Furazolidone- and Nitrofurantoin-ResistantHelicobacter pylori: Prevalence and Role of Genes Involved in Metronidazole Resistance

H. pylori infection is one of the most common infections worldwide and is etiologically related to chronic gastritis, duodenal ulcer, gastric ulcer, gastric adenocarcinoma, and primary gastric lymphoma (3, 17, 18). Approximately one in six H. pylori-infected persons develop peptic ulcer disease. Patients with peptic ulcer disease experience pain and reduced quality of life and risk ulcer complications (1,19). Cure of H. pylori infection prevents ulcer recurrence, heals gastritis, and also prevents progression from mild superficial gastritis to chronic atrophic gastritis (the precursor lesion to gastric cancer), which may reduce the risk, or prevent, gastric cancer (1, 15). Clinical experience has demonstrated that cure of H. pylori infection is difficult due to lack of compliance with the drug regimens and development of antibiotic-resistant H. pylori (8, 15). Metronidazole has been widely used as a critical component of combination therapies for H. pylori infection. Monotherapy with metronidazole results in more than 50% of H. pyloriisolates acquiring resistance (13), and current metronidazole-containing triple therapies are being undermined by development of resistance (4). Because of the high rate of metronidazole resistance in H. pylori, furazolidone and nitrofurantoin have been recommended as alternative agents (2, 9,21).

Although furazolidone, nitrofurantoin, and metronidazole are classified as nitroheterocyclic and nitroaromatic compounds, it is not known whether the drug actions and the resistance mechanisms are similar. Metronidazole resistance among H. pylori strains has been reported worldwide with variable frequencies (6, 13, 16). The mechanism of metronidazole resistance among H. pyloristrains has been related to alterations in gene products having metronidazole nitroreductase activity, including oxygen-insensitive NAD(P)H nitroreductase (RdxA), NAD(P)H flavin oxidoreductase (FrxA), and ferredoxin-like protein (FdxB) (7, 12). This study asked whether furazolidone or nitrofurantoin susceptibility and resistance shared common features with metronidazole susceptibility and resistance.

We evaluated the prevalence of metronidazole, furazolidone, and nitrofurantoin resistance among H. pylori strains isolated from 431 patients, including 297 males and 134 females (median age, 45 years; range, 16 to 82 years) presenting at Guro Hospital in Seoul, Korea, between September 1993 and September 1999. The endoscopic diagnoses were chronic gastritis (101 patients), peptic ulcer diseases (85 patients with gastric ulcer disease and 128 patients with duodenal ulcer disease), and gastric cancer (117 patients). H. pyloriwas isolated from gastric mucosal biopsies that were plated onto selective (containing 1% nalidixic acid, 0.5% trimethoprim, 0.3% vancomycin, and 0.2% amphotericin) and nonselective brain heart infusion (BHI) agar plates containing 5% defibrinated horse blood. All plates were incubated under microaerobic conditions at 37°C for up to 14 days. H. pylori was identified by colony morphology, Gram staining, and catalase, urease, and oxidase reactions. MIC measurements were performed by agar dilution as described previously (12). Only 48% of the strains were susceptible to metronidazole. The MICs for 224 resistant strains ranged from ≥8 to 256 μg of metronidazole per ml; those for 172 strains (40% of the total) ranged from ≥16 to 256 μg of metronidazole per ml. Only 7 of the 431 strains (1.6%) required a MIC of 4 μg of furazolidone per ml. The same seven strains also required a MIC of 4 μg of nitrofurantoin per ml, and all seven strains with furazolidone and nitrofurantoin resistance also showed low-level metronidazole resistance (MICs of 8 or 16 μg/ml) (Table1). All other strains (424 of 431) required MICs of ≤0.5 μg of furazolidone or nitrofurantoin per ml. Similar findings (i.e., low frequency with low-level MICs and cross-resistance of furazolidone and metronidazole) have been recently reported for clinical H. pylori isolates from Brazil (14).

We evaluated the rate of spontaneous development of low-level furazolidone or nitrofurantoin resistance (MIC of 4 μg/ml) usingH. pylori ATCC 700392. Two-day-old H. pyloricells from two 89-mm-diameter plates were harvested and suspended in 4 ml of 5% horse serum–BHI broth. The cells were adjusted to approximately 10⁹ cells and spread onto 5% horse blood–BHI agar plates supplemented with 2 μg of furazolidone or nitrofurantoin per ml or 8 μg of metronidazole per ml. The plates containing the cells were incubated at 37°C under microaerobic conditions for 6 days. No colonies grew on plates supplemented with furazolidone or nitrofurantoin. However, 253 colonies appeared on plates supplemented with metronidazole. These results suggest that furazolidone and nitrofurantoin resistance occurs spontaneously at a much lower frequency than metronidazole resistance. The ease of selection of spontaneous metronidazole-resistant colonies and the difficulty of selecting spontaneous furazolidone- or nitrofurantoin-resistant colonies could be related to clinical observations of high frequencies of metronidazole resistance but low furazolidone or nitrofurantoin resistance among H. pyloristrains.

It was already established that the resistance mechanism for metronidazole is different from that for furazolidone and nitrofurantoin (7, 12, 20). To confirm that the resistance mechanisms for metronidazole and for furazolidone and nitrofurantoin are indeed different, susceptibilities to these drugs were measured inrdxA-, frxA-, and fdxB-inactivated (knockout) H. pylori strains. We used 92 single and/or dual knockout H. pylori strains, including a positive control (furazolidone- and nitrofurantoin-resistant clinical isolate requiring a MIC of 4 μg/ml, as mentioned above) and a negative control (H. pylori ureB knockout strain) to measure MICs of furazolidone and nitrofurantoin. The 92 strains were isolated in the United States and were also used to measure MICs of metronidazole as reported previously (12). None of the parental, knockout, or negative control strains grew on 5% horse blood–BHI plates containing 2 μg of either furazolidone or nitrofurantoin per ml, while the positive controls grew on the same plates. These results were confirmed using five additional clinical isolates from Korea, which are susceptible to furazolidone and nitrofurantoin (MIC of 0.5 μg/ml), including H. pylori ATCC 43629, in which rdxA,frxA, and fdxB were knocked out as described previously (12). The MICs of metronidazole, furazolidone, and nitrofurantoin in these strains were evaluated as described previously (12). As shown in Table2, the metronidazole MICs forrdxA and frxA knockout strains increased from 1 to 32 μg/ml (ATCC 43629, KH220, and KH278) and from 8 to 128 μg/ml (KH259 and KH261). The metronidazole MICs for fdxB knockout strains, however, were unchanged for strains ATCC 43629, KH220, KH278 but increased from 8 to 64 μg/ml for KH259 and from 8 to 32 μg/ml for KH261. In contrast, the furazolidone and nitrofurantoin MICs for all the knockout strains were unchanged (0.5 μg/ml) from those for the parental strains. These results are consistent with those from the 92 knockout strains and confirm the previous data regarding the involvement of these genes in metronidazole resistance. In addition, these results indicate that the genes (rdxA,frxA, and fdxB) are unlikely to be directly involved in furazolidone and nitrofurantoin resistance in H. pylori.

Although the modes of drug action of all three antibiotics are similar and nitroreduction is required for activation, the mechanisms of resistance to metronidazole and to furazolidone and nitrofurantoin may not be the same (20). Nitroreduction of furazolidone and nitrofurantoin may be exerted by nitroreductases other than RdxA, FrxA, and FdxB in H. pylori as described by Whiteway et al. (20) for E. coli, in which nitroreduction of 5-nitrofuran derivatives is exerted by the nsfA andnfsB products. It is possible that the nitroreductase for furazolidone and nitrofurantoin may be essential for H. pylori survival and that resistance may be acquired by partial inactivation of the nitroreductase. Candidates for furazolidone and nitrofurantoin nitroreductases include pyruvate:flavodoxin oxidoreductase (PorCDAB) and 2-oxoglutarate oxidoreductase (OorDABC). The lethal effect and possible metronidazole nitroreductase activity of the porCDAB and oorDABC products have been shown for H. pylori (10, 11). In this view, we postulate that furazolidone and nitrofurantoin may be more specific to PorCDAB and/or OorDABC, while metronidazole may be more specific to RdxA, FrxA, and FdxB, as substrate molecules. Therefore, strains with knocked-out rdxA, frxA, or fdxB were still sensitive to furazolidone and nitrofurantoin because of the existence of fully functional porCDAB and oorDABCgenes. In contrast, partially functional porCDAB andoorDABC genes that confer low-level furazolidone and nitrofurantoin resistance also confer low-level metronidazole resistance, as shown in this study. We are currently testing this hypothesis by purifying PorCDAB and OorDABC from H. pyloriand measuring the nitroreductase activities for furazolidone and nitrofurantoin.

Furazolidone- or nitrofurantoin-containing therapeutic regimens forH. pylori infection have been suggested for use instead of metronidazole to overcome the high frequency of metronidazole resistance among H. pylori strains. The ideal antibiotic forH. pylori therapy would have high efficacy without the development of antibiotic resistance. Several reports published recently indicate that furazolidone or nitrofurantoin is efficacious inH. pylori therapy, although details of the best therapy have not yet been defined (2, 9, 21). The results reported here suggest that furazolidone and nitrofurantoin may be good alternatives, especially in areas where metronidazole resistance is common.

• Copyright © 2001 American Society for Microbiology