—  SYMPOSIUM #48  —

Molecular Pathogenesis of Gastrointestinal Neoplasia
Moderators: Dr. Wataru Yasui and Dr. Jeremy Jass

Section 3 - Adenoma-carcinoma Sequence in the Colorectum

Stanley R. Hamilton


Colorectal carcinoma (CRC) develops as a result of progressive accumulation of genetic and epigenetic alterations that lead to neoplastic transformation of colorectal epithelium. Five overlapping molecular pathways of tumor initiation and progression are known for colorectal carcinogenesis: the chromosomal instability pathway (CIN), the microsatellite instability pathway with high levels of instability (MSI-H), the microsatellite instability pathway with low levels of instability (MSI-L), the hMYH pathway (MYH), and the CpG island methylation pathway/phenotype (CIMP).

Chromosomal Instability Pathway (CIN).
The vast majority (approximately 80%) of colorectal carcinomas develop through the chromosomal instability pathway (CIN) in the conventional adenoma-adenocarcinoma sequence. The hallmark of these CRC is that progression involves structural chromosomal alterations that are reflected in altered total DNA content, cytogenetic aneuploidy, and numerous allelic losses and gains. These CRC usually have inactivation of the adenomatous polyposis coli (APC) gene on chromosome 5q as the initiating event. Familial adenomatous polyposis with germline APC mutation is the inherited form of CRC in this pathway. Point mutations of the K-ras proto-oncogene and the p53 gene are frequent in CIN CRC.

Microsatellite Instability Pathway with High Levels of Instability (MSI-H).
CRC with high levels of microsatellite instability (MSI-H) comprise about 15% of cases. CRC in this pathway are characterized at the molecular level by numerous nucleotide substitutions and insertion/deletion mutations in repeated nucleotide sequences (microsatellites). MSI-H CRC result from inactivation of both alleles of a nucleotide mismatch repair gene, usually hMSH2 or hMLH1. These tumors usually have normal total DNA content, relatively normal cytogenetic karyotype, and infrequent allelic imbalances. Germline mutation of a mismatch repair gene causes hereditary non-polyposis colorectal cancer syndrome (Warthin-Lynch syndrome, Lynch syndrome), while most sporadic MSI-H CRC result from transcriptional silencing of the hMLH1 mismatch repair gene by promoter methylation, as described below. MSI-H CRC have distinctive clinical-pathologic features. These include right-sided location, poor differentiation, unusual histologic types (mucinous, medullary and signet-ring cell histology), absence of "dirty" necrosis, expansile growth pattern, numerous tumor-infiltrating lymphocytes, and prominent peri-tumoral lymphoid nodules, term Crohn's-like lymphoid response. Serrated adenomas may be an important precursor to MSI-H CRC.

Microsatellite Instability Pathway with Low Levels of Instability (MSI-L).
CRC with low levels of microsatellite instability (MSI-L) are heterogeneous. The molecular mechanisms responsible for this small subset of tumors is poorly understood, and the clinical-pathologic characteristics are not yet well defined.

hMYH Pathway (MYH).
CRC in the hMYH pathway have been described relatively recently. These infrequent CRC have mutations in both copies of the hMYH base excision repair gene due to inheritance of one mutated gene from each parent, resulting in bi-allelic alterations. Progression of these tumors is characterized by high frequency of G-to-T transversion mutations and by allelic loss on chromosome 18q, but the tumors have neither chromosomal instability nor microsatellite instability. Understanding of CRC in this pathway is in evolution.

CpG Island Methylation Pathway/Phenotype (CIMP).
About a third of CRC develop in the CpG island methylation pathway/phenotype (CIMP). CpG islands are 0.5 to 2 kilobase regions rich in cytosine-guanosine dinucleotide repeats that are present in the 5` region of approximately half of all human genes. Methylation of cytosine residues within CpG islands of promoter regions and proximal exons is associated with loss of gene expression by repression of transcription. This epigenetic mechanism is observed in physiologic conditions such as X chromosome inactivation and aging, as well as in neoplasia. In CRC with CIMP, transcriptional activation by methylation often occurs in numerous genes that are not usually unmethylated in non-neoplastic colorectal mucosa. Methylation is often concordant among numerous genes, and this concordant methylation defines the CpG island methylation phenotype (CIMP). Because of methylation of hMLH1, many CRC that develop due to CIMP also have MSI-H, whereas CRC with extensive methylation that does not involve hMLH1 have chromosomal instability. Clinical-pathologic features of MSI-H/CIMP tumors are similar to those in the MSI-H pathway, as described above. Microsatellite-stable CRC with CIMP also have characteristic clinical-pathologic findings included right-sided location, poor differentiation, and cribriform gland architecture, whereas corkscrew/serrated gland architecture is not seen in this molecular subtype.

Associations with Pathways.
The characteristics of the various molecular subtypes of CRC have impact on diagnosis, prognosis, and response and resistance to therapies. The best established association is that of MSI-H with hereditary non-polyposis colorectal cancer syndrome. About 95% or more of HNPCC carcinomas have MSI-H, and evaluation for MSI can contribute to diagnosis of the syndrome. Immunohistochemical characterization of the responsible abnormal mismatch repair gene can direct germline sequencing. In addition, stage-specific prognosis is improved in patients with MSI-H colorectal carcinoma in HNPCC and the sporadic setting. Allelic losses in CRC with chromosomal instability have been associated with poorer survival and with poorer outcome after fluoropyrimidine-based chemotherapy than microsatellite-stable carcinomas that lack allelic losses. CRC with CpG island methylation phenotype have also been reported to have adverse prognosis and poor response to chemotherapy.

The major molecular pathways leading to CRC have been identified. Some of the molecular characteristics are now used for clinical decision-making and patient management, although conflicting data are common in the literature. Applications of molecular diagnostics based on the molecular pathways producing CRC are expected to expand as research results continue to evolve. Less clear is the relationship among these molecular pathways and the adenoma-adenocarcinoma sequence and other precursors to CRC.

the Adenoma-Adenocarcinoma Sequence.
Intra-epithelial neoplasia (dysplasia) is a morphologic lesion characterized by histopathologic abnormalities indicating cellular transformation that are confined within the epithelial basement membrane, i.e. adenomatous epithelium. Most CRC develop from grossly visible intra-epithelial neoplasia in the form of an adenoma. The earliest lesions in the development of adenomas are dysplastic aberrant crypt foci. The morphogenesis of adenomas is controversial with "top-down" and "bottom up" mechanisms proposed in the literature.

Top-down Morphogenesis.
This proposed mechanism is described as adenomatous epithelium composed of genetically altered cells located in the superficial portions of the mucosa spreading laterally and downward to form new crypts that first connect to pre-existing crypts and eventually replacing them. Of note, "top" is not equivalent to "luminal surface" because it also includes the upper portions of colorectal crypts.

Bottom-up Histogenesis.
With this proposed mechanism, transformation takes place among the stem cell population at the crypt base, the transformed stem cell expands stochastically, and monoclonal conversion produces the monocryptal adenoma, which expands early by crypt fission and later by overgrowing adjacent crypts as well.

Evidence for Top-down Morphogenesis and Bottom-up Histogenesis.
Several lines of evidence have been cited in support of these two mechanisms. These include observations of histopathology, immunohistochemistry for markers of proliferation (MIB-1) and Wnt pathway abnormalities (b-catenin), and small and unicryptal adenomas, along with molecular analysis of APC gene mutation and chromosome 5q allelic loss in microdissected adenomas.

The study populations in the two major studies of top-down morphogenesis and bottom-up histogenesis are shown in the following table:
Top-down Bottom-up
Small sporadic adenomas N = 35, 1-3 mm N = 40, < 3 mm
FAP monocryptal adenomas N/A 3 patients
Uniformly dysplastic crypts None All (?)
Nuclear b-catenin and MIB-1 Top only Distributed to bottom

The hypothesis that the stem cells are near crypt bases and transformation leads to adenomatous epithelium there, followed by migration up and down the crypt column, is not supported by histopathology. Adenomatous epithelium is usually observed at the orifices of crypts and on the luminal surface in adenomas, sometimes does not involve crypt bases (Cole et al 1963, Lipkin 1974, Wiebecke et al 1974, Bussey 1976, Maskens 1979, and Shih et al 2001).

In molecular analysis of microdissected adenomas, 5q allelic loss was found in 5/10 adenomas, and APC mutation in 8/10 adenomas, including 4/5 with 5q LOH and 3/5 without 5q LOH. The alterations were found only in adenomatous epithelium in the upper crypts, not in underlying non-adenomatous epithelium in the lower portions of crypts. In this series of 35 small adenomas, there was not one example wherein all of the crypts were composed of uniformly dysplastic epithelium throughout their length, nor were any lesions observed in which the dysplastic epithelium was at the base of the crypts with normal overlying epithelium. It was clear that the dysplastic process proceeded from the top downward rather than from the bottom upward.

Implications.
There are three implications of these observations of top-down morphogenesis. First, precursors of dysplastic cells may actually reside in the intercryptal zones at the surface, and the dysplastic cells themselves may migrate laterally and downward into crypts. Secondly, neoplastic cells may originate in stem cells at the bases of the crypts, but the morphologically transformed cells appear as the cells migrate up the crypt column and thereafter become part of the superficial epithelium. Finally, adenomatous transformation of initially uninvolved crypts may occur under the influence of adjoining transformed crypts.

In bottom-up histogenesis, the initial event in the genesis of colorectal adenomas is the monocryptal adenoma. Initial growth is proposed to occur via crypt fission, and spread into adjacent crypt territories is a later, secondary event.

There is strong evidence that both top-down morphogenesis and bottom-up histogenesis occur during adenoma development, and the mechanisms are not mutually exclusive. The relative contributions of and stages in adenoma development when these events occur are not yet clearly defined. Current data suggest bottom-up histogenesis occurs initially and top-down morphogenesis occurs during progression, but additional studies of unicryptal/monocryptal and bicryptal adenomas in dysplastic aberrant crypt foci are needed to identify the earliest morphologic, molecular and clonal events. These studies are accomplished most easily in patients with familial adenomatous polyposis due to the widespread microadenomas in grossly normal mucosa. By contrast, these studies are more difficult but more widely applicable in sporadic patients.

Unanswered Questions.
  • What is the distribution of adenomatous epithelium in the crypt columns and surface epithelium of partially involved unicryptal/monoclonal adenomas?

  • Are the two crypts in bicryptal adenomas clonally identical or independent?

  • Is the surface epithelium between the two glands normal, abnormal but non-dysplastic, or dysplastic?
The answers to these questions will provide the basis for understanding the earliest events in the development of colorectal neoplasia and form the basis for understanding the earliest molecular abnormalities that can be targeted in prevention efforts.