Helicobacter pylori Infection in Colombia: Phylogeny, Resistome, and Virulome

ABSTRACT

Helicobacter pylori is a successful etiologic gastric agent that reaches a prevalence around 80% in Colombia. This bacterium is extremely diverse and has shown a phylogeographic pattern. The objective of this study was to perform an analysis of genomic epidemiology of H. pylori in Colombia. We enriched our set of 29 newly sequenced Colombian H. pylori genomes with additional data from public databases, reaching a total of 221 genomes in our dataset. Phylogenetic characterization was carried out using MLST and whole genome SNP analysis. We also performed a characterized the diversity of virulence factors and mutations associated with antimicrobial resistance. Phylogenetic analyzes showed two new Colombian H. pylori clades. Furthermore, many virulence genotype combinations were found, mutations associated with resistance were found for all the studied antibiotics, highlighting 14.4% of the genomes presented profiles associated with resistance to more than one family of antibiotics. Our analyzes described the genomics of Colombian H. pylori and verify the presence of a population group formed exclusively by Colombian isolates. We demonstrated the great diversity among the isolates and that the analysis by comparative genomics of H. pylori are valuable tools to assess the diversity, virulence, and resistance of H. pylori .

1. Introduction

Helicobacter pylori ( H. pylori ) is a Gram‐negative bacterium that colonizes the stomachs of approximately 4.4 billion people globally, making it one of the most successful human pathogens [1, 2, 3]. Prevalence of H. pylori infection is higher in developing countries, reaching rates around 80% in Colombia [4, 5].

H. pylori is a genetically diverse bacterium, which has co‐evolved with humans for more than 60,000 years [6], this genetic plasticity and the long evolution causes a chronic infection that has a causative association with chronic gastritis, peptic ulcers, MALT lymphoma, or gastric adenocarcinoma [7]. This long‐standing relationship has resulted in the emergence of different genotypes due to the accumulation of host‐specific adaptive changes [8]. The presence of different genotypes that predominate in different geographical regions has been demonstrated [9]. The severity of gastric diseases associated with H. pylori has been associated with the predominant genotype in each region [10].

Several aspects have been described as responsible for the progression of H. pylori infection to severe diseases [11, 12]. Among these, the most studied are a great variety of adhesins where BabA stands out, responsible for the adhesion and colonization of H. pylori in the gastric mucosa [13]; the vacuolizing cytotoxin VacA, present in all strains but with the presence of significant polymorphisms that affect their cytotoxic activity [14]; the CagA oncoprotein, present in some strains and located on the Cag pathogenicity island (cagPAI), which encodes a type IV secretion system that allows the translocation of CagA and other bacterial effector molecules that act directly on gastric cells, resulting in the induction of intracellular signaling pathways that cause aberrant cell motility and proliferation [15, 16]; and the induced by contact with epithelium gene (iceA), which encodes a protein that acts as an endonuclease and is activated when it comes into contact with epithelial cells, which allows the bacterium to protect its genome from different agents [17, 18].

Besides the virulence, another factor that implies a significant challenge in managing H. pylori infection is the emergence of antimicrobial resistance (AMR), which increasingly results in the presentation of treatment failures [5, 19]. The primary resistance to antibiotics like Clarithromicyn (CLA), Levofloxacin (LEV), and Metronidazole (MTZ) has exceeded the acceptable limits in Colombia, presenting resistance rates of up to 17.7%, 27.3%, and 83%, respectively [19, 20, 21]. H. pylori develops AMR mainly through point mutations in the genes that code for the antibiotic site. Therefore, analyzing these genes in the complete genomes of H. pylori reported in Colombia will provide knowledge about the genotype of resistance circulating in the country over time and provide information for the surveillance of the AMR phenomenon in Colombia.

The phylogenetic diversity of H. pylori among New World isolates unveils a tapestry of evolutionary intricacies, showing a dynamic interplay of genetic variation within this microbial environment, is not fully understood. H. pylori can lead to high‐speed local adaptive processes through mutation and homologous recombination with different strains [22]. Bacterial plasticity has been especially evident in Colombia [23, 24, 25]. It has been suggested that in Colombia, bacterial populations evolve rapidly and have formed new subpopulations of European origin [26]. These findings highlight the need to include more Colombian genomes in phylogenetic studies.

Multiple and diverse studies carried out with Colombian H. pylori strains have provided 192 Colombian H. pylori genomes available in public databases. For this study, we used 221 genomes, including 192 previously available and 29 newly sequenced genomes [27]. The objective of this study was to perform genomic epidemiology analysis to expand the knowledge of possible biomarkers of H. pylori infection in Colombia and, on the other hand, the study aims to demonstrate that the WGS (whole genome shotgun) data allows a rapid characterization of several factors that may be involved in the pathogenicity of H. pylori .

2. Materials and Methods

2.1. H. pylori Genomes

A set of 221 Colombian genomes were analyzed, including 29 new genomes sequenced and 192 Colombian genomes obtained from public databases (collected in July 2020). The 192 Colombian H. pylori genomes from public databases were identified through the BV‐BRC: Bacterial and Viral Bioinformatics Resource Center (PATRIC ‐Pathosystems Resource Integration Center 3.6.5) in July 2020. These genomes, reported between 2000 and 2020, can be reviewed in detail at https://zenodo.org/records/4898561#.YMtoPWgzbIV. The new genomes were obtained from strains isolated between 2009 and 2010 from patients residing in Bogotá, Colombia, by Infectious Diseases Group, Sciences Faculty, Pontificia Universidad Javeriana. Bacterial isolates were recovered by culture following the protocols previously standardized by the research group [27]. Total DNA was extracted using a DNeasy Blood & Tissue kit (QIAGEN, Hilden, Germany) and sequenced using the Illumina MiSeq platform (Illumina, San Diego, CA) DNA libraries were prepared using a Nextera XT DNA library preparation kit (Illumina), followed by 2 × 300‐bp paired‐end sequencing resulting in 80× coverage. The low‐quality sequences were removed with the software package Trimmomatic v0.39 [28]. The reads were used for de novo genome assembly with SPAdes v13.3 [29]. The genomes obtained were deposited in the GenBank database of the National Center for Biotechnology Information (NCBI) [27].

The search for the Colombian genomes was carried out on the Bacterial and Viral Bioinformatics Resource Center (BV‐BRC) 3.6.5 [30]. At the search (July 2020), 192 isolates were found. The genomes were downloaded from the NCBI WGS database (Table S1).

2.2. Phylogenetic Analysis by MLST

The MLST analysis facilitates the comparison of results with previous studies. For that the sequences of the seven H. pylori housekeeping (atpA, efp, trpC, ppa, mutY, yphC, and ureI) from the 221 Colombian genomes using BLASTn [31]. For comparison, we obtained the sequences of these genes of 741 H. pylori genomes previously described [32, 33] from the PubMLST database (https://pubmlst.org/organisms/helicobacter‐pylori/) [34].

A multiple alignment was performed using MAFFT v7 [35] using the multi‐FASTA file containing the concatenated seven house‐keeping genes of each strain. Subsequently, a phylogenetic tree was built in the MEGA 7.0 (Molecular Evolutionary Genetics Analysis) software [36], using the Neighbor‐joining method and the Kimura two‐parameter substitution model [37]. The number of bootstrap replications was set at 1000.

In addition, the seven housekeeping genes were used for population structure determination using STRUCTURE 2.3.4 software [38, 39, 40] with the “Admixture” model and performing duplicate runs. In each run, the MCMC (Markov Chain Monte Carlo) method with 10,000 iterations was selected. The highest mean probability value was compared for multiple runs of 5 ≤ K ≤ 7.

2.3. Phylogenetic Analysis by SNP (Single Nucleotide Polymorphisms)

We also inferred phylogeny by SNPs over the complete genome in the online server Genomic Epidemiology CSI Phylogeny [41], with the strain 26,695 (GenBank: NC_000915.1) serving as a reference. In addition to the 221 Colombian genomes, we included another 101 genomes from strains isolated from Nicaragua (45 genomes), Peru (32 genomes), and 45 genomes isolated from worldwide for a total of 332 genomes (Table S2). These genomes were included in our analysis for forming exclusive population groups in these countries that have derived from the European population, as previously reported [9, 42]. We have used a minimum depth of 10×, a minimum relative depth at SNP position of 10%, a minimum distance between SNPs of 10pb, and a Z‐score of 1.96. Phylogenetic trees were visualized in iTol V4 [43].

2.4. Virulence Genotyping

To verify the presence of the virulence factors in Colombian H. pylori genomes, we evaluated the presence of three genes individually (cagA, vacA, and iceA) using BLASTn with threshold a cutoff of 70% and a coverage of 80%. The same tool was used for the confirmation of allelic variants of the vacA gene (s, m, i regions) performing a in silico PCR using previously described primers [44, 45, 46, 47] (Table S3) [31]. For those strains positive for the cagA gene, the EPIYA motif [31] was confirmed by BLASTp analysis [31] using as a query the amino acid sequence presented in Table S4.

Additionally, virulence genes were analyzed using the pathogenic bacteria virulence factors database (VFDB) analysis tool (http://www.mgc.ac.cn/VFs/main.htm).

2.5. Detection of Antibiotic Resistance Mutations

The mutations associated with resistance to CLA were searched on the 23S rRNA gene, requiring at least one allele resistant in both gene copies. For tetracycline (TET) resistance, mutations on16s rRNA gene were examined, and for metronidazole (MTZ) resistance, mutations in the rdxA and frxA genes were analyzed, as these genes have the strongest evidence for conferring resistance. Detection of mutations was performed using the MUMmer tool 48 with the algorithm NUCmer, aligning sequences against the reference genome (26,695, GenBank: NC_000915.1). To identify mutations related to levofloxacin (LEV) and amoxicillin (AMX) resistance, the translated sequences of the genes gyrA and gyrB (for LEV) and pbp1, pbp2 and pbp3 (for AMX), were analyzed with the algorithm PROmer in the software MUMmer [48], as shown in Table S5. Subsequently, the previously reported mutations [20, 49, 50] were filtered and manually analyzed.

2.6. Statistical Analysis

Data were entered and analyzed in the IBM Statistical Package for the Social Sciences (SPSS) v24.0 software. The frequencies of virulence genotypes and AMR‐linked mutations present in the analyzed genomes were analyzed and determined.

3. Results

3.1. H. pylori Colombian Genomes

The 29 new genomes were sequenced from strains of patients aged 18–79 (~48). The majority were women (20/29), and all were diagnosed with chronic gastritis; 65.5% were non‐atrophic gastritis and 34.5% were chronic atrophic gastritis.

The 192 genomes available in the NCBI were obtained from isolates collected between 2000 and 2018; most of them (69.2%) were recovered from patients residing central area of the country, specifically in Bogotá and Cundinamarca; the others from regions such as Cartagena (2), Risaralda (1), Tolima (3), Tunja (10), Valle (1), and Nariño (29). No information related to the geographical area was found in 13 genomes.

70/192 genomes had information on the associated pathology. Of these, 34 belonged from patients with chronic non‐atrophic gastritis, four from chronic atrophic gastritis, and 27 from premalignant lesions, such as intestinal dysplasia or metaplasia. Five genomes were of patients with gastric cancer. Furthermore, regarding demographic data such as age or sex, most of the genomes registered in databases (176/192) did not have related data.

Regarding the genome assembly statistics, for the 29 genomes obtained in this study, the average size of these genomes was 1,641,708 bp. The average GC% was 39.0%, and the average number of coding sequences (CDS) was 1664. Complete assembly statistics for these genomes were previously reported [27]. For the 192 genomes obtained from databases, 1,662,383 bp was the average genome size, with an average GC% of 38.9% and an average CDS of 1624. All demographic and statistical information assembly of these genomes (derived from the query in BV‐BRC [51]) was collected in a database and deposited in a public repository which can be accessed through the following link: https://zenodo.org/record/4898561#.YMtoPWgzbIV.

3.2. Phylogenetic Analysis by MLST

The MLST analysis was performed using the STRUCTURE software 2.3.4 (Figure 1) showed that K = 6 was the best value to define the number of populations. The analysis revealed that the majority (90.5%) of the Colombian genomes belonged to the hpEurope population, while only 9.5% were classified in the hpAfrica1 population. This analysis did not reveal any independent group within Colombian strains, pointing to a lower resolution of the method when compared with whole‐genome phylogenetic analysis. The phylogenetic tree, constructed using the sequences of the seven housekeeping genes (Figure 2), was consistent with the findings obtained in STRUCTURE, evidencing the same groups. In addition, none of the Colombian genomes was grouped into the hpEastAsia or hspAmerind populations.

FIGURE 1.

H. pylori populations based on MLST analysis in Structure software.

FIGURE 2.

Neighbor‐joining tree of Colombian H. pylori (black branch) and 741 reference genomes.

3.3. Phylogenetic Analysis by SNP

The phylogenetic tree (Figure 3) showed that 36 of the 221 Colombian genomes, were classified in the hpAfrica1 population, 46 in the hpEurope population, and 139 Colombian genomes formed two independent groups. In Figure 3 these new groups were called Colombia1 (131 genomes) and Colombia2 (eight genomes). None of the Colombian genomes were grouped with the hpEastAsia or hspAmerind populations.

FIGURE 3.

Phylogenetic tree based on SNP for Colombian genomes.

3.4. Virulence Genotyping

The cagA gene was found in 75.1% (166/221) of the Colombian genomes analyzed. In all these genomes, the western‐type EPIYA motif (C) was found, in agreement with the absence of hpEastAsia population in our set. The most frequent EPIYA motif was ABC (72.3%), followed by ABCC (25.3%), ABCCC motif was not found. The intact cagPAI (26 genes analyzed) was found only in two genomes (1.2%). These two genomes corresponded to genomes downloaded from databases.

Of the 166 isolates positive for cagA gene, clinical information was available only for 63. Considering that the EPIYA‐ABCC motif is the most virulent variant observed, we analyzed its presence in association with the pathologies. Only 12 genomes with the EPIYA‐ABCC motif had clinical information available: five were from patients diagnosed with gastritis, six with dysplasia or metaplasia, and one genome from a patient diagnosed with cancer. Additionally, among the 120 genomes carrying the EPIYA‐ABC motif, clinical information was available for 51. Of these, three were from patients with cancer, 30 from patients with gastritis, and 18 with dysplasia or metaplasia.

For the other cagA‐positive genomes, cagPAI was classified as semi‐intact, partially deleted, and negative, according to the number of genes present in each genome (Table S6). The most frequently absent genes were cagP, cagQ, and cag1, while the cag3 gene was present in all isolates positive for cagA. An analysis of the association between virulence factors and pathology was performed on the 99 genomes with available pathology information. A χ ² test was performed to determine possible associations, and only the variables “disease” and “cagPAI status” showed a statistically significant p value (p = 0.04).

For the vacA gene, 17 different allelic combinations were found. vacAm1i1s1b and vacAm1i1s1c are the most frequent combinations present in 40.7% and 23.9% of the genomes, respectively. For the s region, most of the genomes (92.8%) carried the s1 allele; for the m region, the most frequent allele was m1, present in 75.6% of the genomes, and for region i 76.9% of genomes carried the i1 allele.

For the most recognized adherence gene in H. pylori , babA, we found that the babA2 allele was present in 154 genomes (69.7%). Regarding the other genes for adherence and modulation of the immune response (alpB/hopB, babB/hopT, hpaA, hopZ, sabA/hopP, sabB/hopO, alpA/hopC, napA, and oipA/hopH), we found that all genes were present in 5% of the Colombian genomes. However, the individual frequency of each gene was variable, finding that the most frequent were hpaA and hopZ, both present in 95% of the genomes (210/221). On the contrary, the least frequent were babB and sabB, both present in 38.5% of the genomes (85/221) (Table S7).

For the iceA gene, the most frequent allele was iceA2, present in 52% (115/221) of the analyzed genomes. Finally, possible combinations of virulence genes were analyzed; for this, the vacA, babA, cagPAI, cagA (EPIYA), and iceA genes were considered. The most frequent genotype in the analyzed sample was ‐vacAs1b/i1/m1—iceA2—partially deleted cagPAI—EPIYA ABC—babA2. This genotype was present in 16 genomes (7.2%).

3.5. In Silico Determination of Mutations Associated With Antimicrobial Resistance

Mutations in the 23S rRNA gene related to resistance to CLA were found in 3.61% (8/221) of the Colombian genomes. Of these, six (2.7%) had the A2143G mutation, and two (0.9%) had the A2142G mutation. Regarding the mutations linked to resistance to TET, the A926G mutation was found in the 16 s rRNA gene in 8 Colombian genomes (3.6%). The presence of mutations associated with resistance to AMX was detected in 16 Colombian genomes (7.2%), all with mutations in PBP3, being the most frequent mutation F490Y (6.8%), followed by the A50S mutation in one genome (0.5%).

In the case of mutations associated with resistance to LEV, the N87I, N87K, N87Y, D91G, D91N, D91Y, N97I, and N97K mutations of the GyrA protein and the R484K and S279G mutations in the GyrB protein were analyzed. Some of these mutations were found in 10.26% (23/221) of the genomes.

The presence of stop sequences, insertions, or changes in the reading frame in the rdxA and frxA genes was determined to look for changes associated with MTZ resistance. Twelve genomes (5.40%) showed some modifications, predominantly in the rdxA gene (4.95%, 11/221). In addition, the R90K, D59N, R131K, A118T, I160F, and H97T mutations in the protein sequences encoded by the rdxA gene were also analyzed. Some mutations were present in 160 of the 221 genomes analyzed (72.1%). Thus, the total frequency of mutations related to MTZ resistance was 75.7% (168/221).

The mutations for all antibiotics were analyzed together, and we found that 14.4% of the genomes (32/221) presented profiles associated with resistance to more than one family of antibiotics. Also, 28 (12.6%) genomes had mutations associated with two families, and the most common presentation was LEV/MTZ, present in 12 genomes (5.4%). Additionally, 3 (1.3%) genomes presented mutations associated with resistance to three families, and one genome (NQ4228—Genbank accession number: AKNT00000000) had resistance‐related mutations for all antimicrobials.

The study revealed different mutations in the genomes analyzed. The summary of the frequencies of mutations is presented in Table 1.

TABLE 1.

Frequence of the mutation associated to AMR in the analyzed genomes.

CLA			TET			LEV			AMX			MTZ
Mutation	n	%	Mutation	n	%	Mutation	n	%	Mutation	n	%	Mutation	n	%
ARNr23s A2142G	2	0.9	ARNr 16 s A926G	8	3.6	GyrA D91G	1	0.45	PBP3 A50S	1	0.45	Stop frxA	1	0.45
ARNr23s A2143G	6	2.7	Total	8	3.6	GyrA D91Y	1	0.45	PBP3 F490Y	15	6.76	rdxA H97T	1	0.45
Total	8	3.6				GyrB R484K	1	0.45	Total	16	7.21	rdxA D59N	3	1.40
						GyrA N87K	2	0.90				Stop rdxA	11	4.95
						GyrA N87I	7	3.15				rdxA A118T	18	8.10
						GyrA D91N	11	4.95				rdxA R131K	41	18.50
						Total	23	10.35				rdxA R90K	93	41.90
												Total	168	75.75

4. Discussion

Although hundreds of H. pylori genomes have been published, few studies have focused on analyzing diversity within a single country. We report here a study that compiles all the genomic information of H. pylori of Colombian origin (until 2020), with relevant information considering that Colombia has one of the highest gastric cancer rates in the world [52]. First, our phylogenetic results based on whole‐genome SNPs showed the appearance of two population groups formed exclusively by Colombian H. pylori genomes, suggesting an adaptive evolution and ongoing population segregation in Colombia. Additionally, the absence of the hspAmerind subpopulation in Colombia was evidenced, pointing to a remarkable substitution of the H. pylori population following the European colonization. These findings, along with other recent reports [9, 24, 25, 26], suggest that H. pylori in Colombia has followed unique evolutionary paths [25] and has formed new subpopulations of European origin [9], as shown in the Figure 3. Indeed, recent reports have shown that the new Colombian subpopulation has evolved from the European hspSWEurope subpopulation [26]. Interestingly, a study analyzing Colombian H. pylori genomes found a new subgroup called hspColombia, which was separated into three subgroups: hspColombia_Nariño, hspColombia_Andes, and was done on Colombian. This study supports the evidence of great diversity of was done on Colombian H. pylori hpneurope subpopulation [53, 54].

The results obtained here show that the Colombia 2 group is joined with a group of genomes that had previously been classified as hpEuropaPeru [42] and a group of Nicaraguan genomes that had also been grouped into a new population [25]. It would be valuable in future studies to use a sample that includes several countries in America and Europe and includes mestizo and indigenous populations of the Latin American population to ascertain which subpopulations are circulating in the continent and consider other factors, such as the immune response of the host, dietary habits, that have an impact in the bacterial adaptation to its niche due to the genetic plasticity [55, 56].

Regarding the virulome analysis, we found that the most common genotype was vacAs1b i1 m1/babA2/iceA2/cagPAI partially deleted with cagA + (EPIYA‐ABC). Due to the tremendous genetic diversity of H. pylori that resulted in many allelic combinations, this genotype was only found in 7.2% of the genomes evaluated. Studies of virulence factors are generally characterized by exhibiting a high level of genetic heterogeneity between different strains of H. pylori , mainly explained by patients, geographic location, and ethnicity [46]. It is important to note that none of the virulence markers can be considered an independent factor for disease outcome. Studies have shown that when multiple virulence factors are present, the risk of severe pathologies is higher [13].

One of the main factors that have been associated with virulence and with an increased risk of the development of serious diseases is cagPAI and the cagA gene. Studies have reported that the presence of the cagA gene is associated with a higher risk of developing gastric cancer (OR = 2.28) [57]. In this study, the cagA gene was found in 75.1% of the genomes, like that previously detected in Colombia [58, 59, 60, 61]. Regarding cagPAI, the presence of 26 genes was analyzed, including 17 genes that have been previously described as essential for the translocation of CagA (cag3, cag4, cag5, cagC, cagE, cagG, cagI, cagL, cagM, cagN, cagT, cagV, cagW, cagX, cagY, cagZ, virB11) [62]. We found intact cagPAI in only two of the Colombian genomes and the presence of the 17 essential genes for CagA translocation in 43/221. Global reports have shown a high diversity in what is observed for cagPAI across different H. pylori populations and other geographic regions. According to reports, only between half and two‐thirds of western H. pylori strains carry cagPAI, and a partial cagPAI deletion has been identified in 4%–88% of H. pylori strains [63].

Regarding the EPIYA motifs, previous Colombian reports have described the presence of the ABCCC motif in 21% of the isolates studied [64]. However, we only found the EPIYA‐ABCC motif (25.3%) and the EPIYA‐ABC motif (72%) in this study. Studies have shown that in Western countries, strains harboring multiple EPIYA‐C motifs (ABCC or ABCCC) have a higher phosphorylation capacity and confer an increased risk of gastric cancer (OR = 3.28) compared to only 1 EPIYA motif ‐C [65].

Another virulence factor analyzed in this study was the vacuolizing cytotoxin VacA, which is involved in forming pores in infected cells [15]. Seventeen different allelic combinations were found in the Colombian H. pylori genomes, the most frequent were vacAm1i1s1b and vacAm1i1s1c, present in 40.7% and 24% of the genomes analyzed. These allelic combinations are considered highly virulent; different studies have shown that the vacAs1m1 combination has a greater vacuolizing capacity; however, it is known that within the s1 allele, the s1a subtype is the most virulent, followed by s1b and s1c [66, 67]. When analyzing the iceA gene, a higher frequency of the iceA2 allele (52%) was found in Colombian genomes. The function of the iceA2 product remains unclear. However, studies have shown that iceA2 is more prevalent in Western countries than in Asian countries; additionally, some authors have reported that iceA2 is inversely associated with peptic ulcers (OR 0.76, 95% CI 0.65–0.89) in Western countries [68].

Adherence factors are essential for initiating the H. pylori infection process, particularly by facilitating adhesion to the gastric mucosa and mediating the inflammatory response. The major H. pylori adhesin is BabA [69]. In this study, we found the babA2 allele in 69.7%. Previous reports in Colombia and Latin America showed frequencies of this allele in around 46%–82.3% [58, 70, 71, 72]; however, detecting the babA2 allele does not necessarily reflect the functional status of the gene. The level of BabA protein production and its ability to bind Lewis (Le^b) antigens are also essential for the characterization of the gene's functionality [73]. Therefore, new characterization strategies for this adhesin should be implemented; probably methodologies based on evaluating proteins are more convenient. It is also essential to bear in mind that there are strains of H. pylori that are considered “specialized” due to their ability to bind exclusively to ALe^b antigens, predominantly in South American countries [69].

The clinical information collected from the isolates revealed 27 genomes from patients with premalignant lesions and five with gastric cancer. However, when analyzing their association with virulence factors, no statistically significant association was found. Nevertheless, it is important to consider that one of the limitations of this study is the limited clinical information that we were able to collect.

Here we also studied AMR‐related mutations. Regarding CLA, mutations were found in 3.6% of the analyzed genomes. The available studies of resistance to this antibiotic in Colombia had shown very diverse data with prevalence of resistance ranging from 0% to 46% [21, 74, 75, 76, 77]. Mutations associated with resistance to LEV, these were identified in 10.26% of the genomes analyzed. A previous study in Colombia reported an LEV resistance monitoring, seeing an increase in resistance rates from 2009 with rates of 11.8% until 2014, which reached 27.3% [20]. Another study published a rate of resistance to LEV of 20.3% in the population of the south of the country [76]. Our data are a complete reflection of the situation since the collected genomes were obtained from 2000 to 2018. Together with previous reports, our data show the need to implement surveillance studies of resistance to CLA and LEV in the country. According to the most recent recommendations of the Maastricht V/Florence consensus for empirical treatment, if the CLA resistance is less than 15%, using this antibiotic in a triple therapy of 14 days is recommended. Still, if it is greater than 15%, it should be replaced by LEV if the resistance to it is less than 15% [78].

For TET, a global resistance of between 0% and 7.3% has been reported, being one of the antibiotics proposed as ideal [79]. This study found that 3.61% of Colombian genomes had the A926G mutation in 16 s rRNA. Although few studies have reported resistance to TET in the country, a recent study found resistance in only one of the isolates analyzed [76]. Our data suggest that resistance rates should be monitored to continue considering this antibiotic an excellent therapeutic option. Concerning the antibiotic AMX, resistance rates have been documented in the country at around 7%, being one of the highest in the Latin American region [5]. Although the data found here are consistent with this, we found mutations only in PBP3, which had not been previously analyzed in Colombia. Therefore, previous analyses based on molecular biology have likely underestimated AMX resistance in the country.

Resistance to MTZ has been predominantly associated with mutations in the gene coding for RdxA. However, the precise mechanism of resistance is still debated [80]. We decided to search only mutations for which a structural change in the protein has been reported, such as R90K, R131K, D59N, and H97T [50, 81]. Considering the characteristics of this study, it was not possible to predict whether other mutations found could be relevant in the phenotypic result of resistance.

We found that 75.68% of the genomes analyzed had some change related to MTZ resistance. Previous studies reported high resistance rates to this antibiotic in Colombia (~83%) [5]. However, our results present a broader panorama of the changes related to resistance to MTZ in Colombia and provide a possible alternative for the molecular detection of this resistance since changes in the rdxA and frxA genes were considered.

When analyzing the results of the mutations associated with AMR in all the genomes studied, 32 (14.4%) genomes were found with mutations linked to resistance to more than one antibiotic. The multi‐drug‐resistance (MDR) phenomenon is increasingly observed within the global emergence of H. pylori infection [82]. In Colombia, some studies have demonstrated the presence of H. pylori MDR strains, reporting the presence of strains resistant to CLA and AMX [83], or even four antibiotics (AMX, CLA, LEV, and MTZ) [84]. In Colombia, quadruple therapy continues to be used and two antibiotics are combined (CLA and MTZ generally). Thus, the MDR phenomenon must be monitored with particular attention to avoid unsuccessful treatments that lead to the use of rescue therapies [85].

Here, genomic tools were used to analyze all Colombian H. pylori genomes deposited in public databases to characterize the mutations associated with AMR circulating in the country. Although, in general, data comparable with other studies were obtained, in the case of TET and AMX, our results showed the presence of mutations that other studies in Colombia had not reported.

Our results show that WGS analysis allows the characterization of phylogeny, virulence, and AMR in H. pylori . Also shows that it is an easy‐to‐use method that offers advantages over other techniques. With the massification of sequencing techniques, it will be increasingly easier to access them and use them could provide a solution for the rapid and complete characterization of H. pylori . Although it is possible to use databases to characterize virulome and resistome for many other bacterial models, the tremendous genetic diversity of H. pylori makes it challenging to use the available databases. WGS is valuable for detecting H. pylori resistance mutations but has limitations, especially for amoxicillin and metronidazole, where the correlation between genotypic mutations and phenotypic resistance is inconsistent. This discrepancy may arise because WGS can overlook epigenetic or environmental factors affecting gene expression, leading to inaccurate resistance profiles and potentially impacting clinical decisions. Recent studies suggest that combining WGS with transcriptomic or proteomic data could improve diagnostic accuracy, capturing complex resistance mechanisms beyond single‐gene mutations. Our findings highlight the need for curated databases to rapidly define H. pylori virulence and resistance patterns.

5. Conclusions

It was verified through phylogenetic analysis that there is a new evolutionary line of H. pylori in Colombia, which apparently corresponds to a subpopulation of the hpEurope group. Additionally, we compiled all the genomic information of H. pylori of Colombian origin, relevant information considering the high infection rates in Colombia.

We evidenced high diversity in virulence factors alleles in Colombian H. pylori genomes, being the most frequent vacAs1b i1 m1/babA2/iceA2/cagPAI partially deleted with cagA + (EPIYA‐ABC). However, none of these alleles were found to be significantly associated with any specific geographical region or patient group. Additionally, analyses based on next‐generation sequencing techniques of H. pylori enabled the detection of mutations associated with the generation of resistance mechanisms to CLA, LEV, TET, AMX, and MTZ. However, more research in this field is needed to cover possible rare or previously undescribed mechanisms.

Our results showed that comparative genomics and next‐generation sequencing‐based analyses of H. pylori are helpful tools for assessing the elements contributing to diversity, virulence, and resistance.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1.

Data Availability Statement

The genomes of Colombian strains included in the study are available in the NCBI (https://www.ncbi.nlm.nih.gov/) with the BioProject number PRJNA656306.