INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

In the last 15 years, Helicobacter pylori infection (3, 26) has emerged as an important widespread human gastric bacterial infectious disease. Infection is acquired early in life (7, 15, 44), and although it is frequently asymptomatic, it may produce clinically apparent gastrointestinal discomfort (50). H. pylori is the cause of type B gastritis (50) and most instances of ulcer disease (25, 33, 35, 40) and is an important cofactor in gastric cancers (2, 5, 6, 11, 19, 36, 39, 41, 42).

Effective antimicrobial therapy for this infection presents researchers with a number of challenges (37). There is a consensus that symptomatic individuals should be treated for infection so that the more serious complications of disease can be forestalled (35). In these patients, therapeutic compliance has emerged as the most important proximate reason for treatment failure (13). Treatment of symptomatic patients in the absence of a definitive diagnosis results in unnecessary therapy for microbe-negative disorders such as nonulcer dyspepsia, heightens the risk of antimicrobial resistance (20, 34, 48), and cannot be economically justified in many instances (45).

H. pylori is susceptible to a number of different antimicrobials in vitro (14, 32) but not in vivo. Local (gastric) therapy is complicated by the short transit time of oral drugs and failure to maintain therapeutic drug levels in the stomach when drugs are administered parenterally. Monotherapy is largely ineffective (16, 17, 38, 47). Combination therapies consisting of a bismuth salt, an acid-suppressive agent, and/or one or more broad-spectrum antibiotics (12, 16, 27, 28, 38, 49) have success rates of at least 70% but have the disadvantages of complex drug interactions, unwanted side effects with one or more components, development of antimicrobial resistance (14, 20, 32, 34, 48), and compliance with a lengthy and cumbersome regimen which may last several weeks and require multiple daily dosings (12, 13, 27, 37).

Substantial progress in the understanding of the mechanisms of disease, bacterial virulence factors, and host determinants of disease have been made through parallel study of similar helicobacter infections in laboratory animals. Of these, H. pylori-infected gnotobiotic piglets have been shown to accurately replicate many of the features of the acute disease in humans (1, 21-24). Oral inoculation results in persistent asymptomatic colonization. The infection is multifocally distributed chiefly in the cardia and antrum. Microbes localize within the gastric mucous layer but may also adhere to epithelia; bacterial levels averaging 10⁷ CFU/g of gastric mucosa are consistently obtained. Infection persists indefinitely (tested 90 days after infection and 45 days after conventionalization), in spite of strong local and systemic immunity (8, 10). Gastric lesions are characteristic and consist of diffuse to follicular lymphoplasmacytic inflammation, foci of epithelial swelling, vacuolation, necrosis, and micro- and macroulceration (1, 8, 23, 24). Chronic active (neutrophilic) inflammation occurs in piglets but is dependent upon the strain of microbe (8) and preexisting immunity (11).

The infected gnotobiotic piglet is useful for the in vivo delineation of bacterial virulence factors (8, 9). In addition, overall similarities to humans in diet (omnivores) and gastric physiology make the piglet an attractive nonprimate animal for preclinical testing of antimicrobials and other novel therapeutics. The defined microbial status of gnotobiotes and their environment eliminates the confounding effects of commensal microbes or their products upon antimicrobial therapies. Thus, an antimicrobial effect can be identified under the most optimal in vivo conditions. In this paper, the effects of therapy with a number of different antimicrobials upon the manifestations of H. pylori infection in piglets and the use of the H. pylori-infected piglet as an in vivo method for preclinical evaluations of antimicrobial therapy for this disease are described.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Animals and inoculation. Procedures for derivation and maintenance of gnotobiotic piglets within sterile isolation units are described elsewhere (21). Piglets were fed a 50:50 (vol/vol) mixture of Similac with iron and sow milk replacement diet (Esbilac) three times daily. Size and nutritional requirements limit the duration of experiments with gnotobiotic piglets to 45 days or less. Isolation units were screened for microbial contaminants immediately after derivation, at least once during the experiment, and at the termination of the experiment.

Microbiology. Two- and 3-day-old piglets were orally inoculated with porcine-passaged H. pylori 26695 propagated to the logarithmic phase of growth in brucella broth (2.8% [wt/vol] Bacto; Difco, Detroit, Mich.) and supplemented with 10% (vol/vol) fetal calf serum as described previously (9, 10, 21, 23). For quantitative determinations of gastric bacteria, a 10% (wt/vol) homogenate of gastric mucosa in brucella broth was prepared, and duplicate 10-fold dilutions were plated onto blood agar or on chocolate agar containing trimethoprim, vancomycin, amphotericin B, and polymyxin B (Remel, Lenexa, Kans.). Plates were incubated at 37°C with 95% humidity under microaerobic (5% oxygen) conditions for 4 days prior to evaluation. Isolates were confirmed to be H. pylori by morphology, Gram staining, and the presence of catalase, oxidase, and urease enzyme activities.

MICs. The MICs of test agents for H. pylori 26695 and gastric reisolates were determined by an agar dilution method (32). Mueller-Hinton agar supplemented with 1% IsoVitaleX and 5% chocolatized sheep blood was used. The plates were incubated for 48 h at 37°C in an atmosphere of 85% nitrogen, 10% CO₂, and 5% O₂ prior to determination of bacterial growth (32).

Serologic evaluation. In some experiments, 10-fold dilutions of heat-inactivated sera were tested for isotype-specific antibodies to H. pylori by an enzyme-linked immunosorbent assay as described previously (9, 21).

Pathologic evaluation. Gross determinations of gastric inflammation (submucosal lymphoid follicles), mucus depletion, submucosal edema, and ulcers or erosions (if present) were made on fresh gastric tissue at necropsy. Histopathologic examination was performed with sections of formalin-fixed tissues selected from the gastric cardia, fundus, and antrum and the pyloric antrum including the torus pyloricus. Replicate sets of sections were stained with hematoxylin and eosin and the Steiner's modification of the Warthin-Starry stain for demonstration of organisms. Microscopic inflammatory lesions were scored with a semiquantitative system in which each section was assigned a numerical score of 0 if inflammatory cells were absent and the tissues were no different than the uninfected control gastric tissues, 1 if there was minimal inflammation consisting of focal collections of mononuclear cells in the lamina propria, 2 if there was moderate inflammation consisting of focal and diffuse infiltrates of mononuclear cells into the gastric lamina propria with occasional lymphoid follicles, and 3 if there was severe inflammation consisting of prominent diffuse mononuclear inflammatory cell infiltrates into the gastric lamina propria along with numerous lymphoid follicles. Epithelial changes consisted of acute cellular swelling, cytoplasmic vacuolation, and necrosis. These were variably present in all grades of severity of inflammatory lesions but were most regularly observed with the more severe (grade 3) inflammatory lesions.

Antimicrobial agents. Test antimicrobial agents were selected on the basis of published information on efficacy or lack of efficacy for the eradication of H. pylori in human clinical trials. Piglets were treated orally alone or in combination with the following antimicrobials: amoxicillin (Amoxil; oral suspension; Smith-Kline Beecham, Philadelphia, Pa.), bismuth subsalicylate (Pepto-Bismol; The Procter and Gamble Co., Cincinnati, Ohio), ciprofloxacin (Cipro IV; Miles Laboratories, West Haven, Conn.), clarithromycin (Biaxin; Abbott Laboratories, Chicago, Ill.), erythromycin ethylsuccinate (Barre-National, West Point, Pa.), metronidazole (Flagyl; Schiapparelli-Searle, Chicago, Ill.), nitrofurantoin (Furadantin; The Procter and Gamble Co., Cincinnati, Ohio), and tetracycline (Sumycin; E. R. Squibb & Sons, Princeton, N.J.). Dose and frequency varied, and these are listed in Tables 1 and 2. In addition, omeprazole (Prilosec; Merck, West Point, Pa.) and ranitidine (Zantac; Glaxo, Research Triangle Park, N.C.) were administered alone or in combination with some of the drugs. For omeprazole, beads were removed from the 10-mg capsules and were directly administered with water. For all tested antimicrobials, the dose, frequency, and duration of treatment were adjusted to be equivalent to or lower than those recommended for humans (dosage in milligrams per kilogram of body weight per day) by using the calculated average weight (1,700 g) of piglets during the treatment interval. Five piglets were treated with amoxicillin intraperitoneally twice daily (BID) for 7 days for comparison of the results with those obtained an equivalent orally administered dose of drug.

Experimental design. Two different designs were used to evaluate the antimicrobial efficacies of the test agents. Experiments were performed by litter, whose number varied from 8 to 15 piglets each. A litter was divided into two to four groups of treated piglets. For the experiments with each litter, a group of infected piglets which received either drug vehicle or brucella broth alone were included as infection controls. In the clearance design, piglets were inoculated with H. pylori at 2 to 3 days of age. Seven or 10 days later (9 to 13 days of age), treatment was begun and was continued daily for the next 7 days (16 to 20 days of age). The piglets were killed on the morning after the last treatment, and the effects of therapy upon gross and microscopic lesions and bacterial colonization were determined in gastric mucosal homogenates by a standard plate counting technique.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Drug efficacy. As determined by a combination of serology, pathology, and microbial culture at death, all inoculated pigs became colonized with H. pylori; infection was clinically asymptomatic. The results of treatment with various combination or single agent antimicrobials in both the clearance and the eradication of H. pylori from the infected piglets by using doses and frequencies of administration roughly equivalent to those recommended for humans are given in Table 1. A triple-therapy regimen (bismuth subsalicyclate, metronidazole, and amoxicillin), dual therapy with amoxicillin and omeprazole, and monotherapy with amoxicillin, metronidazole, clarithromycin, or ciprofloxacin cleared and eradicated H. pylori from the stomachs of all treated piglets. In contrast, ranitidine and omeprazole were ineffective at either the clearance or the eradication of H. pylori. Monotherapy with bismuth subsalicylate or erythromycin cleared and eradicated the infection from some piglets; monotherapy with nitrofurantoin or tetracycline resulted in clearance but not eradication. When the infection was not eradicated, colonization levels in treated piglets upon termination of the study were within the range of those for the positive control piglets (10⁶ to 10⁷ bacterial CFU/g).

Pathologic and serologic correlates of clearance and eradication. To determine if the pattern and severity of gastric inflammation correlated with microbial status after treatment, the histopathologic scores at each gastric sampling site of the piglets successfully treated with triple therapy or unsuccessfully with 5.67 mg of bismuth subsalicyclate per kg alone were compared to those for the untreated infected piglets (Table 4). Statistically significant differences between treatment groups by either sample site or total inflammatory score were not observed (P > 0.60). When these same scores were compared by microbial status as either infected (present) or eradicated (absent) at death, regardless of treatment, the mean total inflammatory score for culture-negative piglets was lower than that for the culture-positive piglets, but the difference was also not statistically significant (P > 0.30). In a similar fashion, mean isotype-specific antibody responses in the sera at death were compared between groups. A marginally significant (P < 0.10) reduction in the serum immunoglobulin M (IgM) level was detected in triple therapy-treated piglets, whereas antibody levels in bismuth subsalicylate-treated piglets did not differ significantly from those in control piglets. When these data were arranged by infection status, the trend of diminished IgM levels in successfully treated piglets (P < 0.05) was confirmed.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In the present study, a number of commonly used therapeutic agents were tested for antimicrobial efficacy in H. pylori-infected gnotobiotic piglets. As a positive control, the efficacy of triple therapy consisting of bismuth subsalicyclate, amoxicillin, and metronidazole was used as the "gold standard" of efficacy (12, 27). The oral route of drug delivery was used except in one instance (amoxicillin was given intraperitoneally), and the doses and frequencies of administration were chosen to closely replicate those used in humans. All drugs were well tolerated by the piglets, and the piglets did not have any adverse clinical effects.

Two basic experimental designs were used. The clearance design was most useful for the rapid assessment of antimicrobial efficacy. Therapeutic efficacy was most readily determined by a combination of the eradication design and in vivo dose titration studies. The eradication design contained a 2-week posttreatment observation interval in order to more closely reproduce the pattern of treatment in clinically affected human patients. Qualitative assessment of clearance after treatment was determined by gastric lavage, in which organisms in gastric washes were detected by culture and as a result of their urease activity. While it cannot be assumed that all instances of relapse of infection after treatment would occur within this 10- to 14-day observation interval, several cases of suppression of infection followed by relapse were observed (Table 1).

In piglets, the efficacies of the combined therapy with omeprazole and amoxicillin and also the triple therapy with bismuth subsalicyclate, amoxicillin, and metronidazole are likely attributable to the antimicrobial effects of the amoxicillin component since the dose of the compound as a single agent (5.6 mg/kg) that eradicated H. pylori was approximately 10-fold higher than the doses eventually found to be capable of eradicating H. pylori from swine.

For many monotherapies, the minimal effective dose in piglets was substantially lower than those reported to be effective in humans. The most striking example of the discrepancy between humans and piglets is the efficacy data for amoxicillin. In humans, 500 to 1,000 mg of amoxicillin per kg per day is needed to clear H. pylori from the stomach even temporarily (4, 29). In piglets, the minimally effective oral dose range of 1.0 to 0.1 mg/kg is 1,000- to 10,000-fold less, and yet, a high dose delivered parenterally was ineffective. The latter can be attributed to the fact that the stomach may not possess a mechanism for the efficient translocation of drug from the blood in the vascular system to the mucosa (49). The impressive oral efficacy of amoxicillin was unexpected, and for this reason, titration experiments were conducted to identify the minimally effective doses of amoxicillin and other antimicrobials in piglets. The efficacy of a low dose of amoxicillin may be explained by the fact that there are no competing microbes in gnotobiotes or perhaps by the fact that a component of the normal diet binds, inactivates, or otherwise interferes with the biologic activity of amoxicillin. It is also possible that the total titratable amount (moles) of acid in neonatal porcine gastric juice (versus gastric pH) is substantially less than the moles of acid present in a comparable volume of either adult porcine or human gastric juice. If piglet gastric juice is relatively deficient in moles of gastric acid, then the bactericidal effect of amoxicillin (4, 30) would be magnified such that unexpectedly low doses of amoxicillin exhibit antimicrobial activity in vivo.

Monotherapy with metronidazole, clarithromycin, or ciprofloxacin cleared and eradicated infection from the piglets. Again, it appears that the effective dose for these agents is lower than those ordinarily recommended for humans. For these drugs, their success as monotherapies in piglets was unexpected on the basis of comparable data for humans (20, 31). Similarly, as occurs in humans (28, 47), bismuth subsalicylate monotherapy was moderately effective at bacterial clearance (five of seven piglets were culture negative after therapy) but ineffective at eradication (only two of eight piglets remained culture negative 2 weeks after the cessation of therapy). Thus, for these drugs, piglets are more sensitive indicators of clearance and eradication patterns than humans.

Erythromycin, tetracycline, and nitrofurantoin were ineffective in vivo, in spite of their documented antimicrobial effects in vitro. The explanation for this phenomenon is unknown, but this finding is characteristic of H. pylori. Differential susceptibility may be related to the level of metabolic activity of the microbe within the gastric microenvironment. Organisms in the plateau phase of growth are relatively resistant to the bactericidal effects of commonly used antimicrobials (31). It is possible that the microbes within the gastric niche are physiologically inactive and are thus relatively resistant to bactericidal agents. Alternatively, these antimicrobials might not reach the gastric mucosa in sufficient quantities or may not remain there for a sufficient length of time to exert effective antibacterial activity in vivo (49).

In general, nonmicrobial antemortem indicators of infection were weakly correlated with posttreatment antimicrobial status, a finding likely attributable to the brief posttreatment observation interval of 2 weeks or less. However, as a group, infected piglets could be distinguished from successfully treated piglets by determination of the levels of antibodies of the IgM isotype class in serum by enzyme-linked immunosorbent assay. This is consistent with our previous work which showed that explants prepared from the gastric mucosa of infected piglets rapidly cease local immunoglobulin production (24) following the successful eradication of microbes, even though serum antibody levels remain elevated. It is likely that both IgG and IgA antibody levels would have decreased to undetectable levels over time if the posttreatment observation interval were longer. Like antibody responses, local manifestations of gastric inflammation also regress after successful antimicrobial therapy, although the semiquantitative scores were not statistically significantly reduced in this study. Thus, piglets are similar to humans, in that neutrophilic infiltrates disappear by 6 weeks or earlier after the cessation of treatment; lymphocytic infiltrates also regress but may take several months to return to the normal levels in the gastric mucosa (50). In piglets, microbial culture and reisolation is a more reliable method than either serology or histopathology for determining efficacy.

In humans, the development of microbial resistance after failed therapy or infection with resistant microbes in the absence of a known previous exposure to them is a known complication of therapy with clarithromycin and metronidazole (14, 20). It is estimated that 7% of H. pylori isolates are resistant to clarithromycin (20) and that up to 54% are metronidazole resistant (34). Strains of H. pylori resistant to ciprofloxacin, metronidazole, and erythromycin can be produced in vitro by rapid passage in the presence of these antibiotics (14). Our strain (strain 26695) was uniformly susceptible to all antimicrobials in vitro. Only in metronidazole-treated piglets did resistant strains develop; these were recovered and resistance was confirmed by the substantially elevated MICs determined in antibiotic sensitivity tests performed with the reisolated organisms. Here, the gnotobiotic conditions of the experiment exclude the possibility that the recovered strains acquired resistance by plasmids or other genetic exchange mechanisms which may occur between colonizing microbes and commensal organisms. It thus appears that for this drug at least, the development of resistance in vivo is an inducible event.

Drugs which inhibit gastric acid production either as H-2 antagonists or as proton pump inhibitors were uniformly unsuccessful in clearing or eradicating H. pylori from infected piglets. These findings reflect similar data for humans (18, 43, 46, 47) and indicate that these drugs as monotherapies for the clinical relief of gastric hyperacidity or gastritis exert their beneficial effects without significantly reducing the levels of bacterial colonization in the stomach. As indicated above, the apparent efficacy of combined therapy with omeprazole and amoxicillin in piglets and perhaps humans is likely attributable to the antimicrobial effects of amoxicillin alone.

In infected humans, patient compliance is a major risk factor for therapeutic efficacy (13, 37). In gnotobiotic piglets, compliance can be rigorously controlled. A prominent advantage of piglets as a preclinical treatment trial model is the fact that piglets appear to be remarkably sensitive to antimicrobial therapies for this infection. Because of this, it is reasonable to conclude that if a novel antimicrobial agent or therapeutic approach does not affect colonization levels in piglets, the likelihood of success as a monotherapy when applied to humans is low. Like humans, serologic and histopathologic correlates of infection appear to resolve following eradication of the infection. Gastric physiology and anatomy in piglets also closely resemble those in humans. Thus, the infected gnotobiotic piglet represents an excellent screening model for eliminating ineffective therapies and can be used to rank order the efficacies of single agents by differentiating those which are more effective (e.g., amoxicillin and metronidazole) from those which are less effective (e.g., ciprofloxacin and clarithromycin) and ineffective (e.g., nitrofurantoin and ranitidine). Finally, the model has the potential to predict issues associated with the in vivo development of antimicrobial resistance and possibly unwanted interactions between drugs in combination therapies in an environment devoid of microbial competition or other outside influences. Thus, the porcine model of infection with H. pylori is useful for evaluating the efficacies of test agents for therapy of infection in humans.