| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Letters to the Editor |
Department of Pathology, McGill University, Montreal, Quebec, Canada
To the Editors:
There is clearly a need to identify reliable endpoints in cancer prevention trials (1) . However, the limitations of the adenoma-carcinoma model in the colorectum are underappreciated. The fact that colorectal adenomas may progress to cancer is beyond dispute. Far less clear is the proportion of colorectal cancers (CRCs) that develops within a preexisting adenoma. Increasingly, there is evidence of alternative pathways to CRC implicating lesions other than traditional adenomas (2) . Of special importance in this regard are the serrated polyps initiated by mutation of BRAF and having extensive DNA methylation (3) . Of greater concern in the context of using adenomas as surrogates for CRC is the fact that adenomas outnumber CRCs by approximately 30 to 1 (4) . One can deduce that an individual adenoma has an extremely low probability of evolving into a CRC within the average human life span. Support for this observation is provided by the disappointingly small effect of adenoma removal in the course of three large and protracted population-based randomized controlled trials of fecal-occult-blood testing (5, 6, 7) . By virtue of the design of these trials, study-group patients had far more endoscopies and instances of polypectomy than did control-group patients. Yet the main benefit from the increased investigation of the study- group subjects was not a reduction in the incidence of CRC but the detection of CRC at an early stage (and, therefore, a reduced mortality). Additionally, longitudinal observational studies have demonstrated that colorectal adenomas show very limited growth with time, except in patients with a family history of colorectal neoplasia (8 , 9) .
It is often argued that the demonstration of particular genetic profiles within stages of the adenoma–carcinoma sequence provides absolute proof that this sequence serves as the evolutionary pathway for CRC in the "vast majority" of cases. It is generally acknowledged that mutation of APC, K-ras, and p53 are the cornerstones of this pathway. In fact, only 7% of CRCs have all three mutations, an observation that forces us to question the existence of a dominant linear model of colorectal tumorigenesis and to accept that CRC is very much a multipathway disease (10) . Investigation of unselected series of primary CRC indicates that APC mutation occurs in approximately 60% (11) of cases and that K-ras mutation and p53 mutation are inversely related (10) . Artificial inflation of the frequency of the APC mutation figure is explained by overreliance on nonrepresentative cancer cell line data, underrepresentation of proximally located CRC, or the assumption by APC interest groups that mutations must occur in APC even though they cannot be detected (an analogy with the Emperor’s new clothes). The 40% gap is not filled by ß-catenin, which is mutated only in a subset of CRCs in hereditary non-polyposis CRC (HNPCC; ref. 11 , 12 ).
A further serious concern with respect to regarding adenomas as surrogates for CRCs is the vast difference in the rate of malignant conversion of adenomas in different clinical conditions. In familial adenomatous polyposis, many thousands of colorectal adenomas have developed by the second decade, whereas the mean age for developing CRC is 
40 years. One can estimate that the risk of malignant progression for an individual adenoma is about 1 in 1000. By contrast the risk of malignant transformation of an adenoma in HNPCC is not only around unity but occurs within a short time frame (13)
. An intermediate position is observed in the case of sporadic colorectal neoplasia. Importantly, the adenomas in these three clinical scenarios cannot be distinguished morphologically despite having vastly different natural histories. Conceivably, the latter observation is explained by differing frequencies of genetic instability at the somatic level. Whatever the reason, one can conclude that the clinical behavior of an individual adenoma is highly unpredictable and that this fact will limit the usefulness of the colorectal adenoma as a surrogate for CRC in trials of chemoprevention.
REFERENCES
Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, Maryland
Biological Sciences Division, University of Chicago, Chicago, Illinois
In Response:
Dr. Jass has brought attention to several factors worthy of consideration in evaluating colorectal adenomas as surrogate end points for colorectal cancer (CRC), i.e., CRC arising from lesions other than adenomatous polyps; the small percentage of adenomas progressing to CRC and the prolonged time to progression to CRC; and multiple genetic progression pathways possible within adenomas.
Concerning CRC arising from lesions other than adenomatous polyps, Dr. Jass cited the several large studies that evaluated fecal-occult-blood screening for early detection of cancer as indirect evidence that prevention of adenomas might not be an adequate surrogate end point for colorectal cancer incidence. However, it should be noted that these studies were not intended to evaluate adenoma removal, and one of the studies actually attributed cancer mortality reduction to detection and removal of progressing adenomas (bleeding adenomas) as well as early adenocarcinomas (1) . Furthermore, Dr. Jass did not consider the more compelling evidence provided by the results of the National Polyp Study, which showed that polyp removal resulted in significant reductions in colorectal cancer incidence compared with incidences estimated from population-based studies (2) . The net clinical benefit of preventing a precursor lesion that accounts for at least two-thirds of CRC in the most conservative estimates is obvious (3) . Because we propose that chemoprevention be added to (not replace) standard-of-care screening, such interventions can only result in incremental benefit assuming that they are safe (4) . Additionally, if the target lesion for preventing CRC is not the adenoma, it seems clear that endoscopic screening will be less effective and that developing chemopreventive drugs and strategies may be all the more important.
With regard to the small percentage of adenomas that progress and the prolonged progression period, we have presented trial designs in which patients enter the study postpolypectomy; therefore, the primary end point is the incidence of new polyps (4) . In this setting, the high ratio of adenoma incidence to CRC incidence (typical of ratios found in the conversion from precancers to cancers) does not detract from the importance of preventing the precancerous lesions. Furthermore, the multiyear genetic and phenotypic progression from normal gastrointestinal epithelium to adenocarcinoma is well documented (5 6 7 8 9 10) . Subjects with high-risk adenomas have been characterized—based on the number, size, and architecture of adenomas, as well as on family and personal history of adenomas and CRC and on the presence of germ-line lesions, such as the adenomatous polyposis coli gene (Apc) mutations defining familial adenomatous polyposis (11) . The ability to define such high-risk cohorts is a key factor in designing successful studies for evaluation of drugs to prevent colorectal cancer.
We agree with Dr. Jass that the available evidence does not yet support the use of earlier histologic lesions (e.g., aberrant crypt foci) or the presence of a few specific genetic lesions (e.g., ras, Apc, or p53) as surrogate end points. We recognize and agree that neoplastic progression is multipath and results in increasing cellular disorganization, dysfunction, and heterogeneity. Adenomas, like other intraepithelial neoplasia, are chosen as the optimal surrogate end point because they reflect the process of neoplastic progression and encompass the multiplicity of possible neoplastic events (7 8 9) . We expect that at some point in the future, genomic and proteomic analyses will allow development of molecular profiles that better define subgroups at high risk of neoplastic progression in the colorectal epithelium, and that parallel or surpass adenomas in utility as surrogate end points.
The purpose of our article (4) is to define an effective strategy for clinical development of drugs to prevent colorectal cancer. The clear public health need for this strategy is evidenced by the estimated CRC incidence and death rate for the United States in 2004 of 150,950 and 57,310, respectively (12) . Although we are mindful of Dr. Jass’s concerns, the factors that he cites do not diminish the utility of adenomas as end points in this endeavor.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Cancer Research | Clinical Cancer Research |
| Cancer Epidemiology Biomarkers & Prevention | Molecular Cancer Therapeutics |
| Molecular Cancer Research | Cancer Prevention Research |
| Cancer Prevention Journals Portal | Cancer Reviews Online |
| Annual Meeting Education Book | Meeting Abstracts Online |
