boards of the Universities of Cali and Nariño, and Vanderbilt University Medical Center.

Materials

Stool samples were collected twice with an interval of six months: from 71 children at the first visit (August 2004) and from 83 children at the second visit (February 2005). Parents were asked to collect stool samples from children at home and bring samples to the hospital laboratory the same morning. Stool specimens were stored at −20°C until they were transported to Emory University in Atlanta, GA, USA, where DNA extraction was performed. Samples were stored on dry ice during transportation from Colombia to the USA.

H. pylori status

The H. pylori status of children was determined by the ¹³C-UBT, which was administered at the same time points as stool samples were collected. All 86 children underwent UBT at the second visit, and 73 of 86 underwent UBT at the first visit. Urea-derived ¹³CO₂ was measured utilizing an automated infrared breath ¹³C analyzer. The measured ¹³CO₂/¹²CO₂ ratio was compared with a known ratio of a standard gas. The ¹³CO₂ values were corrected for the international Pee Dee Belemnite (PDB) standard, then averaged and expressed as relative delta per ml difference to the standard gas (δ(‰)). Δδ(‰) was used to express the difference between δ at time 30 min (t₃₀) and at baseline (t₀). Two nurses who were trained in this procedure collected all breath samples. The cut-off delta value of 5 per mille (δ≥5‰), as proposed by Logan and colleagues in their European standard protocol[46], was applied for all children in this study.

DNA extraction

DNA was extracted from stool specimens using the QIAamp® DNA Stool Mini Kit (Qiagen, Valencia, CA, USA), by the protocol for isolation of DNA from stool for pathogen detection, from 180–220 mg frozen stool. DNA from H. pylori reference strains J99 (ATCC 700824, cagA positive, vacA s1m1), Tx30a (ATCC 51932, cagA negative, vacA s2m2), J178 and J63 (isolated from patients from Nashville, TN [39] were used as controls for the presence of cagA, vacA s1 and s2, hopQ types I and II, respectively. The extraction method was validated previously [47].

PCR amplifications

Separate PCRs were performed for 16S rRNA, type I hopQ, type II hopQ, cagA and vacA s region. Five sets of primers were used in the study (Table 1), all generating amplicons not exceeding 200 bp. For the amplification of the types I and II hopQ alleles, two sets of primers were designed to amplify segments of 187 and 160 bp, respectively, from the relatively conserved 3' end of the hopQ gene. Validation of our hopQ primers was performed by amplification of 13 H. pylori DNA samples from Colombian clinical strains using earlier published hopQ type I, region B (OP4070F, BA8705R) and hopQ type II (BA8363F, BA8364R) primers and by amplification of reference strains J178 (for type I hopQ) and J63 (for type II hopQ) [48]. Sequences obtained from amplified clinical and reference genotypes were determined (NCBI, BLAST) as belonging to type I hopQ or type II hopQ alleles. cagA fragments were amplified from the conserved region at the 5' end of the cagA gene using published primers [49]. To amplify vacA s regions, primers VA1F and V136R (the latter designed in this study) were selected, resulting in generation of fragments of 136 bp for the type s1 variants or fragments of 163 bp for type s2 variants [50]. Validation of V136R was performed similarly to that for hopQ primers. Fragments of 109 bp were amplified from the 16S rRNA gene, using primers designed by Ho et al. [51].

Table 1.

PCR primers Used in This Study

Gene or Locus	Primer designation	Primers 5'→ 3'	Amplimer size (bp)	Reference
16S rRNA	Hpl F Hp2 R	CTGGAGAGACTAAGCCCTCC ATTACTGACGCTGATTGTGC	109	Ho et al., 1991
HP1177	hopQ I F R	ACGAACGCGCAAAAACTTTA TTGCCATTCTCATCGGTGTA	187	This study
HP1177	hopQII F R	ACAGCCACTCCAATCCAGAA AACCCCACCGTGGATTTTAG	160	This study
cagA	cagA F R	TTGACCAACAACCACAAACCGAAG CTTCCCTTAATTGCGAGATTCC	183	van Doorn et al 1998
vacA	VA1 F V136 R	ATGGAAATACAACAAACACAC CAACAATGGCTGGAATGAT	136/163	Atherton et al 1995 This study

PCR mixtures for 16S rRNA, cagA, hopQ I and hopQ II consisted of 10 mmol/L Tris-HCl (pH 8.3), 50 mmol/L KCl, 1.5 mol/L MgCl₂, 400 μmol/L of each dNTP, 0.4 μmol of each forward and reverse primer, 0.12 μg/μl of bovine serum albumin, and 0.5 to 2.5 U of HotStar Taq DNA Polymerase, or ABI Hot Start DNA polymerase for the vacA assays, in a final volume of 25 μL. Higher concentrations of dNTP and primers were dictated by use of higher numbers of cycles for stool DNA amplifications. PCR mixtures and handling of samples was performed using protocols designed to reduce the risk of cross-contamination, including use of dedicated areas for pre- and post-PCR, use of plugged pipette tips, and UV irradiation of pipettors and other materials.

The PCR program for 16S rRNA, hopQ, cagA and vacA amplifications was as follows: 15 min at 95°C, and 60 cycles consisting of 1 min at 94°C, 1 min at 55°C (or 52°C for vacA s), and 1 min at 72°C, plus a final 3 min extension step at 72°C. Preliminary experiments using lower cycle numbers produced frequent failures of amplification, presumably due to the small amount of H. pylori DNA present. PCR products (10 μl of each sample) were electrophoresed in 2% agarose gels for 2 hours at 60 volts to confirm the presence of a single band of the expected size. Only samples producing such a band were submitted for sequencing. For quality control of PCR, positive and negative reagent control reactions, including a no-template control, were included in each experiment.

DNA Sequence analysis

Prior to sequencing, PCR products were purified using ExoSAP-it enzyme (Affymetrix, Cleveland, OH, USA). All sequencing for cagA, vacA and hopQ was performed on both strands. The 16S rRNA PCR products were not sequenced. Sequencing was performed with an ABI 3130xl Genetic Analyzer, using BigDye 3.1 terminator chemistry. Sequencing primers were the same as those used for PCR amplification. Sequence construction and alignment were performed using Geneious Pro software, version 4.6.4. (Biomatters, Auckland, New Zealand).

All sequences were compared with reference sequences of H. pylori NCTC 11637 (GenBank AF202973), H. pylori 26695 (GenBank AE000511), H. pylori J99 (GenBank AE001439), H. pylori J178 (prototype of type I hopQ alleles [39], H. pylori J63 (prototype of type II hopQ alleles [39]. Each sequence was also compared with sequences from the same child recovered at a different time point (where DNA was available) and with sequences from other children. Discrimination of s1a, s1b, and s1c vacA genotypes was performed by comparing their sequences with previously published peptide sequences of the vacA s region in reference strains [49]. For purposes of sequence comparison, primer binding sites were omitted. For quality control, 18 randomly selected DNA samples were re-amplified and re-sequenced.

Statistical analysis

The two-sided P value for specificity and sensitivity of 16S rRNA assays versus UBT, and of PCR assay for vacA s or hopQ results versus 16S rRNA PCR results, as well as a P value for the relationship of types I and II hopQ alleles with cagA and vacA genotypes were calculated using Fisher's exact test. P<0.05 was accepted as statistically significant. Calculations were carried out using Interactive InStat software (JavaStat and GraphPad Software Inc, San Diego, CA).

RESULTS

H. pylori status

Fifty-three of 73 children were UBT positive at the first visit (72.6%) and 66 of 86 were UBT positive at the second visit (76.7%; Table 2). Five of 20 children who were UBT negative at the first visit converted to UBT positivity at the second visit. Additionally, of 13 children for whom UBT was unavailable at the first visit, 11 were UBT positive at the second visit. At either visit, sixty-nine (80.2%) of 86 children were classified as H. pylori positive by ¹³C-UBT (Table 2).

Table 2.

Detection of H. pylori Infection by UBT and by PCR for 16S rRNA and for Markers of Virulence

	Positive results by
Test	UBTa		PCR
			16SrRNA		cagA		vacA		hopQ
Visit	1	2	1	2	1	2	1	2	1	2
Positives/Tested (%)	53/73b (72.6)	66/86 (76.7)	45/53c (84.9)	55/66c (83.3)	23/46d (50.0)	33/58d (56.8)	30/46d (65.2)	36/58d (62.1)	27/46d (58.7)	41/58d (70.7)
Positives at either time point (%)	69/86 (80.2)		65/83 (78.3)		43/65 (66.1)		55/65 (84.6)		47/65 (72.3)

^aUBT considered positive when δ≥5.0%

^b13 children were not available for UBT at the 1^st visit

^c16S rRNA positive results of UBT positives

^dmarker gene assay positive results of 16S rRNA positives.

PCR results

Our PCR assays with primers for 16S rRNA, cagA, vacA and hopQ genes amplified the expected fragments of 109 bp for 16S rRNA, 183 bp for cagA, 136 bp for vacA s1/163 bp for vacA s2, and 187 bp for type I hopQ /160 bp for type II hopQ from DNA preparations from the H. pylori J99, Tx30a, J178, and J63 reference strains. No amplification was observed for the no-template controls. Of 53 children who were UBT positive at the first visit, stool DNA samples from 45 (84.9%) were positive for 16S rRNA, and of 66 children who were UBT positive at the second visit, samples from 55 (83.3%) were positive for 16S rRNA (Table 2). At either visit, 65 of 83 UBT positive (78.3%) children were classified as H. pylori positive by the 16S rRNA PCR.

We compared the performance of PCR assays for cagA, vacA and hopQ versus 16S rRNA PCR (Table 2). The cagA assay was positive in samples from 23 of 46 children with positive signals for 16S rRNA at the first visit (50.0%), and in samples from 33 of 58 children with positive signals for 16S rRNA at the second visit (56.8%, Table 2). At either visit, 43 of 65 (66.1%) children had cagA positive stool DNA samples. vacA s1 or s2 was found in 55 of 65 (84.6%); types I or II hopQ alleles were detected in samples from 47 of 65 (72.3%) children with positive signals for the 16S rRNA assay.

Comparing overall the performance of 16S rRNA assays to UBT (See Table 3), the sensitivity and specificity of the 16S rRNA assay varied from 87.3% to 88.2% and from 85.0% to 95.0%, respectively. In samples from four children (one at the first visit and three at the second visit), the 16S rRNA assay was positive, although the UBT was negative.

Table 3.

Comparative Performance of UBT and 16S rRNA PCR

n(%)
Total tested by UBT and 16S rRNA: →	Visit 1	Visit 2
	71a	83b
UBT+, 16S rRNA+	45 (63)	55 (66)
UBT−, 16S rRNA−	19 (27)	17 (21)
UBT+, 16S rRNA−	6 (8.5)	8 (9.5)
UBT−, 16S rRNA+	1 (1.5)	3 (3.5)
p** value	0.000	0.000
PCR Sensitivity (%)	88.2	87.3
PCR Specificity (%)	95.0	85.0

^achildren from whom both stool DNA samples and UBT were available, or stool DNA samples for which both 16S rRNA and virulence marker assay results were available

^bfrom three children, stool DNA samples were unavailable.

The sensitivity and specificity of vacA and hopQ marker gene assays compared to 16S rRNA are shown in Table 4. The sensitivity and specificity of the cagA assay could not be assessed due to the fact that the cagA gene is not present in all H. pylori strains. In samples from 7 children, the vacA assay was positive at either time point in the presence of negative 16S rRNA assays. Of those, 5 samples produced both gel signals and sequences for vacA s1, and 2 produced signals and sequences for vacA s2. For samples from 36 children with results for both cagA and vacA at either time point, all cagA positive signals were accompanied by vacA s1 signals. No sample with a vacA s2 signal was accompanied by a cagA positive signal. In samples from 9 children, the hopQ assay was positive at either time point in the presence of negative 16S rRNA assays. Of those 9, 7 samples produced both gel signals and sequences for type I hopQ, and 2 produced signals and sequences for type II hopQ.

Table 4.

Relationship of 16S rRNA PCR and Virulence Marker Gene PCR Results

n (%)
Samples tested by 16S and Virulence Marker Gene PCRa→	cagA		vacA s		hopQ
	Visit 1	Visit 2	Visit 1	Visit 2	Visit 1	Visit 2
	71b	83c	71b	82d	71b	83b
16S rRNA+, Marker+	26 (36)	33 (40)	36 (51)	36 (44)	32 (45)	41 (49)
16S rRNA−, Marker −	24 (34)	23 (28)	16 (23)	22 (27)	23 (33)	18 (22)
16S rRNA+, Marker −	19 (27)	25 (30)	15 (21)	21 (25)	13 (18)	17 (20.5)
16S rRNA−, Marker +	2 (3)	2 (2)	4 (5)	3 (4)	3 (4)	7 (8.5)
p valuee	0.000	0.000	0.000	0.000	0.000	0.001
Marker PCR Sensitivity (%)	-f		70.6	63.2	71.1	82.0
Marker PCR Specificity (%)	-f		80.0	88.0	88.5	85.7

^aVirulence Marker gene PCR assay=PCR for cagA, or vacA or hopQ.

^bchildren from whom both stool DNA samples and UBT were available, or stool DNA samples for both 16S rRNA and virulence marker assays were available;

^cfrom three children, stool DNA samples were unavailable;

^dstool DNA sample from four 16S rRNA positive children were unavailable;

^ep value from Fisher exact test;

^fsensitivity and specificity of the cagA PCR assay cannot be determined, because not all H. pylori strains carry the cagA gene.

We examined the effect of age on associations between UBT and vacA or hopQ PCR assays by dividing the children into 2 groups: those younger than 6.0 years, and those 6.1 years or older. We found no significant difference between the two age groups.

We evaluated the prevalence of H. pylori infection and of the cagA gene and vacA s1 and type I hopQ alleles in children of three different age groups: 4.0 to 4.9 years, 5.0 to 6.9 years and 7.0 to 9.0 years. The 16S rRNA gene was found 1.77 and 1.87 times more frequently in the second and third group, respectively, compared to the first group.

Diversity of hopQ and vacA s alleles in Colombian children

Type I hopQ alleles were found in DNA samples from 38 of 57 (66.6%) children with positive signals for hopQ PCR, and type II hopQ alleles were detected in samples from 15 (26.3%) children (Table 5). For the remaining 4 children (7.1%), signals for both types I and II alleles of hopQ were found. vacA s genotypes were detected from the stool DNA in 60 of the 83 tested children; the predominant vacA genotype was s1b, which was found in samples from 49 children (81.7%). The other genotypes detected were as follows: s1a for 3 children (5.0%), s2 for 5 children (8.3%) and s1 and s2 at different time points in samples from 3 children (5.0%).

Table 5.

Diversity of hopQ and vacA Genotypes at Two Time Points.

	Genotypes
	hopQ (in 57 of 83 tested children)			vacA s (in 60a of 83 tested children)
Allele	type I	type II	type I+II	s1a	s1b	s2	s1+s2
Visit 1	27	6	2	1	32	3	1
Visit 2	32	12	4	2	34	2	2
Children with same genotype at 2 time points	21	3	2	0	17	0	0
Total children with genotype:	38	15	4	3	49	5	3b
%→	66.6	26.3	7.1	5.0	81.7	8.3	5.0

^aincluding cases negative for 16S rRNA gene;

^bin three children, the s1 and s2 genotypes were found at different time points.

Relationship between hopQ genotypes and cagA or vacA s alleles

The type I hopQ genotype was significantly associated with cagA positivity and vacA s1 genotypes at both visits, whereas the type II hopQ genotype was significantly associated with cagA negative and vacA s2 genotypes (P=0.005 and P<0.0001 for cagA, and P<0.0001 for vacA, Table 6).

Table 6.

Relationship between hopQ Alleles and cagA and vacA.

Gene or allele		hopQ		Two-sided p value *
		type I	type II
cagA	+	46 (60%)	2 (2.5%)	<0.0001
	−	13 (17%)	16 (20.5%)
vacAb	s1	44 (86%)	2 (4%)	<0.0001
	s2	0 (0%)	5 (10%)

^*p value for difference using Fisher's exact test.

Sequencing results for infecting H. pylori strains

To learn to what extent sequencing revealed diversity in infecting strains, we examined all cagA, vacA, and hopQ sequences (from 42, 44, 36 and 16 children, respectively). From common polymorphisms in the portions of the genes sequenced, groups of sequences from the children's strains could be discriminated. Sequence identity for each gene in different children was consistent with harboring a common strain. The detection of sequence polymorphisms in each gene allowed us to discriminate different strains. The highest sequence diversity was observed for the vacA s2 fragments showing 3 different patterns in 7 children (43%), followed by type II hopQ fragments with 6 different patterns in 16 sequences (37%). No H. pylori sequences obtained from the children in the study matched those of the H. pylori strains 26695, 60190 or reference strain J99, Tx30A, J178, J63 sequences, used as PCR controls in this study. The 18 randomly selected DNA samples that were re-amplified and re-sequenced for the purpose of quality control of our assays produced identical sequences in all cases.

H. pylori infection at two time points

The presence of H. pylori infection was determined by detection of a positive result in at least one genotyping assay confirmed by sequencing. Analysis of results obtained from samples collected at two consecutive visits (August, 2004 and February, 2005) allowed us to note differences consistent with status changes between the two time points. Overall in samples from 29 children, sequences for one, two or three genes could be obtained at both time points. In samples from nine of those children (31%), sequences were different at the two time points; in samples from the remaining children (69%), the sequences of one, two or three genes at the two time points were identical. These findings are consistent with the idea that the majority of children (69%) maintained the same strain over a six month period. Considering the presence at least of one positive genotyping result for any of the four genes, eight children of 83 (9.6%) had results consistent with loss of strains, and 9 children (10.8%) showed results consistent with gain of H. pylori infection during the 6 month period. Mixed infection was found in 12 of 70 children (17.1%) as evidenced by PCR and sequencing.

DISCUSSION

Our study is the most extensive evaluation of markers of virulence in H. pylori infections in asymptomatic children from a high-risk population, which we have seen to date. First, the study revealed the high prevalence of H. pylori infection (80.2% by UBT and 78.3% by 16S rRNA PCR) in this population. The high prevalence detected by UBT agrees with our results obtained previously by UBT in non-overlapping study populations of children from this same area of Colombia, where gastric cancer is common [52, 53]. Although published studies on H. pylori infection in children may differ in methodology, it is interesting to note that the prevalence we measured by both PCR and UBT exceeded those reported even for symptomatic children in some studies. Prevalence of H. pylori infection in Spanish children was only 53% as found by Real-Time PCR using gastric biopsies and PCR for the 23S RNA gene [54].

Secondly, our study shows that even when the accrual of participants was not subjected to the potential bias of selection for clinically indicated endoscopy, the prevalence of markers of virulence in this population was high. The prevalence of the cagA gene and vacA s1 alleles associated with increased risk for gastric cancer were found in stool DNA samples from 66.1% and 91.7%, respectively, of H. pylori positive children investigated in our study, percentages surprisingly close to those found in infections in symptomatic Colombian adult volunteers (for cagA, 87%, and for vacA s1 alleles, 93.2%), who may have volunteered due to concern about gastric complaints [43, 44]. These percentages are higher than those reported from children in other countries, even from studies of symptomatic children. In gastric biopsies from symptomatic Portuguese children from a population at high risk for gastric cancer, cagA was detected in 36.1%, and vacA s1 was detected in 32.7% of children [55]. Lower frequencies of both cagA (47%) and vacA s1 (58%) have been reported in gastric biopsy isolates from Mexican children with recurrent abdominal pain, but no peptic ulceration [56]. The high prevalence of virulent H. pylori in young children in the Colombian population we studied is likely to contribute to the high incidence of gastric cancer in adults in the area. The determination that infection by H. pylori is a risk factor for gastric cancer [57] highlights the importance of early detection of H. pylori virulence factors in children, especially in those residing in areas with a high risk for gastric cancer development.

We also examined the presence of hopQ gene alleles in infections in our population. Our findings revealed that the predominant allele of the hopQ gene was type I, found in samples from the majority (72.4%) of children compared to only 42.4% found earlier in symptomatic adults from another area of Colombia [58]. Our results also showed that type I hopQ genotypes were found in all cases with the vacA s1 genotype. Type I hopQ alleles were also significantly associated with cagA-positive status: 81% and 75% of children with type I hopQ positive strains also had cagA positive strains, at the two time points. These associations are in agreement with those found in studies by Cao et al. and Ohno et al. for H. pylori strains isolated from adult patients from Western countries [39, 58]. Predominance of type I hopQ alleles in children from the high risk area for gastric cancer also may be a factor contributing to geographic variation in gastric cancer in Colombia.

Our study demonstrates the feasibility of stool DNA amplification and genotyping of 16S rRNA and hopQ, cagA and vacA markers in a single step PCR assay using a high number of amplification cycles. High cycle numbers in the PCR are generally avoided due to concern about artifacts; however, we have no indication that these occurred with the PCR conditions that we used. Complete reproducibility of results in independent PCRs and a very low percentage of 16S rRNA PCR signals in the presence of negative UBTs (less than 3.6% “false positives”, if UBT is taken to be the standard for the 16S rRNA assay) support the idea that we detected signals accurately. The low percentage of cagA, vacA, and hopQ PCR signals in the presence of negative 16S rRNA PCR (from 2.4% in the cagA assay to 8.4% “false positives” in the hopQ assay, if 16S rRNA PCR is taken to be the standard) also support the accuracy of our assays. Additionally, sequences of multiple positive cagA, vacA and hopQ assays in the absence of 16S rRNA signals suggest that in some instances, it was the 16S rRNA assay that was the false negative. In a previous study examining the presence of cagA and vacA in the stool DNA of other Colombian children, using PCR with the same primers, but without sequencing, we detected no false positive results when we tested DNA samples from Escherichia coli, Enterobacter cloacae, Citrobacter koserii, Enterococcus fecalis, Klebsiella pneumoniae and Campylobacter jejuni subsp. jejuni cultures added to the stool DNA [47]. The present study, in which the identity of the amplified fragments was confirmed by sequencing, also shows that only specific H. pylori DNA was amplified. Additionally, the inclusion of sequencing into the analysis brings the potential of tracking of strains, and confirms exclusion of contamination by reference strains.

We found considerable diversity of sequence in the genes we analyzed, but because other unexamined portions of the genes may contain additional polymorphisms, the variants we detected represent a minimum amount of diversity, and not a complete measure. Identical short sequences between strains are consistent with, but do not establish identity of strains. Despite this limitation, the analyses of sequence diversity of sequences obtained from stool DNA samples allowed us to compare them at different time points, to identify children with identical gene fragments and possibly carrying the same strains, and to exclude laboratory DNA contamination with DNA from reference H. pylori strains. Addition of more and possibly longer amplicons to the analysis may increase the confidence with which particular H. pylori strains may be followed in asymptomatic children.

Because the sequences obtained allowed discrimination of H. pylori strains, our study provides some information that bears upon the question of dynamics of the infection in children. Sequence analysis of polymorphisms within the same segment length allowed us to determine that strains were different. Our findings are consistent with studies showing frequent gain and loss of H. pylori strains in children. Taneike et al. reported intrafamilial changes in infection in two 4 and 9 year old children with duodenal ulcer, as demonstrated by rRNA gene restriction pattern analysis and electron microscopy [59]. Other studies examining infection in Peruvian, Nicaraguan, German, Turkish and Japanese children [6, 9, 10, 60, 61] also found variation, using UBT, antigen-based stool assays or by the presence of IgG antibodies to H. pylori. Higher rates of acquisition and loss of infection reported by these authors (14% acquisition in Turkish children; 7% of acquisition and 35% loss of infection in German children) may be due to longer follow-up periods (6 years and 2 years respectively). The transient infection phenomenon may be an artifact of false positive tests, especially in cases of low prevalence of infection [8, 11, 62], or when using the ELISA test with stool DNA samples without molecular confirmation of positivity. In our study, UBT and our PCR assays for multiple genes all support the high prevalence of infection in the population of children we examined. Our control experiments, including comparison of sequences with our control strains, and replication, as discussed above, support that idea that the risk for false positive results in our study was low.

Our study has some limitations. Although it would have been interesting to compare the sensitivity of our analysis of stool DNA with results from analyses of H. pylori obtained from cultures of gastric biopsies from the same individuals, we were not able to obtain such materials from our set of children, for ethical reasons. Gastric juices from string tests from the same children were not available for comparison of sensitivity of the two methods. Our vacA s region assay was competitive, which means that in the presence of mixed s1/s2 infections, the larger s2 amplicon may be underrepresented. Due to the possibility of mixed infections, we cannot conclude that amplicons corresponding, for example, to vacA s1 and type I hopQ alleles in the same patient both originated from the same strain. For all amplifications, false negatives may occur, which is not surprising, considering the very small proportion of H. pylori DNA mixed with total DNA derived from normal enteric flora and epithelial cells of the gut. Because of the fragmentation inherent to stool DNA, we designed our primers to amplify short sequences. Both hopQ and vacA s1 primers were designed from relatively conserved regions of the respective H. pylori genes. However, in clinical H. pylori strains, variations in those regions are possible, and some negative results may occur due to mismatch of primer binding sites.

In conclusion, the present study based on stool DNA amplification and genotyping of 16S rRNA and of cagA, vacA and hopQ genes reveals the high prevalence of disease-associated H. pylori infection in a group of healthy Colombian children living in a high risk area for gastric cancer. We found type I hopQ alleles in samples from the majority of children; this allelic type was associated with the presence of cagA and vacA s1. High percentages of disease-associated genes in H. pylori infected children may contribute further to the high incidence of gastric cancer in adults in the area. These noninvasive assays may prove to be useful for screening of asymptomatic and symptomatic individuals for early detection of H. pylori disease-associated genotypes in stool samples.