Helicobacter pylori (H. pylori) infection occurs worldwide. The prevalence of infection by H. pylori varies greatly among countries, with over a 70% infection rate in many developing countries compared to 20% to 50% in most industrialized nations. H. pylori infection can lead to gastritis, duodenal and gastric ulcer, gastric carcinoma, and gastric lymphoma 123. Gastric cancer risks can be reduced by up to 60% by eradication of H. pylori infection through primary prevention, thus affecting approximately 800,000 gastric cancer patients worldwide 45.

Currently, a seven-day, triple-drug regimen has been recommended as one of the first-line therapies for H. pylori management 67. This treatment includes omeprazole (a proton-pump inhibitor), clarithromycin (CLR), and amoxicillin. However, this therapy is being investigated because of increased eradication failures due to the prevalence of CLR-resistant (CLR^r) H. pylori infections. Many studies have shown that between 0–50% of H. pylori isolates were CLR^r, which leads to a need for long term assessment of the efficacy of CLR in the triple-drug regimen 89. It is well known that the abuse of macrolide antibiotics including CLR might lead to CLR resistant forms of H. pylori. Whether an optimal therapy containing CLR for H. pylori eradication would contribute to the prevalence of CLR^r H. pylori infection was not well investigated among the Chinese population.

From 1995 to 2003, a randomized, double blind and placebo-controlled intervention trial was conducted in a high-risk population for gastric cancer in Linqu County, Shandong Province of China 10. One of the three treatments of H. pylori infection in this study was a dual-drug therapy with omeprazole and amoxicillin (OA), since the triple-drug regimen was not available in 1995. This study focused on subjects who were H. pylori seropositive at baseline. In 2000, a triple therapy consisting of OA and CLR (OAC) was applied to subjects who were H. pylori seronegative at baseline and subsequently found to be ¹³C-urea breath test (¹³C-UBT)-positive, according to instruction of the Ethic Committee of the Beijing Institute for Cancer Research and the U.S. National Cancer Institute (NCI).

In H. pylori infection, resistance to CLR is mostly due to the presence of A2143G and A2142G/C point mutations of the 23S rRNA gene 81112. We recently reported that the A2143G mutation was a late event in the development of CLR resistance of H. pylori 26695 13. The global average mutation frequency for the A2143G and A2142G mutations causing CLR resistance of H. pylori were 69.8% and 11.7%, respectively 8. Herein, we report the prevalence of the A2143G and A2142G mutations of the H. pylori-specific 23S rRNA gene among these subjects treated with regimens with or without CLR after eradication of H. pylori infection.

Genomic DNA was extracted from 770 of pre-fixed fasting gastric juice samples as described in the Method section. To determine both the presence of H. pylori infection and prevalence of CLR^r, a H. pylori-specific 23S rRNA PCR assay was developed. H. pylori infection was detected in 59.5% of analyzed subjects by the PCR assay as compared to 68.1% and 71.0% by ¹³C-UBT and ELISA, respectively (Table 1). If the combined results of ¹³C-UBT and ELISA assays were used as the gold standard [both positive (H. pylori infection) or both negative (not infected)], the PCR assay's sensitivity and specificity was 83.4% and 75.5%, respectively. Its Youdens' index was 0.59 (Table 2).

Seven years after the OA treatment, the presence of 23S rRNA due to H. pylori infection remained remarkably lower than that in the placebo group [51.1% (157/307) vs. 83.9% (260/310), p = 0.0000] (Table 1). Three years after the triple therapy with CLR, the cumulative H. pylori infection rate in the OAC group was 26.8% (41/153) by PCR (Table 1).

To detect the prevalence of CLR^r H. pylori infection, a BsaI-restriction PCR-RFLP assay was developed using the PCR product of 23S rRNA for detection of the A2143G mutation as illustrated in Fig 1. The A2143G mutation was detected in 65 of the 458 PCR products, which showed the presence of H. pylori, by the RFLP assay (Fig 2A). In 41 of the 65 A2143G-positive cases (63.1%), both the BsaI-sensitive and -insensitive PCR products of 23S rRNA were observed. For 12 representative samples, results of the PCR-RFLP assay were consistent with that of sequencing (Fig 2B).

To detect the A2142G mutation within 23S rRNA, a MboII-restriction PCR-RFLP assay was also developed (Fig 1). No A2142G point mutation was observed among 458 subjects with H. pylori infection. However, two new MboII-RFLP patterns were detected in DNA samples from 17 subjects (Fig 2C): a pattern consisting of a 164 bp and a 125 bp fragment was observed from 11 subjects and another pattern consisting of a 245 bp and 44 bp fragment was detected from 6 cases. Sequencing showed that the former 11 cases had a sequence change of AACC → GAAG (2081–2084) and the latter 6 cases had a sequence change of AAGGG → CTTCA (2222–2226) (Fig 2D).

To investigate the relationship between the prevalence of CLR^r H. pylori infection and administration of CLR as a component of an optimal triple therapy, A2143G mutation rates within 23S rRNA in each therapy group were analyzed (Table 1). The presence of the A2143G mutation in the OAC group was 3.1 times higher than that in the OA group [31.7% vs. 10.2%, p = 0.0005; OR 4.09 (95% CI = 1.64–10.23)] and 2.3 times higher than that in the placebo group [31.7% vs. 13.8%, p = 0.0040; OR 2.89 (95% CI = 1.28–6.46)]. The difference of the A2143G mutation rate due to CLR was 21.5% between the OAC and OA groups. In the OAC group, the co-infection rate (14.6%) of BsaI-insensitive H. pylori among subjects infected with BsaI-sensitive H. pylori was higher than those in the OA group (6.4%) and the placebo group (9.6%) but not significant (p = 0.0838 and 0.3258, respectively). Moreover, the sequence change AAGGG → CTTCA (2222–2226) in the OAC group was 12.3 times of the placebo group (4.9% vs. 0.4%, p = 0.0497; OR 13.28, 95% CI 0.91–379); the AACC → GAAG (2081–2084) in the OAC group was 11.2 times of the OA group (14.6.7% vs. 1.3%, p = 0.018; OR 8.38, 95% CI 1.25–68.84) and 7.7 times of the placebo group (14.6% vs. 1.9%, p = 0.002; OR 5.51, 95% CI 1.18–25.12) (Table 1).

Among Chinese subjects enrolled in the long term population-based trial 10, we found that the proportion of the primary CLR^r related A2143G mutation among 23S rRNA-positive subjects in the placebo group was 13.8% (36/260), which was similar to the average reported CLR-resistant prevalence of 10.5% (35/334) within China: 5.2% (5/96) 14, 9.8% (4/41) 13, 13.5% (12/89) 15, and 14.8% (16/108) 16. That the CLR^r infection rate in the OA group was slightly lower than that in the placebo group (p = 0.274) indicated that the OA treatment for eradication of H. pylori might reduce the prevalence of primary CLR^r among the treated subjects. Heep et al. reported that secondary resistance to CLR was up to 50% in a study of 554 H. pylori isolates after failure of therapy 17. It is well recognized that eradication therapy with CLR led to the prevalence of CLR^r H. pylori infection 89. In the present study, we observed a 31.7% A2143G mutation rate among H. pylori infected subjects three years after treatment with CLR, which is significantly higher than that among subjects without CLR treatment in the OA and placebo groups (OR = 4.09 and 2.89, respectively). An optimal way to validate the exact effect of CLR use on the formation of the secondary CLR resistance was to compare the prevalence of the A2143G mutation among subjects before and after administration of CLR. Because of the lack of gastric juice samples before eradication therapy, the original mutation status of 23S rRNA in samples was not available, thus such comparison could not be carried out in the present study. This is a limitation of our study. The strength of this study was that all of the subjects in the OAC group had the same CLR treatment history in 2000, and those in the OA and placebo group did not. Therefore, the prevalence of CLR^r H. pylori infection among subjects with and without CLR use in the local area was reliably obtained. These results were useful for evaluation of the exact effect of administration of CLR for eradication of H. pylori on the prevalence of the A2143G mutation among Chinese subjects.

Unlike indirect assays such as ELISA and ¹³C-UBT, PCR assays are direct assays to detect H. pylori infection. Thus, these assays should result in higher specificity than the indirect assays. H. pylori colonizes on the apical surface of stomach's epithelial cells and sheds into gastric juice with the regular renewal of gastric mucosa. Several PCR assays based on H. pylori DNA from gastric juice/biopsy sample were reported for detection of H. pylori infection 131819. The possible H. pylori-specific target genes included ureA, ureB, ureC, ureD, 16S rRNA, and 23S rRNA. Point mutations of 23S rRNA were closely related to formation of CLR resistance 81112. Because the PCR product of 23S rRNA described in this study could be used as a biomarker to detect H. pylori infection and CLR resistance simultaneously, this assay has advantages over other PCR-based methodologies. However, the specificity and sensitivity of the 23S rRNA PCR assay was not evaluated extensively. No single test is accepted as the standard for diagnosis of H. pylori infection 2021. In the present study, the status of H. pylori infection was detected by the 23S rRNA PCR assay, ¹³C-UBT, and ELISA for each subject, which provided an opportunity to assess the specificity and sensitivity of these assays. As expected, the highest specificity was observed for the 23S rRNA PCR assay when both CUBT and ELISA assays were negative for H. pylori infection. The Youden's index of the 23S rRNA PCR assay was similar to that of the ELISA assay but lower than that of ¹³C-UBT (Table 2).

Generally, PCR-based assays used to detect H. pylori infection are considered the most sensitive. However, the sensitivity of the 23S rRNA PCR assay was lower than ¹³C-UBT and ELISA in the present study (Table 2). Because the genomic DNA could not be extracted from gastric juice samples immediately after collection and that the shedded gastric epithelial cells easily self-decompose, the quality and amount of DNA extracted is variable between samples. In this study, fixative was added to gastric juice samples at the local clinics to prevent degeneration of DNA, and then the samples were stored at -20°C. These samples were shipped to the laboratory using dry ice and stored at -80°C. Although the fixative was removed with repeated washings before DNA extraction, it is possible that the quality of the DNA was adversely affected. The fixation process could be omitted if gastric juice samples can be stored at -80°C immediately after collection. However, it would be optimal if the DNA were isolated from fresh samples immediately after collection.

CLR^r results mainly from point mutations in the peptidyltransferase loop region of 23S rRNA 89111213. The most frequent mutations are A2143G and A2142G 8. Other mutations such as A2142C, A2115G, G2141A, and T2717C might also be associated with CLR^r. Recent reports demonstrated that the A2143G point mutation of 23S rRNA were detected in average of 90% (44 of 49) CLR^r H. pylori isolates from Chinese subjects 1314151622. In the present study, the A2143G mutation was detectable in 65 of 458 23S rRNA PCR products by the BsaI-restriction PCR-RFLP assay (Table 1). No A2142G mutation was detected in these samples. These results suggest that the A2143G may be a very useful CLR^r indicator for selection of optimal therapy against H. pylori infection in China.

In 41 of the 65 A2143G-positive cases (63.1%), both the BsaI-sensitive and -insensitive 23S rRNA PCR products was observed. Because the PCR products of the A2143G-positive control (CLR^r-1) were digested by BsaI enzyme completely in each experiment, partial restriction could be excluded. Thus, multi-strain H. pylori infection likely happened among these cases. Sequencing results of 12 representative samples supported the deduction (Fig. 2B). Isolation of the sensitive and resistant strains from cultures might provide a direct evidence to show the mixed colonization. Because the experimental conditions at the local clinics in Linqu County was not adequate to support primary H. pylori culture, isolation of H. pylori from biopsies of stomach was not carried out in the clinical trial. The high H. pylori infection rate in the local area might contribute to the high co-infection rate of the CLR sensitive and resistant H. pylori strains.

In addition, two new MboII-RFLP patterns were discovered in 17 subjects in the present study. Sequencing showed that 6 of the 17 cases had a sequence change of AAGGG → CTTCA (2222–2226) and that the remaining 11 cases had a sequence change of AACC → GAAG (2081–2084). The A2143G mutation was detected only in 3 of the 15 GAAG-positive cases (20%). Both the AAGGG → CTTCA (2222–2226) and AACC → GAAG (2081–2084) alterations are located around the conserved domain V of 23S rRNA (2126–2712, equal to Escherichia coli numbering 2042 to 2628), a region closely related to resistance formation of macrolide antibiotics including CLR 2324. It was reported that the point mutation A2224G might not contribute to the CLR resistance among Japanese 27. In the present study, we observed that the presence of the AAGGG → CTTCA (2222–2226) and AACC → GAAG (2081–2084) sequence alterations in OAC group were significantly higher than those in the OA and placebo groups (Table 1). The A2143G mutation was detected only in 2 of the 11 AACC → GAAG (2081–2084) cases (18.2%). Whether these alterations contribute to CLR resistance should be investigated further.

The primary prevalence of the CLR^r-related A2143G mutation within the 23S rRNA gene as determined by a novel PCR-RFLP assay is approximately 10~14% among Chinese subjects without CLR use history. Application of CLR-containing regimen for H. pylori eradication could increase risk of prevalence of the A2143G mutation significantly (OR = 3~4).

ZL carried out experiments including DNA extraction, detection of 23S rRNA and point mutations by PCR-RFLP. JS setup the PCR-RFLP assay. LZ carried out the ELISA and ¹³C-UBT assays. LS collected the gastric juice samples by endoscope. QL validated the quality of DNA from the samples. BZ, JZ, and LG pretreated the samples in the local clinic. GF performed the statistical analysis. JM collected patients' followup data. W-CY conceived the idea of the intervention trial and advised on the manuscript. DD conceived the idea of the present study, analyzed the data, and wrote the manuscript.