AIM: To examine the efficacy and complications of colonoscopic resection of colorectal polypoid lesions.

METHODS: We retrospectively reviewed 1354 polypectomies performed on 1038 patients over a ten-year period. One hundred and sixty of these were performed for large polyps, those measuring ≥ 20 mm. Size, shape, location, histology, the technique of polypectomy used, complications, drugs assumption and associated intestinal or extra intestinal diseases were analyzed. For statistical analysis, the Pearson χ² test, NPC test and a Binary Logistic Regression were used.

RESULTS: The mean patient age was 65.9 ± 12.4 years, with 671 men and 367 women. The mean size of polyps removed was 9.45 ± 9.56 mm while the size of large polyps was 31.5 ± 10.8 mm. There were 388 pedunculated and 966 sessile polyps and the most common location was the sigmoid colon (41.3%). The most frequent histology was tubular adenoma (55.9%) while for the large polyps was villous (92/160 -57.5%). Coexistent malignancy was observed in 28 polyps (2.1%) and of these, 20 were large polyps. There were 17 procedural bleeding (1.3%) and one perforation. The statistical analysis showed that cancer is correlated to polyp size (P < 0.0001); sessile shape (P < 0.0001) and bleeding are correlated to cardiac disease (P = 0.034), tubular adenoma (P = 0.016) and polyp size.

CONCLUSION: The endoscopic resection is a simple and safe procedure for removing colon rectal neoplastic lesions and should be considered the treatment of choice for large colorectal polyps. The polyp size is an important risk factor for malignancy and for bleeding.

The colonoscopic polypectomy was introduced by Wolf and Shinya in the early 1970s and has become the most common therapeutic procedure performed within endoscopic units[1]. Colonoscopic polypectomy is the first approach and the standard treatment for colon rectal polyps and, therefore, a prevention of colorectal cancer. It is a safe technique when performed by expert hands using a cautious technique and equipment that works properly; so that, in these conditions, complications should be an uncommon event.

Some factors can increase the rate of complications such as the type (pedunculated or sessile) and size of the polyp, its location and other factors regarding the comorbidity of the patient (coagulation disorders or drug assumption) and the technique used. The most common complications after polypectomy are bleeding (from 0.3% to 6.1%) and pain due either to excessive gas accumulation or to parietal damage and perforation after current application[2]. Frequently, these complications follow large polyps polypectomies. Several authors have reported complications during the endoscopic removal of large polyps and in the majority of these studies, “large” was defined as equal to or greater than 20 mm.

Large polyps represent a particular challenge for the endoscopist because they are often related to important risks of haemorrhage (0%-22.1%), perforation (0%-1.3%) and inadequate polypectomy[3–19]. The alternative to endoscopic therapy of large polyps is the surgical resection which involves hospitalization and anaesthesia. In addition, the risks of surgery are significant, especially in elderly patients with comorbid diseases. Moreover, the higher mortality (2%-4%) and morbidity (10%) rates reported for surgery with respect to endoscopic polypectomy cannot be overlooked[3].

This study aims to show a retrospective series of endoscopic resections of colon rectal polyps and in particular of large polyps, focusing on the efficacy, complications and risk factors.

From January, 1996 to May, 2006, 1038 patients underwent colonoscopy for the removal of 1354 polypoid lesions. Of these polypectomies, 160 (11.8%) were performed on large polyps, those measuring 20 mm or more. The patients were prepared for colonoscopy following standard protocol, following a fiber and residue free diet within 72 h before the investigation and the assumption of 4000 mL of a polyethylene glycol electrolytic lavage solution 18 h before colonoscopy. The colonoscopy was performed according to standard procedure using two Olympus CF-100 HI and one Pentax EC3830FK videocolonoscopes. An electrocautery Erbe ICC 200 and argon plasma coagulation Erbe APC 300 were also used. The sclerosis needle and endoscopic snare used during the polypectomy procedure were the standard models. All procedures were performed by two expert endoscopists.

For the statistical analysis the polyps were split in four groups: the first group represented polyps measuring between 1-10 mm, the second 11-19 mm, the third 20-39 mm and the fourth group of polyps measuring 40 mm or more. For shape, polyps were grouped in pedunculated (pedunculated and semi-pedunculated) and sessile (sessile and flat-elevated). The method of resection was selected according to the size and morphologic features of each lesion and three groups were created: direct snare (hot or cold biopsy and hot or cold snare in smaller lesions and larger protruding lesions), mucosectomy (en bloc, piecemeal, the inject and cut and the inject, lift, and cut for broad-based lesions) and endoloop.

Pedunculated polyps were transected at the stalk just below the polyp head. In all pedunculated polyps injection into the base of the stalk was not performed, before polypectomy, as a prophylactic measure to prevent the bleeding, except for the cases of pedunculated polyps with a very big stalk where bleeding prophylaxis was performed with the application of endoloop. Sessile polyps were resected using two techniques, en bloc and piecemeal. All resected material was retrieved for histological examination.

Post-polypectomy bleeding was defined as procedural if it occurred during polypectomy, immediate if it occurred within 24 h of polypectomy, and delayed if it occurred more than 24 h after the procedure. Bleeding was treated by injection therapy, with dilute adrenalin at a concentration of 1:10 000 with or without 1% polidocanol, hemoclips and thermal coagulation using argon plasma. The variables evaluated in the study were: polyp size, polyp shape, their location along the colon, complications (bleeding and perforation), histology, technique of polypectomy applied, drugs assumption and associated intestinal or extra intestinal diseases.

Continuous data are described by mean and standard deviation or median and range, according to distribution. Categorical data are presented as numbers and percentages. The Pearson χ² test with Brandt-Snedecor and Kimball’s formula[20] was used to assess the association between categorical variables while to individualize the correlation between categorical variables and some numerical variables a biserial correlation coefficient was used[21]. Differences between the groups were evaluated using the non parametric combination NPC Test, based on permutation tests[22]. A P-value < 0.05 was considered statistically significant. The estimation of a Binary Logistic Regression model allowed for the individualization of the variables which were tied on the bleeding[23]. In this context, the estimation of Log-Likelihood test and G test allows for the obtainment of the measure of goodness-of-fit.

Software used included SPPS, Windows 11.0 (2001) for Pearson χ² test and biserial correlation, Microsoft Excel (2002) for the Brandt-Snedecor method and Kimball’s formula, Methodologica S.R.L. (2001) for nonparametric analysis NPC test and Minitab Release 13.31, Copyright ©2000 Minitab Inc. for Binary Logistic Regression.

The demographic and clinical data of the patients studied and the characteristics of polyps according to size are illustrated in Table 1 and Table 2. Among 1354 endoscopic polypectomies, 907 were lesions less than 10 mm of diameter, 287 between 11 and 19 mm, and 160 more than 20 mm; the size of total polyps was 9.45 ± 9.56 (range 1-100) while the size of large polyps was 31.5 ± 10.8. Macroscopically sessile shapes were prevalent (966/1354 -71.3%) and the pedunculated ones were 388 (28.7%).

The most frequent location of the polyps was the sigmoid colon (559/1354 -41.3%). The most frequent location of large polyps was the sigmoid colon (65/160 -40.7%).

The most commonly used technique was direct endoscopic snare resection in 1294 polypectomies (95.6%) as well as in the group of large polyps 132/160 (82.5%).

The most frequent histological type among all the polyps resected was tubular adenoma, 756/1354 -55.9%), while, for the large polyps, it was the villous type (92/160 -57.5%). Of the removed polyps, 1078 were adenomas with low grade dysplasia (79.6%), 123 with high grade dysplasia (9.1%) and 28 (2.1%) were adenomas containing an area of invasive carcinoma and of these 20 were large polyps. The estimation of biserial correlation coefficient allows us that the cancer and polyp size are correlated (P < 0.0001), but there wasn’t a significant correlation between cancer and age (P = 0.464). To evaluate the association between malignancy and sex, histology, location and shape, we applied a χ² test, where it showed significant results for the association with shape (P < 0.0001), in particular sessile and sex (P < 0.0001), and the association between cancer and location (P = 0.719). Cancer with histology (P = 0.819) was not statistically significant.

The endoloop was used as a prophylactic measure to prevent postpolypectomy bleeding in four cases (0.3%) for pedunculated polyps. The “endoloop” prevented bleeding from the stalk in all cases. Procedural bleeding occurred in 1.3 % (17/1354) of all polyps (11 sessile and 6 pedunculated). In large polyps the bleeding occurred in 10/160 (6.3%). Bleeding was always managed by endoscopic means with the application of hemoclips in 3 cases, adrenalin injection in 13 cases and APC in 1 case. There was no acute or late bleeding. There was no need for blood transfusion.

With the application of the Pearson χ² test, the association between the studied variables was shown to be statistically significant: bleeding and extra intestinal diseases (P < 0.0001) and the histology (P = 0.016). In particular, between bleeding and cardiac diseases (P = 0.034) and tubular adenoma (P = 0.016). The Pearson χ² test was also applied in the sub-groups for evaluation of the association between the bleeding and the studied variables and the P-values are illustrated in Table 3.

Statistical analysis performed by NPC test showed a highly significant difference between the four groups examined in relation to bleeding; the P-values of the analysis are shown in Table 4.

The results of the Binary Logistic Regression (Table 5) allowed confirmation that the associated extra intestinal diseases, histology and the size of the polyps are statistically significant and are linked to the occurrence of bleeding. For the aforesaid model the Log-Likelihood test assumes a value of -62.662, the G test is equal to 57.276 with 9 degrees of freedom and the P-value is 0.000; for this reason we can affirm that the chosen model is adequate to examine the data.

Perforation occurred in one patient (0.07%) after polypectomy of a malignant sessile polyp in the sigmoid colon, but in this lesion of 15 mm there was no suspicion of malignancy. This patient died after surgery repair of perforation from respiratory failure.

In the past two decades the technique of endoscopic polypectomy or mucosal resection has been significantly improved and is the most common therapeutic procedure performed in the endoscopic unit. A particular challenge is the endoscopic treatment of large polyp because the procedure is difficult and reserved for experts since complications rates are very high. In this study, we described a series of 1354 endoscopic resections of colon rectal polyps performed in our endoscopic units between January, 1996 and May, 2006. The gender and age distribution of patients is similar to that described above in other studies regarding the treatment of colorectal polyps[3–19].

From the literature, it emerges that over 80% of polyps resected during colonoscopy are small polyps less than 10 mm in diameter[24] while in our study 67% were 10 mm in diameter. In fact, during a ten-year period we found 11.8% of excised polyps to be 20 mm or more in size out of 1354 total polypectomies. These data illustrate an average of 16 large polyps removed annually. Other authors reported averages of between 8 and 21 large polyps removed annually[3–19]. In the present study, all patients with large colorectal polyps were treated endoscopically and in all cases the complete removal of all pedunculated and sessile polyps was possible.

Several previous publications reported a correlation between malignancy and age of patients, polyp shape, histology, location and higher rates in large polyps up to 50%[625–31]. In the present study, 28 polyps were found to be adenomas containing an area of carcinoma (2.1%) and of these, 20 (12.5%) were in the group of large polyps; however, none of these polyps showed neither vascular or lymphatic invasion.

Statistical analysis, performed by biserial correlation coefficient and the χ² test, showed a correlation between cancer, polyp size (P < 0.0001), sex (P < 0.0001) and sessile shape (P < 0.0001). In accordance with previous publications, we documented a correlation with size, sessile shape and sex. Moreover, our data suggest that invasive carcinoma can appear with an equal probability, both in a tubular or tubulovillous or villous adenoma. Finally, there was no correlation with the site of the polyp along the colon.

Although the complications of polypectomy widely vary as presented in literature, the most frequent remain haemorrhage and perforation, which are often related to the size of the polyp, its morphology (sessile or pedunculated) and location[1432]. The incidence of bleeding during and after the polypectomies has been reported to range from 0.3% to 6.1%, with higher rates in large polyps up to 22.1% (Table 6)[3–19].

In our series, the Pearson χ² test applied in the sub-groups found a correlation between the bleeding and liver diseases (P = 0.003), which usually impairs the coagulation function of the patients and location of polyps in transverse colon (P < 0.0001), in group 2 with histology and in particular with the hyperplastic polyps (P = 0.023) and the malignancy (P < 0.0001). In addition, in group 3 bleeding was correlated with liver diseases (P = 0.001) and in group 4 the correlation was between the bleeding and associated colon rectal cancer (P = 0.036) (before or after surgical resection).

Moreover, the statistical analysis performed by NPC showed that bleeding was related to polyp size because the large polyps bled more than small polyps. These results were confirmed by Binary Logistic Regression. However, the 17 procedural bleeding that we recorded were endoscopically resolved without surgery or blood transfusion, so we can, therefore, consider these cases “non important complications”. In large polyps, the rate of bleeding observed, 6.2%, is similar to that reported in the greater number of the studies (Table 6). Nevertheless, this data cannot be considered as a real complication, due to the immediate resolution during the same procedure, by endoscopic means.

The incidence of perforation during therapeutic colonoscopies has been reported to range from 0.08%-2.2%[33]. The reported incidence of perforation during polypectomy of “normal-sized” polyps ranges from 0.3%-0.5%[34], while the incidence of injury to the colon wall (transmural burn, microperforation, or free perforation) in large polypectomies is 0 to 1.3% (Table 6)[3–19].

In our study, no perforation was reported in the group of large polyps while the only late perforation (0.07%) occurred, after polypectomy, in a small (15 mm) malignant sessile polyp located in the sigmoid colon and without endoscopic sign of suspected malignancy. During the removal of this polyp, a procedural bleeding occurred, which was immediately managed by application of two hemoclips and the patients’ hospitalization. Twenty four hours after the polypectomy, the patient developed lower left quadrant pain, tenderness and radiographic evidence of free air in the peritoneal cavity. The patient underwent surgery for a small (1-2 mm) perforation of the sigma. Unexpectedly, this patient died four days after surgery from respiratory failure secondary to a pleural mesothelioma, unacknowledged and diagnosed the during post mortem examination.

Summarizing, we can conclude that this study, confirming the findings of several others, demonstrates that the endoscopic polypectomy, performed by an expert hand, is safe and effective and should be considered the treatment of choice also for large colorectal polyps.

Polyp size has been identified as an important risk factor for both malignancy and bleeding. Haemorrhage was the most frequent complication even if it remains questionable whether the bleeding should be considered a complication when it occurs during the procedure and is effectively and immediately controlled by endoscopic means. Similarly, in fact, if during a surgical procedure an artery is sectioned and bleeding, this event is not considered a procedural complication because it is treated successfully. It seems likely there will be, in the upcoming future, a growing need for redefinition of the concept of endoscopic operative complications.