Functional and Molecular Surveillance of Helicobacter pylori Antibiotic Resistance in Kuala Lumpur

Background Helicobacter pylori is the etiological agent for diseases ranging from chronic gastritis and peptic ulcer disease to gastric adenocarcinoma and primary gastric B-cell lymphoma. Emergence of resistance to antibiotics possesses a challenge to the effort to eradicate H. pylori using conventional antibiotic-based therapies. The molecular mechanisms that contribute to the resistance of these strains have yet to be identified and are important for understanding the evolutional pattern and selective pressure imposed by the environment. Methods and Findings H. pylori was isolated from 102 patients diagnosed with gastrointestinal diseases, who underwent endoscopy at University Malaya Medical Centre (UMMC). The isolates were tested for their susceptibility on eleven antibiotics using Etest. Based on susceptibility test, 32.3% of the isolates were found to have primary metronidazole resistance; followed by clarithromycin (6.8%) and fluoroquinolones (6.8%). To further investigate the resistant strains, mutational patterns of gene rdxA, frxA, gyrA, gyrB, and 23S rRNA were studied. Consistent with the previous reports, metronidazole resistance was prevalent in the local population. However, clarithromycin, fluoroquinolone and multi-drug resistance were shown to be emerging. Molecular patterns correlated well with phenotypic data. Interestingly, multi-drug resistant (MDR) strains were found to be associated with higher minimum inhibitory concentration (MIC) than their single-drug resistant (SDR) counterparts. Most importantly, clarithromycin-resistant strains were suggested to have a higher incidence for developing multi-drug resistance. Conclusion Data from this study highlighted the urgency to monitor closely the prevalence of antibiotic resistance in the Malaysian population; especially that of clarithromycin and multi-drug resistance. Further study is needed to understand the molecular association between clarithromycin resistance and multi-drug resistance in H. pylori. The report serves a reminder that a strict antibiotic usage policy is needed in Malaysia and other developing countries (especially those where H. pylori prevalence remained high).


Introduction
Helicobacter pylori is an etiological factor for several gastroduodenal diseases with various clinical manifestation ranging from dyspepsia, chronic gastritis, gastric atrophy to peptic ulcer disease [1]. In more severe cases, H. pylori infection may result in distal gastric adenocarcinoma and gastric mucosal-associated lymphoid tissue (MALT) lymphoma [2]. Therefore, it has been classified as the only known class I bacterial carcinogen by the International Agency for Research on Cancer (IARC) [3]. H. pylori is known to colonize almost half of the population worldwide with varying prevalence rates among different geographical regions with higher rates in developing countries [4]. In many developed societies the prevalence of infection rate has been noted to decrease from the late 90's onwards [5].
Early eradication of H. pylori has been proven to reduce the incidence rate of gastric cancer and improve ulcer healing [6,7].
However, eradication of the pathogen has been difficult due to the limited antibiotics available that can withstand the low pH environment in stomach, or the capability of the drug to penetrate the gastric mucous layers to be in contact with the pathogen. To date, the most widely used treatment regimes include the standard triple therapy, which comprises two of three antibiotics, clarithromycin and amoxicillin or metronidazole, together with one proton pump inhibitor (PPI) [8]; and sequential therapy, which initiates treatment with proton pump inhibitor and amoxicillin followed by subsequent treatment with clarithromycin and metronidazole [9]. However, increasing prevalence of clarithromycin resistance greatly reduced the efficacy of those two therapies mentioned [10,11]. As a counter measure, fluoroquinolonecontaining triple and bismuth-containing quadruple therapies have also been proposed as second line therapies after the failure of the clarithromycin-containing treatment [12,13]. In spite of that, a number of studies have shown a low but rapid rise of fluoroquinolone-resistance in countries where fluoroquinolone has been widely prescribed [14,15,16,17,18,19]. To date, the mechanisms underlying the resistance to metronidazole, clarithromycin and flouroquinolones remains obscure.
At the University Malaya Medical Centre (UMMC), the first line eradication regimen for H. pylori consists of two antibiotics (amoxicillin 1 g twice daily and clarithromycin 500 mg twice daily) and pantoprazole 40 mg twice daily for the duration of one week [20]. Due to the high metronidazole resistance rate in Malaysia [21,22], metronidazole is not recommended. In cases where the therapy failed, two consecutive rescue regimens are used [23,24]. The first line rescue regimen consists of rabeprazole 20 mg three times daily and amoxicillin 1 g three times daily for two weeks [23]. Rabeprazole is less susceptible to the influence of genetic polymorphisms for CYP2C19. Thus, it has greater and faster acid suppression compared to older PPIs [25]. Amoxicillin is one of the effective antibiotics to H. pylori with few side effects. The antibiotic resistance to amoxicillin has not been reported in Malaysia [26]. Patients who continue to have H. pylori infection will be put on the second line rescue regimen which consists of a rabeperazole 20 mg twice daily, levofloxacin 500 mg twice daily and amoxicillin 1 g twice daily for a further 2 weeks [23].
As different regimens were used for treatment, it is crucial to monitor the development of resistance for clarithromycin, amoxicillin, metronidazole and fluoroquinolones periodically to ensure the current therapy regimens remain effective. The assessment on resistance rate to other antibiotics used as rescue therapy in H. pylori eradication (such as tetracycline and rifampicin), can also provide a baseline antibiotic resistance profile of locally isolated H. pylori strains.
Recent reports suggest that the underlying antibiotic resistance mechanisms are mainly due to the genetic plasticity of H. pylori that results in genetic mutations. These include point mutation involving transition of adenine to guanine on position 2142 and 2143 in the domain V of 23S rRNA gene, which confers resistance towards clarithromycin [27]. Similarly, mutation in gene rdxA and frxA is reported to confer metronidazole resistance in H. pylori [28]. Unlike mutational patterns in 23S rRNA and gyrA, the impact of mutation in gene rdxA and frxA is caused by the inactivation of the gene function via frameshift mutation, insertions and deletions. On the other hand, mutational changes in quinolone-resistant determination region (QRDR) of gyrA gene greatly reduce the eradication efficacy of fluoroquinolones. Almost 90% of the fluoroquinolones-resistant strains harbour mutation at position 87 and 91 of the translational protein sequence gyrase subunit A (gyrA) [29]. Mutations on gene gyrB known to confer fluoroquinolones resistance have also been reported [17]. Despite that, double mutation in gyrA is reported to have a greater impact on fluoroquinolones resistance [30].
In this study, surveillance of conventional and potential antibiotics -comprising of 10 antibiotics, comprising of metronidazole, clarithromycin, fluoroquinolone (ciprofloxacin, moxifloxacin, levofloxacin, and gemifloxacin), amoxicillin, rifampicin, tetracycline, and nitrofurantoin -were included to provide a comprehensive insight of the current H. pylori antibiotic resistance prevalence in Malaysia. In addition, correlation between the antibiotic susceptibility patterns and the presence of associated genetic mutations in the local population will also be evaluated.

Bacterial strains
This study was approved by the University of Malaya Medical Centre (UMMC) Medical Ethics Committee and biopsy samples for culturing were obtained with informed and written consent from consecutive and non-repetitive patients who presented for endoscopy at UMMC. Gastric biopsy samples obtained from antrum and body of the stomach were homogenized in different tubes and plated directly onto non-selective and selective chocolate agar supplemented with 7% lysed horse blood (Oxoid, UK). Selective chocolate agar contained vancomycin (10 mg/ml), amphotericin B (5 mg/ml), trimethoprim (5 mg/ml) and nalidixic acid (20 mg/ml) [31]. The inoculated agar plates were incubated for 3-10 days in a humidified 10% CO 2 incubator at 37uC. H. pylori were successfully cultured from 110 patients and were confirmed by positivity for urease test. Amongst these, 102 were naive to standard H. pylori therapy whilst the remaining eight were from patients with treatment failure. Clonal strains derived from well-isolated H. pylori colonies of the same patient were tested for homogeneity by random amplification of polymorphic DNA (RAPD) and representative clonal strains were used for this study.

Molecular Detection on Resistant Strains
Mutations in gyrA, gyrB, rdxA, frxA, 23S rRNA were assessed on antibiotic-resistant strains by targeted gene sequencing approach. Bacterial genome DNA was extracted using RTP Bacteria DNA Mini Kit (Invitek, Germany). Amplification of 582-bp and 465-bp region on gyrA and gyrB respectively were performed on fluoroquinolone-resistant strain with thermocycler conditions being described by Wang et al [29]. Regions of 425-bp within bacterial 23S rRNA peptidyl transferase gene were amplified from clarithromycin-resistant strains as used by Ho et al [27]. For metronidazole-resistant strains, rdxA and frxA were amplified as described by Han el al [32]. The primer sequences are shown in Table 1. PCR products were purified using SV Gel and PCR Clean-Up Kits (Promega, USA) and sequenced on the Sanger method. Comparative sequence analysis between resistant and sensitive strains was carried out using BioEdit and MEGA software [35,36]. The DNA sequence results were aligned and coordinated to H. pylori 26695 (Accession: NC_000915.1; GI: 15644634).

Statistical Analysis
The two-tailed Student's t-test with unequal variance function of Microsoft Excel was used to determine the statistical significance of differences between unpaired samples. The minimum level of significance was set at ,0.05.

Prevalence of Antibiotic Resistance
DNA fingerprinting using RAPD demonstrated that of the 102 H. pylori strains isolated from each treatment-naive patient, there was no heterogeneous infection involving more than one genotype of H. pylori occurring in these patients. Among the primary resistance patterns, metronidazole resistance was the highest (33 strains or 32.3%), followed by clarithromycin (7 strains or 6.8%) and fluoroquinolone (7 strains or 6.8%) ( Table 2). Among the fluoroquinolones resistance patterns, all fluoroquinolone-resistant strains were resistant to levofloxacin, ciprofloxacin and moxifloxacin but only four strains (57.1%) were resistant to gemifloxacin. The prevalence of primary multi-drug resistance, where resistance was found towards more than one antimicrobial agent, was demonstrated in five strains (4.9%).
Similarly, among the eight patients with treatment failure, no heterogeneous infection by more than one strain was found based on RAPD analysis. The MIC results revealed that this population of strains consisted of three single-drug resistant (SDR) strains and five multi-drug resistant (MDR) strains (62.5%) ( Table 2). Resis-tance towards amoxicillin, rifampicin, tetracycline, and nitrofurantoin was not detected amongst any of these strains.

Multi-drug Resistance
Among the MDR strains (n = 10), 90% were resistant to clarithromycin, 70% to metronidazole and 50% to fluoroquinolones. Among these, 5/10 (50%) strains were resistant to both metronidazole and clarithromycin followed by 3/10 (30%) strains resistant to both clarithromycin and fluoroquinolone. One of these ten strains (10%) was resistant to all three antibiotics. Only 1/10 (10%) strain was resistant to both metronidazole and fluoroquinolone. The average MIC of MDR were observed to be higher that SDR for all three antibiotics ( Table 3). The two-tailed Student's ttest with unequal variance showed that the differences was statistically significant (P,0.05) for metronidazole and clarithromycin and highly significant for fluoroquinolone (P,0.001).

Genetic Variations of rdxA and frxA in Metronidazoleresistant Strains
Based on rdxA and frxA gene variations and amino acid sequences generated among metronidazole-resistant strains, var- iations of rdxA with its corresponding amino acid alteration in RdxA were identified in 26/37 strains (70.2%) ( Figure S1). Parallel with this, out of these 37 strains, 21 (56.8%) demonstrated the presence of truncated RdxA. The cause of premature termination was due to introduction of a stop codon by single amino acid substitution and frame-shift alteration. Although the other 16 strains encoded for full length RdxA, 3 (8.1%) were found to have nucleotides insertion/deletion resulting in amino acid insertion and deletion and frame-shift mutation at the C-terminal ( Figure  S2). In addition, one strain (2.7%) was found to have the stop codon at the end of rdxA missing. Similarly, variation of frxA and alteration in FrxA were identified in 25/37 (67.6%) metronidazole-resistant strains ( Figure S3). Among these strains premature termination codon (PTC) was present in 20 strains (54.1%) and five out of 17 (13.5%) strains had single or double amino acid deletion or insertion at the frxA ( Figure S4).  (Table 6) ( Figure S6).
On the other hand, two amino acid variants at region of study of gyrase subunit B were observed exclusively in fluoroquinolone resistant strains. Three out of nine strains presented with amino acid substitutions at each position Asp-481 (33.3%) and Arg-484 (33.3%) ( Table 6). One strain with high MIC presented with mutation on gyrB at Phe-438 into Serine and co-occured with mutation at Asn-87 at gyrA. Four out of nine fluoroquinoloneresistant strains did not contain any mutations in gyrB (44.4%). There was no apparent correlation between degree of fluoroquinolone resistance with type and number of mutations in gyrA and gyrB.

Discussion
Generally, in Malaysia, the prevalence of metronidazoleresistant H. pylori strains is high [21]. This sustained high level of metronidazole resistance prior to H. pylori eradication therapy could be attributed to massive prescription of metronidazole as a common anti-parasite drug and for other gynaecologic diseases in many developing countries [37]. Moreover, these high prevalence of metronidazole resistance rate has also been well-documented in neighbouring Southeast Asian countries such as Vietnam (69.9%) [38], Thailand (30.4%) [39], and Singapore (31.7%) [40]. Therefore, in agreement with other previous studies [21,26], we have further confirmed that metronidazole should not be the choice for first line H. pylori eradication therapy in the local population [20].
Unlike metronidazole, the resistance rate of clarithromycin is relatively low in Malaysia compared to other neighboring countries such as Vietnam (33%) [38] and Thailand (13.8%) [41]. The underlying reason on why clarithromycin resistance rate in Malaysia remains low is not clear even though increased clarithromycin resistance, especially in European countries, has been reported [42]. Although also indicated for respiratory infections, skin infections and Lyme disease, clarithromycin has not been popular in clinical practice locally due to high cost and side effects. We speculated that this could be the factor that causes relatively low resistance rate been observed in this study [27]. However, clarithromycin resistance rate is definitely on the rise as shown in this study in comparison to previous study [21,26], which is of growing concern. Clarithromycin is the key antibiotic in the H. pylori therapy regimen. Emergence of resistance rate could eventually reduces the efficacy of clarithromycin-based therapy similar to that of western countries as highlighted by Fischbach [10]. Megraud reviewed that resistance to clarithromycin is highly correlated to use of these drugs for non-H. pylori infectious diseases  [11]. Therefore, even though the emergence of clarithromycinresistant strains in Malaysia is progressively slower than other countries, strict guidelines on the use of clarithromycin should be made to maintain a low level of resistance [43].
In the local population, fluoroquinolone appears to be superior in secondary or tertiary rescue regimens and proven to show high eradication rate [23,24]. In contrast to previous reporting of zero resistance to levofloxacin in Malaysia [26], we found a low but significant increase of fluoroquinolone resistance, which is of concern too. The developing of fluoroquinolone resistance is rapid and closely dependent on increasing usage of this class of antibiotic [11]. Besides, a cross resistance pattern was observed across different generations of fluoroquinolone (moxifloxacin . levofloxacin . ciprofloxacin . gemifloxacin) with high chance of developing fluoroquinolone resistance as a result of other treatments prior to H. pylori eradication. Similar to an early report on the Taiwanese population [15], the degree of resistance to gemifloxacin was low in the Malaysian population suggested that it may have better drug efficacy than levofloxacin in H. pylori eradication in our population. Multi-drug resistance is highly prevalent among secondary resistant strains suggested that MDR strains were likely the result of selective pressure due to previous antibiotic exposure from either H. pylori treatment or treatment of unrelated infections. Interestingly, since most MDR strains were clarithromycinresistant, data from this study further suggested that there may be an underlying mechanism influencing the selection and evolution of MDR strains, which is more prevalently found among clarithromycin-resistant strains. Furthermore, this common underlying mechanism of MDR may directly or indirectly have led to a higher level of resistance, which may explain why high MICs were more commonly associated with MDR than SDR strains. However, these deductions about MDR strains are not conclusive given the small number of MDR strains available in this study and further study is necessary. Nevertheless, it further emphasised the importance to keep clarithromycin resistance under close surveillance.
The correlation between rdxA and frxA mutations and metronidazole resistance in H. pylori has been well-demonstrated [28]. Many point mutations have been shown to exist in rdxA and frxA but these positions were not uniform across all geographical regions [44]. Therefore, rdxA and frxA mutations were analyzed by comparing between local metronidazole-sensitive and metronidazole-resistant strains. As expected, this study demonstrated that nucleotide alterations in rdxA and frxA in resistant strains were random and highly distributed. Accumulatively, 89.1% of metronidazole-resistant strains possessed PTC or insertion/deletion at one or both rdxA and frxA. However, 10.8% of metronidazole-resistant strains did not contain any alteration in both rdxA and frxA. In this small subset of metronidazole-resistant strains, mutations may be present on other redox enzymes. However, screening for mutations in rdxA and frxA should be sufficient to identify about 89% of metronidazole-resistant strains in the local population.
In order to include mutations other than A2142G and A2143G, direct DNA sequencing of the amplicon was applied in this study instead of using BbsI and BsaI restriction method as described in previous study [27]. There was a conflict in stating either A2142G or A2143G mutation as predominant in clarithromycin-resistant strains. Our data agreed with investigators who study on Asian strains whereby the A2143G mutation was prominent. Therefore, this further confirms that the predominant of mutation sites could be strain-specific and differs between Western and Asian strains [45,46]. Apart from that, some investigators proposed that T2183C mutation may play a role in clarithromycin resistance [47]. Contradictory to that, T2183C mutation was detected in both resistant (9 out of 14 or 64.2%) and sensitive (5 out of 8 or 62.5%) strains. Furthermore, dual mutation on A2142G or A2143G and T2183C did not result in higher MIC. Thus, we concluded that T2183C mutation was not directly responsible or synergistically involved in conferring clarithromycin resistance. Screening for A2142G or A2143G mutations of the 23S rRNA was able to detect 92.9% of clarithromycin-resistant strains in the Malaysian population.
Consistent with other studies [48,49,50], sequencing on gyrA revealed the exclusive correlation of Asn-87 and Asp-91 with fluoroquinolone resistance. Apart from that, substitution at codon Val-199 was also observed in our strains. However, the correlation of dual mutation of Asn-87 or Asp-91 and Val-199 with level of fluoroquinolone resistance was poor. The only explanation is that the amino acid substitution on Val-199 did not have much impact to the structure of gyrase subunit A since the substitution from valine to isoleucine or alanine does not change the side-chain polarity. On the other hand, mutational changes on gyrB did not show high correlation with fluoroquinolone resistance too. Our data agreed with previous study by Tankovic et al that demonstrated some amino acid substitution is also present in sensitive strains [30]. In agreement with other studies on newer fluoroquinolone drugs, our data shows that gemifloxacin was active against some but not all strains displaying resistance to ciprofloxacin, levofloxacin and moxifloxacin [15,51]. The phenomenon did not correlate with any mutational patterns detected in our study indicating possibility of other mechanism that act synergistically in fluoroquinolone resistance. The screening for Asn-87 and Asp-91 mutations on gyrA is recommended for our local population.
In summary, this study shows that resistance rate of metronidazole remained high. Despite the current low prevalence of clarithromycin and fluoroquinolone resistance, in the Malaysian population, the trend is on the rise and situation warrants close monitoring.