Feature Story

Colon cancer markers still in limbo

Microsatellite instability as a maker

June 2000
William Check, PhD

Investors in Internet stocks are beginning to learn the same hard lesson that pathologists and clinicians who work with prognostic and predictive markers in colon cancer have taken to heart in the last few years: Faith and promises will take you only so far. When the Nasdaq lost one-fourth of its value during one week in mid-April, Newsweek Wall Street editor Allan Sloan wrote that stocks of many companies in the so-called New Economy "trade largely on the basis of hopes, dreams, hype and momentum because there is little or nothing in the way of profits or assets for investors to weigh." For many years, molecular markers for colorectal cancer were in a similar situation.

In 1994 the CAP sponsored a conference on prognostic factors in colorectal cancer (along with breast and prostate cancers) at which "many factors looked as though they were going to be important," says Carolyn Compton, MD, PhD, professor of pathology at Harvard Medical School and director of gastrointestinal pathology at Massachusetts General Hospital. "It seemed at that time that several tissue-based prognostic factors were poised on the verge of being validated," Dr. Compton told CAP TODAY. But evidence reviewed at a five-year followup conference sponsored by the CAP and held in June 1999 revealed that, "Not a single prognostic factor [for colorectal cancer] had been validated in that time period," says Dr. Compton, chair of the group that wrote the conference summary on prognostic factors in colorectal cancer, which will be published next month in the Archives of Pathology & Laboratory Medicine.

As with dot-com stocks, there was a dearth of necessary information. "What prevented us from accumulating a critical mass of data that would either validate or eliminate these factors from further consideration," says Dr. Compton, "was that all studies were too small to achieve the statistical power needed and they varied in methodology and statistical analysis so you couldn’t amalgamate the data from different studies."

Another conference participant, Clive Taylor, MD, PhD, chair of the Department of Pathology at the University of Southern California School of Medicine, cites the same obstacle. "The problem with prognostic markers," Dr. Taylor says, "is that when they first appear in the literature, you get two or three positive reports and people get enthusiastic. But many years later you still have only inconclusive datasets," which won’t support a definitive evaluation, much less a positive recommendation.

Two other conferences have reached similar conclusions in the last two years. Mark Welton, MD, associate professor of surgery at the University of California-San Francisco School of Medicine, took part in the 1999 CAP conference and a 1998 conference on the same topic sponsored by the American Joint Commission on Cancer, or AJCC. (The summary report from that conference was published in the April 1, 2000 issue of Cancer.) The outcomes and recommendations of the two conferences were similar, Dr. Welton says. For prostate cancer, PSA is indicated. "But there is not an equivalent molecular marker for colon cancer," he says. "That is the exact place where we came down from both the CAP meeting and the AJCC meeting."

Gershon Locker, MD, chief of hematology/oncology at Evanston (Ill.) Northwestern Healthcare, sits on the American Society of Clinical Oncology’s tumor marker panel, which last year reviewed evidence for the clinical utility of molecular markers in colorectal cancer. "We only advocated the use of carcinoembryonic antigen [CEA] and only for monitoring in limited circumstances, and even that was somewhat controversial," Dr. Locker reports. "That coincides with my own practice. It is my view that until a marker is prospectively shown to change a patient’s prognosis and our clinical practice or treatment, that marker shouldn’t be advocated. Retrospective analysis is not the right way to choose what markers should be used clinically," he adds.

What has changed fundamentally during the past few years in the tumor marker field is what may now be occurring for Internet stocks-a demand that promises be redeemed with proof of performance. "Just as molecular biology and the discovery of molecular markers have been a major scientific advance," Dr. Locker says, "there has been a major advance in how we look at evidence. We have moved toward evidence-based medicine. It is no longer sufficient to adopt in clinical practice based on retrospective studies what looks like a hopeful marker." Today long-term prospective clinical outcomes are the standard of adoption. And in colorectal cancer no molecular marker has met that standard.

"Molecular markers are a big field, but when it comes down to it very few are used in management decisions," agrees Mark Redston, MD, staff pathologist at Mt. Sinai Hospital, Toronto. "To justify clinical adoption, we need to show that a marker predicts response to treatment, rather than just predicting behavior."

"I don’t think at the present time any study or series has led to a change in clinical management and treatment on the basis of molecular markers," states Chris Willett, MD, professor of radiation oncology at Massachusetts General Hospital, who took part in the CAP conference. "I have been taking care of patients with bowel cancer for 20 years," Dr. Willett continues. "There have been a lot of promises based on markers, but the accepted methods for guiding therapy remain essentially the same."

Currently accepted methods for guiding therapy made up the second major topic at the CAP conference on prognostic factors in colorectal cancer. Were this part of the conference made into a movie, it could star Michael J. Fox and be titled "Forward Into the Past." For at the outset of the 21st century, the best prognostic factors are related to the traditional, decades-old TNM staging system-local extent of tumor (T), regional lymph node status (N), and presence or absence of distant metastases (M). "Evaluation for those three factors remains the single most powerful indicator for any solid tumor," Dr. Compton states.

"At the present time, all clinical decisions are based on the TNM staging system or classical surgical pathology staging," Dr. Willett agrees. "That is still the fundamental basis of treatment."

This system is hardly static, however. Conference participants made many substantive recommendations for improving the TNM system. In fact, most of the conference turned out to be "an intense discussion enumerating the reasons on a very specific point-by-point basis why each of the identified tissue-based factors had not advanced in terms of usefulness in patient care," Dr. Compton says. Potentially useful but unproven histopathologic factors include, among others, vascular invasion, tumor border configuration, and host immunologic response. "The problem was dissected for each factor as to why we weren’t closer to knowing whether this factor was important or not," says Dr. Compton.

As M. Elizabeth Hammond, MD, chair of pathology at LDS Hospital, Salt Lake City, succinctly summarizes it, "The main conclusion of the conference was that significant work needs to be done in improving the way we stage colon cancer. In all three of these cancers we made significant recommendations about how to do grading so we would get better information. In colon cancer, for instance, we recommended changing to a two-grade rather than a three- or four-grade system."

Standardization was also emphasized. "What has come to the fore," adds conference participant Stanley R. Hamilton, MD, head of pathology and laboratory medicine at the University of Texas M. D. Anderson Cancer Center, "is the recognition that there are sets of criteria that have to be met as to how evaluation of specimens should be undertaken for results to be reproducible and useful in patient management."

Standardization of basic surgical pathology is essential to the clinician as well. "It is hard to make innovations if the fundamentals of surgical pathology are not done well," Dr. Willett says. "Proximal and distal margins plus radial or soft tissue resection margins are very relevant in colorectal cancer," he emphasizes, "as are adequate number of lymph nodes and the presence or absence of lymphatic or venous invasion."

If widely adopted, the suggested improvements will have a further advantage: They will make it easier to evaluate molecular markers. Major obstacles to interpreting the data in this field are that investigators follow their own methods and there are no standards for interpreting or reporting data or for statistical analysis. "Some good ideas came out of the conference about what needs to be done to quality control testing of molecular markers and to accumulate data for their clinical validity," Dr. Hamilton says. "Molecular markers have tended to be in the purview of molecular biologists and laboratory researchers, who don’t have the skill set to be able to evaluate them or to apply them in clinical settings. We as pathologists need to get involved in that process."

In the remainder of this article we will discuss proposed enhancements to traditional histopathologic staging, followed by a look at molecular markers-why they remain in clinical limbo, and a few of the most promising factors under investigation.

"Histopathology has been in a state of evolution," Dr. Hamilton points out, "with the addition of important pathologic features over and above the standard ones such as depth of invasion and lymph node status that have been the mainstay of the TNM classification for nondisseminated disease. We have now begun to recognize that some newer features are useful for identifying subsets of patients at higher risk. Tumor differentiation, extramural venous invasion, and depth of invasion beyond the muscularis propria have now begun to move into the classification with synoptic reports, so that they are emphasized and provide maximal information for clinicians who need to make decisions."

Another notable change, Dr. Hammond points out, is in thinking about how to stage a colon cancer patient who has chemotherapy and then surgery. "A number of [colon cancer] patients have chemotherapy prior to surgery," she notes. "A tumor that hasn’t been treated before surgery can’t be compared directly with one treated with chemotherapy prior to surgery. It is in a different stage category."

Dr. Welton cites several ways in which the conference’s recommendations have already helped him manage patients.

"Our pathologists’ reports became significantly more useful soon after the conference," he says. (Whether these changes were due to his input or some other route is not important, he says.) "Perineural and vascular involvement, large vessel involvement, and inflammatory response-all issues we discussed at the CAP conference-are now itemized on my final pathology report and available on my computer," Dr. Welton says.

"One of my major goals," he adds, "was to dig in my heels and ask CAP to set an adequate number of lymph nodes to be reported in the surgical specimen. This is an ongoing argument between pathologists and clinicians and is really important for us surgeons and radiation and medical oncologists." He acknowledges that this is "a very sticky issue," due to constraints on how much time a pathologist can spend on each specimen. "In the end, we got a minimum of 12 lymph nodes," Dr. Welton reports. If fewer nodes are seen, the pathologist might go back and use visual enhancement techniques such as fat clearing. "Clinically that has already been important for me," Dr. Welton says. "It gives me as a surgeon making treatment decisions something to stand on when I come up short of lymph nodes. And it takes some of the acrimony out of the discussion."

Guidelines on how many lymph nodes or tissue sections are acceptable are important to pathologists as well, Dr. Hamilton points out. These numbers represent a "target," rather than a new standard of care. "Especially in this era of such severe constraints on cost, pathologists are under pressure to hold down costs," Dr. Hamilton notes. "But we don’t want to compromise patient care. Having a document of this sort that provides guidelines means that now it is not just one individual pathologist doing battle with an administrator about the number of blocks being processed for a given colon cancer case."

Dr. Welton notes that the conference participants also spent time clarifying T3 and T4 lesions. A lesion that comes up to the serosa but does not penetrate it is a T3 lesion; if it involves the serosa, that is a T4 lesion. "There was a lot of discussion on this," Dr. Welton reports, "even a drawing by [Dr. Compton] illustrating that you can have sloughing of serosa on the colon because of inflammation. Is that T3 or T4? And what is T4 with residual disease-when you know you have left residual tumor behind?" These questions are clarified in the conference report.

But even with improved histopathologic staging, there will still be clinical roles for molecular markers. Dr. Compton cites one possible application. For patients with stage 2 colorectal cancer, who have no regional lymph node metastases (N0), the current standard of care is surgery alone. Cure is achieved by complete excision of the tumor in about 75 percent of such patients, but 25 percent have recurrent tumor and die of their disease. "We can’t predict which 25 percent of stage 2 patients those are," Dr. Compton points out. A molecular marker that reliably identifies stage 2 patients at high risk of recurrence could guide the use of adjuvant chemotherapy.

In a reverse situation, Dr. Willett notes the tendency to overtreat stages 2 and 3 (N0 or N+) rectal cancer after surgical removal. For these patients, postoperative chemotherapy and radiation are now recommended. Some, however, are cured by surgery alone. "It would be valuable to have markers to elucidate which patients are more likely to have a good outcome and so can be spared further treatment," he says.

Dr. Hamilton agrees that the "most exciting and encouraging" aspect of molecular markers is their potential ability to individualize therapy. For instance, in stage 3 colon cancer, where adjuvant chemotherapy (5-fluorouracil plus leucovorin) has become standard, it would be useful to be able to identify which tumors will respond and which will be resistant. "Some patients appear not to get a survival benefit from that regimen," Dr. Hamilton says. "It would be good to be able to pick those patients in whom newer, more toxic regimens should be tested as first-line adjuvant therapy." One such regimen is 5-fluorouracil plus leucovorin with CPT-11 (irinotecan), which looks advantageous in clinical trials. "Doing that will require input from molecular techniques," Dr. Hamilton believes. "It is essential that pathologists be involved in that process."

Given the perceived value of adding molecular markers to the staging repertoire, why are the needed studies still lacking?

"One reason," says Dr. Hammond, "is that we have not been able to be precise enough in how we collect routine information about tumors-how far the tumor has spread through the wall, what constitutes metastasis in a lymph node, and being precise about the histologic grade of the tumor. There is so much variation in the way we examine specimens, the way we interpret results, and how we report those results that it has confounded our ability to get adequate amounts of data to tell us whether one marker or another is important. If we don’t learn to standardize the way we examine, report, and interpret things," she predicts, "we will never get enough information to make conclusions about molecular markers."

Dr. Compton describes another obstacle: the need for large multicenter studies. "We already have very strong prognostic factors related to TNM classification," she notes. "To know whether a newer factor is important, you have to validate it by determining whether it has independent prognostic significance over and above staging. You have to stratify patients according to stage, then look at the factor in a defined patient population in a randomized prospective fashion. You can’t do that unless you have many institutions working together."

But only about three percent of cancer patients go on to cooperative studies. "Even funded studies are having trouble accruing adequate numbers of patients," Dr. Compton says. Cancer trials are labor-intensive and grants pay for ancillary support but not for physicians who enter patients. "So," she points out, "you are asking physicians who are already beleaguered in this time of managed care to do more work for no additional pay."

Pharmaceutical companies tend not to fund such studies because most markers are not patentable; government grants tend to emphasize treatment. "Prognostic factors are not high on the fundability list," Dr. Compton notes wryly.

She sees a possible change in attitude. "People are starting to recognize now that we are not going to get to the goal as efficiently, we are not going to improve the care of the cancer patient fast enough, unless the emphasis comes off the RO1 grant-the independent investigator, my-favorite-marker approach-to the cooperative group mechanism," she thinks. "That is the only way to get data conclusively and efficiently."

At the CAP conference, data were examined for a large number of markers, among them p53, K-ras, c-myc, bcl-2, thymidylate synthase, mitotic index by flow cytometry, loss of heterozygosity (LOH) at chromosome 18q (also called deleted in colon cancer, DCC), and microsatellite instability. A few newer markers are considered worthy of further investigation. Dr. Hammond says, "Microsatellite instability and loss of heterozygosity at 18q are promising, but there are not enough clinical data to recommend them for routine clinical use" at this time. (See "Microsatellite instability as a marker," page 19.)

At M. D. Anderson, Dr. Hamilton and colleagues are looking at microsatellite instability [by DNA analysis and immunohistochemistry for the products of genes responsible for nucleotide mismatch repair] and 18q allelic loss, as well as p53 gene status. "This is a field that is still very much in the development stage," Dr. Hamilton says. "We are trying to move from the research laboratory into the clinical setting." As for the two methods to measure microsatellite instability, he says, "Obviously IHC is particularly appropriate, because that is a standard technique available in the vast majority of pathology laboratories."

Dr. Taylor’s clinical research at USC concerns proliferation markers, which, he believes, "have a promising future, although they are not proven yet." In his view, "Topoisomerase II will be of greater value in the long run than mitotic counts." Tumor proliferation is assessed now by morphologic factors, such as tumor grade and counting mitotic figures. Flow cytometry is thought to be more accurate, but that method is not always available and is relatively expensive. Thus the interest in immunostaining for molecular markers of proliferation.

One of the first proliferation markers described was PCNA, which did not pan out. Next was Ki 67, which was better but still had a fairly long half-life, so it was not specific for the proliferative phase of the cell cycle. "Now we have topoisomerase II, which is most closely restricted to S phase, though not totally so," Dr. Taylor says. "It provides a pretty good index of proliferating cells. One remaining challenge is to find a scoring system that is reproducible across institutions as well as defining the best antibodies." Work on proliferation markers is spurred by the increasing number of papers appearing across a number of tumor types showing that rate of division relates to tumor aggressiveness and eventual outcomes.

Of course, even these newest and most promising markers will still be question marks in five years if the correct studies are not done. But progress is being made. Most trials groups are now requiring tissue blocks so that various molecular markers can be assayed, according to Dr. Willett. "Cooperative groups are running tests on batteries of markers to determine which are valuable in a consistent way," he says. "So far results vary among studies and are very heterogeneous." If the improvements in methodology recommended by the CAP prognostic markers conference report are adopted, perhaps more convincing data will emerge.

William Check is a freelance medical writer in Wilmette, Ill.