The influence of oxygen-derived free radical scavengers on survival in gastric cancer, with serosal invasion and metastases to the lymph nodes surrounding the stomach, was assessed in a prospective randomized controlled double-blind trial conducted for 5 years. To this end, allopurinol (inhibits the enzyme xanthine oxidase which is responsible for the formation of superoxide radicals and scavenges hydroxyl radicals) and dimethyl sulphoxide (DMSO; scavenges hydroxyl radicals) were used. Following potentially curative distal two-thirds partial gastrectomy, 228 patients making an uneventful recovery from surgery were randomized to the control group or to receive allopurinol (50 mg by mouth 4 times a day) or DMSO (500 mg by mouth 4 times a day). In 160 fully evaluable patients who were studied for 5 years, allopurinol and DMSO incurred a significant (p <l 0.01) survival advantage over the whole period of study. The similarity in efficacy between allopurinol and DMSO and the fact that the only action they share is scavenging oxyradicals suggest that these radicals mediate the aggressiveness of gastric cancer by producing tissue damage, thus allowing the cancer to spread. Consequently, oxygen-derived free radicals are implicated in the mechanism of gastric cancer, and removing them provides patients with a survival advantage.

The majority of gastric cancer patients have widespread disease at the time of diagnosis. consequently in a considerable proportion surgery is often palliative.1 The overall 5-year survival of patients with gastric cancer is approximately 5%.2 The early stages of this disease, defined as tumor limited to the mucosa and submucosa with or without lymph node metastases3 have a 5-year survival following surgical resection of 70-90%, compared with 30-40% in locally advanced stages.4-6After almost all non-curative, gastrectomies and some curative ones, gastric cancer recurs. Over half of the patients show peritoneal dissemination followed in order of importance by distant metastases via lymphatic and/or haematogenous spread, and reappearance of the cancer at anastomotic stomata.2, 3-5 Recently,7 it has been clearly shown that the 5-year survival rate in gastric cancer patients is inversely related to the degree of serosal invasion and that patients with this invasion at resection will develop disease recurrence. Furthermore, the presence of lymph node metastases in conjunction with serosal invasion had poor prognosis causing intraperitoneal dissemination followed by death of most of the patients within 2 years of their operation.7The results of adjuvant chemotherapy trials in gastric cancer are less favorable in European and Anglo-American countries than in Japan.8 The majority of findings in Western countries leads to the conclusion that the routine application of adjuvant chemotherapy for gastric cancer cannot be recommended.8 On the other hand, intraoperative radiotherapy has been shown to improve the survival only in patients with locally advanced gastric carcinoma.9Oxygen-derived free radicals are cytotoxic and play a key role in the mechanism of tissue injury.10-12 Studies in the rat (to be published) demonstrated that these radicals are permissive for gastro-intestinal carcinogenesis. Exposure to doses of carcinogenics that are capable of producing gastrointestinal cancer failed to exhibit any carcinogenic activity when the rat was treated with radical scavengers. Conversely, the doses of these carcinogenics which do not produce cancer under normal circumstances were allowed to do so when the overall pool of oxygen-derived free radicals was increased. It was also noted that in the rat bearing gastrointestinal cancer scavenging oxygen-derived free radicals delays hepatic metastases and prolongs survival. These studies propose that oxyradicals are implicated in the mechanism of gastrointestinal cancer in the rat and mediate its aggressiveness probably by producing the destruction of tissues associated with this cancer and which, in turn, allows it to spread. The present investigation was, therefore, designed to examine whether removing oxygen-derived free radicals incurs any survival advantages in patients with carcinoma of the stomach.

Drugs

A 1% solution of allopurinol (Burroughs Wellcome Co, Research Triangle Park. N.C., USA) was prepared by dissolving the powder in double-distilled water containing the molar equivalent of 0.1 M NaOH. A 10% solution of dimethyl sulphoxide (DMSO; Sigma. St. Louis, MO, USA) was prepared by diluting the stock solution with double-distilled water. The vehicle solution of allopurinol was given to controls. Solutions were placed in 600-ml capacity dark-colored glass bottles of identical appearance. The patients were issued a fresh supply of solutions every 30 days and were treated with identical volumes of solution throughout the 5-year period of the study.

Study Design

This was a prospective randomized controlled double-blind trial conducted for 5 years on consecutive patients making an uneventful recovery from a 'potentially' curative distal two-thirds partial gastrectomy for carcinoma of the distal third of the stomach. Randomization was carried out by drawing sealed envelopes. Treatment started on the 5th postoperative day and continued to the end of the study (5 years).

Recruitment Criteria

Patients were considered to be suitable for the study if all the following conditions applied:

1. The carcinoma had invaded the serosa but not any contiguous structures.
2. The gastrectomy specimen showed tumour-free proximal resection lines (within 5 mm from resection lines) for at least 2 cm.
3. Complete excision of all the regional lymph nodes of the stomach according to the location of cancer, i.e. lymph nodes along the greater and lesser curvatures, the supra- and subpyloric nodes, the right paracardial lymph nodes and the lymph nodes along the course of the left gastric, the common hepatic and around the coeliac arteries.
4. There were metastases to the lymph nodes surrounding the stomach but no metastases to the nodes along the main arteries including the coeliac axis.
5. The lymph, node metastases were in the form of cancer cells within the lymph node system but without extension beyond the node into the perinodal fatty tissues.
6. No macroscopically evident peritonea dissemination and no free cancer cells within the peritoneal cavity immediately after the gastrectomy (the peritoneal cavity was washed with 100 ml of physiological saline at 37°C immediately after the gastrectomy, then 50 ml of this fluid was centrifuged, stained and examined as described in detail elsewhere.13
7. No evidence of hepatic or distant metastases. This means that the stage of the patients studied was S₂ N₁ P₀ according to the Japanese Research Society for Gastric Cancer.

Patients were not recruited into the trial if one or more of the following were present:

1. age over 80 years
2. risk factors (adenomatous polyp(s) removed]
3. admission as an emergency failing to respond to conservative treatment or septicaemia or peritonitis on presentation
4. pregnancy
5. alcoholism
6. taking prohibited drugs or regularly taking any form of medication (to avoid unknown therapeutic effects and drug interactions)
7. hypertension
8. diabetes
9. hepatic (including cirrhosis) or renal disorders
10. serious underlying disease (for example: cardiorespiratory problems)
11. rheumatoid arthritis or any form of collagen diseases
12. Zollinger-Ellison syndrome or other gastrointestinal disorders like the irritable-bowel syndrome which would make it difficult to assess patients and the significance of their signs and symptoms
13. previous gastrointestinal surgery
14. history of radiotherapy, chemotherapy or any malignancy
15. synchronous carcinomas
16. invasion of the abdominal wall or any adjacent organs or structures found during laparotomy
17. postoperative complications (cardiovascular, pulmonary -- pneumonia or atelectasis --, hepatic, wound infection or dehiscence, gastrointestinal hemorrhage, anastomotic failure, ileus, significant reflux esophagitis or psychiatric problems)
18. cases with hypersensitivity to penicillin and its derivatives.

Survival was measured from the time of tumor resection until death from any cause. Postoperative deaths (mortality from any cause during hospital stay irrespective of its duration) were not included in the study. Survival calculations were only carried out on patients who had died from tumor recurrence, and those dying from other causes (e.g.cerebrovascular accidents, myocardial infarction, bronchopneumonia) were excluded.

Clinical Management

Gastric carcinoma was diagnosed from the history, physical examination, endoscopy with biopsies and a barium meal. Other investigations were undertaken to determine the extent of the disease or to serve as a baseline for future reference, and these included: standard haematology, biochemistry measurements, urine examination, chest X-rays and liver ultrasonography. Once the diagnosis of gastric carcinoma had been made, patients were admitted to hospital and prepared for surgery; correction of any anemia or hypoproteinaemia, and chest and lower limb's physiotherapy. At induction of anesthesia the patients were given intravenous chemoprophylaxis, 500 mg of ampicillin and 80 mg of gentamicin, and prophylaxis against deep-vein thrombosis was by peri-operative pneumatic compression leggings. In all cases the abdomen was opened by an upper midline incision, the liver was palpated for any metastases, the various lymph node groups (surrounding the stomach, in the hepatoduodenal ligament, in the retropancreatic area, in the mesenteric root, around the left gastric, common hepatic, coeliac and middle colic arteries, and around the abdominal aorta) were carefully examined and the peritoneum was inspected for evident dissemination. At least a 2-cm tumor-free proximal resection line was achieved, excision of all the regional lymph nodes as detailed above was carried out, gastric reconstruction was by a Roux-en-Y gastrojejunostomy, all the anastomoses were performed by double-layer continuous suturing with 2/0 polyglycolic material (Dexon), and the peritoneal cavity was examined for free cancer cell as stated above. Post-operatively patients were hydrated intravenously for at least 3 days during which period their oral intake of fluids was gradually increased. The return to solid food was not permitted before the 4th postoperative day. The patients were supplied with special charts to mark their compliance with the therapeutic regimen and to record all adverse events.Following discharge from hospital, the patients were reviewed every 3 months on an out-patient basis. At each visit a detailed assessment of any symptoms and possible adverse events coupled with a full physical examination, endoscopy with biopsies of any suspicious gastric mucosa, full blood counts, liver function tests and a liver ultrasound scan were performed. A chest X-ray was taken annually, unless indicated earlier.

Ethical Considerations

This study was approved by the Ethical Committee on Human Experimentation of the hospital, and every patient gave written informed consent.

Study Groups

Two hundred and twenty-eight consecutive patients with an uneventful recovery from a distal two-thirds partial gastrectomy for carcinoma of the stomach were allocated to one of three groups and treated for 5 years. In the first group patients were given the vehicle solution of allopurinol by mouth, 5 ml 4 times a day (controls). In the second group patients were given allopurinol by mouth, 5 ml (50: mg) 4 times a day. In the third group patients were similarly treated with 5 ml DMSO (500 mg) 4 times a day.

Exclusion Criteria

Exclusion of patients from evaluability was based on the following rules:

1. Significant adverse events to the therapeutic regimen.
2. Intolerance of this regimen.
3. Failure to comply with the regimen.
4. Concomitant treatment during the study with medicines.
5. Postoperative complications or mortality occurring after randomization of the patients an the 5th post-operative day.
6. Missed synchronous malignant tumours (detected up to 6 months after surgery) or occurrence of metachronous malignant tumours.
7. Death from other than tumour-related causes.
8. Failure to attend for or lost to follow-up.

The decision to regard patients as nonusable for analysis was undertaken before breaking the treatment code.

Statistical Analysis

A sample size of 150 patients with 50 patients in each group was chosen initially. Based on a two-tailed test, such a size will detect a significant difference of 30% between active and no therapy (p <0.05) with a probability of 80% for the overall sample. Because of the anticipated problems of non-evaluability of a proportion of patients inherent in a long-term study which could weaken any conclusions drawn, the aim was to enter at least 75 patients in each group. Results are expressed as percentages or the mean ±SEM unless stated otherwise. The statistical significance (p < 0.05) of observed differences in percentage values was assessed using the 2 test with Yates' correction. and the Mann-Whitney U test for non-parametric data was employed to examine whether differences in mean values were significant. Kaplan-Meier's product limit methodwas used to estimate the survivor functions for the three study groups. The difference between these functions was assessed using the Mantel-Cox statistics. Proportional hazard models were then used to investigate the effect of treatment on survival when account was taken of the other patient factors as covariates. Additional intention-to-treat analyses were carried out re-including the patients who were excluded from the study and using various theoretically possible outcomes to assess any influence their exclusion might have had on the conclusions of the study.

Patient Characteristics

Seventy-seven patients (30 women and 47 men with an age range of 41-79 years, mean 59) were randomized to the control group. Seventy-five patients (30 women and 45 men with an age range of 44-78 years, mean 58) were randomized to the allopurinol group. Seventy-six patients (27 women and 49 men with an age range of years, mean 61) were randomized to the DMSO group. The patients excluded from evaluability are presented in table 1, and the characteristics of those remaining shown in table 2. The three groups were well matched for age and sex, for duration of symptoms before presentation, type of presentation, weight loss, and for tumor differentiation.

General Observations

Seventeen controls (31%), 20 patients in the allopurinol group (39%) and 16 patients in the DMSO group (30%) were smokers, and all the men studied were social drinkers who did not indulge heavily. Ten patients in the control group (18.2%; 6 vomiting and 4 acute gastrointestinal haemorrhage), 8 patients in the allopurinol group (16%; 3 vomiting and 5 acute gastro-intestinal hemorrhage) and 11 patients in the DMSO group (20.4%; 6 vomiting, 1acute epigastric pain, 4 acute gastro-intestinal hemorrhage) presented as an emergency which settled with conservative treatment. The remaining cases presented with one or more of abdominal pain, dyspepsia, anorexia, or signs and symptoms of iron deficiency anemia, and the frequency of these symptoms was similar among the three groups. The mean length of symptoms for the emergency and non-emergency cases was comparable among the groups. In the non-emergency cases, iron deficiency anemia was seen in 10 controls, in 8 members of the allopurinol group, and in 11 members of the DMSO group. Of these patients, 5 controls, 4 members of the allopurinol group, and 6 members of the DMSO group were given pre-operative blood transfusions, whereas the remaining patients were given oral ferrous sulphate. Consequently, the total number of patients in each group who received pre-operative blood transfusions were 9 controls (16.4%), 9 patients in the allopurinol group (17.6%), and 10 patients in the DMSO group (18.5%). During surgery, 10 patients in the control group (18.2%), 12 patients in the allopurinol group of (23.5%), and 11 patients in the DMSO group (20.4%) received blood transfusions.The overall operative mortality (death from any cause during hospital stay irrespective of its duration) was 3.1%. Of these patients 1 control died because of myocardial infarction, and 1 patient in the DMSO group died because of bronchopneumonia after randomization.

One patient in the control group (bronchopneumonia), 3 patients in the allopurinol group (1 deep-vein thrombosis and 2 wound infections) and 2 patients in the DMSO group (1 wound infection and 1 anastomotic failure) developed postoperative complications after entry into the study. One patient given DMSO and 2 patients given allopurinol were intolerant to their therapeutic regimen. Four patients given the vehicle solution of allopurinol and 5 patients in each of the remaining groups reported drug-related side-effects consisting of headache, nausea and general malaise. In addition, allopurinol treatment caused skin hypersensitivity and allergic reactions. However, these effects were severe enough to warrant cessation of treatment only in 2 controls, in 2 patients treated with allopurinol and in 1 patient treated with DMSO.

Tumor Characteristic