Duodenoesophageal reflux induces esophageal adenocarcinoma without exogenous carcinogen

Duodenoesophageal Reflux Induces Esophageal Adenocarcinoma Without Exogenous Carcinogen Martin Fein, Ph.D., M.D., Jefiey H. Peters, M.D., Para Chandrasoma, M.D., Adrian l? Ireland, M.D., Stefan Oberg, 111.D., Man.ed I? Ritter, 211.D., Cedric G. Bremner, M.D., Jefiey A. Hagen, M.D., Tom R. DeMeester, M.D.

In the rat model, esophageal adenocarcinoma reproducibly develops following surgically induced duodenal reflux into the esophagus and administration of nitrosamine. In addition, decreasing gastric acid via partial or total gastrectomy increases the prevalence of adenocarcinoma in this model. We questioned whether carcinogen was necessary for cancer development in the gastrectomized model and whether esophageal acidification could reverse the effect of gastrectomy. Three groups of 26 rats each were randomized to a surgical procedure to produce one of the following reflux models: gastroduodenal reflux by esophagojejunostomy, duodenal reflux by total gastrectomy and esophagojejunostomy, or no reflux by Roux-en-Y reconstruction. In a second experiment, 42 rats were operated on to induce duodenal reflux. One week following surgery, they were randomized to receive acidified water (pH 1.8) or tap water. The animals were killed at 24 weeks of age, and the esophagus was evaluated histologically. All animals with reflux had severe esophagi& and 87% developed columnar lining of the distal esophagus. Nearly half (48%)d eve lpd o e ad enocarcinoma at the anastomotic site 16 weeks postoperatively and without carcinogen administration. Cancer prevalence did not differ between animals with gastroduodenal or duodenal reflux but tended to be lower in animals receiving acidified water. Duodenoesophageal reflux is carcinogenic in the rat model. Exogenous carcinogen is not necessary for cancer development in gastrectomized rats. (J GASTROINTEST SURG 1998;2:260-268.)

The incidence of esophageal adenocarcinoma has risen faster than that of any other cancer, and is now estimated to be 1 per 100,000.1,2 The reasons for this change are unclear. It is known that the presence of metaplastic specialized intestinal, or Barrett’s epithelium, is the most important etiologic factor in the development of esophageal adenocarcinoma.3 Longitudinal studies have documented progression from metaplasia to dysplasia to carcinoma, thereby linking the common malady of gastroesophageal reflux disease with esophageal cancer.4 Evidence from both clinical studie+l2 and animal modelsi3-19 have shown esophageal exposure to duodenal juice to be a key factor in the genesis of specialized intestinal metaplasia and likely the development of adenocarcinoma. The pathophysiology of esophageal carcinoma has been studied in the rat model. Previous studies

have

shown

compounds (MNAN)

that

the

including

administration

of N-nitroso

methyl-N-amylniuosamine

and dimethyl-N-amylnitrosamine

@MNM)

results

in the induction of squamous cell carcinoma of the esophagus in a dose-dependent fashion.20 Surgically induced reflux of gastroduodenal juice into the esophagus increases the tumor yield and alters the histology to adenocarcinoma in most tumors.14 Animals with reflux of gastroduodenal juice who did not receive a carcinogen rarely developed esophageal adenocarcinoma.14J7 Recently it has been shown that decreasing gastric acid via partial or total gastrectomy increases the prevalence of adenocarcinoma in this model.‘* We questioned whether carcinogen was necessary for cancer development in gastrectomized rats with duodenoesophageal reflux and whether esophageal acidification could reverse the effect of gastrectomy.

From the Department of Surgery, University of Southern California School ofMedicine, Los Angeles, Calif.; and the Chirurgische Universitatsklinik (MI and MXR.), Wurzburg, Germany. Presented at the Thirty-Eighth Annual Meeting of The Society for Surgery of the Alimentary Tract, Washington, D.C., May l l-14,1997. Reprint requests: Jeffrey H. Peters, M.D., Department of Surgery, University of Southern California School of Medicine, 15 10 San Pablo Street, Ste. 514, Los Angeles, CA 90033-4612. 260

Vol. 2, No. 3 1998

Adenocarcinomain Rats

METHODS Reflux Models One hundred nineteen 8-week old male SpragueDawley rats (Harlan, Sprague-Dawley, Inc., Indianapolis, Ind.) were studied in two different experiments. Experiment 1. Three groups of rats were operated on in a randomized fashion to produce the following reflux models (Fig. 1, A): 1. Gastroduodenal R&‘ux (n = 26). Esophagojejunostomy (EJ) with gastric preservation to induce reflex of gastric and duodenal juice into the esophagus. The anastomosis was performed 4 cm distal to the ligament of Ti-eitz. The posterior vagal trunk and the left gastric artery remained uncut. 2. Duodenal RefEzlx(n = 26). Esophagojejunostomy and total gastrectomy (EJTG) to induce esophageal reflux of duodenal juice alone. 3. No Refux (n = 25). Total gastrectomy and a Roux-en-Y reconstruction (EJRY) to divert duodenal content away from the esophagus. After completion of the end-to-side esophagojejunostomy, the afferent jejunal loop was ligated with silk, divided, and anastomosed end to side to the

Gastroduodenal

reflux

Duodenal reflux-no

jejunum at least 15 cm from the esophagojejunal anastomosis.21 This group served as a control group. Experiment 2. This experiment was designed to study the effect of exogenous acid added to the drinking water of rats with esophagojejunostomy and total gastrectomy (Fig. 1, B). Forty-two animals were randomized 1 week after surgery to drink either acid syrup (duodenal reflex-acid) or tap water (duodenal reflex-no acid). Acid syrup was prepared with hydrochloric acid at a pH of 1.8 (0.016 mol/L) and sucrose 68 mg/ml and given ad libitum.

Surgical Procedure Operations were performed after an acclimatization period of 3 days. Rats were kept in metal cages on a 12-hour light/dark cycle at a temperature of 70” F and a humidity of 60%. Water and standard solid chow (Teklan rodent diet 8604, Harlan) were given ad libitum. Before surgery, the animals were starved overnight, and water was discontinued on the morning of surgery. Rats were anesthetized with an intramuscular injection of xylazine hydrochloride (12 mg/kg) and ketamine (75 mg/kg). In all animals, an

Duodenal reflux

acid

261

No reflux

Duodenal reflux-acid

Fig. 1. A, Reflux models: Experiment 1. Esophagojejunostomy inducing gastroduodenal reflux, total gastrectomy and esophagojejunostomy inducing duodenal reflux, and total gastrectomy and Roux-en-Y reconstruction to divert reflwr. B, Reflex models: Experiment 2. Total gastrectomy and esophagojejunostomy inducing duodenal reflux in both groups. Acidification of the esophagus by acid syrup given orally in one group.

262

Journal of Gastrointestinal Surgery

Fein et al.

end-to-side esophagojejunostomy was performed. The esophagus was divided above the gastroesophageal junction, and a 4 mm jejunostomy was made 4 cm distal to the ligament of Treitz. The esophagojejunal anastomosis was constructed in an antecolic manner with eight interrupted full-thickness 7-O polypropylene sutures.13~16~22Before abdominal wall closure, 1 ml of 0.9% sodium chloride was instilled into the peritoneal cavity. Water was permitted when the rats awoke, and chow was provided the next day. Rats were weighed every 2 weeks during the course of the experiment. Any rats that became ill were killed. The experimental course was 16 weeks (24 weeks of age). The total mortality rate was 8% (9 of 119). Two rats died from an accidental intravascular injection of the drugs for anesthesia. Three rats died within the first week after surgery for unknown reasons. Four animals died of reflux disease, two with severe pneumonia and two with upper gastrointestinal bleeding. Seventy-three rats were evaluated in experiment 1 (22 EJ, 26 EJTG, and 2 5 EJRY) and 3 7 rats in experiment 2 (18 EJTG-no acid and 19 EJTG-acid).

Gross and Histologic Evaluation Sixteen weeks after surgery, the rats were killed with an overdose of phenobarbital. The esophagus was resected from below the larynx to the jejunum, and 2 mm of jejunal mucosa was left attached to the specimen. The esophagus was opened longitudinally and the length, the circumference, and the proportion of macroscopically inflamed to normal esophagus was measured and the specimen photographed. The esophagus was divided into two longitudinal strips that were rolled up similar to a jelly roll, fixed in 10% buffered formalin, and embedded in paraffin blocks. Macroscopically visible tumors were cut in half and embedded separately. Sections were stained with hematoxylin and eosin. Histologic sections were read by the pathologist (P.C.) in a blinded fashion. Histologic changes in the squamous epithelium were classified as (1) hyperplastic, defined as an increased thickness of the squamous epithelium with normal maturation and hyperkeratosis, or (2) regenerative, where the squamous epithelium showed increased height of lamina propria papillae greater than 70% of mucosal thickness, basal cell hyperplasia greater than 20% of mucosal thickness, and absence of hyperkeratosis. Regenerative changes are identical to changes described for reflux esophagitis in humans. Hyperplastic changes were scored on a scale from 0 to 3 ranging from absent to severe. Inflammation was classified as (1) intraepithelial, based on the presence of eosinophils and/or neutrophils in the squamous epithelium, or (2) subep-

ithelial, based on the degree of inflammatory cell infiltration in the lamina propria below the epithelium. Both types of inflammation were scored on a scale from 0 to 3 (absent to severe). The presence or absence of epithelial ulceration was also noted. Columnar metaplasia was defined as the presence of a histologically abnormal glandular mucosa proximal to the anastomotic line. This columnar lining was required to be different from the jejunal mucosa by virtue of having markedly shortened or absent villi and distorted crypt architecture. The presence of this metaplastic columnar epithelium admixed with squamous epithelium was recorded. Neoplasms observed in this study were of a uniform histologic type with the characteristic features of a mutinous adenocarcinoma. These consisted of malignant infiltrative glands associated with lakes of extracellular mucin.

Statistics The study population was designed based on a sample size analysis (power 0.8; P = 0.05). In experiment 1 a prevalence of 4% tumors in the control group and 40% tumors in animals with reflux was assumed. To detect a difference at these levels, 25 animals had to be analyzed. Data are reported as mean ? standard deviation unless otherwise stated. The chi-square test was used to compare proportions among three groups and the Fisher’s exact test for two groups. The Kruskal-Wallis test was used to compare continuous data among three groups and the MannWhitney U test for two groups. A P value ~0.0.5 was considered significant. The study protocol was approved by the Institutional and Animal Care and Usage Committee of the University of Southern California.

RESULTS Experiment 1 Macroscopic Findings. Mean baseline and subsequent body weight for each group of rats is shown in Fig. 2. Animals with no reflux grew significantly faster than animals with either gastroduodenal or duodenal reflux. Esophageal specimens of rats with reflux were significantly shorter and more dilated than those of rats without reflux (Table I). Shortening of the esophagus was unlikely to be due to the differences in the growth of the animals, as the anastomotic site was pulled up to or above the hiatus in animals in the reflux groups and remained inn-a-abdominal in those without reflux. There was gross evidence of severe esophageal mucosal injury in all animals with surgically induced reflux. This included epithelial thickening and extensive hyperplasia of the lower two thirds

Vol. 2, No. 3 1998

Adenocarcinomain Rats 263

of the esophagus. Ulceration was frequently present in the area above the anastomosis. A nodular intraluminal tumor or thickening of the anastomotic site was seen in some animals (Fig. 3). T h e esophagus of animals without reflux was grossly normal, except for a short area o f epithelial t h i c k e n i n g adjacent to the anastomosis seen in some of the animals. HistologicFindings. Histologic findings are shown

in Table II. In b o t h groups with reflux, histologic changes became progressively more severe from the pharynx to the anastomosis. T h e specimens showed marked hyperplastic changes with increased thickness of the squamous epithelium, hyperkeratosis, regenerative changes with papillomatosis, and basal cell hyperplasia (Fig. 4). Inflammation was severe and typically involved the complete esophageal wall. Epithe-

Table I. Experiment 1: Macroscopic findings

Length (mm) Circumference (mm) Esophagitis (%)

Gastroduodenal

Duodenal

reflux (n = 22)

reflux (n - 26)

N o reflux (n = 25)

55 -+ 8 10.4 + 2.3 69 + 23

53 -+ 7 11.0 + 2.1 68 -+ 20

67 -+ 9 4.5 +- 1.6 2 + 10

All parameters differbetweenthe group with no refluxand the groupswith reflux(P