Endoscopists' estimation of size should not determine surveillance of colonic polyps.

OBJECTIVE: Current British Society of Gastroenterology guidelines use adenomatous polyp size as one of the key factors in determining polyp follow-up. This study aimed to compare polyp size assessment by colonoscopists and pathologists before and after fixation to determine the optimal method for measurement. METHOD: Thirty-five colorectal polyps were found during pre-arranged colonoscopies in one centre. Polyp size was measured to the nearest 1 mm by three different methods: 1. by the endoscopist at colonoscopy; 2. by the pathologist fresh, following removal; 3. by the pathologist fixed, following fixation. The endoscopist and the pathologist were blinded to each other's measurements. RESULTS: Seventeen men, eighteen women with mean age of 66.2 years (SD: 9.4, range: 38.7-85.5) underwent polypectomy/s with all polyps removed intact. Polypectomies were performed by consultants (43%), nurse specialists (34%) and specialist registrars (23%). The median size (mm) of polyps measured were endoscopically, 6.5 (2-25 mm); fresh specimen 7.0 (4-28 mm) and fixed 7.0 (4-28 mm). Endoscopic measurements were significantly lower than that of fresh and fixed sizes (P < 0.001 and P = 0.003 respectively), with poor correlation [correlation of variance (CV): 21.0% and intraclass correlation coefficient (ICCC): 0.841 for endoscopic and fresh measurements; CV: 21.1% and ICCC: 0.838 for endoscopic and fixed measurements]. There was no statistical difference between fresh and fixed specimen measurements (P > 0.05; CV: 4.2%, ICCC: 0.974). In three patients, the endoscopic measurement was < 1 cm in polyps that were found to be >or= 1 cm on pathological measurement. CONCLUSIONS: Endoscopists consistently underestimated polyp size. Fixation had no effect on polyp size. Pathologists' measurement of polyp size on fixed specimens should determine the need for further colonoscopic follow-up.

Positive lymph node retrieval ratio optimises patient staging in colorectal cancer.

Alternative lymph node (LN) parameters have been proposed to improve staging in colorectal cancer. This study compared these alternative parameters with conventional TNM staging in predicting long-term survival in patients undergoing curative resection. A total of 295 consecutive patients (mean age 70 years; range 39-95; s.d. 10.4) underwent resection for colorectal cancer from 2001 to 2004. Age, sex, primary tumour site, TNM stage and chemotherapy/radiotherapy were recorded. Patients with colon and rectal cancers were analysed separately for LN parameters: LN total; adequate LN retrieval (> or =12) and inadequate (<12); total number of negative LN; total number of positive LN and the ratio of positive LN to total LN (pLNR). Univariate and multivariate survival analysis was performed. The median number of LN retrieved was 10 (1-57) with adequate LN retrieval in 147 cases (49.8%). For each T and N stage, inadequate LN retrieval did not adversely affect long-term survival (P>0.05). On multivariate analysis, only pLNR was an independent predictor of overall survival in both colon and rectal cancers (HR 11.65, 95% CI 5.00-27.15, P<0.001 and HR 13.40, 95% CI 3.64-49.10, P<0.001, respectively). Application of pLNR subdivided patients into four prognostic groups. Application of the pLNR improved patient stratification in colorectal cancer and should be considered in future staging systems.

Malignant disease in peptic ulcer surgery patients after long term follow-up: a cohort study of 1992 patients.

AIMS: To assess the effect of previous peptic ulcer surgery on subsequent malignant events, in particular in relation to previous vagotomy, a historical cohort study was conducted. METHODS: All patients undergoing surgery for peptic ulcer disease with accurate follow-up data at a large peptic ulcer clinic in the Western Infirmary, Glasgow, from 1965 to 1983 were assessed. All cancer events and specific cancer events (gastric, bronchial, laryngeal, colorectal, bladder, breast, prostate, pancreas, kidney, oesophageal cancers) were determined as outcome measures and expressed as standardised incidence ratio (SIR). RESULTS: Vagotomy and drainage accounted for 67% of all procedures for peptic ulcer disease. Eighty-three percent were habitual smokers. For all peptic ulcer surgery patients, the SIR for all cancer events was 0.86. For specific cancers, the SIRs were bronchial cancer (SIR 1.13); laryngeal cancer (SIR 2.17), colorectal cancer (SIR 0.67). For vagotomised patients the risk of gastric cancer was significantly elevated (SIR 1.50). CONCLUSIONS: An excess of cancers attributable to smoking have been found in peptic ulcer surgery patients. Vagotomised patients have a higher risk of gastric cancer after long term follow-up. This finding may have implications for screening and the safety of long term acid suppression with agents such as proton pump inhibitors.

Bile carcinoembryonic antigen levels and occult hepatic metastases from colorectal cancer.

PURPOSE: Up to 30 percent of patients will have occult hepatic metastases at the time of curative surgery for colorectal cancer. The ability to predict this group of patients would allow better targeting of appropriate therapy. It has been shown previously that patients with overt hepatic metastases have significantly high levels of carcinoembryonic antigen in gallbladder bile compared with serum levels. The aim of this study was to assess the accuracy of bile carcinoembryonic antigen levels taken at the time of operation in predicting patients with occult hepatic metastases. METHODS: Bile and serum carcinoembryonic antigen samples were collected from 37 patients undergoing surgery for colorectal cancer, 26 of whose procedures were deemed curative and who were followed up for a median of 63.5 months. RESULTS: Twelve patients were alive with no evidence of recurrent disease, and two had recurrent disease, whereas 12 died of disease. The median (interquartile range) serum carcinoembryonic antigen in the disease-free group was 2.8 (1.1-6.1) ng/ml, and in the recurrent group it was 6.35 (4.3-30) ng/ml (P = 0.006), whereas bile carcinoembryonic antigen in the disease-free group was 7 (5-39) ng/ml as compared with 31 (5-383.7) ng/ml in the recurrent group (P = 0.210). The accuracy of serum carcinoembryonic antigen in predicting occult hepatic metastases was 77 percent compared with 72 percent for bile carcinoembryonic antigen. CONCLUSION: Intraoperative bile carcinoembryonic antigen levels are no more accurate than serum carcinoembryonic antigen levels in predicting occult hepatic metastases in patients undergoing potentially curative colorectal cancer surgery.

Cytokines secreted by lymphokine-activated killer cells induce endogenous nitric oxide synthesis and apoptosis in DLD-1 colon cancer cells.

IL-2-activated killer lymphocytes (LAK cells) secrete inflammatory cytokines such as interferon-gamma (IFN-gamma) and tumor necrosis factor alpha (TNFalpha) that can induce nitric oxide (NO) synthesis. We evaluated whether LAK cells could activate NO synthesis in human cancer cells. LAK cells and their culture supernatants induced NO synthesis in DLD-1 colon cancer cells in a dose-dependent manner. NO synthesis was inhibited completely by blocking antibodies to IFN-gamma, demonstrating a key role for this LAK cell cytokine in regulating NO synthesis. The addition of TNFalpha antibodies resulted in partial inhibition. Induction of iNOS mRNA and protein expression in DLD-1 cells was detected. Endogenous NO production inhibited DLD-1 cell proliferation and induced apoptosis, processes that were inhibitable by the NO synthase inhibitor N(G)-monomethyl-l-arginine. Our study has identified a novel, non-contact-dependent LAK cell cytotoxic mechanism: induction of growth inhibition and programmed cell death due to endogenous NO synthesis in susceptible human cancer cells.

Immunosuppressive effects of factor IX products: an in vitro study.

The effects of a recombinant factor IX product (BeneFix), and of five plasma-derived factor IX products, AlphaNine, Immunine, Konyne, Mononine and Replinine on in vitro peripheral blood mononuclear cell (PBMC) immune function were compared in a blinded study. We assessed the effects of these products on Con-A-induced lymphocyte proliferation and interleukin-2 and interleukin-10 secretion, expression of lymphocyte activation markers, and nitric oxide secretion by stimulated mouse peritoneal macrophages. At 1 mL-1 for 48 h, Konyne reduced Con-A-induced mitogenesis by 50% (P < 0.05); AlphaNine, Mononine and BeneFix had no effect. At 10 IU mL-1, Con-A-induced mi- togenesis was at control levels with Mononine and BeneFix, but was reduced to <15% (P < 0.05) with each of the other products. IL-2 and IL-10 secretion by Con-A-stimulated lymphocytes was also markedly depressed by all the products tested except Mononine and BeneFix. Dialysis of these products did not substantially affect these results. Flow cytometric analysis of lymphocyte activation markers following Con-A stimulation showed that Konyne also decreased IL-2 receptor alpha and beta chain (CD25 and CD122) induction on PBMC. Konyne also inhibited nitric oxide secretion to levels <18% of controls. These results indicate that certain factor IX products, including some of purported higher purity, substantially depress in vitro immune function. The importance of these findings to in vivo immune function in haemophilia B patients remains to be established.

Receptor-mediated activation of murine peritoneal macrophages by antithrombin III acts as a costimulatory signal for nitric oxide synthesis.

We evaluated the effect of antithrombin III (ATIII), a serine protease inhibitor (SERPIN), on induction of nitric oxide (NO) synthesis in murine peritoneal macrophages. Incubation of macrophages with ATIII plus interferon-gamma (IFN-gamma) but not ATIII alone induced nitrite accumulation (a metabolite of NO) in a dose-dependent manner. Expression of the inducible nitric oxide synthase isoform was confirmed by Western blot. NO synthesis was inhibited by NG-monomethyl-l-arginine, by complexing ATIII with thrombin or by rabbit anti-human ATIII antiserum. Addition of polymyxin B to macrophage cultures failed to inhibit ATIII/IFN-gamma-induced NO synthesis, excluding lipopolysaccharide contamination. 125I-ATIII bound to macrophages in a dose-dependent, specific, and saturable manner, with a Km of approximately 7.1 nM. Our results demonstrate that ATIII, but not ATIII/thrombin complex, acts to costimulate macrophage activation and NO synthesis via a novel receptor mediated mechanism, which may indicate a role for SERPINs in macrophage activation.

Nitric oxide exposure inhibits induction of lymphokine-activated killer cells by inducing precursor apoptosis.

Nitric oxide synthesis is strongly induced during IL-2 treatment of mice and humans. While this free radical can act as an antitumor mechanism by inhibiting cellular respiration and DNA synthesis in cancer cells, immunosuppressive effects have also been suggested. We evaluated the effects of NO exposure on the induction of murine lymphokine-activated killer (LAK) cells from splenocytes by IL-2 (6000 IU/ml). When splenocytes were exposed to pure NO gas for 30 min prior to the addition of IL-2, complete abrogation of LAK cell cytotoxicity was observed. In contrast, cytolytic activity of already activated LAK cells was only minimally affected by NO exposure. NO exposure markedly depressed cellular proliferation in response to concanavalin A or IL-2. Immunostaining of LAK cell cultures following NO exposure revealed a marked decrease in CD8+, and peanut lectin (PNA+)/CD56+ subsets (48 and 69%). Dual staining of LAK cells for DNA strand breaks and either PNA or CD8+ identified the induction of programmed cell death in these subsets 12-24 h following NO exposure. These experiments demonstrate that NO has the capacity to inhibit LAK cell induction by inducing apoptosis of cytolytic lymphocyte precursors.

Nitric oxide synthesis contributes to IL-2-induced antitumor responses against intraperitoneal Meth A tumor.

IL-2 therapy is a potent inductive stimulus for nitric oxide (NO.) synthesis in mice and humans. It is not yet clear whether NO. can contribute to IL-2-induced therapeutic responses. The murine skin cancer Meth A is relatively resistant to lymphokine-activated killer (LAK) cell killing, allowing evaluation of the role of IL-2-induced NO. synthesis in vivo, without contribution by LAK cells. Subcutaneous IL-2 treatment of mice bearing i.p. Meth A tumor increased nitrite production by cells derived from ascites (63 +/- 14 microM vs 3.2 +/- 1.5 microM in untreated controls). N omega-monomethyl-L-arginine (MLA), NO. synthase inhibitor, prevented this increase. NO. production correlated in an inverse fashion with tumor cell proliferation in vitro. Evidence for IL-2-induced heme nitrosylation was demonstrated in tumor cells by electron paramagnetic resonance spectroscopy. By immunomagnetic depletion experiments, macrophages were implicated as a major source of NO. synthesis. Cytologic and flow-cytometric evaluation revealed that IL-2 treatment resulted in enhanced lymphocyte and macrophage recruitment into malignant ascites, and decreases in tumor cell recovery. MLA administration further increased host cell recovery. Subcutaneous IL-2 therapy increased urinary nitrate excretion up to eightfold in mice, and appeared to produce a significant survival advantage that was prevented by MLA administration.

Collagen promotes perianastomotic tumour growth in an experimental animal model.

Local application of growth factors promote wound healing and may find clinical application for use in high-risk intestinal anastomoses such as that following anterior resection. Since viable tumour cells are present in the bowel lumen and circulation after curative colorectal cancer surgery, it is unclear what effect such factors may have on tumour recurrence. The aim of this study was to examine the effect of epidermal growth factor (EGF) and basic fibroblast growth factor (bFGF) in a collagen suspension on perianastomotic tumour growth in an animal model. Significantly (P < 0.05) more animals in the collagen treated groups developed anastomotic tumours. The area of tumour growth at the anastomosis was also significantly greater for the collagen (median 14.7 mm2) and collagen + EGF (median 10.8 mm2) groups compared with controls (median 0.78 mm2). We were unable to demonstrate any promotion of tumour by growth factors alone. Collagen promotes perianastomotic tumour growth in this model and is not a suitable vehicle for growth factor application in colorectal cancer surgery.

Effectiveness and toxicity of protracted nitric oxide synthesis inhibition during IL-2 treatment of mice.

The current study was designed to characterize nitric oxide (NO.) synthesis during interleukin-2 (IL-2) treatment of mice, and to determine whether NO. mediated IL-2-induced "vascular leak." We developed a technique for chronic subcutaneous infusion of the NO. synthase inhibitor N omega monomethyl-L-arginine (MLA) via osmotic minipump to aid in further study of these processes. After IL-2 administration to C3H/HeN mice (180,000 IU i.p. b.i.d. for 5 days), NO. synthesis increased two-to-three fold, peaking on days 5-8. Administration of MLA reduced NO. synthesis in both IL-2-treated mice (from 2.7 to 1 microM/mouse/day), and normal mice (from 1 to 0.5 microM/mouse/day). This agent decreased IL-2-induced radiolabeled albumin accumulation in the liver after i.p. IL-2 administration (p < 0.02). MLA infusions resulted in minimal systemic toxicity in mice, as reflected by complete blood counts or serum chemistries. MLA also did not impair lymphokine-activated killer cell induction in vitro or in vivo, or alter IL-2-induced tumor responses in a 3-day pulmonary metastasis model. These experiments demonstrated that NO. is a mediator involved in the genesis of vascular permeability induced by IL-2 treatment. Studies designed to further evaluate the toxicity and usefulness of MLA infusions to modify this IL-2 induced toxicity appear to be warranted.

Efficacy of tumoricidal agents in vitro and in vivo.

Implantation of exfoliated tumour cells can give rise to local recurrence of colorectal cancer and it has been recommended that the bowel lumen be lavaged with a tumoricidal agent. This study identified which tumoricidal agents are currently used in Scotland and investigated their efficacy in vitro and in vivo. Cytotoxic efficacy was tested in vitro by a clonogenic assay and in vivo by a rat model with viable intraluminal tumour cells. Overall 70 per cent of surgeons used a tumoricidal agent during colorectal cancer surgery. Povidone-iodine, sodium hypochlorite and chlorhexidine-cetrimide were all effective at killing tumour cells in vitro but were all inactivated by the presence of 25 per cent whole blood in vitro. With 10(5) cells in vivo povidone-iodine and sodium hypochlorite significantly (P < 0.02) reduced the incidence of tumour growth while chlorhexidine-cetrimide had no significant effect. With 10(6) cells povidone-iodine had no effect on the incidence of tumour growth. Tumoricidal agents have effective cytotoxicity in vitro but are only weakly cytotoxic in vivo.

Comparison of manually constructed and stapled anastomoses in colorectal surgery. West of Scotland and Highland Anastomosis Study Group.

OBJECTIVE: The authors compared both the initial and the long-term outcomes of patients undergoing stapled and sutured colorectal anastomoses. SUMMARY BACKGROUND DATA: Sutured and stapled large bowel anastomoses are perceived to be equally safe, but concern has been raised about increased rates of tumor recurrence with the use of stapling instruments. METHODS: The outcome of patients with sutured and stapled colorectal anastomoses were compared in a prospective, multicenter, randomized study. Factors affecting long-term outcomes were assessed by both univariate and multivariate analysis. RESULTS: Seven hundred thirty-two patients were recruited. There was a significant increase in radiologic leakage in the sutured group (14.4% vs. 5.2%, p < 0.05), but there was no difference in clinical anastomotic leak rates, morbidity, or postoperative mortality. Tumor recurrence and cancer-specific mortality were higher in the sutured patients (7.5% and 6.7%, respectively) and in patients with anastomotic leaks. CONCLUSIONS: This study shows that suturing or stapling are equally safe in large bowel surgery. However, it also shows a long-term benefit of stapling in colorectal cancer patients.

Effect of the surgeon's specialty interest on the type of resection performed for colorectal cancer.

PURPOSE: The aim of this study was to examine the type of resection performed for colorectal cancer by surgeons with a colorectal interest and compare this with the type of resection performed by surgeons with other specialty interests. METHODS: One hundred sixteen patients had curative surgery performed for primary colorectal cancer over a one-year period by ten surgeons with four different specialty interests. RESULTS: Surgeons with an interest in colorectal cancer resected twice as much colon (280 mm vs. 130 mm; P > 0.0001, Mann-Whitney U test) and were more likely to remove adjacent clinically involved organs (15 percent vs. 0 percent) for left-sided colon and rectal cancers compared with surgeons with vascular or transplant interests. Surgeons with an interest in gastroenterology performed a resection that was intermediate between the colorectal and other specialty groups for left-sided cancers. Distal resection margins were significantly greater (55 mm vs. 20 mm; P > 0.001) for sigmoid cancers in the colorectal group, but were similar in all groups for rectal cancer. Resection lengths and margins for right-sided cancers were similar in all groups, although the number of lymph nodes retrieved from the mesentery was greater in the colorectal group (13 vs. 7.5; P = 0.08). CONCLUSION: This study shows a wide variability in the type of resection performed for colorectal cancer and illustrates the need for clinical trials to evaluate the effect of such variability on patient outcome.

Use of N-acetyl cysteine to increase intracellular glutathione during the induction of antitumor responses by IL-2.

IL-2 therapy can induce marked oxidative stress via reactive oxygen and nitrogen intermediates. Glutathione, the major intracellular reductant, may become rate limiting to cytotoxic lymphocyte activation and proliferation under these circumstances. N-Acetyl cysteine (NAc-cys) was used to increase intracellular glutathione levels during lymphokine-activated killer (LAK) cell activation by IL-2. Incubation of splenocytes with NAc-cys (0.6 to 1.0 mM) resulted in significant changes in intracellular reduced and total glutathione (92% and 58% increase, respectively) at 96 h. These levels correlated with markedly enhanced cell proliferation (threefold) and cytolytic effector cell generation (> fivefold increase in LU/10(6) cells) induced by the combination of NAc-cys with IL-2. IL-2 exposure by itself unexpectedly increased intracellular reduced glutathione by 43%. IL-2 and NAc-cys were synergistic in increasing glutathione levels (reduced glutathione: 292% increase; total: 251% increase). Inhibition of glutathione synthesis, using L-buthionine-(S,R)-sulfoximine reversed the effects of NAc-cys on intracellular glutathione, as well as cellular proliferation and cytotoxicity. This experiment established that the effects of NAc-cys required de novo glutathione synthesis. In conjunction with IL-2/LAK treatment, oral NAc-cys administration (260 to 900 mg/kg/day for 7 days) significantly decreased tumor progression in a refractory s.c. tumor model. A small fraction of mice (11 to 17%) had complete tumor regressions. NAc-cys may be useful as an adjunct to increase the antitumor activity of IL-2/LAK therapy.

Plasma gastrin concentrations are normal in patients with colorectal neoplasia and unaltered following tumor resection.

BACKGROUND/AIMS: Previous studies have found that colorectal cancer patients have hypergastrinemia, but most have been inadequately controlled. Preoperative fasting and meal-stimulated gastrin levels were measured in patients with colorectal tumors (n = 42) and in carefully matched controls (n = 34). Helicobacter pylori status was assessed because it causes significant hypergastrinemia. METHODS: Plasma gastrin levels were measured by radioimmunoassay. Helicobacter status was assessed using the [14C]urea breath test and serology (immunoglobulin G). RESULTS: Preoperatively, fasting plasma gastrin levels were similar in patients with tumors (median, 55 ng/L; interquartile range, 45-82.5) and controls (77.5 ng/L; 53.7-137.5; P = 0.10). Similarly, peak gastrin levels were not significantly different in tumor patients (200 ng/L; 137.5-312.5) and controls (247.5 ng/L; 147.5-375; P = 0.21). The prevalence of H. pylori infection in patients with tumors (60%) and controls (53%) was similar in both groups. Five (20%) tumor patients who were H. pylori-positive preoperatively were negative postoperatively, and their median peak plasma gastrin level decreased from 200 ng/L to 140 ng/L. After these patients were excluded, fasting and peak plasma gastrin concentrations were similar preoperatively and postoperatively. CONCLUSIONS: When confounding factors are controlled for, plasma gastrin levels are not increased in colorectal cancer and do not decrease after curative resection. Previously noted decreases in gastrin levels after tumor resection may be attributable to loss of H. pylori infection in some patients, as noted here.

Omeprazole inhibits colorectal carcinogenesis induced by azoxymethane in rats.

Numerous clinical and experimental studies suggest that gastrin plays an important part in the development of colorectal cancer in humans. This study was done to assess the influence of omeprazole induced hypergastrinaemia on the development of colorectal tumours in an experimental animal model. Forty female Sprague-Dawley rats received either omeprazole (40 mumol/kg) or vehicle (0.25% methylcellulose) by once daily oral gavage throughout the experiment. All animals received 12 consecutive weekly subcutaneous injections of azoxymethane (10 mg/kg/week) beginning at week 6. Serum gastrin concentrations were measured during weeks 1 and 5 and at death (week 27). Chronic omeprazole treatment resulted in appreciable hypergastrinaemia during the study, mean gastrin concentrations in omeprazole treated rats being raised by up to nine to 10 fold, compared with vehicle treated control rats (p < 0.001). Despite this, tumour incidence in the omeprazole group was significantly lower at 63%, compared with 95% in the vehicle only group (p < 0.02). The median number of tumours in the omeprazole group (1) compared with the vehicle group (3) was also significantly lower (p = 0.02). Average tumour size, site distribution, and the comparative frequencies of adenomas and adenocarcinomas were similar in the two groups. This study shows that omeprazole protects against colorectal carcinogenesis in this model despite causing appreciable hypergastrinaemia. The mechanism by which this occurs is unclear and merits further investigation. Because of the compounding protective effects of omeprazole, this model is not a suitable one for studying the longterm trophic effects of gastrin on the colon.

Effect of fibrin sealant on perianastomotic tumor growth in an experimental model of colorectal cancer surgery.

Viable intraluminal tumor cells can penetrate a clinically intact rodent colonic anastomosis and give rise to perianastomotic tumor growth. The aim of this study was to determine whether transanastomotic cell migration can be prevented by fibrin-based tissue sealant. Following distal colonic transection and reanastomosis with 5/0 silk sutures, Fischer F344 rats were randomly allocated to three experimental groups. In Group A, a circumferential ring of tissue sealant was placed around the serosal surface of the anastomosis; in Group B, sealant was limited to 50 percent of the anastomotic circumference; and, in Group C, no sealant was applied. All rats then had 10(5) Mtln3 carcinoma cells injected into the proximal colonic lumen via a rectal catheter. The incidence of perianastomotic tumor at 21 days was significantly lower in Group A (3 of 14 animals) than in Group B (11 of 16 rats) (P = 0.012; Fisher's exact test) or Group C (10 of 14 rats; P = 0.011). A further experiment demonstrated that sealant did not protect the anastomosis when tumor cells were instilled directly into the peritoneal cavity. A topical carcinocidal action therefore appears unlikely, but our results suggest that a circumferential anastomotic ring of fibrin sealant forms an effective mechanical barrier preventing intraluminal tumor cells from reaching the peritoneal cavity.