CLINICAL SAMPLES

Gastric biopsy samples from 61 patients who had not receivedH pylori eradication treatment were obtained at the time of endoscopy. Forty seven were classified asH pylori positive on the basis of culture, and 14 were found to be H pylori negative and were used as negative controls in further experiments. The patients were successively included until the number with resistant strains reached 33. However, for the cases that turned out to beH pylori negative as well as for those where an H pylori strain susceptible to clarithromycin was isolated, only the first 14 in each category were included. The H pylori positive gastric biopsy specimens were from 20 patients with duodenal ulcer and 24 with non-ulcer dyspepsia. No clinical data could be obtained for the three remaining patients.

BACTERIAL STRAINS AND CULTURE

H pylori was cultured from gastric biopsy samples ground for 2–3 seconds with an electric tissue homogeniser (Ultraturax; LaboModerne, Paris, France) before inoculation on to a selective in house medium, pylori agar (bioMérieux, Marcy l’Etoile, France), and a non-selective Columbia medium (Oxoid; Unipath, Basingstoke, Hants, UK) enriched with 10% human blood. The plates were incubated under microaerobic conditions at 37°C for 12 days. A subculture of H pylori strains was performed under the same conditions for two to three days.

DETERMINATION OF MINIMAL INHIBITION CONCENTRATION (MIC)

The MICs of clarithromycin on H pyloristrains were determined by Etest (AB Biodisks, Solna, Sweden) and by an agar dilution method, as previously described.6 The plates were incubated in a microaerobic atmosphere (jars with Gaspack H₂ + CO₂ without catalyst (Oxoid)) at 37°C for two days. A strain was considered resistant when the MIC was higher than 2 mg/l.

DNA EXTRACTION

The biopsy samples were ground as described above and centrifuged for five minutes at 10 000 g. The pellet was resuspended in 300 μl extraction buffer (20 mM Tris/HCl, pH 8, 0.5% Tween 20), and proteinase K was added at a final concentration of 0.5 mg/ml. The mixture was incubated at 56°C for one hour. Finally, the enzyme was inactivated by boiling for 10 minutes.

AMPLIFICATION OF THE HP23S FRAGMENT

The Hp23S fragment was obtained by PCR amplification ofH pylori extracted DNA with primers Hp23–1 (5’-CCACAGCGATGTGG TCTCAG-3’) and Hp23–2 (5’-CTCCAT AAGAGCCAAAGCCC-3’), which are complementary from nucleotide 1820 to 1839 and from nucleotide 2244 to 2225 respectively, as previously described,6 ,10 except for the amount ofTaq polymerase (2 units) (Eurobio, Les Ulis, France) and the concentration of primers (1 μM) (Eurogentec, Seraing, Belgium). This fragment corresponds to the peptidyltransferase region of the 23S rRNA gene. The amplified 425 bp fragment was visualised after electrophoresis on a 1.5% agarose gel stained with ethidium bromide.

RESTRICTION FRAGMENT LENGTH POLYMORPHISM (RFLP) OF THE HP23S FRAGMENT

As previously described,6 ,10 digestion of the Hp23S fragment with the restriction enzymes BbsI and BsaI permitted the discrimination between two kinds of mutant genotype (A2143G and A2144G respectively). Digestions were performed according to the instructions of the manufacturer (New England Biolabs Inc, Beverly, Massachusetts, USA). Restriction products were analysed on a 2% agarose gel.

SEQUENCING OF AMPLIFIED PRODUCTS

Before sequencing, the PCR products were purified by using the Wizard PCR preps (Promega, Madison, Wisconsin, USA). The sequencing reaction was performed as described by Sanger et al 11 using a Dye Terminator Ready Reaction kit as indicated by the manufacturer (Perkin Elmer, Applied Biosystem Division, Foster City, California, USA). The same primers as were used for PCR were used for sequencing. The sequencing reaction took place in a Perkin Elmer 9600 thermal cycler. Twenty five cycles were applied, each of 10 seconds at 96°C, five seconds at 50°C, and four minutes at 60°C. Sequencing was performed on the two strands of each PCR product with the automated DNA sequencer ABI Prism 377 (Perkin Elmer).

DNA ENZYME IMMUNOASSAY (DEIA)

This colorimetric hybridisation assay was performed with the GEN-ETI-K DEIA kit as previously described.10 ,12 Four different 5’ biotinylated probes were designed to be specific for each genotype (wild type and the three mutants A2143C, A2143G, A2144G). Optimal hybridisation conditions were previously determined.10 Data presented in this study are the means of three independent experiments.

DETERMINATION OF CUT OFF VALUE

Cut off values for the probes wild type, 2143G, and 2144G were determined by the receiving operating characteristics curve construction method13 as previously described.10 ,12 Briefly, different virtual cut off values between 0 and 4 (OD) were assigned, and for each cut off value, the numbers of true positives, true negatives, false positives, and false negatives were calculated in comparison with the standard method. Thus each cut off gives a sensitivity and a specificity value. These data are then used to determine the cut off value that provides both the best sensitivity and the best specificity.

Owing to the low number of positive cases for the 2143C probe, the cut off value was arbitrarily defined as the mean of the values corresponding to a negative result by the reference method plus three standard deviation values.

AMPLIFICATION OF THE HP23S FRAGMENT

Amplification with DNA extracted from biopsy samples for whichH pylori culture was positive yielded a fragment of the expected size. In contrast, no fragment was obtained with DNA from H pylori negative biopsy samples (data not shown). The specificity of the amplification with primers Hp23–1 and Hp23–2 has previously been shown.6 ,10 The patients were not selected on the basis of the CLOtest nor heavy growth. There was complete correlation between culture and PCR results. We compared the semiquantitative results of culture (+, ++, +++) with the OD obtained by DEIA for each probe when a positive result was obtained. No correlation was found (data not shown).

The reproducibility of the three OD measurements was satisfactory, with standard deviations of less than 10% of the mean and even less than 4% for probes 2144G and 2143C.

DETERMINATION OF THE CUT OFF VALUES

The receiving operating characteristic curves were constructed for the wild type, 2143G, and 2144G probes. They were close to the ideal curves. The cut off value for the 2143C probe was 0.967 (mean of negative cases 0.746 plus three standard deviations). Table 1 gives the cut off values as well as the sensitivity and specificity for each probe.

DISTRIBUTION OF DIFFERENT 23S rDNA GENOTYPES DIRECTLY DETECTED IN BIOPSY SPECIMENS

The results from the DEIA were analysed on the basis of the cut off values defined above. They were compared with the results obtained from MIC determination, PCR/RFLP, and/or PCR product sequencing performed on the strains isolated from the corresponding biopsies (table 2). The Etest was used as a screening test, and the agar dilution test was the confirmatory test; the results are reported in table 2. Both methods were applied to all strains. None of the 14 biopsy samples for which H pylori culture was negative gave an OD₄₅₀ higher than the cut off values with any of the probes (data not shown).

Of the 33 biopsy specimens from which resistant strains were isolated, 25 allowed the amplification of a 425 bp fragment which hybridised with only one probe: one PCR product gave a positive signal with the probe detecting the A2143C mutation (biopsy sample 1), eight with the 2143G probe (samples 5–12), and 16 with the 2144G probe (samples 17–32). The other eight samples yielded the amplification of a fragment hybridising with more than one probe: 2143C and wild type (samples 2, 3, and 4), 2143G and 2143C (samples 13 and 14), 2143G and wild type (samples 15 and 16), and 2144G and wild type (sample 33). The OD₄₅₀ values were significantly higher than the cut off values needed to consider them as multiple genotypes. Fourteen samples from which clarithromycin susceptible strains were isolated gave a single positive signal with the probe detecting the wild type genotype (samples 34–47).

When the DEIA results were compared with those obtained by the reference methods, complete agreement between the different techniques was found. All the specimens classified as resistant on the basis of MIC gave a positive signal with a probe that detected a resistant genotype (specimens 1–33). Likewise, those containing susceptible strains were found to hybridise with the wild type probe (specimens 34–47). In one case (specimen 4), it should be noted that the MIC determined by Etest was 0.047 mg/l and some resistant mutants were observed. As expected, the MIC determined by the agar dilution method corresponded to a resistant strain.

Concerning the PCR/RFLP, the restriction byBsaI was always associated with hybridisation with the 2144G probe. A similar observation was made for the 2143G probe and the BbsI digestion. When two genotypes were present in the same biopsy specimen, the restriction profiles also correlated. The restriction enzyme assay was not able to discriminate between the A2143C genotype and the wild type genotype, as expected. Nevertheless, when mutant genotypes (A2143G and A2144G) were associated with either the A2143C or the wild type genotype, the restriction enzyme assay showed both genotypes (samples 13–16, and 33). Indeed, PCR/RFLP showed three bands of 331, 94, and 425 bp corresponding to BbsI digestion products and an unrestricted fragment respectively (samples 13 to 16). For sample 33, three fragments were detected corresponding toBsaI restriction products (319 and 106 bp) and an undigested amplicon (425 bp). In the case of specimens 34–47, the result of the DEIA correlated with that of the restriction enzyme assay; indeed, the PCR products were digested by neitherBbsI nor BsaI.

Sequence analysis of PCR products was performed in some samples of each mutant genotype (samples 1–8, 16–21, 32 35, 37, 46, and 47). In these cases, the sequence data confirmed the results provided by the other methods.