Helicobacter felis Eradication Restores Normal Architecture and Inhibits Gastric Cancer Progression in C57BL/6 Mice

Transcription

1 GASTROENTEROLOGY 2005;128: Helicobacter felis Eradication Restores Normal Architecture and Inhibits Gastric Cancer Progression in C57BL/6 Mice XUN CAI,* JANE CARLSON,* CALIN STOICOV,* HANCHEN LI,* TIMOTHY C. WANG,*, and JEANMARIE HOUGHTON* *Division of Gastroenterology, Department of Internal Medicine, University of Massachusetts Medical School, Worcester, Massachusetts; and Division of Digestive and Liver Diseases, Department of Medicine, College of Physicians and Surgeons, Columbia University, New York, New York Background & Aims: The impact of Helicobacter eradication therapy on the progression or regression of gastric lesions is poorly defined. This study examined the effects of eradication therapy on inflammation, atrophy, metaplasia, dysplasia, and cancer progression. Methods: C57BL/6 mice were infected with Helicobacter felis and received bacterial eradication therapy after 2, 6, or 12 months of infection. The gastric mucosa was examined at early, mid, and late intervals after eradication and graded for histology, expression pattern of -catenin and -catenin, and IQGAP1. Results: Eradication of Helicobacter infection after 2 or 6 months of infection led to a regression of inflammation, restoration of parietal cell mass, and reestablishment of normal architecture. Progression to adenocarcinoma was prevented. Bacterial eradication at 1 year was associated with the reappearance of parietal cells, partial regression of inflammation, and restoration of architecture. Hyperplasia scores significantly improved, and dysplasia did not progress. Infected mice developed antral adenocarcinoma and gastric outlet obstruction by 24 months. Only 30% of the mice receiving bacterial eradication therapy at 12 months developed antral carcinoma. Bacterial eradication at any time during the first year of infection prevented death due to gastric outlet obstruction. The expression pattern of -catenin, -catenin, and IQGAP1 varied with cell type and paralleled histologic changes. Conclusions: Inflammation, metaplasia, and dysplasia are reversible with early eradication therapy; progression of dysplasia was arrested with eradication therapy given as late as 1 year and prevented gastric cancer related deaths. The relationship between Helicobacter infection and gastric carcinoma is well known. 1 Bacterial eradication leads to regression of mucosa-associated lymphoid tissue lymphoma 2 ; however, the effects of bacterial eradication therapy on gastric adenocarcinoma are more controversial. Epidemiologically, eradication of Helicobacter pylori is linked to a decrease in the incidence of gastric adenocarcinoma 3,4 and appears to have the most profound effects if eradication therapy is given before the development of precancerous lesions. 5 The effects of eradication therapy are less clear when infection has persisted for long periods or when mucosal changes are severe, 6 suggesting there is a stage when premalignant changes are irreversible and progressive despite bacterial eradication. These studies raise important questions: at what point are mucosal lesions reversible, and what is the point of no return when they are irreversible and/or progressive? Human studies suggest improvement in gastric atrophy and intestinal metaplasia in some patients 7,8 but not in others 9 who have received successful eradication therapy. Because of limitations in human studies, investigation has turned to animal models to further address the role of bacterial eradication on reversibility of premalignant lesions and the effect on progression to gastric adenocarcinoma. Bacterial eradication in the Mongolian gerbil model appears to be effective in preventing gastric cancer if given in the early stage of inflammation 10 ; however, this model is limited in that infection alone does not reliably induce tumors and use of the cocarcinogen N-methyl-N-nitrosourea does not produce tumors in all infected animals. Mouse models of infection recapitulate human disease where infection causes gastric adenocarcinoma without the need for a chemical carcinogen, making the murine model very attractive for further study. Infection of the C57BL/6 strain of mice with SS1, a mouse-adapted H pylori strain, results in chronic colonization and hypertrophy, but these mice do not appear to consistently develop dysplasia and carcinoma. 15 On the other hand, Helicobacter felis infection in the C57BL/6 mouse model reproducibly results in the classic sequence of histologic changes seen in human infection: chronic gastritis, atrophy, metapla- Abbreviations used in this paper: BrdU, bromodeoxyuridine; IHC, immunohistochemistry; LI, labeling index by the American Gastroenterological Association /05/$30.00 doi: /j.gastro

2 1938 CAI ET AL GASTROENTEROLOGY Vol. 128, No. 7 Figure 1. Eradication protocol. (A) Mice were infected and received bacterial eradication therapy at 2, 6, or 12 months or were left infected and killed at set time points after eradication as shown by the arrows. The black bar represents the length of infection, and the white bar represents noninfection. (B) flab polymerase chain reaction was used to confirm infection status, with glyceraldehyde-3-phosphate dehydrogenase as eukaryotic DNA control. All mice were evaluated, and a representative gel is shown. sia, dysplasia, and adenocarcinoma, with adenocarcinoma occurring in 100% of mice by 15 months of infection However, longer-term studies in this model and the effects of eradication on histologic regression or progression of lesions are lacking. The purpose of this study was to establish the long-term natural history of H felis infection (beyond 15 months) and determine the effects of successful bacterial eradication on the progression and/or resolution of histologic changes and to define the expression pattern of 3 potential markers of neoplastic transformation during natural progression of H felis infection and after successful bacterial eradication. Materials and Methods All work was performed at the University of Massachusetts Medical School. Animals Approval was obtained from the Institution Animal Care and Use Committee of the University of Massachusetts Medical School before initiation of the study. Six-week-old male C57BL/6 mice that were viral antibody free, parasitefree, and bacterial pathogen free inclusive of Helicobacter species were purchased from Jackson Laboratories (Bar Harbor, ME), housed in microisolator cages, fed standard chow, allowed free access to water, and acclimated for 2 weeks before entry into the study. Bacterial Culture H felis (strain 49179) was grown as recommended; bacteria were enumerated, 19 diluted with culture medium to colony-forming units/ml, and used immediately in animal experiments. H felis Infection and Eradication Therapy Eight-week-old C57BL/6 mice were infected with H felis ( colony-forming units) by oral gavage 3 times at 2-day intervals. Bacterial eradication was performed at 2, 6, or 12 months with tetracycline HCl (0.5 mg/30-g mouse/day), metronidazole (0.675 mg per 30-g mouse/day), and bismuth subsalicylate (0.185 mg per 30-g mouse/day) dissolved or suspended in a total volume of 500 L sterile water and given by oral gavage daily for 14 days or sham eradicated with 500 L sterile water daily for 14 days. Mice were killed 2, 6, or 12 months after eradication therapy. Uneradicated mice and control/sham-infected mice served as controls for each time point (Figure 1). Necropsy and Histology Mice received 1 mg/kg bromodeoxyuridine (BrdU) exactly 1 hour before they were killed. Strips of gastric tissue along the lesser curvature from the squamocolumnar junction through the pylorus were fixed in 10% neutral buffered formalin or Prefer fixative (Anatech Ltd, Battle Creek, MI) for 4 hours, processed by standard methods, embedded in paraffin, cut into 5- m sections, and stained with H&E and Giemsa stain or processed for immunohistochemistry (IHC). Histology was scored by 2 blinded reviewers using a numeric scale of 1 4 using the MIT scoring criteria as described previously. 20 Mice found moribund or who had lost 15% of their greatest achieved weight were killed; the tissues were evaluated and mice were counted as mortalities. Quantitative Analysis of H felis Colonization by Real-Time Polymerase Chain Reaction A2 2 mm piece of gastric mucosa taken at the fundus/antral border was snap frozen at the time of necropsy. All

3 June 2005 H FELIS ERADICATION AND GASTRIC CANCER RISK 1939 samples were processed together as described previously. 20 Each sample was analyzed in triplicate. Mice were scored as infected if the 225 base pair flab fragment was amplified with 40 cycles and scored as negative if 45 cycles fail to produce a band of appropriate size in the eradicated mouse while producing an appropriate size band in a known infected control. Proliferation Determination Proliferation was assayed using the BrdU in situ detection kit (BD Biosciences Pharmingen, San Diego, CA) according to the manufacturer s protocol. Well-oriented sections were visualized and photographed using the Olympus BX51 microscope and analyzed using Olympus Microsuite 835V (Melville, NY). Sections were scored and reported as BrdU-positive staining cells per 20 pit/glands (labeling index [LI]). Three sections per mouse were scored, and the average number for each animal was used in calculations. IHC Paraffin-embedded tissue was cut into 5- m sections, rehydrated, washed, and endogenous peroxidase inhibited with 3% hydrogen peroxide in methanol for 20 minutes. Subsequently, slides were heated to 89 C for 20 minutes in citrate buffer (0.1 mol/l sodium citrate, ph 6.0) for antigen retrieval. After blocking with 5% normal goat serum, slides were incubated with primary rabbit antibody H,K adenosine triphosphatase (H,K -ATPase) 1:800 dilution (EMD Biosciences Inc, San Diego, CA), IQGAP1, and -catenin or -catenin 1:200 dilution (Santa Cruz Biotechnology, Santa Cruz, CA) at 4 C overnight. Slides were washed with phosphate-buffered saline and incubated with biotinylated antirabbit secondary antibody (Vectastain ABC kit; Vector Laboratories, Burlingame, CA) for 30 minutes at room temperature followed by an additional 30-minute incubation with ABC reagent (Vectastain ABC kit) at room temperature. Color was developed with 3,3=-diaminobenzidine tetrahydrochloride using the DAKO Liquid DAB Substrate-Chromogen System (DAKO Corp, Carpinteria, CA), counterstained with hematoxylin (Fisher Chemicals, Springfield, NJ), and analyzed. Statistical Analysis Pathologic data are compared using the Mann Whitney analysis of nonparametric data and considered significant at a value of P.05. Survival data are shown using the Kaplan Meier curve. Data for the BrdU LI are reported as the mean SD and considered significant if P.05. Results Natural History of Helicobacter Infection The infection and eradication protocols are outlined in Figure 1A. H felis infection was eradicated according to schedule using triple therapy for 14 days. Infection status was assessed in all mice at the time of necropsy by a combination of H&E staining, Giemsa staining, and flab polymerase chain reaction (Figure 1B). Uninfected control mice ranging in age from 8 weeks to 24 months did not differ in stomach/body weight ratios or any histologic parameter evaluated and are therefore discussed together. Under control, noninfected conditions, the mucosa of the fundus (Figure 2A) had rare submucosal and mucosal mononuclear leukocytes (Figure 2A, thick arrow). Chief cells (Figure 2A, thin arrow, darker cell) populate the lower third of the fundic mucosa. Parietal cells, located in the lower two thirds of the mucosa, are larger than chief cells and are seen wedged between the chief cells (Figure 2A, thin arrow, pink cells). Parietal cells are shown at higher power using IHC directed toward the H,K -ATPase (Figure 2C, brown staining). The antrum (Figure 2B) consists of well-ordered glands extending down the entire thickness of the mucosa to abut the muscularis mucosae. There is a lack of inflammatory cells. Early Helicobacter infection. At 4 months of infection (Figure 2D), inflammation was prominent and seen as both submucosal (thin arrow) and intramucosal lymphoplasmacytic infiltrates (thick arrow). There was antralization of glands, with loss of parietal and chief cells (Figure 2D [bar] and F [IHC]). Hypertrophy was prominent, with infected mucosa 50% thicker than the uninfected specimen (compare Figure 2D and A). The antrum shows hypertrophy and metaplasia of glands (Figure 2E [bar]) and mild inflammation in both the mucosa and submucosa (arrow). Neither metaplasia nor dysplasia were prominent at this early time point of infection in either the fundus or the antrum. Midpoint of Helicobacter infection. After 8 months of infection, there was moderate to severe inflammation and hyperplasia, with near complete parietal and chief cell loss (Figure 2G and I). Metaplastic glands are seen as corkscrew-shaped and/or dilated glands, extending through the mucosa, with basally located nuclei and abundant cytoplasm (Figure 2G, arrow), resembling Brunner s glands. 21 Late Helicobacter infection. We defined the group of long-term infected mice as those infected for months. In these mice, inflammation remained severe (Figure 2J [12 months], M [18 months], and P [22 months]). Architectural distortion became increasingly severe, with marked hypertrophy, widespread metaplasia, and a significant increase in dysplasia. From 12 months onward, there was a complete absence of parietal cells in the sections studied by histology (Figure 2J, M, and P) orihc(figure 2L and O). Dysplasia progressed from large intramucosal cystic glands lined with uniform cuboidal cells (low grade) 22 to invasive cystic glands penetrating beneath the muscularis mucosa with marked cellular pleomorphism and piling up of

4 1940 CAI ET AL GASTROENTEROLOGY Vol. 128, No. 7 nuclei (high grade) 22 (Figure 2J, M [thick arrow], and P). Fifty percent of mice examined at 12 months had adenocarcinoma of the fundus. At 18 and 22 months, 100% of mice had adenocarcinoma in the fundus with invasion through the muscle layer to the serosa (Figure 2M). Changes within the antrum lagged behind those of the fundus, became prominent at 12 months, and progressed as time of infection lengthened. There was marked antral hypertrophy, a mucous cell metaplasia similar to that seen in the fundus along with the formation of exophytic polypoid lesions consisting of severely dysplastic glands with loss of epithelial cell nuclear polarity and pronounced nuclear atypia. These lesions progressed in architectural complexity to include cribriform corkscrew

5 June 2005 H FELIS ERADICATION AND GASTRIC CANCER RISK 1941 glands, followed in time by bizarre glands embedded in a desmoplastic stroma with extension of the lesions below the muscularis mucosa (Figure 2K, N, and Q). These antral masses became large (Figure 2R), were found in all mice examined from the 18-month and later infection groups, and caused gastric outlet obstruction. A total of 75% fit the criteria of polypoid adenocarcinomas, defined as polypoid lesions composed of tubular and/or tubulovillous glands penetrating below the muscularis mucosa into the stalk of the polyp. 22 The remaining 25% were deemed adenomas, because they had all of the above criteria except extension through the muscularis mucosa. 22 Histologic changes are summarized in Figure 3. Early H felis Eradication Restores Mucosal Integrity and Prevents Progression to Adenocarcinoma We next examined the impact of eradication on the natural progression of mucosal changes. All infected mice tested positive for H felis by a combination of polymerase chain reaction, detection of organisms by routine histology, and Giemsa staining. There was no evidence of infection in the mice that received eradication therapy. Mice were infected for 2 months, received sham or bacterial eradication therapy, and were examined at early, mid, and late time points after eradication. Within 2 months after successful bacterial eradication, inflammatory infiltrates disappeared completely, parietal cell mass was restored to control levels, and mucosal architecture was reestablished (Figure 4B). These changes were durable, because the mice examined 12 months after successful eradication had architecture indistinguishable from control mice of the same age (Figure 4C and D). There was no progression to carcinoma in situ or invasive carcinoma and no increase in mortality when compared with uninfected control mice. Histology scores are shown in Figure 3B. Eradication Therapy at the Midpoint of Infection Restores Mucosal Architecture and Prevents Progression to Gastric Adenocarcinoma Mice were infected for 6 months (midpoint infection), received sham or bacterial eradication therapy, and were examined at early, mid, and late time points after eradication. Infection in this group resulted in architectural distortion with metaplasia, antralization of glands, and complete loss of parietal and chief cells (Figure 4E). Inflammation, hyperplasia, and dysplasia scores were substantially elevated (Figure 3C) relative to control values. As early as 2 months posteradication, inflammatory infiltrates markedly improved (Figure 4F); the inflammatory scores decreased to less than half of peak levels, returned to uninfected control levels at 6 months, and remained there for the duration of the study (Figure 3C). Hyperplasia and dysplasia scores decreased rapidly, achieved control levels at 2 months posteradication, and remained there for the duration of the study (Figure 3C). Additionally, within 2 months of treatment, parietal cell mass was reestablished (Figure 4F). This reversion to normal architecture, although slightly delayed compared with the earlier group (described previously), was durable to 1 year posteradication (Figure 4G) when parietal cell repopulation was confirmed by IHC (Figure 4H). There was no progression to carcinoma in situ or invasive carcinoma and no increase in mortality when compared with uninfected control mice. 4 Figure 2. Natural progression of H felis infection in the C57BL/6 mouse model. (A C) Sham infected. (A) Fundic mucosal strip taken from the lesser curvature at the squamocolumnar junction. (B) Antral mucosa. (C) IHC for H,K -ATPase. The arrow indicates large parietal with central nuclei and brown staining cytoplasm. (D F) H felis infection for 4 months. (D) Fundic mucosal strip taken from the lesser curvature at the squamocolumnar junction. Moderate to severe submucosal chronic inflammatory infiltrate (thin arrow) and intramucosal chronic inflammatory cells (thick arrows) and antralization of glands (bar) are seen. (E) Antral mucosa with intramucosal inflammatory infiltrate (arrow) and mucous metaplasia of glands (bar). (F) IHC for H,K -ATPase. (G I) H felis infection for 8 months. (G) Fundic submucosal chronic inflammatory infiltrates (thick arrow), intramucosal inflammation, marked mucous cell metaplasia (thin arrow), and complete loss of parietal cells are prominent. (H) Antral mucosa. (I) IHC for H,K -ATPase confirms a complete absence of parietal cells. (J L) H felis infection for 14 months. (J) Dysplastic dilated cystic fundic gastric glands (gastrointestinal intraepithelial neoplasia) (thin arrow), and dilated cystic glands invading into the submucosal layer surrounded by chronic inflammatory cells (thick arrow). (K) Markedly thickened antral mucosa comprised of dysplastic glands (surface) and mucous metaplasia of deeper glands (bar). (L) IHC for H,K -ATPase confirms complete loss of parietal cells. (M O) At 18 months postinfection, there is progressive architectural distortion of the fundus (M) with dilated cystic glands both intramucosal (thin arrow) and submucosal (thick arrow). (N) Architectural distortion of the antral mucosa is progressive with severe metaplasia and dysplasia of glands, profound thickening, and loss of all normal gland structure. (O) IHC for H,K -ATPase confirms a complete absence of parietal cells. (P) The fundus at 22 months of infection contains bizarre dilated cystic glands invading into the submucosal layer, surrounded by chronic inflammatory cells. The arrow indicates the squamocolumnar junction. (Q) Frank adenocarcinoma of the antrum. Cells have nuclear to cytoplasmic ratio alterations, loss of nuclear polarity, and architectural distortion. (R) Gross pathology at 22 weeks: (1) squamous forestomach; (2) fundus, greater curvature; (3) fundus, lesser curvature, more pronounced mucosal thickening; (4) large antral polypoid lesion causing gastric outlet obstruction; (5) proximal duodenum. All sections of fundus were taken at the squamocolumnar junction. Fundus and antral histology; H&E stain, original magnification 200. H,K -ATPase IHC counterstained with hematoxylin.

6 1942 CAI ET AL GASTROENTEROLOGY Vol. 128, No. 7 Figure 3. Histology scores. Mice were sham infected or infected with H felis and killed at the time points indicated. Inflammation, hyperplasia, and dysplasia were scored on a 0 4 scale. Two sections through the squamocolumnar junction at the lesser curvature were evaluated for each mouse, and the worst score was recorded for each section evaluated. (A) Natural history of H felis infection from 4 to 22 months of infection. (B) Mice infected with H felis for 2 months received sham eradication therapy or effective eradication therapy and were killed after 2, 6, or 12 additional months. (C) Mice infected with H felis for 6 months received sham eradication therapy or effective eradication therapy and were killed after 2, 6, or 12 additional months. (D) Mice infected with H felis for 12 months received sham eradication therapy or effective eradication therapy and were killed after 2, 6, or 12 additional months. n 5 for each group. All experimental groups are compared with infected controls. A horizontal bar extending over more than one group indicates that each group was independently compared with infected control and shares a P value. Data are reported as mean SD. *P.05. Eradication of Long-term Infection Decreases Inflammation and Halts Progression to Adenocarcinoma of the Fundus Mice infected for 12 months developed severe inflammation, with marked architectural distortion, metaplasia, severe dysplasia, and cystic dilation of fundic glands extending into the submucosa (Figure 4I), with 50% meeting the criteria for gastrointestinal neoplasia. 22 Eradication therapy given at this point substantially decreased the level of inflammation (Figures 3D, 4J, and 4K). However, even 12 months after eradication therapy, residual inflammatory infiltrates remained and inflammation scores had not returned to basal levels. Indeed, the mucosa retained a substantial number of inflammatory cells within both the mucosa and the submucosal space. Hyperplasia decreased but not to a significant degree until 6 and 12 months after eradication and never achieved uninfected levels. Dysplasia scores remained high despite successful eradication therapy and did not decrease significantly even at 12 months after eradication (Figure 3D). Interestingly, though, dysplasia scores did not progress further in the eradicated group compared with a continual increase in the

7 June 2005 H FELIS ERADICATION AND GASTRIC CANCER RISK 1943 Figure 4. Eradication of Helicobacter infection restores parietal cell mass. Mice at early (2 months), mid (6 months), or late (12 months) points of infection received triple eradication therapy and were killed 2, 6 (data not shown), or 12 months later. (A) Sham eradication therapy at 2 months of infection. (B and C) Two months of infection, followed by eradication therapy, and killed (B) 2 months later or (C) 12 months later shows complete restoration of architecture and regression of inflammatory cell infiltrates. (D) H,K -ATPase IHC demonstrates complete restoration of parietal cell mass at 12 months after eradication therapy. (E) Six months of infection, sham eradication therapy. (F and G) Six months of infection, followed by eradication therapy, and killed (F) 2 months later or (G) 12 months later. Complete restoration of architecture is noted, with mild residual inflammation at 2 months and complete regression of inflammation at 12 months. (H) H,K -ATPase IHC shows complete restoration of parietal cell mass at 12 months posteradication. (I) Twelve months of infection, sham eradication therapy. (J and K) Twelve months of infection, followed by eradication therapy, and killed (J) 2 months later or (K) 12 months later. (L) H,K ATPase IHC shows partial restoration of parietal cell mass. untreated infected mice (Figure 3A), suggesting that eradication therapy halted the progression of dysplasia. At 12 months of infection, the incidence of severe gastrointestinal intraepithelial neoplasia was 50%; 2 months after eradication, the rate remained 50% (3 of 6). The rate of severe GIN declined to 40% (2 of 5) 6 months after eradication and to only 30% (3 of 10) 10 months after eradication, suggesting a regression of established lesions with successful eradication therapy. In mice that did develop gastrointestinal intraepithelial neoplasia, lesions were fewer in number and of lower histologic grade when compared with lesions in mice actively infected. The lesions remained as large dilated cystic glands with less nuclear stratification and fewer bizarre mitoses. Gland structure of the gastrointestinal intraepithelial neoplasia was more preserved, and there was no extension to the serosa. In addition, eradication therapy restored parietal cell mass to near-normal levels at late stages of infection, as shown by both histology (Figure 4K) and IHC (Figure 4L), despite the persistence of dysplasia. Parietal cells are seen intercalating between dysplastic cells and were found abundantly in areas of dilated cystic glands and gastrointestinal intraepithelial neoplasia (Figure 4K and L). Eradication of H felis Infection at 12 Months Reduces the Incidence and Size of Antral Adenocarcinoma and Prevents Gastric Outlet Obstruction All mice infected for 12 months developed moderate to severe dysplasia; by 16 months, all developed invasive lesions (dysplasia grade 3.5 4) in both the fundus and the antrum. Fundus lesions, while becoming

8 1944 CAI ET AL GASTROENTEROLOGY Vol. 128, No. 7 more aggressive histologically, remained grossly small and did not metastasize or become locally invasive beyond the stomach wall. In contrast to this were lesions in the antrum. Antral dysplasia was present in all mice infected for 12 months, and adenocarcinoma was microscopically detected in all mice infected for 16 months or more. Antral lesions rapidly increased in size, becoming obstructing in 70% at 22 months and in 100% at 24 months. Eradication therapy given at 12 months markedly decreased the progression from antral dysplasia to carcinoma, with only 30% of the eradicated mice developing invasive antral carcinoma by 24 months. The lesions in these eradicated mice were not grossly apparent and were only seen microscopically with meticulous examination of serial section. There were no mortalities in this group, and there was no incidence of gastric outlet obstruction (Figure 5). It is not clear if lesions would eventually progress, albeit at a slower rate than those in the actively infected group, because mice were not followed up beyond 24 months posteradication. Eradication Therapy Impacts Proliferation Rate of Gastric Mucosal Cells Increased proliferation and widening of the proliferative zone is a hallmark of Helicobacter infection and inflammation and is believed to underlie the increased cancer risk associated with Helicobacter infection. We next addressed the effect of eradication therapy on gastric mucosal proliferation at early, mid, and late time points of infection. These alterations in proliferation were measured 2 months after effective therapy was completed to determine short-term growth effects and after 1 year to ascertain if these effects were durable. Under normal conditions, proliferating cells in the fundus are confined to the isthmus region of the gland, where the presumed multipotent gastric stem cells reside. 23 Helicobacter infection caused a marked increase in proliferation, as measured by BrdU incorporation. Mice receiving eradication therapy early in infection (Figure 6A) had significantly lower BrdU incorporation when compared with mice that were infected and were indistinguishable from uninfected control values (LI: control, 3 1; bacterial eradication, 5 0.7; infected, 78 8). This decrease in proliferation was sustained at 1 year posteradication (LI: control, 4 1; bacterial eradication, 8 1.5; infected, ). In addition, the proliferating cell zone was markedly expanded (Figure 6C) in the infected mice and was restored to the normal location after bacterial eradication (Figure 6B), suggesting that mitosis is not confined to the multipotent gastric stem cell zone during infection but involves additional cell populations. 18 Figure 5. Bacterial eradication at 1 year of infection prevents death due to obstructing antral adenocarcinoma/gastric outlet obstruction. (A) Mice were infected with H felis, received sham eradication or eradication therapy at 12 months of infection, and were followed up until 26 months. At 24 months of infection, all infected mice had died; at 26 months (14 months after eradication therapy), all mice that had received eradication therapy were alive. (B) Incidence of microscopic invasive antral adenocarcinoma at 8, 14, 16, and 22 months of infection and in mice infected for 12 months that received bacterial eradication and were evaluated at 24 months (12 months after eradication therapy). For 8, 14, and 16 months of infection, n 5; for 22 and 24 months of infection, n 10. Mice infected for 6 months who received effective eradication therapy had a 50% reduction in the LI 2 months after successful eradication therapy (LI: control, 3 1; bacterial eradication, ; infected, ). This reduction in proliferation was durable; however, even at 1 year, proliferation levels did not decline to baseline levels (LI: control, 4 1; bacterial eradication, ; infected, ) (Figure 6D F). After 12 months of infection, bacterial eradication did not alter proliferation rates when measured at 2 months (Figure 6G) and BrdU LI was unchanged between the eradicated and infected groups (LI: control, 3 1;

9 June 2005 H FELIS ERADICATION AND GASTRIC CANCER RISK 1945

10 1946 CAI ET AL GASTROENTEROLOGY Vol. 128, No. 7 bacterial eradication, ; infected, ). Approximately 1 year after therapy, BrdU incorporation was decreased by one third in the eradicated group compared with the infected group (LI: control, 4 1; bacterial eradication, 59 9; infected, 92 12). However, the proliferative zone remained expanded (Figure 6H and I). Later times were not evaluated; therefore, it is not clear if this trend of decreasing proliferation would continue. Under normal conditions, proliferating cells in the antrum are confined to the distal third of the gland and are seen in sections as BrdU-positive staining cells tightly associated and contiguous with one another. Mice that received eradication therapy early in infection (Figure 6M) had a significantly lower level of BrdU incorporation following eradication therapy when compared with infected mice but higher than in control mice (LI: control, 42 1; bacterial eradication, 56 6; infected, ). At 1 year after eradication therapy, proliferation rates were substantially lower than in infected mice; however, proliferation rates remained significantly higher than in control mice (LI: control, 38 1; bacterial eradication, 62 9; infected, 156 8) (Figure 6J L). Mice infected for 6 months who received effective eradication therapy had a 50% reduction in the antral LI shortly after successful eradication therapy when compared with those mice still infected. Despite this decrease, proliferation levels remained substantially elevated relative to control values (LI: control, 38 1; bacterial eradication, 83 8; infected, 156 8). This reduction in proliferation was durable; however, even after 1 year, proliferation levels did not decline to baseline levels (LI: control, 41 1; bacterial eradication, 85 15; infected, ) (Figure 6M O). After 12 months of infection, bacterial eradication did not decrease antral proliferation rate relative to infected mice (Figure 6P) (LI: control, 35 1; bacterial eradication, ; infected, 196 7). Approximately 1 year after successful bacterial eradication, BrdU LI had decreased to approximately half of that seen in the infected mice (LI: control, 38 1; bacterial eradication, 94 6; infected, ) yet remained substantially higher than in uninfected controls. At these later times of infection, architectural distortion in the antrum was significant, resulting in disorientation of glands within sections. In these cases, serial sections were examined; however, the BrdU LI may misrepresent the level of proliferation. Later times were not evaluated; therefore, it is not clear if the trend of decreasing proliferation would continue. Expression Pattern of Cell Adhesion Molecules Is Altered in Helicobacter Infection and Restored by Early Successful Eradication Therapy Altered cell adhesion is an important factor for the development of gastric cancer, yet it is not clear when a cell acquires these alterations. We first evaluated the expression pattern and overall levels of expression of -catenin (Figure 7), -catenin (Figure 8), and IQGAP1 (Figure 9) in the gastric mucosa as Helicobacter infection progressed; in parallel experiments, we evaluated the effects of successful eradication therapy on the expression pattern. In the absence of infection, -catenin expression is predominantly membrane bound. Parietal cells consistently express lower protein abundance, which remains confined to the area of cell-cell junctions (Figure 7A, thin arrow). Smaller mucus cells have prominent surface expression but also show abundant intracytoplasmic staining (Figure 7A, thick arrow). As infection progresses and architectural distortion becomes increasingly prominent, parietal cells are lost and a metaplastic mucus cell lineage emerges. The expression pattern of -catenin in these mucus cells is similar to control mucus cells in that they show dual cytoplasmic and surface expression (Figure 7B D). The few isolated parietal cells (Figure 7C, thin arrow) maintain the membrane-only pattern. Thus, the apparent shift from surface to surface/cytoplasmic expression and the apparent increase in abundance reflects the change in cell type within the mucosa rather than an actual shift in pattern within a cell. The expression 4 Figure 6. Bacterial eradication decreases gastric mucosal proliferation. Mice infected with H felis for 2, 6, or 12 months received bacterial eradication and were killed 2 or 12 months later. All mice received intraperitoneal BrdU 1 hour before they were killed. (A I) Fundus. (J R) Antrum. BrdU LI counted as the number of positive nuclei per 20 glands. (A) Mice were infected for 2 months (early infection) and killed 2 months later (4-month time point) or 12 months later (14-month time point). IHC for BrdU in (B) eradicated and (C) uneradicated mice. (D) Mice were infected for 6 months (midpoint), received eradication therapy, and were killed 2 months later (8-month time point) or 12 months later (18-month time point). Representative IHC for BrdU in (E) eradicated and (F) uneradicated mice are shown. (G) Mice were infected with H felis for 12 months (late infection), received eradication therapy, and were killed 2 months later (14-month time point) or 10 months later (22-month time point). IHC for BrdU in (H) uneradicated and (I) eradicated mice are shown. BrdU LI in the antrum for (J L) early, (M O) mid, and (P R) late time points of infection as described for the corresponding sections of fundus described previously. n 5 mice per group. Twenty glands were counted per histologic section. Two histologic sections were counted per mouse and reported as the average SD. Results were considered statistically significant at P.05.

11 June 2005 H FELIS ERADICATION AND GASTRIC CANCER RISK 1947 Figure 7. -catenin expression pattern is restored with successful bacterial eradication. -catenin expression was evaluated by IHC (brown staining), counterstained with hematoxylin (blue). (A) Control mice. Mice infected wiwth H felis for (B) 2 months, (C) 6 months, (D) 12 months, and (E) 24 months. Short-term effects of eradication therapy on -catenin were evaluated in tissues 2 months after sham or effective eradication therapy at (F and G) 2 months, (H and I) 6 months, or (J and K) 12 months of infection. Mice eradicated at 12 months were evaluated at a late time point of 22 months (10 months posteradication, L) along with (M) infected control. (A M, original magnification 400.) pattern within the antrum remains membrane bound throughout infection before cancer formation. Cells within both the fundus and the antrum maintain the same -catenin expression level and distribution early on, and the pattern and quantity of protein is altered only after invasion occurs, where in some glands it is completely lost (Figure 7E, fundus, antrum not shown). The alterations in -catenin expression with successful Helicobacter eradication therapy parallel the restoration of normal architecture. Indeed, Helicobacter eradication therapy at 2 or 6 months restores parietal cells with their predominant membrane expression. Admixed with parietal cells are mucus cells, which maintain both membrane and cytoplasmic staining (Figure 7F M). -catenin is expressed predominantly at cell-cell junctions in parietal cells (Figure 8A, thin arrow), with cytoplasmic and surface expression seen in smaller mucus cells (Figure 8A, thick arrow). With Helicobacter infection, the membrane-predominant expression pattern is preserved until very late times of infection (Figure 8B

12 1948 CAI ET AL GASTROENTEROLOGY Vol. 128, No. 7 Figure 8. -catenin expression pattern is restored with successful bacterial eradication. -catenin expression was evaluated by IHC (brown staining), counterstained with hematoxylin (blue). (A) Control mice. Mice infected with H felis for (B) 2 months, (C) 6 months, (D) 12 months, and (E) 24 months. Short-term effects of eradication therapy on -catenin were evaluated in tissues 2 months after sham or effective eradication therapy at (F and G) 2 months, (H and I) 6 months, or (J and K) 12 months of infection. Mice eradicated at 12 months were evaluated at a late time point of 22 months (10 months posteradication, L) along with (M) infected control. (A M, original magnification 400.) E). After the appearance of severe dysplasia and carcinoma, the overall -catenin abundance is decreased and the expression pattern switches from clearly membrane bound to sparse intracytoplasmic expression (Figure 8E). In keeping with the findings of -catenin alterations occurring after the initiation of carcinoma, eradication at 2 months (Figure 8F and G) and 6 months (Figure 8H and I) restores architecture and the expression pattern for -catenin is indistinguishable from the pattern seen in control mice. At 12 months of infection, -catenin remains predominantly membrane bound, and this is not altered after eradication therapy (Figure 8J and K). At later times of infection, expression dramatically decreases and shifts from surface expression to a predominantly cytoplasmic pattern. In mice that had their Helicobacter infection eradicated at 12 months, those areas of the gastric mucosa that showed restoration of normal architecture had a -catenin expression pattern indistinguishable from control mice and areas of progressive architectural distortion had decreased expression levels in a

13 June 2005 H FELIS ERADICATION AND GASTRIC CANCER RISK 1949 Figure 9. IQGAP1 expression pattern is restored with successful bacterial eradication. IQGAP1 expression was evaluated by IHC (brown staining), counterstained with hematoxylin (blue). (A) Control mice. Mice infected with H felis for (B) 2 months, (C) 6 months, (D) 12 months, and (E) 24 months. Short-term effects of eradication therapy on IQGAP1 were evaluated in tissues 2 months after sham or effective eradication therapy at (F and G) 2 months, (H and I) 6 months, or (J and K) 12 months of infection. Mice eradicated at 12 months were evaluated at a late time point of 22 months (10 months posteradication, L) along with (M) infected control. (A M, original magnification 400.) cytoplasm-predominant pattern. Changes in both -catenin and -catenin occur late in disease, suggesting that they do not play a role in cancer initiation but may contribute to progression and invasion of lesions once formed. IQGAP1 is an F-actin binding protein that has been implicated in the modulation of cadherin-based cell adhesion and has been suggested to maintain the integrity of the gastric mucosa in older mice. 24 The expression pattern in the gastric mucosa did not change in uninfected mice as they aged (data not shown) and consists of a predominantly speckled cytoplasmic pattern (Figure 9A). With infection, there is a progressive shift from cytoplasmic expression to cell surface expression, with the majority of protein expressed at the apical surface (Figure 9B E). Eradication therapy restored the expression pattern to the same degree that normal histology was restored. Metaplastic and dysplastic cells maintained the apical cell surface expression, while restoration of parietal and chief cell populations in the fundus was accompanied by a reversion to cytoplasmic expression. In the an-

14 1950 CAI ET AL GASTROENTEROLOGY Vol. 128, No. 7 trum, metaplastic and dysplastic glands maintained the apical cell surface expression and restoration of normal glandular structure was accompanied by a shift from apical expression to more diffuse cytoplasmic expression. In contrast to -catenin and -catenin, alterations in the expression pattern of IQGAP1 occurred before invasion and were consistently found in advanced lesions. These findings argue for a role for IQGAP1 in cancer initiation and progression. Discussion H pylori infection causes gastric adenocarcinoma; whether eradication of this bacterium prevents gastric cancer is not known. Human studies using premalignant lesions as surrogate markers suggest that treatment is effective in preventing disease; however, these studies are not conclusive. To address the question of the effects of bacterial eradication on cancer initiation and progression, we used the well-described C57BL/6 mouse model. There is a clear systematic progression from parietal and chief cell loss to hypertrophy of glands accompanied by the emergence of a mucus cell metaplasia and varying degrees of antralization within the fundic mucosa. These changes were reversible with bacterial eradication, which, if given at early and mid points of infection, completely prevented progression to gastric cancer. Dysplasia increased in severity and progressed to invasive glands such that 50% of mice infected for 12 months had invasive fundic lesions. This increased to 100% by 15 months of infection. Additional independent studies from our laboratory confirm this progression to invasive carcinoma in all mice infected beyond 15 months (personal observation and Houghton et al, 18 September 2003). Antral adenocarcinoma occurred late in infection and resulted in gastric outlet obstruction and death by 24 months of infection in all infected mice. Eradication therapy given at 12 months prevented progression of the fundic lesions and caused a regression of established lesions. More importantly, lesions within the antrum were prevented in 70% of mice. In the 30% that developed antral lesions, these lesions remained small and were only microscopically visible. Bacterial eradication at 12 months completely protected the mice from death due to gastric outlet obstruction, which was the inevitable outcome in mice that remained infected. One proposed mechanism for the initiation of the metaplasia/dysplasia/carcinoma sequence is loss of parietal cells and the subsequent loss of critical cellular cross-talk. Mature parietal cells seem necessary to influence the direction of maturation of gastric epithelial cell precursors and to orchestrate the migration-associated differentiation programs of zymogenic and pit lineages. 25 In keeping with this, early loss of parietal cells precedes the development of metaplasia and dysplasia in the C57BL/6 model of infection and experimental ablation of parietal cells is associated with metaplastic alterations. 25,26 Bacterial eradication at early and mid time points restored parietal cell mass concomitant with the restoration of architectural integrity, supporting the notion of a central role for the parietal cell in normal cellular differentiation. After 12 months of infection, eradication therapy restored parietal cell mass despite the persistence of metaplasia and dysplasia, raising the possibility that parietal cell loss and metaplasia may be temporally but not causally associated. Conversely, parietal cell loss may initiate metaplasia yet with accumulated genetic damage in the long-term infected mice may no longer be necessary for sustained metaplasia. Helicobacter infection increases proliferation within the gastric mucosa. Increased proliferation is a hallmark of cancer in other organs, and current dogma suggests that a return of normal proliferation is needed to eliminate cancer risk. 27 Eradication therapy given at early times completely restores proliferation rates and at the midpoint of infection significantly reduces but does not completely restore proliferation to normal. The decrease in proliferation as a result of bacterial eradication at late time points, while substantial, remains significantly higher than basal levels, suggesting that a continued cancer risk exists. Cell-cell interaction is critical for the maintenance of normal cell morphology, differentiation, and control of growth. In addition to cellular context (such as alterations in parietal and chief cell populations), alterations in binding proteins have been implicated in the progression of cancer. -catenin and -catenin are important for the function of E-cadherin. -catenin is believed to link the actin cytoskeleton to E-cadherin and colocalizes with E-cadherin at the adherens junctions. 28 Abnormal expression of these proteins, either as decreased quantity or by an altered cellular location, has been implicated in early phases of malignant transformation and local invasion as well as enabling tumor cells to metastasize. In our studies, -catenin and -catenin expression paralleled changes in histology and total protein expression decreased only after glands became malignant, suggesting a role later in the invasion process. IQGAP1 alterations were seen earlier in disease, implying that these alterations may play a role in earlier stages of carcinogenesis. Based on our studies, evaluation of protein expression did not provide additional information relative to standard H&E staining and histology, and expression pattern could be predicted based on histology. Expression pat-

15 June 2005 H FELIS ERADICATION AND GASTRIC CANCER RISK 1951 terns paralleled changes in cell lineage and were reversible with bacterial eradication, which induced restoration of normal histology. Dramatically decreased protein expression was seen only in frank adenocarcinoma, and these abnormalities could be predicted based on histology alone. These findings suggest that evaluation of -catenin, -catenin, and IQGAP1 in histologic sections does not add to the interpretation of reversibility of lesions. Eradication therapy given early or at the midpoint of infection prevented gastric cancer formation. Eradication therapy given at late times of infection when metaplasia and dysplasia were severe resulted in partial reversion to normal histology and a substantial decrease in progression of fundic and antral cancer. Bacterial eradication prevented cancer-related gastric outlet obstruction and death in all groups studied. These findings suggest strongly that bacterial eradication is effective for preventing gastric cancer; in those cases in which gastric cancer exists before eradication, effective antibiotic therapy substantially slows progression of malignant lesions. References 1. IARC Working Group on the evaluation of carcinogenic risks to humans. Schistosomes, liver flukes and Helicobacter pylori. IARC Monogr Eval Carcinog Risks Hum 1994;61: Zucca E, Cavalli F. Are antibiotics the treatment of choice for gastric lymphoma? Curr Hematol Rep 2004;3: Sepulveda AR, Coelho LG. Helicobacter pylori and gastric malignancies. Helicobacter 2002;7: Ley C, Mohar A, Guarner J, Herrera-Goepfert R, Figueroa LS, Halperin D, Johnstone I, Parsonnet J. Helicobacter pylori eradication and gastric preneoplastic conditions: a randomized, doubleblind, placebo-controlled trial. Cancer Epidemiol Biomarkers Prev 2004;13: Wong BC, Lam SK, Wong WM, Chen JS, Zheng TT, Feng RE, Lai KC, Hu WH, Yuen ST, Leung SY, Fong DY, Ho J, Ching CK, Chen JS, China Gastric Cancer Study Group. Helicobacter pylori eradication to prevent gastric cancer in a high-risk region of China: a randomized controlled trial. JAMA 2004;291: Leung WK, Lin SR, Ching JY, To KF, Ng EK, Chan FK, Lau JY, Sung JJ. Factors predicting progression of gastric intestinal metaplasia: results of a randomised trial on Helicobacter pylori eradication. Gut 2004;53: Zhou L, Sung JJ, Lin S, Jin Z, Ding S, Huang X, Xia Z, Guo H, Liu J, Chao W. A five-year follow-up study on the pathological changes of gastric mucosa after H. pylori eradication. Chin Med J 2003; 116: Ito M, Haruma K, Kamada T, Mihara M, Kim S, Kitadai Y, Sumii M, Tanaka S, Yoshihara M, Chayama K. Helicobacter pylori eradication therapy improves atrophic gastritis and intestinal metaplasia: a 5-year prospective study of patients with atrophic gastritis. Aliment Pharmacol Ther 2002;16: Ruiz B, Garay J, Correa P, Fontham ET, Bravo JC, Bravo LE, Realpe JL, Mera R. Morphometric evaluation of gastric antral atrophy: improvement after cure of Helicobacter pylori infection. Am J Gastroenterol 2001;96: Nozaki K, Shimizu N, Ikehara Y, Inoue M, Tsukamoto T, Inada K, Tanaka H, Kumagai T, Kaminishi M, Tatematsu M. Effect of early eradication on Helicobacter pylori-related gastric carcinogenesis in Mongolian gerbils. Cancer Sci 2003;94: Watanabe T, Tada M, Nagai H, Sasaki S, Nakao M. Helicobacter pylori infection induces gastric cancer in Mongolian gerbils. Gastroenterology 1998;115: Honda S, Fujioka T, Tokieda M, Satoh R, Nishizono A, Nasu M. Development of Helicobacter pylori-induced gastric carcinoma in Mongolian gerbils. Cancer Res 1998;58: Ogura K, Maeda S, Nakao M, Watanabe T, Tada M, Kyutoku T, Yoshida H, Shiratori Y, Omata M. Virulence factors of Helicobacter pylori responsible for gastric diseases in Mongolian gerbil. J Exp Med 2000;192: Zhen Q, Chen YX, Shi Y, Xiao XD. Development of gastric adenocarcinoma in Mongolian gerbils after long-term infection with Helicobacter pylori. J Gastroenterol Hepatol 2004;19: Thompson LJ, Dannon SJ, Wilson JE, O Rourke JL, Salama NR, Falkow S, Mitchell HJ, Lee A. Chronic Helicobacter pylori infection with Sydney strain 1 and a newly identified mouse-adapted strain (Sydney strain 2000) in C57BL/6 and BALB/c mice. Infect Immun 2004;723: Wang TC, Goldenring JR, Dangler C, Ito S, Mueller A, Jeon WK, Koh TJ, Fox JG. Mice lacking secretory phospholipase A2 show altered apoptosis and differentiation with Helicobacter felis infection. Gastroenterology 1998;114: Rogers AB, Fox JG. Inflammation and cancer. I. Rodent models of infectious gastrointestinal and liver cancer. Am J Physiol Gastrointest Liver Physiol 2004;286:G361 G Houghton J, Stoicov C, Nomura S, Rogers AB, Carlson J, Li H, Cai X, Fox JG, Goldenring Jr, Wang TC. Gastric cancer originating from bone marrow derived cells. Science 2004;306: Houghton J, Macera-Block LS, Harrison L, Kim KH, Korah RM. Tumor necrosis factor alpha and interleukin 1 up-regulate gastric mucosal Fas antigen expression in Helicobacter pylori infection. Infect Immun 2000;68: Stoicov C, Whary M, Rogers AB, Lee FS, Klucecsek K, Li H, Cai X, Reza S, Ge Z, Khan I, Luster A, Fox JG, Houghton J. Co-infection modulates inflammatory response and clinical outcomes of Helicobacter felis and Toxoplasma gondii infection. J Immunol 2004; 173: Kushima R, Manabe R, Hattori T, Borchard F. Histogenesis of gastric foveolar metaplasia following duodenal ulcer: a definite reparative lineage of Brunner s gland. Histopathology 1999;35: Boivin GP, Washington K, Yang K, Ward JM, Pretlow TP, Russell R, Besselsen DG, Godfrey VL, Doetschman T, Dove WF, Pitot HC, Halberg RB, Itzkowitz SH, Groden J, Coffey RJ. Pathology of mouse models of intestinal cancer: consensus report and recommendations. Gastroenterology 2003;124: Karam SM, Straiton T, Hassan WM, Leblond CP. Defining epithelial cell progenitors in the human oxyntic mucosa. Stem Cells 2003;21: Li S, Wang Q, Chakladar A, Bronson RT, Bernards A. Gastric hyperplasia in mice lacking the putative Cdc42 effector IQGAP1. Mol Cell Biol 2000;20: Li Q, Kasam SM, Gordon JI. Diphtheria toxin-mediated ablation of parietal cells in the stomach of transgenic mice. J Biol Chem 1996;271: Yamaguchi H, Goldenring JR, Kaminishi M, Lee JR. Association of spasmolytic polypeptide-expressing metaplasia with carcinogen administration and oxyntic atrophy in rats. Lab Invest 2002;82: Biasco G, Paganelli GM, Miglioli M, Barbara L. Cell proliferation biomarkers in the gastrointestinal tract. J Cell Biochem 1992; 16G:73 78.