Endoscopy 2001; 33(12): 1042-1064
DOI: 10.1055/s-2001-18938
Review

© Georg Thieme Verlag Stuttgart · New York

Early Diagnosis and Prevention of Sporadic Colorectal Cancer[1]

R. Lambert 1 , D. Provenzale 2 , N. Ectors 3 , H. Vainio 1 , M. F. Dixon 4 , W. Atkin 5 , M. Werner 6 , S. Franceschi 1 , H. Watanabe 7 , G. N. Tytgat 8 , A. T. Axon 9 , H. Neuhaus 10
  • 1 International Agency for Research on Cancer, Lyon, France
  • 2 Institute for Clinical and Epidemiological Research, Durham, North Carolina, USA
  • 3 Department of Pathology, Leuven, Belgium
  • 4 Academic Unit of Pathology Leeds, UK
  • 5 Colorectal Cancer Unit, St Mark's Hospital, Harrow, UK
  • 6 Institute of Pathology, Munich, Germany
  • 7 Division of Molecular and Diagnostic Pathology, Niigata, Japan
  • 8 Academisch Medisch Centrum, Amsterdam, The Netherlands
  • 9 Leeds General Infirmary, Leeds, UK
  • 10 Evangelisches Krankenhaus, Düsseldorf, Germany
Further Information

Publication History

Publication Date:
07 December 2001 (online)

Preview

Early Colorectal Neoplastic Lesions

The Adenoma-Carcinoma Sequence

Most cases of colorectal (CR) cancer are sporadic; a small percentage occur in heritable syndromes such as familial adenomatous polyposis (FAP), attenuated adenomatous polyposis coli, flat adenoma syndrome, hereditary nonpolyposis CR cancer (HNPCC), juvenile polyposis syndromes, Peutz-Jeghers syndrome and hereditary non-FAP, non-HNPCC.

A large body of clinical evidence supports the belief that a majority of CR cancers arise from precursor lesions, benign adenomatous polyps [1]. The classical adenoma-carcinoma sequence postulates that adenomas contain dysplastic epithelium which arises from mutations in either the adenomatous polyposis coli gene (APC) or the DNA mismatch repair genes.

There is some confusion in the use of the terms dysplasia and adenoma: in the West, protruded or slightly elevated noninvasive neoplastic lesions are called adenomas, while flat or depressed neoplastic noninvasive lesions are called dysplasia. In the East, both types of lesions are called adenomas and described as protruding, flat or depressed. There is now a trend to use the terms polypoid and nonpolypoid. Polypoid adenomas correspond to protruded polyps, including both sessile and pedunculated types. Nonpolypoid adenomas include sligthly elevated polyps and the flat or depressed areas of dysplasia (or adenoma).

In polypoid adenomas, the risk of malignant transformation increases over time and with size and/or villous architecture. In populations with a high prevalence of adenomas, 40 to 50 % of individuals have multiple lesions. The degree of dysplasia is greater in patients with multiple rather than with solitary adenomas, which may explain the higher risk for cancer in patients with multiple adenomas.

De Novo Cancer

The term “de novo” cancer was initially introduced to describe those CR cancers which did not develop from a pre-existing adenoma, along the adenoma-carcinoma sequence. In Japan, the concept of “flat or depressed adenoma” was elaborated by Muto in 1985, and increasing numbers of nonpolypoid cancers (mainly early cancers) have since been reported in that country [2]. The macroscopic morphology of CR adenomas is determined by the balance between proliferation and apoptosis; the high apoptosis found in depressed adenomas correlates with low net growth [3]. In depressed adenomas, the tubular architectural pattern is distinct and the villous architectural pattern is nonexistent. These lesions were originally thought to be unique to the Japanese population. A prospective study of 1000 colonoscopies in the UK revealed that 36 % (117/321) of the adenomas found were nonpolypoid and that the likelihood of high-grade dysplasia or cancer increased from 4 % (3/70) in small nonpolypoid lesions, to 6 % (9/154) in small protruded polyps, 16 % (8/50) in larger polyps, 29 % (14/49) in large nonpolypoid lesions, and 75 % (3/4) in nonpolypoid depressed lesions [4]. These Western data corroborate the notion expressed by Kudo et al. that although depressed lesions have a low frequency, they are the aggressive ones [5] [6] .

Nonpolypoid early cancers are believed to adopt a distinct molecular route in carcinogenesis, and some claim these lesions may develop de novo, without the adenomatous step. Genetically APC and p53 gene mutation rates are similar in polypoid and nonpolypoid lesions, while K-ras mutation rates are lower in nonpolypoid early cancer. This may suggest a specific sequence of mutations [7] [8] , but does not necessarily indicate de novo carcinogenesis. In fact, more than 50 % of nonpolypoid and nondepressed early cancers are accompanied by residual adenoma, whereas nonpolypoid depressed early cancers have no adenomatous component [9]. Kudo [1] has adopted the concept of two distinct routes for carcinogenesis in the colorectal mucosa, where the “alternative” direct route relates to nonpolypoid precursors and thus to some of those lesions described as de novo cancer. A similar low K-ras mutation rate is proposed for the development of malignancy in serrated adenomas or in the mixed type of serrated adenoma and hyperplastic polyp [10]. The difficulty resides in assessing the relative proportion of polypoid, nonpolypoid (flat or depressed) and serrated adenomas as precursors in carcinogenesis. However, it is extremely likely that the adenoma-carcinoma sequence still holds for the majority of cases.

Intramucosal Neoplasia

Some confusion occurs in the distinction between dysplasia and cancer with respect to intramucosal lesions. In the absence of invasion of the lamina propria, the same neoplastic lesion may be called high-grade dysplasia in the West and cancer in Japan. The Vienna classification was elaborated by a group of Western and Eastern pathologists in 1998, in an attempt to resolve those differences, largely by adopting the terminology proposed there [11]. A group of gastrointestinal pathologists and gastroenterologists also met in Houston in February 1998 and in Padova in spring 1998, with the same objective, and proposed the Padova international classification [12]. Both articles were published within a short interval of time. Since the introduction of the Vienna classification, the new WHO classification of neoplasias of the gastrointestinal tract has been published and modifications to the Vienna classification have been suggested; however, the differences in nomenclature are merely semantic.

Although no overall agreement has been achieved yet on terminology, there is a tendency to use the term “low- or high-grade intraepithelial neoplasia (IEN)”, in the absence of stromal invasion. Intramucosal neoplasia corresponds to intraepithelial lesions as well as lesions invasive into the mucosa only. The absence of lymphatics in the colonic mucosa, and thus of a potential risk for lymphatic metastasis, justifies this lumping.

Depth of Tumor Invasion into the Submucosa

Neoplastic lesions restricted to the colonic mucosa never metastasize to lymph nodes. Submucosal cancers metastasize to lymph nodes in approximately 15 % of cases, and the risk correlates with the degree of submucosal invasion in depth and in width. Most Japanese specialists classify the degree of submucosal invasion, either by a relative value (Figure [1]), or by an absolute value, for safe endoscopic resection of early CR cancer.

Figure 1 Classification of the depth of invasion of colorectal cancer into the submucosa. The risk of lymph node involvement increases from the sm1a to the sm1c category Relative Evaluation of the Degree of the Submucosal Invasion Depth sm1: submucosal invasion within the superficial third of the depth of the submucosa; sm2: submucosal invasion reaching the middle third; sm3: submucosal invasion reaching the deepest third. Width This applies only to the sm1 degree of invasion. The width of the tumor component in the superficial third of the submucosa (B) is compared with the width of the tumor component in the mucosa (A), resulting in: sm1a: submucosal invasion with a B/A ratio of less than Œ sm1b: submucosal invasion with a B/A ratio of less than Œ to œ sm1c: submucosal invasion with a B/A ratio of more than œ. Absolute Evaluation of the Degree of the Submucosal Invasion Depth This is the vertical length of the submucosal invasion from the lowest edge of the muscularis mucosae: sm1: less than 0.5 mm; sm2: 0.5 to 1 mm; sm3: over 1 mm. Width This is the horizontal length of the submucosal invasion: sma: less than 1 mm; smb: 1 to 3 mm; smc: over 3 mm. It has been reported 13 that overt lymph node metastasis was negative in sm1 lesions (0/12) and sm2 lesions (0/2), regardless of the degree of horizontal submucosal invasion; it was also negative for sm3a (0/2) and sm3b lesions (0/5), but positive at 21 % (23/110) for sm3c cancers. When a CR neoplastic lesion has been resected at endoscopy, if there is cancer invasion into the submucosa, additional bowel resection should generally be considered. The risk of metastasis is particularly high when there is obvious lymphatic/venous invasion, massive cancer invasion near the cut margin, and when the adenocarcinoma is poorly differentiated or undifferentiated. However the criteria which suggest additional therapy may be equivocal. This is why endoscopic mucosectomy should be avoided when there is high suspicion for submucosal invasion; indeed the endoscopic morphology of the lesion may predict the risk of submucosal invasion. Polypoid lesions are most frequently sm1 or sm2, and infrequently sm3. Nonpolypoid lesions are less frequently sm1 and often sm2 or sm3.

Serrated Adenomas

Serrated adenomas are uncommon lesions. Three retrospective studies based on polypectomies agreed on an frequency of 0.5 to 0.6 % among polyps [14]. However, since serrated adenomas are often small and have an “hyperplastic appearance” at endoscopy, they are not likely to be removed [15], thus leading to an underestimation.

The clinical features of hyperplastic polyps, traditional adenomas and serrated adenomas are not distinctly different, with regard to sex distribution, age and location in the colorectum [14]. Serrated adenomatous polyposis has been described. Serrated adenomas range from 0.2 to 7.5 cm in diameter, and 66 % are small (0.2 to 0.6 cm). Of the larger ones approximately 10 % are pedunculated. Nonpolypoid variants have been described.

Serrated adenomas are composed of serrated glands, resembling hyperplastic polyps at low-power histological examination. However at higher magnification they are seen to be lined by a monotonous epithelium which contains more mucus than most adenomas but fewer mature cells than hyperplastic polyps. Serrated adenomas, as hyperplastic polyps, do express a similarly altered distribution pattern of glycoproteins antigens (MUC2, MUC4, MUC5AC and pS2) compared with tubular adenomas and carcinomas [16]. Traditional adenomatous glands may be admixed in one out of four cases. The subepithelial collagen table is not thickened, in contrast to hyperplastic polyps, and inflammation may be present.

Based on the description of variable degrees of dysplasia and the development of invasive adenocarcinomas within serrated adenomas, the latter have a role as precursor for CR cancer. Molecular biological studies have involved Ki-67 and Bcl-2 expression, p53 gene mutations [10]. Data on p53 overexpression, APC gene and K-ras mutations [5] and DNA microsatellite instability are variable. Loss of heterozygosity (LOH) of the DCC gene was not found [17]. Mutations of the p53 gene seem to be the most characteristic genetic alteration, as an event in a multistep carcinogenic pathway, which might be distinct from the traditional adenoma-carcinoma sequence or from carcinogenesis via mutations of mismatch repair genes.

Markers of Neoplastic Evolution p53 Status In order to clarify the issue of the influence of mutant p53 on treatment outcome in CR cancer Petersen et al. performed a meta-analysis of published studies 18. A total of 28 studies were evaluable for a total of 4416 patients, using either immunochemistry to detect p53 overexpression or DNA sequencing to detect TP 53 mutations. Both negative and positive influences on survival were reported with mutant p53, and it is unlikely that p53 status can be applied in a routine clinical test. Microsatellite Instability in Hereditary Cancer The microsatellite loci may be classified according to stability into those with a low or a high frequency of instability. Microsatellite instability (MSI) is shown by the positive replication error test (RER+) using the polymerase chain reaction. When instability occurs at 40 % or more of microsatellite loci, this is defined as high instability (hMSI); this feature has been found in most cases of HNPCC and has been included in the Amsterdam-Bethesda criteria for HNPCC 19. At present six human genes have been identified as participating in the function of mismatch repair. Inherited germ-line mutations of mismatch-repair genes have been found in approximately 50 % of persons with a family history which fulfils the Amsterdam criteria. Alterations of the MSH2 and MLH1 mismatch-repair genes account for more than 90 % of these cases. Another mismatch repair gene, hPMS2, is involved in approximately 6 % of HNPCC cases. Microsatellite Instability in Sporadic Cancer hMSI occurs in approximately 15 % of sporadic cases of CR cancer. Acquired, noninherited alterations of the MLH1 gene occur in most sporadic cases of CR cancer with hMSI. A number of studies have shown that hMSI occurs relatively frequently in CR cancers which arise proximal to the splenic flexure, in poorly differentiated cancers or those of mucinous cell type, and in cancers with peritumoral lymphocytic infiltration; these are associated with longer survival. In a recent study, based on the Ontario Cancer Registry (607 patients treated for CR cancer) hMSI was found in 17 % of cases, low MSI in 3 % and microsatellite stability in 80 % 20. In a multivariate analysis MSI was associated with a significant survival advantage, independently of all standard prognostic factors, including tumor stage (Odd's ratio 0.42, P < 0.001). Furthermore, regardless of the depth of tumor invasion, CR cancer with hMSI had a decreased likelihood of metastasizing to regional lymph nodes (Odd's ratio 0.33, P < 0.001) or distant organs (Odd's ratio 0.49, P = 0.02). From other studies it would seem that these patients are at risk of metachronous CR cancer. Finally there is accumulating (in vitro) evidence to suggest that these patients respond differently to chemotherapy. This clinically distinct subtype of sporadic CR cancer needs to be identified. Immunochemical Testing of Microsatellite Instability Immunohistochemical findings to detect loss of MLH1, MSH2 or also MSH6 protein expression has been included in the Amsterdam-Bethesda criteria for HNPCC 19. A report 21 on the immunohistochemical analysis of 501 cases of CR cancer, with antibodies directed against hMSH2 and hMLH1 showed an excellent correlation with the hMSI status as detected by DNA sequencing. Immunochemistry is an initial simple laboratory test, which could be performed on all CR cancers. Thus, it could provide a way for selecting patients who should be investigated for HNPCC, as well as those patients with a distinct subtype of sporadic cancer. A recent study in Belgium 22 on patients with CR cancer with familial clustering indicates that in the case of missense mutations, the immunohistochemical analysis of sections of the tumor can differ from the MSI results: if the substituted amino acid is located outside the epitope of the hMLH1 or hMSH2 antibodies, the staining will not be affected.

1 Text prepared at the Third European Endoscopy Forum, which was supported by an educational grant from Olympus, in Faro, Portugal, June 8 - 10, 2001.

References

1 Text prepared at the Third European Endoscopy Forum, which was supported by an educational grant from Olympus, in Faro, Portugal, June 8 - 10, 2001.

R. Lambert, M.D.

Descriptive Epidemiology Unit
International Agency for Research on Cancer

150 cours Albert Thomas
Lyon 69372 cedex 08
France


Fax: + 33-4-72738650

Email: lambert@iarc.fr