Long Term Outcomes of Colon Polyps with High Grade Dysplasia after Endoscopic Resection


 Background

The clinical characteristics of patients with high-grade dysplasia (HGD) in colon polyps at baseline colonoscopy in relation to the risk of recurrent colonic adenomas have been unclear. We conducted this hospital- based cohort study recruiting patients who had undergone polypectomy at screening colonoscopy to assess the risk factors of recurrent colonic adenomas.
Methods

11,565 subjects had undergone screening colonoscopic examinations between September 1998 and August 2007. The patients who had HGD in colon polyps and had undergone surveillance colonoscopy were eligible.
Results

A total of 211 patients were recruited. The rates of metachronous adenomas and advanced adenomas at surveillance colonoscopy were 58% and 20%, respectively, during the mean follow-up period of 5.5 ± 1.8 (3–12) years. On univariate logistic regression analysis, the number of adenomas ≥ 3 at baseline colonoscopy was strongly associated with the rates of overall recurrence, multiple recurrence, advanced recurrence, proximal recurrence, and distal adenoma recurrence with odds ratios of 4.32 (2.06–9.04 95%CI), 3.47 (1.67–7.22 95%CI), 2.55 (1.11–5.89 95%CI), 2.46 (1.16–5.22 95%CI), 2.89 (1.44–5.78 95%CI), respectively. On multivariate analysis, male gender [P = 0.010; OR 3.09(1.32–7.25 95% CI)] and number of adenomas ≥ 3 [P = 0.002; OR 3.08(1.52–6.24 95%CI)] at index colonoscopy were significantly associated with recurrent advanced adenomas.
Conclusions

Recurrence of colonic adenomas at surveillance colonoscopy is common in patients who had undergone polypectomy for colon adenomas with HGD at baseline colonoscopy. The risk of developing future advanced adenomas is associated with the number of colon adenomas and male gender.


Background
Colorectal cancer (CRC) is the third leading cause of cancer-related death in Taiwan [1]. CRCs were found to originate from the initially benign colon adenomas that followed an adenoma-carcinoma sequence.
Colonoscopy and polypectomy interrupted this adenoma-carcinoma sequence and reduced the incidence of CRC [2][3][4][5]. The occurrence of CRC was also effectively prevented by detection and concomitant removal of advanced adenomas that were de ned as adenomas greater than 10 mm, with high grade dysplasia (HGD) or ≥ 20% villous component [6]. The rates of recurrent adenomas after polypectomy have been reported to be 29-58% ve years after screening [5,7]. Previous studies have suggested that the characteristics of adenomas at index colonoscopy are closely related to recurrence of adenomas. Some studies have shown that location, size, histological type, presence of atypia, and number of adenomas detected at the index colonoscopy were risk factors for adenoma recurrence [8][9][10][11].
Whether advanced adenomas (villous adenoma, severe dysplasia or adenoma size ≧ 10 mm) at index colonoscopy have a more aggressive behavior leading to a faster growth and an earlier recurrence than the normal counterparts are still unclear [11]. In the univariate analysis from the NCI Pooling Project, adenomas with HGD were shown to be strongly associated with the risk for advanced neoplasia during surveillance colonoscopy (OR, 1.77; 95% CI, 1.41-2.22) [11] but the result was not demonstrated by other studies. There were few studies addressing the long-term outcomes of patients with HGD in colon polyps after polypectomy at index colonoscopy [12,13]. We therefore conducted this retrospective cohort study.
The primary goals of this study were to nd out the cumulative incidence rate of recurrent advanced neoplasia in patients with HGD at baseline screening colonoscopy and the association between baseline endoscopic characteristics and the subsequent risk of advanced neoplasia.

Patients selection
After obtaining approval from the Institutional Review Board of Keelung Chang Gung Memorial Hospital (IRB No. 104-6993D), an extensive chart review was undertaken to review the colonoscopic reports and pathology reports of 11,565 patients who had undergone consecutive colonoscopic examinations at the Chang Gung Memorial hospital (Keelung, Taiwan) between September 1998 and August 2007. Patients who had adenoma at screening and had undergone surveillance colonoscopic examinations were identi ed. Patients with HGD in colon adenomas were selected and reviewed for their eligibility.
The patients with previous colon cancer or metachronous colon cancer at the index study, in ammatory bowel diseases (including Crohn's disease or ulcerative colitis), familial adenomatous polyposis (FAP), no surveillance total colonoscopy in three years after screening polypectomy, incomplete colonoscopy study, other systemic diseases causing death during surveillance studies were excluded.
A complete colon visualization was required for any patient who had undergone a colonoscopic examination at the colonoscopy unit of this hospital. For the patient who had colonic neoplastic lesions at colonoscopy, polypectomy was routinely performed with biopsy forceps (small lesions less than 5 mm), cold snare polypectomy, hot snare polypectomy or endoscopic submucosal resection. The gross appearance, size, histology, and location of the adenomas at baseline colonoscopy of the recurrence group were compared to that of the recurrence free group to illuminate the risk factors for future adenoma recurrence. The adenomas were categorized as protruding (0-Ip), subpedunculated (0-Ips), sessile (0-Is), lateral spreading (0-IIa) and depressed lesions (0-IIc) according to the Paris endoscopic classi cation of super cial neoplastic lesion [13]. The size, morphology and location of the adenomas were recorded in the reports. The size of the adenomas was measured by an open biopsy forceps (diameter = 8 mm) or by a rule for the resected specimens. The locations of the adenomas were de ned as proximal colon (caecum, ascending colon, hepatic exure and transverse colon), distal colon (splenic exure, descending colon, sigmoid colon and rectum).
The removed adenomas were obtained for pathologic examination. Advanced adenomas were de ned as any adenoma greater than ( ≥ ) 10 mm, with villous or villotubular histology(≥ 20% villous component), or adenoma with high grade dysplasia (severe dysplasia or intramucosal carcinoma). The polyps were de ned as tubular, tubulovillous, villous adenoma, or serrated polyp according to the histological characteristics.
A second look colonoscopy was performed 6 to 12 months after the index colonoscopy when the adenomas were removed by the piecemeal method or an incomplete resection was suspected. The patients who had undergone surveillance colonoscopy at least three years after the initial polypectomy were eligible for inclusion in this study. The clinical characteristics including gender, age, indication of colonoscopy, family history of colon cancer, tumor number, size, location and histology of the adenomas were analyzed in this study.
Continuous variables were expressed as mean ± standard deviation (mean ± SD) and categorical variables as frequencies or percentages. Independent t -test or chi-square test were used to compare the differences between all the patients with high grade dysplasia with different demographic characteristics (Table 1) and between recurrence and recurrence free groups ( Table 2) as appropriate. Univariate logistic regression analysis was performed to obtain the odds ratios of all predictors for polyp recurrence (Table 3 and  Table 4). Multivariate logistic regression analysis was performed to nd out the risk ratios of any adenoma recurrence and advanced colorectal neoplasia recurrence base on baseline characteristics of patients and colon adenoma (Table 5). A p value < 0.05 was considered statistically signi cant and all calculations were two sided. All statistical analyses were performed using IBM SPSS 22.0 (IBM Inc., Armonk, New York). *The Paris endoscopic classi cation of super cial neoplastic lesion. ** Distal colon (splenic exure, descending colon, sigmoid colon and rectum), ***Proximal colon (caecum, ascending colon, hepatic exure and transverse colon),     Forty-seven of 258 patients failed to complete surveillance colonoscopy and were excluded. Figure 1 shows the selection and exclusion results. A total of 211 patients were included for analysis. The mean age was 64 ± 12 years. The mean follow-up period was 5.5 ± 1.8 (3)(4)(5)(6)(7)(8)(9)(10)(11)(12) years after index colonoscopy. 129 patients were male and 82 patients were female. 123 patients (58%) had recurrent adenomas at surveillance colonoscopy.
There was no statistical difference between the recurrence and recurrence free group in terms of gross appearance, histologic nding, location, and size of adenomas at baseline colonoscopy ( Table 2). The number of polyps at index colonoscopy was associated with future adenoma recurrence (P = 0.03) A subgroup analysis for the impact of polyp number and location of polyp on the recurrence of polyps showed that patients who had adenoma number ≥ 3 were associated with a greater risk of metachronous adenoma recurrence (Table 3). For the distal colon polyps with number ≥ 3 at the index colonoscopy, the rates of overall recurrence, multiple recurrence, distal colon recurrence, proximal colon recurrence and advanced adenomas recurrence were 93%, 60%, 87%, 27% and 40%, respectively. In contrast, for the distally located polyps with number < 3, the overall recurrence rate, multiple recurrence rate, distal colon recurrence rate, proximal colon recurrence rate and advanced adenomas recurrence rate were 49%, 35%, 36%, 22% and 12%, respectively. For the adenomas located at the proximal colon at index colonoscopy, a similar odds ratio was also observed in the group of adenomas ≥ 3 in number but not in the group with the number < 3 ( Table 3). The odds ratios for the overall recurrence, multiple recurrence, distal recurrence, proximal recurrence and advanced adenoma recurrence of adenomas ≥ 3 in number and < 3 in number were 13 (Table 3).
On univariate logistic regression analysis, the number of adenomas greater than 3 at the baseline colonoscopic examination was strongly associated with risks of overall recurrence, multiple recurrence, advanced recurrence, proximal recurrence, and distal adenoma recurrence (

Discussion
The rates of developing metachronous adenoma and advanced adenoma after polypectomy for colon polyps with HGD were 58% and 20%, respectively, in the current study. The clinical features of age, gender, stool occult blood positivity, size of polyp, morphology, pathology and location of polyps were not different between the patients who had developed recurrent adenomas and those who had not.
There was no metachronous adenocarcinoma observed during the period of this cohort. The incidence rate of adenomas was 35.9% in our patients who had undergone screening colonoscopy. The rate is higher than that of the Western population of 20% [14] and 8.13 to16.5% of Taiwanese and Chinese ethnic groups, respectively [15][16][17], because our study subjects were mostly referred on account of positivity for stool occult blood or symptomatic of gastrointestinal disorders.
The reported reduction rates of colon cancer after polypectomy were around 50-80% and the occurrence of interval cancers was located mostly in the right colon [18][19][20]. The reason accounting for higher recurrent rates of interval carcinoma in the right colon included incomplete colonoscopy study due to poor endoscopic technique [21], incomplete removal of the polyps [22], di culty in visualization of smaller adenomas in the right colon [23][24], or sessile serrated adenoma in the proximal colon.
At screening colonoscopy, the prevalence of polyps with HGD was 2.2% (258/11565) in our patients, which is slightly lower than that of 3.5% in the general population of Chinese ethnic [26]. There was no interval cancer during the follow up period in the current study, because we performed the second look colonoscopy 6 to 12 months after piecemeal resection for advanced polyps or suspected incomplete resection to ensure complete resection of polyps and to decrease the rate of undetected adenomas at the index colonoscopies, as has been suggested by some endoscopists in treatment using the piecemeal method [25][26]. Recurrent adenomas after polypectomy were reported to be about 36 to 64% within 2 to 6 years [12,[27][28][29][30]. In the current study, the overall recurrence rate of metachronous adenomatous polyps was 58%, which is similar to that of 64% in Toll's study [12]. The rate of recurrent advanced adenomas was 20% in the current study which is lower than that of 40% in Toll's report.
HGD in adenoma was considered following the adenoma-adenocarcinoma sequence and a precursor of adenocarcinoma [31]. However, the risk of future advanced adenomas in relation to HGD at index colonoscopy was reported to be small and variable.
In the meta-analysis study of Saini [32], the patients with HGD in polyps had a relative risk of advanced adenomas of 1.84 (95% CI 1.06-3.19) compared to those who had no HGD. Some other studies including a randomized controlled trial revealed no association of HGD with subsequent advanced adenomas during surveillance colonoscopy [33][34][35][36].
Two meta-analyses [37][38] have also shown that the presence of HGD had a small association with the risk of future advanced adenomas. On the multivariate analysis in these studies, the presence of HGD did not confer recurrence of metachronous adenomas. Most of the advanced adenomas are ≥ 1 cm [39]. However, if the adenomas with HGD can be removed completely, the risk for metachronous recurrence is probably dependent on the number of colon polyps and male gender as shown in this study. The rationale is also applicable to the association of villous histology of the resected adenomas with the risk of recurrent advanced adenomas. A cohort study had shown that the presence of villous adenomas at baseline colonoscopy had a relative risk of 1.8 of future adenomas [33]. Two other studies including one metaanalysis and a pooled analysis had found no association of villous histology with future adenomas [32,14], which is in accordance with the present study result. The size of colonic adenomas has been shown to be closely related to the histological features of HGD and invasive cancer [40]. HGD in histology or prominent villous component was seen in 87.5% of large polyps (≥ 1 cm) [95% CI = 86-89.4]) [41]. The risk of recurrent advanced adenomas was shown to be corresponding with the size of polyps at index colonoscopy. In Martinez's study [14], with polyp size of 0.5 cm as the reference group, the adenoma size of 1.0 to 1.9 cm or ≥ 2.0 cm at the baseline colonoscopy had a relative risk of 2.3 and 3.0, respectively, of recurrent advanced adenomas. There were four other studies showing an increase of relative risk of recurrent advanced adenomas in colonic adenomas greater than 1 cm at base-line [22,32,[34][35][36]. The relative risk of polyps with HGD at baseline was 6.87 (95% CI 2.61-18.07) of interval advanced neoplasia in Lieberman's study, in which 6 among 11 patients who had recurrent cancers or high-grade dysplasia were found in the portion of colon where the polyps were resected [22]. In the current study, the size of adenomas greater than 1 cm does not increase the risk of advanced adenomas on surveillance colonoscopy, which is in line with Saini's meta-analysis [32]. Incomplete removal or missed adenomas [21][22] might account for the contradictory observations.
The number of adenomas ≥ 3 has been shown to increase the risk of recurrent advanced adenomas. The relative risk was between 1.5 and 5.0 [22,32,[34][35][36] with one adenoma as the reference group. The current study in concert with the previous studies has found odds ratios of 2.45 (1.22-4.93) on univariate analysis and 2.11(1.00-4.43) on multivariate analysis, respectively, when comparing adenoma number ≥ 3 to adenoma number ≤ 2. Male gender is also associated with advanced adenoma recurrence in his study with an odds ratio of 3.09, which is consistent with Zhang's study that reported male gender and the number of adenoma ≥ 3 had high risk of advanced adenoma recurrence [17].
The natural history of colonic adenomas is still elusive. Two longitudinal follow up studies for small polyps (6-9 mm) using computed tomography showed the tumor progress rates of 22% [42] and 35% [43] during the follow up periods of 8 and 3 years, respectively. Advanced histology was seen in 47% of progressing polyps [43], which is similar to the rate of 40% in our patients with the baseline adenoma number ≥ 3. The high incidence rate of recurrent adenomas and advanced adenomas in patients with multiple adenomas is hard to explain solely on incomplete resection of adenomas. The multiple small polyps that were not detected on optical colonoscopy at baseline might have progressed slowly with time and became detectable on surveillance colonoscopy. The multiplicity or polyclonicity in patients with adenoma number ≥ 3 in number is a reasonable explanation but a longer observation with other non-invasive study modalities such as computed tomography or capsule endoscopy to detect missed adenomas may be required to elucidate the natural history of recurrent adenomas. The surveillance recommended by the United States Preventive Services Task Force (USPSTF) was 3 years after removal of advanced adenoma, traditional serrated adenoma, or advanced sessile serrate adenoma [44]. The European guidelines [45] recommended a more aggressive surveillance at 1 year for high risk polyps (≥ 20 mm). In the treatment with piecemeal resection, Walsh et al revealed a rate of 14% polyp recurrence after at least one negative examination, and the rate of CRC development was 17% in 65 patients with large at polyps [46]. A second look examination for patients who underwent piecemeal resection or suspected incomplete resection may be warranted.
The current study had some limitations. We did not take into account the impact of diet change or life style change such as abstinence of smoking, alcohol drinking and the reduction of body mass index after polypectomy, as which might have some in uences on the recurrence of adenomas or advanced adenomas [47][48]. The current study did not include the low risk patients as control group and the surveillance time of patients recruited was not the same. This might in uence the rate of recurrent adenomas.

Conclusions
In conclusion, colon polyps recurrence is common in the patients with high grade dysplasia at index colonoscopy. The number of polyps greater than 3 (≥ 3) and male gender have a higher risk for future occurrence of adenomas and advanced adenomas.

Declarations
Ethics approval and consent to participate The study received approval from the Institutional Review Board of Keelung Chang Gung Memorial Hospital (IRB No. 104-6993D), Informed consent was not required, because this study involved chart reviewing of colonoscopy reports and pathology report.

Consent for publication
Not applicable Availability of data and material The data of colonoscopy reports and pathology reports between September 1998 and August 2007 can be available from the electronic medical record at the Chang Gung Memorial hospital (Keelung, Taiwan) between September 1998 and August 2007.
The datasets generated and/or analysed during the current study are not publicly available due the hospital's principle and the right of privacy of patients but are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests" Funding: The study is an original research supported by a research grant from the Chang Gung Memorial Hospital

CMRPG2303
There was no other funding body in the design of the study and collection, analysis, and interpretation of data and in writing the manuscript

Authors Contribution
JJC study design, analysis and drafting of manuscript.
SWC, LWC, CJ L provision of clinical data CHC provision of clinical data, analysis and interpreted the data CLY data collection, study design and revision of the manuscript.
All authors read and approved the nal manuscript.

Figure 1
Diagram showing selection and exclusion of eligible patients