Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
See article on page 818
Konishi et al (see page 818) have documented morphological findings in 4147 colorectal neoplasms detected colonoscopically in a consecutive series of 5025 average risk subjects. The importance of their report lies in the understated and sobering perspective it brings to the concept of de novo colorectal cancer. They used high resolution magnifying endoscopy to detect, remove, and classify all epithelial neoplasms. The neoplasms were grouped as mucosal lesions (3605), submucosal cancers (91), and advanced cancers (451), although the advanced cancers were not considered further. The mucosal lesions included 3353 adenomas and 252 intramucosal carcinomas. In the West, the latter would be classified as adenomas with high grade dysplasia1 and for the purposes of this study were grouped with the adenomas, and are subsequently referred to as such.
The neoplasms were also grouped into depressed and non-depressed, the latter including flat, slightly raised, and polypoid lesions. The rationale for this was the previous demonstration of de novo cancer associated genetic changes specifically within early depressed cancer.2 The findings seem to ratify this decision. Among the 86 non-depressed submucosal cancers, residual adenoma was found in 63 (73%); it is conceivable that an adenomatous component was destroyed in the remainder. None of the depressed submucosal cancers (five out of the 91 found) showed residual adenoma. If it is assumed that there was no precursor lesion in these five cases, it must nevertheless be accepted that de novo carcinoma is uncommon, even in Japan. The distribution of these putative de novo submucosal cancers is also of interest: three in the right colon (17% of right sided submucosal cancers), two in the left colon (5% of left sided submucosal cancers), and none in the rectum.
The fact that depressed submucosal cancers are more common in the right colon and show a low frequency of K-rasmutation3 ,4 fits with the recent description of DNA microsatellite instability in flat appearing cancers of the proximal colon.5 K-ras mutation is uncommon in cancers showing extensive microsatellite instability (MSI-H phenotype).6 As adenomas in hereditary non-polyposis colorectal cancer (HNPCC), which are often MSI-H, are thought to show an accelerated progression to cancer,7 it is possible that the lack of residual adenoma in flat or depressed carcinoma is explained by malignant conversion of an adenoma when it is still small and therefore susceptible to complete obliteration.
The study by Konishi et al offers indirect evidence that depressed carcinoma may indeed develop within depressed adenoma. The percentage of lesions with a depressed appearance is similar for both adenoma (4%) and carcinoma (5%). Furthermore, depressed neoplasms (benign and malignant) are vanishingly rare in the rectum and most common in the proximal colon. To these points may be added the earlier observations of infrequent K-ras mutation in flat or depressed adenoma as well as carcinoma.3 ,4 ,8
It has been known for many years that colorectal adenoma and cancer are distributed differently in the colon. Autopsy studies show a relatively even distribution of adenomas throughout the colon whereas cancer is more frequent in the distal colon and rectum.9 Presumably, the microenvironment in the distal colon and rectum favours multistep neoplastic progression (or promotion). Konishi and colleagues support this notion by showing a higher frequency of >10 mm polyps, a higher frequency of villosity among the non-depressed adenomas of the distal colon and rectum, and a higher carcinoma:adenoma ratio in the distal colon and rectum.
The often repeated assertion that proximal and distal large bowel cancers are fundamentally different is alluded to in this study. It seems that the main basis for this observation is the predilection of MSI-H cancers for the proximal colon.10 As noted previously, this fact may explain the distribution of flat or depressed neoplasms. Nevertheless, other factors may determine the growth characteristics of colorectal neoplasms. Mechanical shearing forces may be maximal in the distal colon and rectum, encouraging mucosal prolapse and the development of a polypoid mass. Less mundane explanations may lie in the growth characteristics of neoplastic crypts. Crypt elongation and branching will produce an expansile lesion that is more likely to become exophytic. When neoplastic crypts remain short, branch infrequently, and spread laterally, lesions will be flat or depressed. At this time, however, there is limited information on the mechanisms underlying the growth of neoplasms.
This study supports the adenoma–carcinoma concept, highlighting this as the major morphogenetic pathway for colorectal cancer in the left colon and rectum. The de novo pathway, by contrast, is uncommon and operates mainly in the proximal colon. Even this suggestion may be overstating the importance of the de novo pathway if the explanation for an apparent de novo origin lies in the rapid conversion of flat or depressed adenomas showing DNA microsatellite instability.