Metabolic factors accelerate colorectal adenoma recurrence

This multicenter retrospective study was conducted with the participation of 5 community hospitals in Japan, including Chigasaki Municipal Hospital, Yokohama Rousai Hospital, Tokyo Metropolitan Hiroo Hospital, Machida Municipal Hospital, and NTT Medical Center Tokyo. We analyzed the medical records of 43,195 patients who had undergone colonoscopy between January 2005 and December 2011 at any one of the 5 hospitals. Among these, only those who had undergone complete colonoscopy for the purpose of screening were enrolled in this study (n = 32,566); the remaining 10,629 patients were excluded either because they had undergone colonoscopy for diagnostic purposes, or did not undergo complete colonoscopy. Next, 24,123 subjects who had not undergone removal of adenomas ≥5 mm in size at the time of the screening colonoscopy were excluded, and the remaining 8443 subjects who had undergone removal of adenomas ≥5 mm in size were included in this study (smaller adenomas have been shown to be of minimal clinical significance [11]). Furthermore, from among the 8443 subjects, we excluded 6703 subjects who did not undergo surveillance colonoscopy at 1 year and 2 years after the initial colonoscopy (that is, all the eligible patients enrolled in this study had undergone colonoscopy 3 times; colonoscopy is very cheap in Japan, and annual surveillance colonoscopies are common). To be eligible for this study, the participants had to have undergone complete colonoscopy at all the three examinations. A complete colonoscopy was considered to include: colonoscopy up to the level of the cecum, good bowel preparation, and removal of all the detected adenomas. Thereafter, further participants were excluded for the following reasons (total n = 629): no history or no laboratory data available (n = 327); age less than 50 years old or greater than 85 years old (n = 120); history of colorectal resection or appendectomy (n = 88); history of removal of carcinoma (n = 28); history of inflammatory bowel disease (IBD), familial adenomatous polyposis (FAP) or hereditary non-polyposis colorectal cancer (HNPCC) (n = 18); history of regular use or use for over 7 days a month of aspirin or non-aspirin steroidal anti-inflammatory drugs (NSAIDs) (n =31); and patients with a life expectancy of less than 2 years because of severe diseases such as cancer, severe liver dysfunction, severe renal dysfunction, severe infection, etc. (n = 17); Finally, 1111 participants were included for the analysis in our study (Figure 1). To minimize the selection bias, all of the patients who had undergone colonoscopy at the 5 hospitals over the 7-year period were assessed, and all of the patients who fulfilled the inclusion criteria but not the exclusion criteria were enrolled in this study.

The study protocol was approved by the individual ethics committees of each of the 5 participating hospitals, that are Chigasaki Municipal Hospital, Yokohama Rousai Hospital, Tokyo Metropolitan Hiroo Hospital, Machida Municipal Hospital, and NTT Medical Center Tokyo.

A total of 20 data points at the initial colonoscopy and colorectal adenoma recurrence over the 2-year observation period were obtained from the medical records. The 20 items evaluated were: age at initial colonoscopy, gender, body mass index (BMI), current alcohol consumption, current cigarette smoking, the present history of hypertension, diabetes mellitus and dyslipidemia, fasting blood glucose (FBG), hemoglobin A1c (HbA1c), serum total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C) and triglycerides (TG), morphology/villous features/grade of dysplasia/location/size of the largest removed adenoma, and number of removed adenomas.

Siegel et al. showed that the prevalence of colorectal tumors increased in adults aged 65 and older [1]. So, age was categorized into ≥65 and <65. BMI was calculated as the weight (in kg) divided by height squared (in m²). Obesity was defined as BMI ≥25 [12]. BMI was classified into ≥25 and <25. Current alcohol intake history was defined as positive when the consumption was estimated to be in excess of 30 g/day. Present history of hypertension was defined as blood pressure ≥130/85 mmHg and/or current use of antihypertensive medication [13]. Present history of diabetes mellitus was diagnosed if the patients had a random glucose level of ≥200 mg/dl, FBG of ≥126 mg/dl, HbA1c of ≥6.5% or were taking antidiabetic medication [14]. Present history of dyslipidemia was defined as TG ≥150 mg/dl and/or HDL-C <40 mg/dl or were taking antidyslipidemic medication [13]. The FBG, HbA1c, TC, HDL-C, LDL-C, and TG were determined by laboratory testing at the time of the initial colonoscopy. FBG, HbA1c, HDL-C, and TG were classified into two groups with the above cutoff values. TC ≥220 and LDL ≥ 140 were adopted as cutoff values according to the Japanese definition [15]. For the largest among the removed adenomas at the time of the initial colonoscopy, we determined the presence/absence of >20% villous components, grade of dysplasia (high/low), location, and size. The location of the largest adenoma was classified as right-sided if it was in the cecum, ascending colon or transverse colon, and as left-sided if it was in the splenic flexure, descending colon, sigmoid colon or rectum. Colorectal adenoma recurrence was diagnosed if an adenoma(s) ≥5 mm in size was found in the surveillance colonoscopy, and both newly developed colorectal adenomas and colorectal adenomas recurring at the same location as that of the previously removed adenoma were included.

Statistical analysis was conducted using the JMP 10.0 software. Chi-square analysis was used to compare differences in variables. Univariate analysis was performed by logistic regression analysis. Multivariate analysis was performed by multiple logistic regression analysis. The correlation between the risk score and colorectal adenoma recurrence were assessed by analysis of the receiver operating characteristic (ROC) curves. Differences were considered statistically significant at p <0.05.

The baseline clinical characteristics of the study participants with and without colorectal adenoma recurrence developing within 2 years of the initial colonoscopy are provided in Table 1 and clinical characteristics of the removed adenomas are shown in Table 2.

During the 2-year follow-up after the initial colonoscopy, 471 of the 1111 patients (42.4%) were identified as having recurrence of colorectal adenomas whereas the remaining 640 patients (57.6%) were free of recurrence (Table 1). Of the 471, the recurrence was identified at the 1-year surveillance colonoscopy in 313 patients (28.2%), and at the 2-year surveillance colonoscopy in the remaining 158 patients (14.2%).

The characteristics of recurrent adenomas are demonstrated in Table 3. Among recurrent adenomas, advanced adenomas occupied 10.8% (51/471). Location of the largest recurrent adenomas tended to exist in the right-sided colon.

Each of 20 risk factors evaluated was found to be statistically significant, except for the age, gender, current alcohol consumption, TC and HDL-C. Although older males were more likely to undergo adenoma removal at the time of the initial colonoscopy, age and gender were not identified as statistically significant factors influencing the risk of colorectal adenoma recurrence. While current alcohol consumption was not found to be statistically significant, current cigarette smoking was statistically significant. Metabolic factors such as the BMI, hypertension, diabetes mellitus and dyslipidemia were found to be statistically significant factors, as were the FBG, HbA1c, LDL-C and TG. Among the metabolic factors, TC and HDL-C were not statistically significant. In regard to the adenoma-related factors, non-polypoid lesion, presence of >20% villous components, presence of high-grade dysplasia, a larger maximum diameter and a larger number of removed adenomas were found to be statistically significant risk factors for colorectal adenoma recurrence. Right-sided adenomas were associated with a higher risk of recurrence than left-sided adenomas. The odds ratio and p value of each factor with univariate analysis are shown in Table 4.

Multiple logistic regression analysis was performed with 15 of 20 risk factors. There was collinearity among a present history of diabetes mellitus, FBG and HbA1c. Among these 3 factors, FBG was used in the multivariate analysis because the odds ratios of FBG were maximum among these in the univariate analysis. There was also collinearity among a present history of dyslipidemia, TC, HDL-C, LDL-C and TG. The odds ratios of LDL-C and TG were greater than that of others, and the correlation coefficient between LDL-C and TG was low (correlation coefficient = 0.0992). So, among these factors, LDL-C and TG were used in the multivariate analysis.

The results from the multivariate analysis identified 7 factors, namely, age, BMI, FBG, dysplasia grade/location/size of the largest adenoma, and number of the removed adenomas as being statistically significant factors influencing the risk of colorectal adenoma recurrence. Villous features were almost statistically significant (p = 0.063). The odds ratio and p value of each factor from the multivariate analysis are shown in Table 5.

Analysis was performed using 8 factors (Table 5): For adenoma-related factors, 5 factors, namely, villous features/grade of dysplasia/location/size of the largest adenoma, and number of the removed adenomas were included. For metabolic factors and other factors, 3 factors, namely, age, BMI, FBG were included. Seven of 8 factors were adopted because they were statistically significant in the multivariate analysis. Villous features were included because it was an established risk factor in the AGA guidelines, and was also nearly statistically significant in our study in the multivariate analysis including metabolic risk factors.

One point was scored for if each positive factor. But, because the odds ratio of the number of removed adenomas in the multivariate analysis was almost 3 times bigger than that of other factors, the number of removed adenomas counted as 3 points if it was positive. The risk score was calculated by summing up each individual score, and ranged from 0 to 10 points.

The correlation between the risk score (0–10 points) and colorectal adenoma recurrence during the 2-year follow-up after the initial colonoscopy was analyzed with a logistic regression analysis, which revealed a statistically significant correlation (p <0.001).

ROC curve analysis revealed an AUROC for colorectal adenoma recurrence of 0.80 (Figure 2). The sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for each risk score were calculated (Table 6). The recurrence rate of colorectal adenomas during the 2-year follow-up period after the initial colonoscopy, the number of cases showing adenoma recurrence and the total number of cases for each score are also shown in Table 6. When the cutoff value for the risk score was set at 3, the odds ratio of ≥3 points (3–10 points) relative to <3 points (0–2 points) for adenoma recurrence was determined to be 7.07 (95% CI 5.30–9.43). In those cases where the risk score was ≥3 (3–10) points, the colorectal adenoma recurrence rate during the 2-year period was 59.8%, and the sensitivity, specificity, PPV and NPV of the detection of adenoma recurrence were determined to be 83.2%, 58.8%, 59.8% and 82.6%, respectively (Table 7).