Microarray-Based Detection and Clinical Evaluation for Helicobacter pylori Resistance to Clarithromycin or Levofloxacin and the Genotype of CYP2C19 in 1083 Patients

Background. Helicobacter pylori (H. pylori) is one of the most frequent and persistent bacterial infections that affect nearly half of the world's population. Antibiotic resistance is a constantly evolving process and local surveillance of antibiotic resistance is warranted to guide clinicians in their choice of therapy. The aim of this study was to establish a microarray-based detection to identify H. pylori infection, clarithromycin and levofloxacin susceptibility, and CYP2C19 genetic polymorphism and guide to potential choice of proton pump inhibitor (PPI), antibiotic administration for tailored H. pylori eradication therapy. Methods. By analyzing the sequence of human genomic CYP2C19⁎2 and CYP2C19⁎3 and mutations within the 23S rRNA and gyrA gene regions conferring clarithromycin and levofloxacin resistance, respectively, we developed a microarray for individual therapy detection of H. pylori infection. Plasmids were established as positive or limit of detection (LOD) reference materials. The specificity and sensitivity of the microarray had been performed. And a total of 1083 gastric biopsy samples were tested and the Kappa value had been calculated between the array and Sanger sequencing. We also analyzed the resistance to clarithromycin and levofloxacin in China, as well as the CYP2C19 polymorphisms. Results. The LOD of detecting H. pylori was 103 CFU/mL and human genome DNA was 2 ng/μL. The detection results of 1083 gastric biopsy samples showed that 691 (63.80%) were H. pylori positive, of which 266 (38.49%) were resistant to clarithromycin, 192 (27.79%) were resistant to levofloxacin, and 61 (8.83%) were resistant to both of them. For the type of CYP2C19 polymorphism, 412 (38.04%) were homozygous fast type (HomEM), 574 (53%) were heterozygous EM (HetEM), and 97 (8.96%) were poor metabolizer (PM). Conclusions. The proposed microarray-based detection has high specificity, sensitivity, and reproducibility for detecting the resistance of clarithromycin or levofloxacin as well as CYP2C19 polymorphism, which may help to improve the clinical eradication rate of H. pylori.


Introduction
More than 50% of the world's population is infected with Helicobacter pylori (H. pylori), and the rate of infection among Chinese adults is 40%-60% [1]. H. pylori infection is associated with a variety of gastrointestinal diseases such as gastritis, peptic ulcer and promotes the accumulation of various mutations and might promote gastric carcinogenesis [2]. At present, triple or quadruple therapy based on proton pump inhibitors (PPIs) combined with amoxicillin and clarithromycin (or metronidazole) is recognized as a firstline treatment of H. pylori. PPI-clarithromycin-amoxicillin with a bismuth quadruple regimen was recommended as the second-line after prior treatment failure in the Maastricht V/Florence Consensus Report [3] and the Fifth Chinese Consensus Report [1]. The efficacy of therapy is mainly affected by the following two factors: one hand is antibiotic resistance and primary drug-resistance rate of three out of six kinds of antibacterial drugs recommended by consensus report has raised, metronidazole (40%-70%), clarithromycin (20%-50%), and levofloxacin (20%-50%), and even multidrug resistance also occurred. The antibiotic resistance has significantly reduced the eradication rate [1,4,5]. Due to resistance, eradication rate of standard triple therapy has less than the 80%, and even the treatment time of standard triple therapy increased from 7d to 10d or 14d, and eradication rate can only improve about 5% [6]. Another hand is that PPI inhibitors pay an important role for the success of triple therapy. PPIs are serials of pharmaceutical agents that target H + /K + -ATPase and have high potency to increase gastric pH value coupled with minor side effects, which made them very popular in gastrointestinal disorders, particularly in H. pylori infection eradication regimen [7]. PPIs' high acid suppression make acid-labile antibiotics such as clarithromycin more stable by increasing gastric pH value, thereby increasing concentration and H. pylori sensitivity to antibiotics [8]. PPIs (except rabeprazole) are mainly metabolized by cytochrome P450 2C19 (CYP2C19), whose highly polymorphic genotype can cause variability response in omeprazole-based or lansoprazole-based triple therapies [9]. CYP2C19 * 2 (681 G > A, rs4244285) and CYP2C19 * 3 (636 G > A, rs4986893) are most frequent variant and the two types alleles associated with loss of function [10]. Other known factors that contributed to the increased failure rate of H. pylori eradication include poor compliance, high gastric acidity, and high bacterial load [11].
According to CYP2C19 * 2 and CYP2C19 * 3, the CYP2C19 phenotype has been classified into three groups: homozygous extensive metabolizers (Hom-EMs), heterozygous extensive metabolizers (Het-EMs), and poor metabolizers (PMs) [12]. Research shows that, in the first-line treatment based on PPI, PM patients can get the highest H. pylori eradication rate. In remedial therapy based on PPIs, the H. pylori eradication rate between HomEM and HetEM is significantly different [7].
In this study, we established a visual DNA microarray method to detect susceptibility of clarithromycin and levofloxacin, polymorphisms of CYP2C19 in patients simultaneously. After validating accuracy, specificity, and sensitivity of the microarray, we applied it in clinical research by detecting gastric mucosa samples of 1083 cases and assessed the antibiotic resistance of H. pylori and genetic polymorphism of CYP2C19 in China. Gastric biopsy specimens were crushed or cut into pieces using a sterile glass homogenizer by the Full Automatic Tissue Grinding Instrument (Tuhe Mechatronics Co., Shanghai, China) and mixed with 200 L normal saline. Total DNA mixtures from biopsy specimens were extracted using the QIAmp DNA Mini Kit (Qiagen, Hilden, Germany) following the protocol.

Multiplex PCR.
DNA regions involved in clarithromycin (23S rRNA), quinolone (gyrA) resistance, CYP2C19 * 2 (681 G > A, rs4244285), and CYP2C19 * 3 (636 G > A, rs4986893) were amplified by PCR as previously reported [18][19][20]. Primers were picked in the conserved upstream or downstream regions and probes were designed at positions of mutations (Table 1). All the primers and probes were verified by BLAST (http://blast.ncbi.nlm.nih.gov/). All reverse primers for target genes were labeled by biotin at the 5 -ends and multiplex PCR was performed in one tube. Reaction mixtures (25 L) contained 12.5 L of 2× Multiplex PCR Mix (cwBiotech, Beijing, China) and 3 L of DNA template. The concentrations of forward and reverse primers all were 0.12 M and 1.2 M, respectively. PCR was performed on a Thermal Cycler PCR system (Applied Biosystems, Foster City, US) using the following conditions: 15 min at 95 ∘ C; 45 cycles of 30 s at 94 ∘ C, 60 s at 65 ∘ C, and 30 s at 72 ∘ C; and a final extension of 5 min at 72 ∘ C.

Construction of Reference
Plasmids. The international standard strain H. pylori 26695 was used to construct the wild-type control plasmid and normal human genomic DNA was used as template for PCR amplification. Mutant control plasmids were constructed using PCR site-directed mutagenesis to amply fragments of mutations in specific sites. The purified amplification products were ligated with PGM-T vector and cloned into DH5 . The target fragments were identified by sequencing the entire regions. Then the positive clones were picked to culture. After the determination of the initial concentration, the bacteria liquid was diluted to 1 × 10 5 , 1 × 10 4 , 1 × 10 3 , and 1 × 10 2 CFU /ml by using phosphate buffer saline. Plasmids of human genomic DNA were extracted and diluted to 10 ng/ L, 5 ng/ L, 2 ng/ L, and 1 ng/ L in order to determine the LOD of the microarray.
TGCGGTTTATAATGCRCTAT Quality control e TTTTTTTTTTTTTTTTTTTT a F: forward primer; b R: reverse primer; reverse primers used for microarray have biotin conjugated at 5'-ends; c W: wild genotype probe; d M: mutant genotype probe, an oligonucleotide of 12 T's with an amino-labeled 3'-end was conjugated to the 3'-ends of all probes; e an oligonucleotide of 20 T's with an amino-labeled 3'-end, biotin-labeled 5'-end was used as microarray quality control.

Microarray Fabrication.
A repeat sequence of 12T-with an amino labeled 3 -end was connected to the 3 -end of all the probes and fixed on the aldehyde-chip surface. All modified microarray probes were synthesized by Sangon Biotech Co., Ltd. (Shanghai). Probes, at 50 M final concentration, were spotted thrice repeatedly by using a noncontact inkjet Nanoplotter 2.1 (GeSim, Dresden, Germany) on the aldehydechip after mixing with uniform proportional printing buffer (5% glycerol, 0.1% sodium dodecyl sulfate (SDS), 6×salinesodium citrate buffer (SSC), and 2% (wt/vol) Ficoll 400). Quality control (QC) probe was spotted eight times repeatedly in the horizontal direction at a final concentration of 12.5 M in order to manage the standard operation and calibrate the signal values. Each aldehyde slide was divided into 10 blocks (11 × 11 mm) by a waterproof film that can detect 10 different samples. The layout was shown in Figure 1(a) and the principle of hybridization was shown in Figure 1(b). The fabricated microarray was placed in a dryer for 24 h at room temperature. By washing once with 0.2% SDS and once with distilled water for 30 s at room temperature, the unbound probes were removed before use.

Hybridization and Signal Detection.
After amplification, the products were denatured at 95 ∘ C for 5 min and immediately placed on ice for 5 min and then mixed with 5 L of hybridization buffer (8× SSC, 0.6% SDS, 10% formylamine, and 10× Denhardt). A total of 10 L hybridization mixture were hybridized on the microarray in a hybrid-box for 45 min at 45 ∘ C.  According to the sequencing results, the coincidence rate of the microarray results had been calculated. McNemar Chisquare test and Kappa consistency check were performed to identify the difference and consistency between the microarray and the sequence. The SPSS 17.0 software package (SPSS Inc. Chicago, USA) was used for all analyses.

Determination of Threshold Signal
Intensity. The signal intensities values were quantified by Gel-Pro analyzer and the signal intensities were calibrated as follows: the calibrated value of a probe = mean signal intensities value of the probe / mean signal intensities value of the QC probe (the same detection block) ×100.
For the two probes of CYP2C19 * 2, the ratio of the calibrated value of 2C19 * 2W to that of 2C19 * 2 M was determined. If the ratio was greater than 2.0, the locus was considered to be wild type and homozygote, while if the ratio was less than 0.5, the loci were considered to be homozygous mutation. If the ratio was between 0.5 and 2.0, we determined this locus was heterozygous. Two probes of CYP2C19 * 3 (2C19 * 3W and 2C19 * 3M) criteria were the same as CYP2C19 * 2. For the four probes of 23S rRNA (42W43W, 42M43W, 42W43M, and 42M43M), the one of calibrated value two times greater than the other three ones was considered to be positive. Criteria of the four 87-locus probes (87WT, 87WC, 87MA, and 87MG) and three 91-locus probes (91W, 91MG, and 91MA) of gyrA were the same as 23S rRNA.

Specificity and Sensitivity of Microarray Test.
To evaluate the specificity of the microarray method, we detected the reference plasmids. The result images covering 15 probes were shown in Figure 2 and showed that the microarray was able to exactly distinguish the mutation sites of these nucleotides. The negative control H. pylori 26695 showed the negative microarray results which also demonstrated the specificity of this assay (Figure 2).
To determine the absolute LOD of this strategy, serially diluted recombinant plasmids had been quantified by the copy numbers served as the reference materials (i.e., from 1 ×10 5 , 1 × 10 4 , 1 × 10 3 , and 1 × 10 2 CFU/mL for plasmid bacteria and 10, 5, 2, and 1 ng / L for human genomic DNA). The absolute LODs for the plasmids were 10 2 CFU/mL and 2 ng / L for the human genomic DNA (Figure 3(a)). 23S rRNA and gyrA of H. pylori detection kit (PCR-Fluorescence Probing, provided by our laboratory) were also used to detect 23S rRNA and gyrA reference plasmids (1 × 10 1 to 1 × 10 5 copies/ L), respectively. The results of sensitivity comparison between microarray assay and real-time PCR showed that they had similar sensitivities (Figures 3(b) and 3(c)).  CYP2C19 * 2 and Kappa=0.884 (P < 0.0001) for CYP2C19 * 3, which revealed that the consistency between microarray and sequencing was statistically significant.

Detection of 23S rRNA and gyrA Mutations.
Determined by sequencing, there were 691 cases as H. pylori positive and 266 cases that showed mutations of 23S rRNA. Part of the microarray and sequencing results are shown in Figure 6. The sensitivity, specificity, and concordance were 92.13%, 93.75%, and 95.94%, respectively, between the microarray and sequencing (    (Z=23.323, P < 0.0001) at codon 91, indicating there is no statistical difference between the results of microarray and sequencing.

Correlation between H. pylori Mutations and CYP2C19
Genetic Polymorphism. Table 7 showed a correlation between the mutations of H. pylori and CYP2C19 genetic polymorphism. Chi-square test was used to assess differences and there was no significant correlation that was observed between CYP2C19 genetic polymorphism and mutations of  H. pylori to clarithromycin (P = 0.079) and levofloxacin (P=0.17).

Discussion
In China, prevalence of H. pylori strains resistance to clarithromycin, metronidazole, and levofloxacin (fluoroquinolone) has been increasing over the last decades. Studies show that clarithromycin resistance was higher than 10% and the success of standard eradication (first-line therapy) was less than 85% [18]. Many studies had confirmed that the eradication rate of clarithromycin sensitive strains was 87%-92%, while the eradication rate decreased to 18%-21% in clarithromycin resistant strains [19]. Levofloxacin was proposed as a salvage treatment regimen after the failure of clarithromycin-based treatments [20]. However, the frequency of levofloxacin resistance was widely recognized throughout the world, which may reduce the efficacy of levofloxacin-based therapeutic regimen [21,22]. On the whole, the resistance rate of these antibiotics was every high in China; however, regional differences were exit. On the contrary, the resistance rates of H. pylori to amoxicillin  (0%∼5%), tetracycline (0%∼5%), and furazolidone (0%∼1%) were still very low [23][24][25]. By assessing sensitivity of clarithromycin, levofloxacin, and genotypes of CYP2C19 before using the triple therapy, the choice of drugs for individual medication can ensure the success rate of treatment and reduce the overuse of antibiotics adverse reaction and waste of resources.
In this study, we designed specific probes to DNA regions involved in clarithromycin (23S rRNA), levofloxacin (gyrA) resistance, CYP2C19 * 2 (681G>A, rs4244285), and CYP2C19 * 3 (636G>A, rs4986893). The DNA microarray could be used to provide a type of antibiotics sensitivity of clarithromycin, levofloxacin, and polymorphisms of CYP2C19 before the patients' treatment with triple or quadruple therapy. The microarray was fast so the entire experiment could be finished within 6 h from DNA extraction of samples to obtain visual results by naked eyes. The specificity and sensitivity of the microarray had been validated: the specificity concordance was 97.49% compared with the sequencing, and the sensitivity was 10 3 CFU/ml equivalent to fluorescence quantitative PCR (Figure 3).
DNA microarray technology was high throughput, efficiently, and low cost. Quantum dots served as a semiconductor nanocrystalline material due to their good fluorescence characteristics [23], which was widely used in bioimaging. Quantum dots acted as similar electron transfer and could convert a silver ion to a metallic silver atom in the presence of reductive reagents (such as hydroquinone). Silver atom deposited on the surface of quantum dots and could further catalyze the conversion of silver around. The cascade catalysis made silver particles accumulate together and form into silver shell, and then formation of black particles was visible by naked eyes. In this study we developed a new microarray visualization technology based on tyramine signal amplification coupled with quantum dotcatalyzed silver deposition (TSA-QDS) to detect and evaluate H. pylori resistance to clarithromycin or levofloxacin and the genotype of CYP2C19. The TSA-QDs possessed similar detection sensitivity as real-time PCR of 10 3 CFU/mL [26,27] and the signal point of the method can be scanned by general scanners which are less expensive than fluorescence scanner. Without equipping sophisticated and expensive scanner, the microarray visualization technology greatly reduces the cost (less than six dollars per sample) of chip technology platform to popularize. Compared with traditional methods to detect gastric biopsy specimens such as phenotypic culture [28] and real-time PCR, the method was high throughput and accessible to detect coinfections with clarithromycin and/or levofloxacin-resistant and susceptible strains.
This study showed that distribution of CYP2C19 gene polymorphism in population from both Beijing in northern China and Zhejiang province in south eastern China was consistent with that of Chongqing in southwestern China [24] and Asia [7]. PMs and EMs accounted for a significant proportion in Asian population, while EMs accounted for one in Europeans [25]. CYP2C19 polymorphism could influence the PPIs metabolization and H + concentration in stomach and thus affect activity and stability of antibiotics. H. pylori was not completely destroyed or inhibited to grow by these antibiotics in the microenvironment, which might lead to microevolution of H. pylori even development of drug resistant strains [24]. Our study indicated that CYP2C19 genetic polymorphism was not significant associated with susceptibility of clarithromycin (P = 0.079) or levofloxacin (P=0.17); however there was a trend of correlation between CYP2C19 genotypes and clarithromycin resistance. When comparing the 23S rRNA mutations in CYP2C19 EMs and CYP2C19 PMs, the number of mutations was significantly higher in the CYP2C19 EM than in the CYP2C19 PM (P=0.037 < 0.05).
This study has some limitations. Firstly, the sequencing results of some samples were not consistent with the results of microarray detection. The false negative results of microarray may be due to the difference of the PCR amplification system. For microarray detection, all target genes were amplified in one tube while single PCR amplification was performed for sequencing, which may reduce the sensitivity of microarray detection. Secondly, the DNA microarray could not cover all genes of antibiotics for the eradication of H. pylori. Therefore, future studies could address the genes that account for antibiotic resistance of H. pylori.

Conclusions
In summary, we established a visible microarray-based detection that could simultaneously and specially determine the A2142G and A2143G mutation associated with clarithromycin, mutations at amino acids 87 and 91 in the levofloxacin, and human CYP2C19 genotype. The microarray-based detection was rapid, reliable, and high throughput and could be used easily for clinical applications in guidance of individual H. pylori eradication therapy.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Ethical Approval
This clinical trial strictly adhered to the ethical standards of human medical research in the Declaration of Helsinki. This clinical trial used surplus samples after daily detection from Wenzhou Central Hospital, Huzhou Central Hospital, and Chinese Center for Disease Control and Prevention. All testing was out of touch with patients. All the results were only used as the raw data of this clinical trial and did not provide to the patients and did not serve for clinical diagnosis. There was no risk for patients. According to the "guidelines for clinical research technology of in vitro diagnostic reagents" issued by the China Food and Drug Administration, this clinical trial could not submit informed consent and ethical review.