Molecular walk-through for CRC testing

Karen Titus

March 2013—For pathologists and clinicians alike, molecular testing can carry shades of a Pinter play: a seemingly straightforward situation with a whiff of discomfort, if not outright menace, and bedeviled by shifting characters and layers of meaning that defy snap interpretations.

Yet molecular testing is fast becoming part of standard repertoire in clinical laboratory testing, and pathologists who shy away from it aren’t doing themselves—or patients, or clinical colleagues, or the profession—any favors. Deciding what tests need to be ordered and interpreting results are sine qua nons of the pathologist’s job. That includes molecular testing, says Wendy L. Frankel, MD. “We don’t want clinicians talking about it at their meetings, then looking at us as technicians and ordering us to set up this or that,” says Dr. Frankel, distinguished professor of pathology, and vice chair and director of anatomic pathology, Ohio State University Wexner Medical Center (OSUWMC).

No indeed. Nor should the imminent arrival of next-generation sequencing be seen as a get-out-of-jail-free card. “If we start sequencing every colon cancer, a lot of what we do will change,” Dr. Frankel concedes. “But that’s OK. Everybody needs to know what to do in the meantime.” It’s not as if all labs will begin sequencing every tumor by month’s end.

So how should pathologists fulfill their role as molecular consultants? Let’s start with colorectal cancer, what Kevin Halling, MD, PhD, calls bread and butter testing for molecular pathologists. Whether they do the testing themselves or send it out, pathologists need to be familiar with the tests, their interpretation, and how to explain them to clinicians (and perhaps the occasional patient).

Dr. Halling, vice chair of research and development for the Department of Laboratory Medicine and Pathology, Mayo Clinic, uses a simple list of questions to organize his approach to solid tumor testing. What are the diagnostic, prognostic, theranostic, and monitoring tests, and what are the tests that identify individuals with a hereditary predisposition to cancer? “I think about this regularly, and ask what’s available for any given tumor type,” says Dr. Halling, who co-directs the clinical molecular genetics and molecular anatomic pathology laboratories at Mayo.

“It’s funny—it seems like there are probably 10, 15 different tests for any tumor,” he says. “But in fact, there are currently only a few with proven clinical utility for any given tumor type.”

With colorectal cancer, surgical pathologists don’t need a test to help establish a diagnosis. “That’s easily done with the microscope alone. So there’s really nothing to do there,” he says. For monitoring CRC recurrence, he knows of no molecular test.

The main areas in which molecular testing aids colorectal cancer patients now are identifying those who have the hereditary CRC syndromes known as Lynch syndrome familial polyposis and MYH-associated polyposis, identifying which patients with advanced CRC are candidates for anti-EGFR therapy, and predicting prognosis and guiding therapy, especially in those patients with stage II CRC.

Lynch syndrome is the most common hereditary CRC syndrome, and the first step in evaluating CRC patients for this disorder is to assess their tumor for defects in DNA mismatch repair (MMR), using microsatellite instability (MSI) testing and/or immunohistochemical analysis for expression of the four DNA MMR genes known as MLH1, MSH2, MSH6, and PMS2. If the MSI or DNA MMR immunohistochemical results show evidence of defective DNA MMR, then the patient may be evaluated further for Lynch syndrome by assessing a peripheral blood sample for germline mutations of one of the four DNA MMR genes responsible for Lynch syndrome.

Laboratories use various numbers of MSI markers, but most tend to use five markers, says Dr. Halling, who presents a course at USCAP titled, “Colorectal Cancer: Molecular Testing for the Surgical Pathologist.” If two or more of the five markers are positive, then the tumor is said to exhibit high-level microsatellite instability, or an MSI-H phenotype. If none of the markers show instability, the tumor is said to be microsatellite stable or to show an MSS phenotype. If only one of the five markers shows instability, then the tumor is said to show low levels of MSI, or an MSI-L phenotype. About 15 percent of sporadic CRC exhibit high levels of microsatellite instability, and these tumors have been shown—even after adjusting for stage—to have a better prognosis than CRC that are microsatellite stable. It’s also been recently discovered, says Dr. Halling, that stage II CRC patients whose tumors exhibit an MSI-H phenotype and who are treated with 5 fluorouracil (5FU), a standard adjuvant treatment, do no better than if they had been treated with surgery alone, suggesting that these patients should not receive 5FU.

IHC, too, is used to assess a CRC or other Lynch syndrome-associated tumor for defective DNA MMR. As noted, four genes are involved in encoding proteins that participate in DNA mismatch repair: MLH1 (3p21), MSH2 (2p22-p21), MSH6 (2p16), and PMS2 (7p22). Tumors in patients with Lynch syndrome have a germline mutation of one allele of a DNA MMR gene (for example, MLH1) and a somatic mutation in the other allele. The presence of two inactivating mutations results in defective DNA mismatch repair, Dr. Halling explains. This, in turn, increases accumulation of mutations across the genome, especially in microsatellites. Some of the mutations promote tumorigenesis by inactivating tumor suppressor genes or activating oncogenes. Most microsatellites are in noncoding regions of the genome and instability of these microsatellites does not contribute to tumorigenesis. “Nonetheless,” he says, “MSI is a convenient hallmark for identifying that a tumor exhibits defective DNA MMR.”

The goal, then, of assessing tumors for defective DNA MMR is first and foremost to identify patients with Lynch syndrome, but it can also be useful in predicting prognosis and guiding therapy in patients with sporadic CRC.

For colorectal cancer testing, is one—MSI or IHC—better than the other? Do you need both?

“Surgical pathologists, of course, like immunostains, so I think the field is drifting toward doing immunostains,” predicts Dr. Halling, in part because they have the advantage of being accessible to almost any laboratory. Based on CAP proficiency testing subscriptions, he says, DNA MMR IHC has already surpassed MSI testing, even though it’s only been available for several years. “But microsatellite instability will continue to be an important complementary and for some cases more robust technique,” he says.

The bottom line, according to Dr. Frankel, is that both are good ways to screen, though each may miss the occasional patient. “You have to make an institutional decision as to which one is the best way to go. They’re both very good.” Few places do both as a screen because it’s expensive. If a patient is clinically suspicious despite a negative test by one method, however, consider testing by the other method.

Regardless of method, pathologists need to convey some basic information to their clinical colleagues. “Clinicians might not understand that when we do microsatellite instability testing, we need both tumor and normal tissue,” says Dr. Halling. And not every clinician grasps that this testing is used to identify patients who are more likely to have Lynch syndrome but that it is not the last step. They oftentimes don’t understand that the next step, if results show either high microsatellite instability and/or loss of protein stain, would be to provide a peripheral blood sample for germline testing to look for mutations in one of the four DNA mismatch repair genes.

“They also need to be aware that the immuno-stains act as a surrogate genetic test,” Dr. Halling continues, something neither clinicians nor pathologists typically think about in relation to immuno-stains. Certain patterns of staining have a very high positive predictive value for the presence of a germline mutation. For example, a patient whose tumor shows loss of MSH2 and MSH6 has a very high likelihood of having a germline MSH2 mutation and thus a diagnosis of Lynch syndrome. Pathologists and clinicians should discuss whether informed consent is necessary. For her part, Dr. Frankel says such consent is unnecessary because these stains evaluate protein expression and not genes. All colon cancer patients at OSUWMC are educated about the tests prior to surgery.

Pathologists should know how to interpret the clinical significance of the five common staining patterns (at right). When all the immunostains show normal expression, “it basically means that the tumor has normal DNA MMR and the patient is very unlikely to have Lynch syndrome.” In such a case, “you shouldn’t see microsatellite instability in the tumor,” Dr. Halling says. The most common abnormal pattern shows simultaneous loss of MLH1 and PMS2, with normal staining of MSH2 and MSH6. This could indicate either Lynch syndrome or a sporadic colon cancer. The other three abnormal staining patterns (MLH1/PMS2 normal, MSH2/MSH6 absent; MLH1/PMS2/MSH2 normal, MSH6 absent; and MLH1/MSH2/MSH6 normal, PMS2 absent) are highly suggestive of Lynch syndrome. The stains are also good at identifying which gene is likely to carry a germline mutation, and that can be used to guide subsequent gene sequencing.

Dr. Frankel says her lab screens all resected colorectal cancers. In addition to helping identify patients (and possibly family members) with Lynch syndrome, it provides prognostic information—again, patients whose tumors are microsatellite unstable have a better prognosis than those with stage matched microsatellite stable tumors. It also predicts possible response to chemotherapy.

This approach grew out of a study she and her colleagues did at OSUWMC (Hampel H, et al. N Engl J Med. 2005;352:1851–1860) looking at more than 1,000 patients with colorectal cancer to determine whether it would be feasible and useful to screen all patients. The short answer: Yes. Subsequent studies have reinforced the point, Dr. Frankel says, as has the CDC’s Evaluation of Genomic Applications in Practice and Prevention (EGAPP) working group (Genetics in Medicine. 2009;11:35–41). The approach is being adopted by comprehensive cancer centers and academic centers, with community hospitals starting to follow.

The OSUWMC study found that one out of every 35 patients who have colorectal cancer has Lynch syndrome. For every patient, there will average an additional three family members who also have LS.

In the past, it was common to look for Lynch syndrome using Amsterdam and Bethesda criteria, looking at age, family history and, more recently, histology. “But you still miss patients,” she says.

If the IHC stains are intact, or if the microsatellites are stable—which will be true in about 85 percent of cases—it’s reasonable to halt further testing unless there’s strong clinical suspicion, says Dr. Frankel.

In the 15 percent of cases with MLH1 and PMS2 missing, three-quarters are due to a sporadic hypermethylation. “So, they’re not a genetic syndrome, but they’re still microsatellite unstable,” she says. The next step is to find out if LS is involved.

Some of the stains can be finicky, she cautions. In rectal cancer cases, MSH6 can be harder to read after neoadjuvant chemoradiation. Pathologists may see low levels of staining, or very weak staining, or even just the nucleoli. Many of these patients don’t have an MSH6 mutation. Dr. Frankel and her colleagues now circumvent the issue by testing the preoperative biopsy in resections with equivocal staining. In fact, post-treatment testing in rectal cancer cases is problematic in some cases because insufficient tumor may be available for testing. “More and more patients are getting neoadjuvant therapy for rectal cancer, and that can be challenging for some testing,” she says.

Molecular testing is nothing if not dynamic. Says Dr. Frankel: “Every month a new paper comes out, and we reassess what we do.” The MSH6 issue was just one example of the flexibility required of labs. “Initially we were going right to molecular sequencing after obtaining proper patient consent. We stopped doing that after, in five to 10 patients in a row, there was no MSH6 mutation.”

If MSI or IHC suggest that the patient might have Lynch syndrome, the pathologist will next want to consider testing for germline mutations, either in one or all of the DNA MMR genes—MLH1, MSH2, MSH6, or PMS2. Somatic BRAF mutations are generally not seen in tumors from those with Lynch syndrome, so BRAF mutation testing can be used to help filter out these cases from follow-up genetic counseling, testing, or both.

Another option for determining if a patient with MLH1 and PMS2 loss has a germline MLH1 mutation, instead of BRAF testing, would be to look for methylation of the MLH1 promoter. If methylation is present, the tumor is presumed to be sporadic, and not due to Lynch syndrome. “So they don’t need additional testing,” says Dr. Frankel.

In patients with a BRAF mutation or methylation of the MLH1 promoter, LS can be ruled out in most cases. Once again, unless there’s strong clinical suspicion, “You can stop further testing,” says Dr. Frankel. If they don’t have a BRAF mutation or MLH1 promoter hypermethylation, then they’ll need additional testing. At this point, patients at OSUWMC are contacted by genetic counselors or nurses in the cancer genetics program, who obtain consent and discuss different testing options, which will vary depending on the patient’s history but basically involve some kind of gene sequencing.

Dr. Frankel’s lab started using BRAF mutation analysis routinely in 2009 in all MLH1-/PMS2-negative patients. It also occasionally offers MLH1 methylation testing.

KRAS testing is frequently performed on CRC patients as a way to determine which patients with advanced CRC are likely to benefit from anti-EGFR therapies such as cetuximab and panitumumab. BRAF testing is to a lesser extent performed for the same reason. Pathologists do need to decide which test to offer or recommend. And even once a choice has been made, it’s not necessarily chiseled in stone. Dr. Frankel says her lab has been getting requests for both KRAS and BRAF prior to treatment on patients who have metastatic cancer and are candidates for anti-EGFR therapy. “I’m letting the clinicians drive it, but some places are starting to feel getting BRAF and KRAS up front on the primary tumors may even change the initial drugs they use.”

Pathologists can choose from a variety of methods to assess for mutations in these genes. Controversies over the best method (see “KRAS mutations—which method is best?” CAP TODAY, May 2009) have yet to be resolved, says Dr. Halling, who also notes the FDA has shown a heightened interest in having labs use agency-approved assays for targeted therapies, rather than laboratory-developed tests.

Even if pathologists choose not to perform molecular testing themselves, they need to be involved in front-end testing steps, says Dr. Halling. For any type of solid tumor testing, in fact, “They have to identify the area of tumor to evaluate,” from, say, an H&E-stained slide. “They have to give an estimate of the percentage of tumor cells in the circled area that are truly tumor cells,” and not nonneoplastic cells. If the area isn’t rich enough in tumor percent, there’s risk of a false-negative result.

He offers as an example a slide from his lab with an estimated 60 percent of tumor circled (see page 52). The assay Mayo uses for microsatellite instability has an estimated sensitivity of 30 to 40 percent. “If we cannot identify an area that has at least 30 to 40 percent tumor cells for a patient, then we would not perform the testing.” It’s not uncommon, he adds, to get samples where the tumor cells are scattered and rare.

In the meantime, the steady drip of new markers—including PI3KCA and PTEN—continues. With them come a chasm and a decision: When should labs leap to next-generation sequencing?

Dr. Halling and his Mayo colleagues are developing a next-gen sequencing panel that will be used not only for colon cancer but also for other solid tumors, including lung. Despite the push from various quarters to offer whole genome or whole exome sequencing, clinical labs are required to validate all their targets—a very difficult task, at least in the traditional sense of validation. “So we’re focusing on a set of genes that we know have clear clinical utility, such as KRAS and BRAF for CRC, and EGFR for lung cancer.” The final panel will likely contain 15 to 20 genes linked to targeted therapies.

It’s possible that next-gen platforms will eventually be affordable for even smaller labs, putting molecular testing within easy reach. Not only will it enable labs to test more targets, but—and this is no small consideration—it will also improve turnaround times and potentially lessen the problem of small tissue specimens.

In cases of advanced metastatic colorectal cancer, turnaround times have to be short, says Marina Nikiforova, MD, associate professor of pathology and director, molecular anatomic pathology laboratory, University of Pittsburgh Medical Center. “Five to seven days is just the amount of time clinicians can tolerate,” says Dr. Nikiforova, who says she hears anecdotal evidence of CRC send-outs taking about two weeks.

That could be seen as an argument for laboratories to set up molecular testing in-house if they haven’t already done so. Dr. Halling agrees—up to a point. If a lab has the expertise, then certainly it’s worth considering, he says. “But I don’t think this is something to dabble in.”

It’s a complex decision, says Dr. Nikiforova. Small labs might find it hard to offer advanced levels of testing and to stay on top of new analytes. On the other hand, they shouldn’t automatically reject it. Pathologists should be keeping abreast of molecular targets anyway, she says. Good commercial kits are available, and if a laboratory does any type of molecular testing, it may not be all that difficult to bring KRAS or BRAF testing in-house.

Dr. Frankel advises that labs look at their volumes before making a decision. “It’s one of those cost-benefit analyses that every lab has to look at.” At her own lab, MSI and methylation testing were initially send-outs. Now, as requests and evidence of clinical value have grown, “We feel it is absolutely worthwhile to set up. But that’s very lab-dependent. Just like any other test. Many smaller labs don’t do even immunohistochemistry that we consider second nature. You can only do what you can do. If money and time weren’t an issue, we’d all do everything, right?” She and colleagues at OSUWMC are setting up next-gen sequencing now.

Dr. Nikiforova and others are convinced that next-gen sequencing will soon eclipse more traditional molecular methods. Even several years ago, she notes, it was sufficient to test for one individual mutation, such as KRAS. “But now that’s not the case,” she says, “because more and more mutations are linked to specific targeted pathways or have specific prognostic or diagnostic information.” It’s apparent to all that the future will be filled with multiple targets.

At Pittsburgh, that future arrived last November. Instead of offering molecular testing for KRAS, BRAF, and PI3KCA for advanced colorectal cancers, the lab runs a next-gen sequencing panel of 46 cancer genes and more than 740 mutations. The cost is the same as, if not less than, what it would be to analyze all these genes by conventional techniques, Dr. Nikiforova says. “But the next-gen sequencing panel we use helps us to screen for all these mutations at once, quantitate the prevalence of mutation, and detect mutation quickly and cost-efficiently,” she says.

That should be strong incentive for smaller labs to start looking at next-gen sequencing, she says. And if they’re considering bringing molecular in-house, it might “be useful to wait a year or two, and then jump directly into these techniques.”
Even though next-gen sequencing is the future, it won’t blot out the past, Dr. Halling says. He doesn’t, for instance, see it being used for MSI testing. The pathologist who hopes next-gen sequencing will obviate the need for single-analyte assays is guilty of wishful thinking.

Molecular testing requires pathologists to be intellectually engaged. It doesn’t matter if they’re sending the tests out or doing them in-house, nor does it matter if they’re using traditional methods or next-gen sequencing. They can’t unknit results from interpretations.

That’s the never-ending refrain from molecular experts, but the message may not have sunk in with everyone. Dr. Halling says he has pathologist acquaintances who think it’s sufficient to provide a result and leave interpretation to the clinician. That’s only safe to do with knowledgeable clinicians—and not all of them are knowledgeable.

Molecular results, like toddlers, should not travel solo. It’s not useful to tell clinicians a patient has a KRAS mutation, then stop. They need to be told that the mutation predicts resistance to anti-EGFR therapies; that absence of a mutation predicts the tumor may respond to therapy; that high levels of MSI indicate possibility of Lynch syndrome. “We provide a fair amount of information in our reports about what it means, and next steps to take, if next steps are necessary,” Dr. Halling says.

This can be done with boilerplate comments on a report, says Dr. Halling, though occasional free texting is required for unusual cases. A good starting point for guidance on what molecular pathology reports should look like can be found in “Clinical laboratory reports in molecular pathology” (Gulley ML, et al. Arch Pathol Lab Med. 2007;131:852–863).

With next-gen sequencing reports, says Dr. Nikiforova, labs typically categorize mutations into several levels.

The first is for a mutation with known clinical significance. For example, with a KRAS mutation, the report would include information about the mutation, links to references, and a short comment indicating the mutation is found frequently—in 40 percent of colorectal cancers—and that it does not respond to certain drugs, including cetuximab.

Another category is for mutations or variants with uncertain clinical significance. In addition to explaining that the mutations do not have well-defined responses to specific therapies, the report will again provide links to relevant literature, and comment on potential importance of the mutations to the particular genes involved.

The next category is for mutation-negative results. In each case, Dr. Nikiforova says, “We try to spell things out as much as possible to help clinicians understand what it means.” It goes without saying that she and others in the lab are also prepared to discuss results with clinicians, either via phone or face-to-face.

Labs that send out their testing might be tempted to take themselves out of the equation. Shouldn’t the brunt of interpretation fall on reference labs that provide the results?

No. Dr. Halling says explanations accompanying lab results tend to be short—clinicians don’t read long ones. So pathologists must be prepared to discuss findings. Ditto for discussing which tests to order.

He provides a few scenarios. “A pathologist should understand that if it’s a 45-year-old with colon cancer, that’s out of the ordinary. And that microsatellite instability testing should be considered. Where if it’s a 65-year-old, it’s more of a garden-variety colon cancer, and unless there’s a strong family history, they may not do that testing.

“Or they may see a certain kind of histology that suggests the possibility of Lynch syndrome,” he continues. “Or, for instance, for KRAS testing, the drug is FDA-approved for patients with advanced cancer, so it doesn’t make any sense to do KRAS testing on somebody who has early-stage colon cancer.”

Dr. Frankel says she hears from pathologists who are reluctant to do pathologist-directed testing. How can they make sure appropriate tests are ordered? “Call the clinician,” she says. “Identify the histologic features that suggest MSI in a young patient and suggest additional testing, if you’re uncomfortable ordering it yourself.”

Pathologists will also need to begin thinking about validation in broader terms, Dr. Halling says, especially with the advent of next-gen sequencing and with the FDA indicating it wants to boost oversight of LDTs. “In the past, just showing analytical validity was sufficient, and we left it up to our clinicians to decide whether they thought the test was useful. But I think it’s going to be more incumbent on the labs themselves to show that they’re clinically useful,” he says.

Molecular testing, in short, means pathologists have to engage with their clinical colleagues more, not less. That relationship should be as transparent as a Scandinavian government. At OSUWMC, Dr. Frankel says, everyone—pathologists, surgeons, oncologists, cancer genetics—is in the loop, every step of the way. Whenever a change is made to the CRC testing menu, it doesn’t take anyone by surprise. “Everyone knows what we’re doing.”

There’s a reason for that. When OSUWMC first began its colorectal screening program, she says, “We were concerned to learn that patients who were suspicious for LS were falling through the cracks. Some reports were not being followed up. That’s why you can’t do this testing unless your surgeons and oncologists are involved.” A member of the cancer genetics team is now meeting with these patients at one of their post-op visits, which has improved matters tremendously. “In order for this to work, this is not just pathology-driven,” she says.

This was also another example, she says, of how molecular testing undergoes constant change. It arrives as regularly as promises from politicians.

Pols can be ignored; colorectal cancer testing can’t. Certainly laboratories may find it cheaper, sooner rather than later, to sequence all the mismatch repair genes, and thus be tempted to view current screening and molecular methods as a waste.

“But it isn’t,” Dr. Frankel says. “In the meantime we saved lives.”

Karen Titus is CAP TODAY contributing editor and co-managing editor.