Frequent co-occurrence of high-grade dysplasia in large flat colonic polyps (>20 mm) and synchronous polyps

By retrospectively investigating a large single center cohort, we show that flat colonic polyps >20 mm have a distinct distribution of histology and localization. Overall, our results demonstrate that large flat colonic polyps preferentially occur in the proximal colon, as shown by others [11, 13, 16]. Large flat colonic polyps containing invasive and non-invasive adenocarcinoma are found at a frequency of 2.5–7 % [11, 12, 15, 17]. The percentage of adenocarcinoma in our cohort was slightly higher (11.2 %), possibly due to the high fraction of rectal polyps, which we found to contain adenocarcinoma more frequently. The percentage of polyps containing adenoma with high-grade dysplasia varied greatly between different studies, ranging from 7.5 % to 40.5 % [14, 15, 17]. This divergence among studies may reflect discrepancies in selection criteria of polyps. For example, some studies excluded rectal polyps [16] while others did not [14]. The fraction of HGD in our cohort (26.7 %) was within the reported range. The rate of sessile serrated adenomas/polyps was shown to be less than 5 % in patients undergoing screening colonoscopy [21, 26]. We found a higher rate of serrated polyps in our cohort (9.6 %). This may reflect the true prevalence of SP within large colonic polyps, which is supported by a recent study that also found a higher frequency of 20 % [12]. The correlation between histology and location of large colonic polyps showed that large adenomas with HGD and adenocarcinomas are predominantly located in the rectosigmoid colon, which is in line with previous observations [11, 13, 16]. While small hyperplastic polyps are mostly left sided, sessile serrated adenomas/polyps are located predominantly in the proximal colon [5]. Indeed, the majority of SP in our cohort were sessile serrated adenomas/polyps and we found a preference towards a right-sided location. We identified polyp size and age to be independently associated with advanced dysplasia and carcinoma in large polyps. Polyp size has been known to correlate with malignant conversion, with rates of HGD increasing significantly for polyps >20 mm [7, 27]. Advanced patient age has also been previously described to be associated with HGD in adenomatous polyps [10]. Our data shows that patients with large flat, serrated polyps were significantly younger than patients with large polyps of other histology. The majority of SP in our cohort harbor no dysplasia or only low grade dysplasia and this subgroup of SP is known to preferentially occur in younger patients [26]. In summary, our cohort of patients represents clinicopathological features that support previous observations for large flat adenomas.

A main focus of this study was to investigate how specific features of large polyps affect prevalence and characteristics of synchronous polyps. Synchronous polyps are reported to occur in 21–29 % of patients undergoing screening colonoscopy [28, 29] and 36 % with colorectal cancer [19]. The rate of synchronous polyps was much higher in our cohort (67.4 %), indicating that large polyps are generally associated with synchronous polyps. A potential link between the histology of large polyps and synchronous polyps was proposed by reports showing that large proximal SP were associated with synchronous advanced dysplasia [20, 22]. Our study demonstrates that presence of HGD in the large polyp is strongly associated with co-occurrence of synchronous polyps with the same histology. This observation is supported by a study showing that occurrence of adenomas with advanced dysplasia in the distal colon is predictive of advanced lesions in total colonoscopy [30]. In addition, we found that HGD in any synchronous polyp was also associated with an increased overall number of synchronous polyps. These observations provide further clinical evidence for the theory of field carcinogenesis, which implies that the environmental milieu that leads to carcinogenesis is not a local event, but affects larger parts of the colon [31].

The majority of large flat polyps in our cohort was detected by primary or secondary (gastroenterological) centers and then referred to our institution for removal. This might introduce a selection bias towards polyps that are most difficult to resect, due to their morphology or location, e.g., in the proximal colon. In some cases, there was documented removal of small, synchronous polyps by external gastroenterologists prior to admission. We must therefore assume that the fraction of patients with synchronous polyps and the number of polyps per patients are underestimated by our study. In addition, a considerable proportion of colonoscopies could not be included in our analysis of synchronous polyps, as they were considered incomplete. This may introduce a selection bias, but comparison of clinic-pathological characteristics of polyps between the complete and incomplete colonoscopy group did not reveal significant differences (data not shown). Due to the retrospective design, there was no predefined scheme for the documentation of polyp morphology. In some cases, polyp morphology was determined by post hoc review of endoscopic images which is prone to inter-observer variation and thus imprecise. Furthermore, quality of bowel preparation and polyp detection rate of endoscopists were not documented in a standardized manner. Therefore, these parameters could not be comprehensively assessed in our retrospective approach. In addition, this study includes data from a period of ten years. During this time, improvements to endoscopy imaging technologies have been introduced which partially increased adenoma detection rates [32, 33]. In summary, these limitations indicate that the rate of synchronous polyps is likely underestimated due to our retrospective approach.

What is the impact of this finding for clinical practice? First, our data underlines the importance of a complete colonoscopy prior to removal of large polyps, especially if high grade dysplasia or malignancy is highly suspected due to the presence of “red flags” (advanced age, localization in the rectosigmoid, large size). Secondly, surveillance after resection should always include a complete colonoscopy and not be restricted to the site of the large polyp. This is of particular importance as the incidence of metachronous cancer is higher in patients with synchronous polyps [34‐36]. Lastly, our data supports that endoscopic instead of surgical removal of large polyps should be considered first. Surgical removal of large polyps is associated with a considerable level of post-operative complications [37] and may not be ideal to remove synchronous or metachronous adenomas at multiple sites. In contrast, endoscopic resection allows for repetitive and multilocal removal. If endoscopic removal fails, surgery can still be performed without compromising oncological outcome [38].