| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Annual Oration |
1 From the Department of Radiology, McMaster University, Hamilton, Ontario, Canada. Received July 13, 1999; revision requested August 17; revision received September 8; accepted September 24. Address reprint requests to the author, 336 Lloyminn Ave, Ancaster, Ontario, Canada L9G 3X4 (e-mail: stevenso@fhs.mcmaster.ca).
Index terms: Cancer screening, 75.321, 757.321 • Chromosomes, abnormalities • Colon, neoplasms, 75.3111, 75.3114, 75.321 • Radiology and radiologists, 75.321, 757.321 • Rectum, neoplasms, 757.3111, 757.3114, 757.321
President Fraser, Dr Linda Warren, Ladies, and Gentlemen, I am grateful for the privilege of presenting the 1998 RSNA Annual Oration in Diagnostic Radiology, and it is a particular pleasure that this lecture is dedicated to Dr Joachim Burhenne. Dr Burhenne not only was personally responsible for many key advances in gastrointestinal radiology but also stimulated a generation of young radiologists to take up the specialty. My topic today is colorectal carcinoma and the opportunities and challenges it presents for radiology. Before considering how radiology may help in the diagnosis, management, and prevention of colorectal cancer, it will be helpful if I describe some recent developments in the understanding of its pathogenesis.
PATHOGENESIS OF COLORECTAL CARCINOMA
It has become clear that colorectal carcinoma develops as a result of a number of mutations, and studies by Vogelstein and others (1,2) have shown that a cascade of mutations, which has become known as the chromosome instability pathway, is responsible for most colorectal carcinomas. No less than seven mutations affecting four genes on chromosomes 5, 12, 18, and 17 must occur for invasive colorectal carcinoma to exist.
Initiation of the sequence starts with mutations in the APC gene, the same gene whose dysfunction is responsible for the inherited syndrome of familial polyposis coli. The function of the APC gene appears to be to sweep excess free ß-catenin from the cytoplasm of the cell. ß-Catenin has two normal roles: It activates cadherin, whose extracellular hooks attach to hooks from other cells, thus restraining cells from premature cell shedding; and it also enters the nucleus and increases cell proliferation. An intact APC gene thus helps to ensure that cells are shed after the normal 6 days, and so it restrains growth.
Familial polyposis coli is an autosomal dominant condition in which one copy of the APC gene is missing or inactive in every cell, and it is the loss of the second allele in a cell, caused by mutation, that promotes the action of cadherin by allowing an excess of ß-catenin to accumulate. This leads to increased cell proliferation and delayed shedding so that an excess of daughter cells begins to accumulate, which causes the formation of a small adenoma. In familial polyposis coli, the occurrence of this loss of the second allele at multiple sites throughout the colon leads to the development of hundreds or thousands of adenomas. In most people, however, the loss of both alleles of the APC gene in a single cell is unusual, so that small adenomas are less common, but nevertheless 35% of us will develop at least one adenoma by the age of 70 years (3). The adenoma is genetically unstable, and over 2 or more decades, further chromosome deletions inactivate crucial genes, which liberates successive waves of clones of cells that are progressively damaged, with each new clone displacing the preexisting one.
The second gene involved in the development of colorectal cancer lies on chromosome 12 and is called the K-ras gene. Unlike APC, which is a tumor suppressor gene, K-ras is a proto-oncogene. Mutations in the K-ras gene may either diminish or increase its function. Mutations that lead to an increase in function effectively convert this proto-oncogene into an oncogene with "the accelerator pedal stuck to the floor" (2) for a variety of growth factors. If a K-ras mutation occurs without a preceding APC mutation, the result is a tiny aberrant crypt focus that is nonneoplastic. However, if it occurs in cells in which the APC gene is inactive, growth to a larger adenoma occurs, and an accelerator mutation on only one allele of the K-ras gene is sufficient to trigger this growth.
The third gene is on chromosome 18 and was thought for some years to be a gene known as DCC, which codes for a cell adhesion protein. However, inactivation of the second allele of DCC has not been demonstrated in colorectal carcinoma, and it is not clear that the DCC gene itself is relevant. A mutation in this general region of chromosome 18 is critical and greatly affects the prognosis of the subsequent colorectal cancer, but the actual tumor suppressor gene involved has not been identified yet.
The final gene in this sequence is on chromosome 17 and is the P53 gene. P53 is also a tumor suppressor gene, and the normal function of P53 is to activate proteins that can prevent DNA replication. The gene thus appears to recognize damaged DNA and normally ensures that cells containing damaged DNA die. Inactivation of P53 is usually critical in the transformation from adenoma to carcinoma, and P53 is inactivated in more than 75% of all colorectal carcinomas (2).
APC is one of the genes known as a gatekeeper, and such genes prevent tumors by inhibiting growth or promoting cell death. Their dysfunction leads directly to tumor formation. However, it has become clear in the last few years that there is also another completely different class of genes known as caretakers. The mismatch-repair genes are examples of this, and they permit tumor formation indirectly, as their dysfunction leads to genetic instability in other genes, including the gatekeeper APC.
MISMATCH-REPAIR GENES
It is estimated that approximately one error occurs in DNA replication for every 105 base pairs and that 99% of these mistakes are corrected by enzyme proofreading, leaving one error in every 107, or one in 10,000,000 (2). The mismatch-repair system repairs almost all remaining replication errors, mainly by excising the sequence of nucleotides that is incorrect, copying the correct sequence, and inserting it into the DNA chain. This improves the accuracy by about a factor of a thousand, leaving one error in every 1010 replications. Because the human genome contains about 3 billion base pairs, this intact mismatch-repair system means that there are normally only one or two surviving errors per cycle of cell division, which gives a low probability for the development of carcinoma. However, if the mismatch-repair system fails, then DNA errors will accumulate by hundreds and thousands, mainly in noncoding DNA, but about 5% of the errors will accumulate in genic DNA. This leads to genes that are truncated and inactive, and in the early 1990s, it was found that inactive truncated repair genes were highly associated with hereditary nonpolyposis colorectal cancer (HNPCC).
HNPCC was described by Henry Lynch and associates (4). The discovery was the result of careful follow-up of a chance remark made in 1895 by a seamstress to her customer, who was a professor of pathology. The seamstress mentioned that she would die at an early age of uterine cancer because many of her relatives had done so. Intrigued, the pathologist (Warthin) constructed a family tree and noted a large number of tumors in the family. In due course, the seamstress did indeed die of endometrial carcinoma.
In the 1950s, Henry Lynch revisited and analyzed the extended family and described what has become known as the Lynch syndrome, or HNPCC, in which colorectal carcinoma develops at an early age. In this syndrome, there are multiple primary cancers, there is predominance in the right colon, there are few adenomas, and there is an autosomal dominant pattern of inheritance, although with incomplete penetration. In some families, there is an association with a variety of other carcinomas, particularly endometrial carcinoma, gastric carcinoma, and uroepithelial cancer, while in other families, only colorectal carcinoma is involved. HNPCC is now known to be caused by inheritance of inactive mismatch-repair genes, and no less than four discrete genes have been identified, the first in 1993 (5). Together, they are responsible for some 76% of the known HNPCC families, so more genes remain undiscovered.
In the past few years, it has become clear that the chromosome instability pathway caused by the cascade of genetic mutations that starts with the APC gene is responsible for about 85% of all cases of colorectal carcinoma (including the 1% that is caused by familial polyposis coli). The hypermutability pathway, caused by errors in mismatch-repair genes, is responsible for some 15% of the cases of colorectal cancer, including the 5% that is caused by HNPCC.
In patients with familial polyposis coli, the mean age at onset of adenomas is 16 years, and the mean age at onset of carcinoma is 42 years. The same slow development from small adenoma to carcinoma is almost certainly true in the vast majority of sporadic cancers that start with a mutation in the APC gene, with an "adenoma dwell time" of 2–3 decades, which allows ample opportunity for the detection and removal of precursor adenomas. In the hypermutability pathway, however, the situation is different, and both in patients with HNPCC and in those with sporadic carcinomas associated with mismatch-repair gene problems, the cascade from no visible abnormality to advanced carcinoma may occur rapidly, sometimes in less than 2 years (6). Clearly, the prevention of such lesions would require a different strategy from that suited to the slower chromosome instability pathway.
TUMOR DETECTION
We are now in a position to consider the various tools available for the detection of colorectal cancer and to assess how well they perform in the two types of colorectal cancer. Performance of the barium enema examination has been exhaustively reviewed, usually retrospectively, and the double-contrast barium enema examination has been clearly shown to have a detection rate of better than 70% for polyps larger than 7 mm in diameter (7). The detection rate for colorectal cancer, however, has been extraordinarily variable, from 70% to greater than 96% (8–10). In Hamilton, Ontario, we reviewed the pathology records of the performance of double-contrast enema examinations at the four teaching hospitals and found a detection rate of colorectal cancer of 94% (11). Ten years later, we took the opportunity to revisit this and to assess the detection rate for all colorectal carcinomas diagnosed in the region, whether at clinics, small hospitals, or teaching hospitals. We found an overall detection rate of only 76%. At the same time, 1990, we noted that colonoscopy had a detection rate of 85% of colorectal carcinomas (12). In 1997, the results of a careful multihospital study from Indiana revealed a colorectal cancer detection rate of 85% for double-contrast barium enema examinations, while for colonoscopy, the detection rates were 95% when the examination was performed by gastroenterologists and 87% when performed by nongastroenterologists (13).
Studies that are based on pathology records are likely to show most fairly the relative sensitivities of barium enema examinations and colonoscopy (12–14), and it is clear that the barium enema study is capable of performing at a high level of sensitivity. The barium enema examination and colonoscopy are both superb investigations, neither of which is performing to its potential standard in routine clinical practice. In such a situation, we can either take steps to improve the performance of the investigations, or we can look for another test; and another test is becoming available for radiologists—helical computed tomography (CT).
HELICAL CT AND CT COLONOGRAPHY
By 1990, several authors had outlined the key criteria for examination of the colon on the CT scan, indicating the importance of air distention, bowel preparation, and the possible role of intravenously administered contrast agent (15). With the advent of helical CT, there were further reports on the value of CT as an alternative to the barium enema examination or colonoscopy, especially in the frail and elderly, for whom CT might function as an excellent first-line investigation (16). CT is also a simple test for problem solving in some situations, such as assessing whether a large bulky ileocecal valve is caused by tumor or fatty infiltration. In 1996, David Vining (17) reported on the development of three-dimensional computer reconstruction from a volumetric data set using a high-end workstation after distending the clean colon with gas. Vining was able to demonstrate not only axial and three-dimensional static images, but also sequential three-dimensional images, to produce what is now known as virtual colonoscopy or CT colonography.
I am grateful to David Vining and to other pioneers who have allowed me to show their work, with which I shall illustrate to you the present status of CT colonography—Helen Fenlon, Joe Ferrucci, Brooke Jeffrey, Amy Hara, Dan Johnson, Abe Dachman, and Vassilios Raptopoulos. Time does not permit discussion of technical details, but suffice it to say that with appropriate preparation and by using equipment available from several different manufacturers, it is now possible with one or two breath holds to produce images that allow a "fly-through" view of the entire colon from rectum to cecum, in both directions (18–24). Large-bowel obstruction caused by carcinoma poses no problem because the colon can be imaged well with CT both above and below the obstruction, a major advantage over both colonoscopy and the barium enema examination.
Several technical problems remain to be solved, including those of retained stool, retained barium, respiratory artifacts, stool-filled diverticula, and fluid in the colon. The advent of multisection helical scanners is reducing the time required, so that a single-breath-hold scanning procedure will be achievable in almost all patients, but the issue of retained fluid may mean that two scanning procedures are required for each patient.
One of the current stumbling blocks is the length of time required for analysis of all of the images, and this time is being reduced by automated midline identification software and by a policy of routine review of two-dimensional images only, reserving intraluminal analysis for problem solving. Some companies are developing multiple display formats so that the reader can view axial, vertical, and horizontal orthogonal two-dimensional images; true cross-sectional images; volume-rendered images; and three-dimensional images simultaneously. A mathematically straightened view of the colon in longitudinal section can be provided, and software programs to provide a split-open view of the colon that present a pathologic view of the flattened colonic mucosa are also becoming available. It is not yet clear which of these display protocols will be the most useful in routine practice.
Polyp detection rates with CT that are competitive with barium enema examination and colonoscopy for lesions larger than 7 mm have been documented by some observers. To our knowledge, there are only two reports at present of the clinical utility of CT compared with colonoscopy, and the results are promising. The findings from one report in Radiology on 65 patients, of whom 41 had carcinoma, showed that all 41 cases were detected with CT but only 34 with colonoscopy and that a complete colon study was provided in 61 of the 65 patients by using CT and in 49 patients by using colonoscopy (25). The results of another report, which was published in the American Journal of Roentgenology (26), compared the cancer detection rate with barium enema examination, colonoscopy, and CT in a total of 52 patients and showed that of the 40 cancers present, 24 were detected with barium enema examination, 29 with colonoscopy, and 37 with helical CT. There were five incomplete barium enema examinations, five incomplete CT studies, and 21 incomplete colonoscopy procedures (26).
These initial results are promising, but unfortunately, early reports usually turn out to be a little overoptimistic, and the scene is now set for a series of carefully controlled studies to compare the sensitivity of helical CT and colonoscopy for the detection of cancers and polyps. If CT can match the sensitivity of colonoscopy, then this will provide an examination of much greater safety and comfort for patients and will allow colonoscopy to be reserved for a more focused confirmatory and therapeutic role.
Magnetic resonance (MR) imaging has been following close behind CT in most areas of imaging, and the colon is no exception. This year has produced reports of MR colonography from Debatin and colleagues in Switzerland, and axial images, volume-rendered images, and luminal views have been demonstrated that compete with helical CT, with sensitivities for lesions larger than 5 mm of just over 90% (27,28). There is good reason to be optimistic that cross-sectional imaging will make a major contribution to the diagnosis of colorectal cancer in the next few years. Perhaps the single most useful advance to make life easier for patients would be techniques to allow the electronic subtraction of residual stool so that complete and arduous bowel preparation might no longer be necessary.
RADIOLOGY IN STAGING AND THERAPY
Radiology has other roles apart from diagnosis, and there are three areas of recent development in which radiology is poised to make useful contributions to the care of patients with colorectal carcinoma—one diagnostic and two therapeutic.
Ultrasonography (US) has lagged somewhat behind CT and MR imaging in the demonstration of metastatic disease in the liver. Although contrast agents for US have been developed primarily for intravascular purposes, an observation on the behavior of US contrast agents in the liver and spleen has opened up a new area of possible contribution (29). When microbubbles are scanned with a high-energy beam, they burst. The sudden disappearance of the reflector produces phase changes, and these changes, together with transient disruption emissions, produce an intense pseudo-Doppler signal.
It seems that some microbubble contrast agents gradually concentrate in hepatic and splenic tissue, possibly in Kupffer cells. If the liver is scanned 5 minutes after the intravenous injection, there is an intense flash of color from the normal liver as the microbubbles burst, which disappears after only a few frames. Whenever there is tissue lacking in Kupffer cells, such as in metastases, there will be a signal void, and the results of early studies from the Hammersmith Hospital have demonstrated that questionable lesions at normal scanning and also some lesions invisible at normal scanning become clearly apparent at scanning performed some 5 minutes after intravenous injection of contrast microbubbles because of the increased conspicuity of the lesion (29). However, with color imaging, there is some loss of spatial resolution.
Gray-scale imaging with pulse inversion can be used to achieve a similar enhancement of conspicuity of metastases without the color but at higher spatial resolution. The technique of pulse inversion in this context relies on the fact that when two sound signals are emitted 180° out of phase, the returning signals reflected from a flat surface cancel each other out and will show as a void. However, in the presence of microbubbles, with their curved surfaces, the returning signals come at all angles and are somewhat enhanced. Thus, when there are microbubbles in the liver parenchyma, the signal from normal tissue, where the bubbles are present, enhances. However, metastases, in which there is an absence of bubbles, show loss of echogenicity, and the enhancement of metastases as black holes in a bright liver is striking. The clinical use of these new discoveries in the follow-up of high-risk patients remains to be demonstrated.
Interventional radiology is proving useful in patients with colorectal carcinoma in the management of obstruction and in the treatment of liver metastases. Large-bowel obstruction is an emergency, and patients who are undergoing surgery for malignant large-bowel obstruction do not do as well as those undergoing surgery electively. The findings from an initial report from Spain on 120 patients treated with Wallstents (Schneider, Minneapolis, Minn) placed radiologically showed that 82 patients were relieved of their obstruction and went on to elective resection, while 38 ended up with the Wallstent as their definitive palliation. Seventeen patients failed to respond to placement of the Wallstent and went on to emergency surgery. In all, 126 stents were placed in the 120 patients, and 103 patients had complete relief of their obstruction (30). It seems likely that over the next few years, placement of expandable stents, radiologically or endoscopically, will become the mainstay of emergency treatment of patients with malignant colorectal obstruction.
A number of methods have been used for treatment of metastatic liver lesions in patients with colorectal cancer. These methods have met with partial success, and one of the particular problems has been the recurrence of tumor around the margin of the treated area. Two different companies have produced equipment for radio-frequency therapy for liver metastases by using small needles, 18–20 gauge. These needles are placed easily with US, CT, or MR imaging, and the development of internal cooling of the electrode has allowed treatment of lesions larger than 5 cm, which may translate into much greater long-term effectiveness (31). Goldberg and associates (32) and Dodd et al (33) in the United States and Lees and co-workers (34) in the United Kingdom have recently reported on their early results with radio-frequency treatment. MR imaging appears to be the most effective, though least available, method for monitoring the progressive destruction of the lesion, and complete treatment of 5-cm lesions now appears to be a reasonable goal. Surgical skill in partial hepatic resections has also moved forward quickly over the past few years, and the combination of radio-frequency therapy for some lesions and surgical resection for others promises a potentially curative advance in the care of patients with early metastatic disease.
COLORECTAL CANCER PREVENTION
Finally, with regard to the issue of prevention of colorectal cancer, the ideal would be primary prevention to reduce the incidence of the disease. Although it is known that those who eat a diet high in fruit and fiber and who take their main protein in chicken and fish have an incidence of colorectal cancer six times lower than those who habitually eat well-charcoaled red-meat steaks, the introduction of this knowledge into a coherent and successful public policy remains an unlikely dream (35). In the short term, however, there is one drug that can have an effect. Research on patients with familial polyposis coli and small adenomatous polyps in their rectal stump has shown that sulindac (an aspirin derivative) can make small adenomas disappear (36). Aspirin itself has also been convincingly demonstrated to reduce the risk of colorectal carcinoma (37) and, under appropriate safeguards, can be strongly recommended to all the members of this audience who are reaching the age at which colorectal carcinoma becomes a matter of interest, not to mention the benefits of aspirin in reducing the incidence of myocardial infarction, stroke, and deep venous thrombosis on long-distance air travel in economy seats!
SCREENING
Given that there are limitations, however, in primary prevention, is screening for colorectal carcinoma worthwhile? Several questions have to be asked and answered in considering screening for a disease. Is the disease common, and are its consequences serious? Five percent of us will develop colorectal carcinoma, and half of those who develop it will die from it. Are high-risk populations identifiable, and do they behave similarly to low-risk groups? The answer is yes, high-risk populations are indeed identifiable, and there is in fact a cascade of risk factors. The prodromal phase of most colorectal carcinoma is clearly the adenoma. It is present in 35% of us at the age of 70 years, although adenomas larger than 1 cm in diameter are only present in 6% by the age of 70 years (3). Adenomas smaller than 5 mm have no statistical association with increased risk of cancer later on (38), so low is the probability that any given small adenoma will develop into cancer, so that it is only the larger adenomas that are of interest in screening. In pathogenetic terms, the larger adenomas are probably those in which not only has an APC mutation occurred, but also the K-ras mutation. Most likely, it is the adenomas containing only the APC mutation that may be reversible with sulindac or aspirin administration.
The dwell time of adenomas has been estimated in the literature as 15–18 years, but we now know that this figure includes a number of cancers that will have arisen by mismatch-repair gene errors, so that in patients with the chromosome instability pathway, the dwell time of adenomas is probably longer, around 25 years. We also know that the chance of carcinoma being present in a polyp increases with size, with the percentage of the lesion that is villous or tubulovillous, and with multiplicity of polyps (39). The next question to be asked is whether early detection of either the disease or the prodromal phase can lead to reduced mortality. The answer to this question is also clear. There are three tests for fecal occult blood (40–42) and two flexible sigmoidoscopic studies (43,44) that have demonstrated, reasonably conclusively, an approximately 30% reduction in mortality from screening with either of these methods followed by appropriate treatment, despite rather poor compliance.
Before rushing into screening for colorectal cancer, however, we might do well to sit back and review some of the results of the other type of cancer screening in which radiologists are heavily involved and see what we can learn from our breast cancer screening colleagues. The first thing we can learn is the considerable difficulty in interpreting the statistics with which we are showered. The six randomized controlled studies on breast cancer screening with mammography show reductions in mortality that barely reach statistical significance, and when breast cancer deaths annually are plotted, it becomes apparent that the main effect of screening is not to reduce the rate of accrual of breast cancer deaths, but to defer it by about a year, after which the rate of accrual of deaths from breast cancer rises at the same slope in the screened group as in the nonscreened group (45,46). Ideally, what we would like to see after a screening intervention is the rate of accrual of deaths from the disease continuing to diverge for several years in the screened and nonscreened groups.
In addition, examination of the Ontario Breast Screening Program data from 1990 to 1994 reveals that the positive predictive value of an abnormal mammogram is 7%; in other words, 93% of positive results are false-positive (47). A number of harms result from such a high rate of false-positive results, and in addition, some 19%–25% of breast cancers are not detected with high-quality mammographic screening (48,49). There is thus, despite the undoubted benefits from mammography, also considerable emotional and financial distress caused to women with abnormal mammograms and no cancer (7% of all those examined), and a false hope of normality in those with false-negative results (19%–25% of all those with cancer).
Can we expect the same results from colorectal cancer screening? Comparison of the two diseases suggests that we are probably on much surer ground with colorectal cancer than with breast cancer. In the first place, there is an identifiable prodrome, and for 85% of the patients, the prodrome is present for probably more than 2 decades before cancer develops. We have tests that enable detection of the presence of colorectal cancer with greater than 95% sensitivity (even if they do not always perform at that level) and tests that enable detection of the prodrome with 80%–90% sensitivity (again, even if they do not always perform to such a level). Even more important, early detection does mean cure with much more certainty with colorectal carcinoma than with breast cancer, and 10-year survival means cure, which it does not with breast cancer.
Test Selection
So what tests should we use for screening for colorectal carcinoma? Fecal occult blood tests and flexible sigmoidoscopy (followed by therapeutic colonoscopy) have been proved to be effective, reducing mortality by up to 30%. Radiologists have produced no data as to whether barium enema examination is effective, and in lieu of data, we have produced careful economic predictions that suggest that all strategies (fecal occult blood tests, flexible sigmoidoscopy, colonoscopy, and double-contrast barium enema examination) are cost-effective at 1 year of life gained for less than $25,000 (5).
There are no colonoscopy data on reduction of mortality in average-risk individuals. There are five studies of the yield of lesions with colonoscopic screening (50–54) but none on the effect on mortality. The results of two surveillance studies have provided calculations that colonoscopy and polypectomy should reduce the incidence and mortality of colorectal carcinoma during the surveillance period (55,56). However, the point has been made that "if the frequency for surveillance colonoscopy were to be decreased . . . survival might improve, considering that deaths related to colonoscopy complications currently exceed deaths from colorectal cancer in surveillance studies" (57). The national polyp study provides no direct data. In 6 years of follow-up, five cancers were found, but there was no randomized control group, and it is only by calculation from three other sets of data that the authors conclude that 48, 43, and 21 cancers might have been expected (56). Most, but not all, authors agree that colonoscopy is simply too dangerous for the screening and surveillance of average-risk populations.
A suggestion was made from Britain in 1993 that a single flexible sigmoidoscopy between the ages of 55 and 60 years might prevent one-third of all colorectal carcinoma deaths at a cost of $11,000 per cancer detected and $17,000 per death prevented, compared with the current cost for breast and cervical cancer death prevention of $40,000–$60,000 each. The proposal was based on the known prolonged dwell time of adenomas in patients developing colorectal cancer in middle and old age and on the belief that a highly focused and targeted flexible sigmoidoscopy program could achieve a high compliance of 70% (58).
These authors obtained funding for their study which, after a pilot project, was implemented in 13 centers in the United Kingdom. By the end of 1998, there were 40,000 individuals who had been screened, with approximately 130,000 control subjects, and data acquisition is now occurring. The findings from the pilot project were published in 1998 (59), and the predictions from the economic analysis paper in 1993 that 8%–10% of the subjects would have adenomas, 3%–5% would have high-risk adenomas, and 0.15% would have carcinoma were met, except that the actual number with carcinoma was higher than expected at 0.65%. Data on mortality reduction from this intervention, in comparison with the control group, will accrue over the next 3–5 years.
This method of one-time flexible sigmoidoscopic screening makes no attempt to eliminate all colorectal carcinoma from the population (because almost one-half of colorectal cancer arises beyond the reach of the sigmoidoscope) but rather, on the basis of a focused targeted approach to controlled populations, aims to eliminate 30% of the disease in a cost-effective and affordable manner.
Given this information then, how should we screen for colorectal cancer today? We have to consider separately the mismatch-repair gene carcinoma and the average-risk carcinoma because the one occurs earlier in life and develops fast, and the other occurs later in life and develops slowly. We also have to distinguish funded population screening from advice to a single concerned individual. The single individual wants assurance close to the 100% level of freedom from disease and may be willing both to pay for it and to accept the increased cost and complications that result from a total colonic examination. The funded program starts with a budget paid for with your taxes and aims to provide the greatest reduction in population mortality achievable within that budget: Total colon examination of a small segment of the population does not achieve that goal.
High Risk
Strategies that rely on detection of large adenomas, such as the barium enema examination, clearly have no place in surveillance for HNPCC because so many of these tumors develop rapidly, either de novo or from flat adenomas. Colonoscopy is the only method currently appropriate for screening and surveillance for these families. The findings from one study reported on the effect of surveillance colonoscopy in HNPCC—a study of 251 subjects (mean age, 38 years) in Finland. Over 10 years of surveillance, cancer developed in 4.5% of the 133 patients who were followed with colonoscopy, and all of these cancers were Duke A and B lesions. In the 118 patients who did not undergo colonoscopy, cancer developed in 11.9%, and half of the lesions were Duke C and D tumors (60).
It is clear that colonoscopy is a most effective screening and surveillance tool for these high-risk families, and the issue then is not whether we should screen HNPCC family members, but how to identify them in the general population. Asking family physicians to go through their records and identify patients with appropriate histories is probably impractical, and the simplest and most effective method is for interested surgeons, gastroenterologists, radiologists, and pathologists to work through local pathology records. Five percent of all cases of colorectal cancer occur in HNPCC families. If we identify from the hospital pathology records all those patients who are under the age of 55 years and all those patients who already have another member of the family with colorectal carcinoma, the resultant list will comprise some 15% of all of the patients with colorectal cancer in the pathology records (61).
Such a list can form the basis for development of a regional registry of families at high risk for colorectal cancer. Of these 15%, approximately one-third will turn out to be HNPCC families, and their relatives will have a 45%–50% chance of developing colorectal cancer. The other two-thirds are at increased risk to a level not exactly known but probably somewhere between 5% and 20%. Such registries constructed and followed over 10 years will identify within each region almost all of the patients in HNPCC families, as well as many other families at increased risk.
Average Risk
What about screening for the average-risk population in whom we are looking for adenomas larger than 5 mm? The first principle of all screening is the old medical adage, Primum non nocere—first, do no harm. Colonoscopy then, for the time being, with a sensitivity for adenoma detection of 80%–90%, perforation rate of 1/500, and mortality of 1/5,000, is unacceptable as a primary screening test.
Is barium enema examination any better? Most of our mistakes are made because of observer perceptive error, although technical errors are also relevant (9,10). Most mistakes also occur in the sigmoid colon, and many believe that supplementing barium enema examination with flexible sigmoidoscopy produces a powerful combination test that is much safer than colonoscopy (62–65). However, we radiologists have produced no data to document the effectiveness of this strategy in screening. If we wish barium enema examination to be used for screening, we need to carry out the studies to document that it can perform in a screening context, and there are many who believe that to produce consistent standards, the examination will need to be carried out in accredited screening centers—accredited in much the same way that we currently accredit mammography screening. Having the barium enema examinations double-read by two radiologists but performed by supervised radiology technologists rather than by radiologists may facilitate such quality control (66,67).
For the time being, fecal occult blood testing, followed by colonoscopy or barium enema examination when positive, appears to be the most likely scenario, unless the data accruing from the United Kingdom on one-time flexible sigmoidoscopy become overwhelmingly convincing. The recent approval of the barium enema examination as a screening modality under Medicare by the U.S. Health Care Finance Committee does provide us with the opportunity to mount the sort of studies required, which will allow us to produce data to show that the barium enema examination is capable of being a positive influence in a screening program. In the United Kingdom, the Ministry of Health has approved two centers to conduct screening of the general population, and in one of them, the double-contrast barium enema examination is to be evaluated as the follow-up test after a positive fecal occult blood test.
However, radiologists and gastroenterologists together will watch with interest the results of the comparative projects that will be undertaken in the next year or two to compare helical CT and colonoscopy. The American Society for Gastrointestinal Endoscopy (68) has written that helical CT scanning appears to be a safe, noninvasive modality for colorectal polyp screening and that further studies are needed to establish its clinical efficacy and cost-effectiveness. It is absolutely essential that radiologists work together closely with gastroenterologists and surgeons to assess the relative merits of our technologies.
"TART" SUMMARY
In closing, I would like to summarize briefly what I have told you, and to help you to remember the take-home messages, I have put them together in the acronym TART. The first T stands for "two diseases." It is important to remember that there is a fast disease responsible for some 15% of colorectal cancers, which is caused by problems with mismatch-repair genes, and that this disease occurs in young patients and develops quickly. There is a slow disease responsible for 85% of colorectal cancers, which is caused by the chromosome instability pathway, in which adenomatous polyps stay in the bowel and are readily detectable with imaging methods for some 2 decades or more before cancer develops.
The A stands for aspirin and is a reminder that aspirin is a modern wonder drug and can be useful to all of you in diminishing your chance of getting cancer, in preventing cardiovascular accidents, and in stopping you from getting a deep venous thrombosis when you fly for long distances in economy seats.
The R stands for registries, to underline the importance of using the pathology records to develop registries of those patients who get colon cancer when they are younger than 55 years and of those patients who already have another family member with colorectal cancer. These groups will form the basis of our assault on the problem of HNPCC.
Finally, the last T in the acronym TART is to remind us that we need to conduct trials to demonstrate that our methods can truly be useful and do more good than harm. We need trials of barium enema examination in a screening context, trials comparing CT scanning with colonoscopy, and the results of the current flexible sigmoidoscopy trials. The data from such studies will be a major help in selecting strategies to decrease the mortality of colorectal carcinoma.
Footnotes
Abbreviation: HNPCC = hereditary nonpolyposis colorectal cancer
References
This article has been cited by other articles:
|
|
 |
|
  P. Veit-Haibach, C. A. Kuehle, T. Beyer, H. Stergar, H. Kuehl, J. Schmidt, G. Borsch, G. Dahmen, J. Barkhausen, A. Bockisch, et al. Diagnostic Accuracy of Colorectal Cancer Staging With Whole-Body PET/CT Colonography JAMA, December 6, 2006; 296(21): 2590 - 2600. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P Veit, C Kuhle, T Beyer, H Kuehl, C U Herborn, G Borsch, H Stergar, J Barkhausen, A Bockisch, and G Antoch Whole body positron emission tomography/computed tomography (PET/CT) tumour staging with integrated PET/CT colonography: technical feasibility and first experiences in patients with colorectal cancer Gut, January 1, 2006; 55(1): 68 - 73. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. N. Klabunde, E. Jones, M. L. Brown, and W. W. Davis Colorectal Cancer Screening with Double-Contrast Barium Enema: A National Survey of Diagnostic Radiologists Am. J. Roentgenol., December 1, 2002; 179(6): 1419 - 1427. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. T. Ferrucci Colon Cancer Screening with Virtual Colonoscopy: Promise, Polyps, Politics Am. J. Roentgenol., November 1, 2001; 177(5): 975 - 988. [Full Text] [PDF]
|
|
|
|
 |
|
 P. M. McMahon, J. L. Bosch, S. Gleason, E. F. Halpern, J. S. Lester, and G. S. Gazelle Cost-Effectiveness of Colorectal Cancer Screening Radiology, April 1, 2001; 219(1): 44 - 50. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
